1 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3 * tree.c (make_tre_binfo_stat): Remove _stat from name.
4 * tree.h (make_tree_binfo_stat): Adjust prototype.
5 (make_tree_binfo): Remove.
7 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9 * tree.c (copy_node_stat): Rename to copy_node.
10 (build_distinct_type_copy): Adjust.
11 * tree.h (copy_node_stat): Adjust prototype.
12 (copy_node): Remove macro.
14 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16 * tree.c (make_node_stat): rename to make_node.
17 (build_tree_list_stat): Adjust.
18 (build0_stat): Likewise.
19 (build2_stat): Likewise.
20 (build3_stat): Likewise.
21 (build4_stat): Likewise.
22 (build5_stat): Likewise.
23 (build_decl_stat): Likewise.
24 * tree.h (make_node_stat): Adjust prototype.
25 (make_node): remove macro.
27 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
29 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
30 (PPC_FEATURE2_SCV): Likewise.
31 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
33 2017-07-28 Tamar Christina <tamar.christina@arm.com>
35 * config/aarch64/aarch64.c
36 (aarch64_internal_mov_immediate): Add new special pattern.
37 * config/aarch64/aarch64.md (*movdi_aarch64):
38 Add reg/32bit const mov case.
40 2017-07-28 Tamar Christina <tamar.christina@arm.com>
41 Richard Sandiford <richard.sandiford@linaro.org>
43 * config/aarch64/aarch64.md (mov<mode>): Generalize.
44 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
45 Add integer and movi cases.
46 (movi-split-hf-df-sf split, fp16): New.
47 (enabled): Added TARGET_FP_F16INST.
48 * config/aarch64/iterators.md (GPF_HF): New.
49 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
51 2017-07-28 Tamar Christina <tamar.christina@arm.com>
53 * config/aarch64/aarch64.c
54 (aarch64_simd_container_mode): Add prototype.
55 (aarch64_expand_mov_immediate): Add HI support.
56 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
57 (aarch64_can_const_movi_rtx_p): New.
58 (aarch64_preferred_reload_class):
59 Remove restrictions of using FP registers for certain SIMD operations.
60 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
61 (aarch64_valid_floating_const): Add integer move validation.
62 (aarch64_simd_imm_scalar_p): Remove.
63 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
64 (aarch64_legitimate_constant_p): Expand list of supported cases.
65 * config/aarch64/aarch64-protos.h
66 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
67 (aarch64_reinterpret_float_as_int): New.
68 (aarch64_simd_imm_scalar_p): Remove.
69 * config/aarch64/constraints.md (Uvi): New.
70 (Dd): Split into Ds and new Dd.
71 * config/aarch64/aarch64.md (*movsi_aarch64):
73 (*movdi_aarch64): Add SIMD mov case.
75 2017-07-28 Bin Cheng <bin.cheng@arm.com>
77 * tree-predcom.c: (struct chain): Handle store-store chain in which
78 stores for elimination only store loop invariant values.
79 (execute_pred_commoning_chain): Ditto.
80 (prepare_initializers_chain_store_elim): Ditto.
81 (prepare_finalizers): Ditto.
82 (is_inv_store_elimination_chain): New function.
83 (initialize_root_vars_store_elim_1): New function.
85 2017-07-28 Bin Cheng <bin.cheng@arm.com>
87 * tree-predcom.c: Revise general description of the pass.
88 (enum chain_type): New enum type for store elimination.
89 (struct chain): New field supporting store elimination.
90 (struct component): Ditto.
91 (dump_chain): Dump store-stores chain.
92 (release_chain): Release resources.
93 (split_data_refs_to_components): Compute and create component
94 contains only stores for elimination.
95 (get_chain_last_ref_at): New function.
96 (make_invariant_chain): Initialization.
97 (make_rooted_chain): Specify chain type in parameter and record it.
98 (add_looparound_copies): Skip for store-stores chain.
99 (determine_roots_comp): Compute type of chain and pass it to
101 (initialize_root_vars_store_elim_2): New function.
102 (finalize_eliminated_stores): New function.
103 (remove_stmt): Handle store for elimination.
104 (execute_pred_commoning_chain): Execute predictive commoning on
106 (determine_unroll_factor): Skip unroll for store-stores chain.
107 (prepare_initializers_chain_store_elim): New function.
108 (prepare_initializers_chain): Hanlde store-store chain.
109 (prepare_finalizers_chain, prepare_finalizers): New function.
110 (tree_predictive_commoning_loop): Return integer value indicating
111 if loop is unrolled or lcssa form is corrupted.
112 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
114 2017-07-28 Bin Cheng <bin.cheng@arm.com>
116 * tree-predcom.c (initialize_root): Delete.
117 (execute_pred_commoning_chain): Initialize root vars and replace
118 reference of non-combined chain directly, rather than call above
121 2017-07-28 Bin Cheng <bin.cheng@arm.com>
123 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
124 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
126 2017-07-28 Bin Cheng <bin.cheng@arm.com>
128 * tree-predcom.c (struct chain): New field init_seq.
129 (release_chain): Release init_seq.
130 (prepare_initializers_chain): Record intialization stmts in above
132 (insert_init_seqs): New function.
133 (tree_predictive_commoning_loop): Call insert_init_seqs.
135 2017-07-28 Bin Cheng <bin.cheng@arm.com>
137 * tree-predcom.c (determine_roots_comp): Skip trivial components.
139 2017-07-28 Richard Biener <rguenther@suse.de>
141 * match.pd: Remove superfluous :c.
142 * genmatch.c (simplify::id): Add member.
143 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
145 (current_id): New global.
146 (dt_node::parent): Move from ...
147 (dt_operand::parent): ... here. Add for_id member.
148 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
149 (decision_tree::find_node): Relax order requirement when
150 merging DT_TRUE nodes to ones inbetween the current simplify
151 and the one we try to merge with. Add diagnostic whenever
152 we need to enforce pattern order by not merging.
153 (decision_tree::insert): Set current_id.
154 (decision_tree::print_node): Dump parent node and for_id.
155 (parser::last_id): Add member.
156 (parser::push_simplify): Assign unique id.
157 (parser::parser): Initialize last_id.
159 2017-07-28 Martin Liska <mliska@suse.cz>
162 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
163 gimple_build_debug_bind.
165 2017-07-28 Richard Biener <rguenther@suse.de>
167 PR tree-optimization/81502
168 * match.pd: Add pattern combining BIT_INSERT_EXPR with
170 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
172 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
173 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
174 for BIT_FIELD_REF args.
175 * fold-const.c (make_bit_field_ref): Likewise.
176 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
178 2017-07-28 Jakub Jelinek <jakub@redhat.com>
181 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
182 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
183 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
184 Or it into SANITIZER_UNDEFINED.
185 * ubsan.c: Include gimple-fold.h and varasm.h.
186 (ubsan_expand_ptr_ifn): New function.
187 (instrument_pointer_overflow): New function.
188 (maybe_instrument_pointer_overflow): New function.
189 (instrument_object_size): Formatting fix.
190 (pass_ubsan::execute): Call instrument_pointer_overflow
191 and maybe_instrument_pointer_overflow.
192 * internal-fn.c (expand_UBSAN_PTR): New function.
193 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
194 * sanitizer.def (__ubsan_handle_pointer_overflow,
195 __ubsan_handle_pointer_overflow_abort): New builtins.
196 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
197 * internal-fn.def (UBSAN_PTR): New internal function.
198 * opts.c (sanitizer_opts): Add pointer-overflow.
199 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
200 * fold-const.c (build_range_check): Compute pointer range check in
201 integral type if pointer arithmetics would be needed. Formatting
204 2017-07-28 Martin Liska <mliska@suse.cz>
207 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
208 parameters that are of a variable-length.
210 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
212 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
214 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
215 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
216 (CRT_CALL_STATIC_FUNCTION): Likewise.
217 (ASM_DEFAULT_SPEC): New define.
218 (ASM_SPEC32): Likewise.
219 (ASM_SPEC64): Likewise.
220 (ASM_SPEC_COMMON): Likewise.
221 (ASM_SPEC): Likewise.
222 (INVALID_64BIT): Likewise.
223 (LINK_OS_DEFAULT_SPEC): Likewise.
224 (LINK_OS_SPEC32): Likewise.
225 (LINK_OS_SPEC64): Likewise.
226 (POWERPC_LINUX): Likewise.
227 (PTRDIFF_TYPE): Likewise.
228 (RESTORE_FP_PREFIX): Likewise.
229 (RESTORE_FP_SUFFIX): Likewise.
230 (SAVE_FP_PREFIX): Likewise.
231 (SAVE_FP_SUFFIX): Likewise.
232 (SIZE_TYPE): Likewise.
233 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
234 (TARGET_64BIT): Likewise.
235 (TARGET_64BIT): Likewise.
236 (TARGET_AIX): Likewise.
237 (WCHAR_TYPE_SIZE): Likewise.
238 (WCHAR_TYPE): Undefine.
239 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
240 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
241 (CPP_OS_RTEMS_SPEC): Delete.
242 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
243 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
245 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
247 2017-07-28 Jakub Jelinek <jakub@redhat.com>
249 PR tree-optimization/81578
250 * tree-parloops.c (build_new_reduction): Bail out if
251 reduction_code isn't one of the standard OpenMP reductions.
252 Move the details printing after that decision.
254 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
256 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
257 related to reload_in_progress.
258 (splat_input_operand): Likewise.
259 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
261 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
263 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
264 (TARGET_INSTANTIATE_DECLS): Likewise.
265 (legitimate_indexed_address_p): Delete reload_in_progress code.
266 (rs6000_debug_legitimate_address_p): Likewise.
267 (rs6000_eliminate_indexed_memrefs): Likewise.
268 (rs6000_emit_le_vsx_store): Likewise.
269 (rs6000_emit_move_si_sf_subreg): Likewise.
270 (rs6000_emit_move): Likewise.
271 (register_to_reg_type): Likewise.
272 (rs6000_pre_atomic_barrier): Likewise.
273 (rs6000_machopic_legitimize_pic_address): Likewise.
274 (rs6000_allocate_stack_temp): Likewise.
275 (rs6000_address_for_fpconvert): Likewise.
276 (rs6000_address_for_altivec): Likewise.
277 (rs6000_secondary_memory_needed_rtx): Delete function.
278 (rs6000_check_sdmode): Likewise.
279 (rs6000_alloc_sdmode_stack_slot): Likewise.
280 (rs6000_instantiate_decls): Likewise.
281 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
282 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
283 Delete reload_in_progress.
284 (*vec_reload_and_plus_<mptrsize>): Likewise.
285 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
286 (vsx_div_v2di): Likewise.
287 (vsx_udiv_v2di): Likewise.
289 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
291 * config/rs6000/rs6000.opt (mlra): Replace with stub.
292 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
293 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
294 (rs6000_debug_reg_global): Delete print of LRA status.
295 (rs6000_option_override_internal): Delete dead LRA related code.
296 (rs6000_lra_p): Delete function.
297 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
299 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
301 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
302 * config/riscv/rtems.h: New file.
304 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
305 Sudakshina Das <sudi.das@arm.com>
307 * config/aarch64/aarch64.md
308 (define_split for and<mode>3nr_compare): Move
309 non aarch64_logical_operand to a register.
310 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
311 register immediate operand to a register.
312 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
314 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
317 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
319 2017-07-27 Richard Biener <rguenther@suse.de>
321 PR tree-optimization/81573
322 PR tree-optimization/81494
323 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
324 multi defuse cycle case.
326 2017-07-27 Richard Biener <rguenther@suse.de>
328 PR tree-optimization/81571
329 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
332 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
334 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
335 earlier and only if MASK_FPU is set. Adjust formatting.
337 2017-07-27 Martin Liska <mliska@suse.cz>
339 * opt-functions.awk: Add validation of value of Init.
340 * optc-gen.awk: Pass new argument.
342 2017-07-27 Martin Liska <mliska@suse.cz>
344 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
347 2017-07-27 Martin Liska <mliska@suse.cz>
349 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
350 BBs and edges seen by autoFDO.
352 2017-07-27 Richard Biener <rguenther@suse.de>
354 PR tree-optimization/81502
355 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
356 with incompatible but same sized type.
357 (execute_update_addresses_taken): Likewise.
359 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
361 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
362 flag_tree_loop_vectorize rather than flag_tree_vectorize.
364 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
367 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
368 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
369 Change s_operand to memory_operand.
371 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
373 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
374 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
375 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
376 Emit instructions rather than returning an expression. Handle TFmode
377 and KFmode by casting to TImode.
378 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
379 (rs6000_emit_le_vsx_store): Likewise.
380 * config/rs6000/vsx.md (VSX_TI): New iterator.
381 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
382 (*vsx_le_undo_permute_<mode>): Likewise.
383 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
384 emit the split sequence.
385 (*vsx_le_perm_store_<mode>): Likewise.
387 2017-07-27 Jakub Jelinek <jakub@redhat.com>
389 PR tree-optimization/81555
390 PR tree-optimization/81556
391 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
392 if true, force CHANGED for the recursive invocation.
393 (reassociate_bb): Remember original length of ops array, pass
394 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
396 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
397 attributes for noipa attribute. For naked attribute use
398 lookup_attribute first before lookup_attribute_spec.
399 * final.c (rest_of_handle_final): Disable IPA RA for functions with
401 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
402 for functions with noipa attribute.
403 (cgraph_externally_visible_p): Return true for functions with noipa
405 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
406 for functions with noipa attribute.
407 * doc/extend.texi: Document noipa function attribute.
408 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
409 also for functions with noipa attribute.
410 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
412 2017-07-26 Andrew Pinski <apinski@cavium.com>
414 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
415 vec_unalign_load_cost and vec_unalign_store_cost.
417 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
419 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
420 -mvsx-small-integer option.
421 (ISA_3_0_MASKS_IEEE): Likewise.
422 (OTHER_VSX_VECTOR_MASKS): Likewise.
423 (POWERPC_MASKS): Likewise.
424 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
425 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
426 code, only testing for DImode being allowed in non-VSX floating
428 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
429 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
431 (rs6000_option_override_internal): Delete -mvsx-small-integer.
432 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
433 TARGET_P8_VECTOR test.
434 (rs6000_secondary_reload_simple_move): Likewise.
435 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
436 since TARGET_P9_VECTOR was already tested.
437 (rs6000_opt_masks): Remove -mvsx-small-integer.
438 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
439 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
441 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
442 test for TARGET_VEXTRACTUB was used, and that uses
444 (p9 extract splitter): Likewise.
445 (vsx_extract_<mode>_di_p9): Likewise.
446 (vsx_extract_<mode>_store_p9): Likewise.
447 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
448 for TARGET_P9_VECTOR was used. Delete code that is now dead with
449 the elimination of TARGET_VSX_SMALL_INTEGER.
450 (vsx_extract_<mode>_p8): Likewise.
451 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
452 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
453 (vsx_set_<mode>_p9): Likewise.
454 (vsx_set_v4sf_p9): Likewise.
455 (vsx_set_v4sf_p9_zero): Likewise.
456 (vsx_insert_extract_v4sf_p9): Likewise.
457 (vsx_insert_extract_v4sf_p9_2): Likewise.
458 * config/rs6000/rs6000.md (sign extend splitter): Change
459 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
460 (floatsi<mode>2_lfiwax_mem): Likewise.
461 (floatunssi<mode>2_lfiwzx_mem): Likewise.
462 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
463 since a test for TARGET_P9_VECTOR was used.
464 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
465 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
466 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
467 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
468 TARGET_P8_VECTOR test.
469 (fix_trunc<mode>si2_stfiwx): Likewise.
470 (fix_trunc<mode>si2_internal): Likewise.
471 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
472 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
474 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
475 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
476 TARGET_P8_VECTOR test.
477 (fixuns_trunc<mode>si2_stfiwx): Likewise.
478 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
479 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
481 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
482 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
483 since a test for TARGET_P9_VECTOR was used.
484 (splitter for loading small constants): Likewise.
486 2017-07-26 Andrew Pinski <apinski@cavium.com>
488 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
491 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
494 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
495 (fp_valid_at): Likewise.
497 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
499 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
500 (qdf24xx_addrcost_table): Likewise.
501 (cortexa57_tunings): Update to use generic_branch_cost.
502 (cortexa72_tunings): Likewise.
503 (cortexa73_tunings): Likewise.
504 (qdf24xx_tunings): Likewise.
506 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
508 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
509 (thunderx2t99_branch_cost): Likewise.
510 (cortexa35_tunings): Update to use generic_branch_cost.
511 (cortexa53_tunings): Likewise.
512 (cortexa57_tunings): Likewise.
513 (cortexa72_tunings): Likewise.
514 (cortexa73_tunings): Likewise.
515 (thunderx2t99_tunings): Likewise.
517 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
519 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
520 (sparc_option_override): Honour MASK_FSMULD.
521 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
522 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
523 * config/sparc/sparc.opt (mfsmuld): New option.
524 * doc/invoke.texi (mfsmuld): Document option.
526 2017-07-26 Marek Polacek <polacek@redhat.com>
529 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
531 2017-07-26 Richard Biener <rguenther@suse.de>
533 * gimple-match-head.c (do_valueize): Return OP if valueize
535 (get_def): New helper to get at the def stmt of a SSA name
537 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
538 do_valueize to get at the def stmt.
539 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
541 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
544 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
546 2017-07-26 Martin Liska <mliska@suse.cz>
549 * function.c (expand_function_start): Make expansion of
550 nonlocal_goto_save_area after parm_birth_insn.
552 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
554 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
555 from all CPU target flags enable members.
557 2017-07-26 Richard Biener <rguenther@suse.de>
559 * genmatch.c (dt_simplify::gen): Make iterator vars const.
560 (decision_tree::gen): Make 'type' const.
561 (write_predicate): Likewise.
563 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
565 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
566 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
567 (rs6000_option_override_internal): Likewise.
568 (rs6000_expand_vector_set): Likewise.
569 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
570 (TARGET_UPPER_REGS_SF): Likewise.
571 (TARGET_UPPER_REGS_DI): Likewise.
572 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
573 (TARGET_DIRECT_MOVE_64BIT): Likewise.
574 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
575 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
576 (Splitters for DI constants in Altivec registers): Likewise.
577 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
578 (vsx_set_v4sf_p9): Likewise.
579 (vsx_set_v4sf_p9_zero): Likewise.
580 (vsx_insert_extract_v4sf_p9): Likewise.
581 (vsx_insert_extract_v4sf_p9_2): Likewise.
583 2017-07-25 Carl Love <cel@us.ibm.com>
585 * doc/extend.texi: Update the built-in documentation file for the
586 existing built-in functions
587 vector signed char vec_cnttz (vector signed char);
588 vector unsigned char vec_cnttz (vector unsigned char);
589 vector signed short vec_cnttz (vector signed short);
590 vector unsigned short vec_cnttz (vector unsigned short);
591 vector signed int vec_cnttz (vector signed int);
592 vector unsigned int vec_cnttz (vector unsigned int);
593 vector signed long long vec_cnttz (vector signed long long);
594 vector unsigned long long vec_cnttz (vector unsigned long long);
596 2017-07-25 Andrew Pinski <apinski@cavium.com>
598 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
599 accesses where the use is for the first operand of a BIT_INSERT.
601 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
604 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
605 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
607 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
609 * config/i386/gstabs.h: Delete.
610 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
612 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
614 * config/i386/i386.c (ix86_decompose_address): Do not check for
615 register RTX when looking at index_reg or base_reg.
616 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
618 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
620 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
621 to update EH info here.
623 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
625 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
627 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
629 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
631 2017-07-25 Torsten Duwe <duwe@suse.de>
633 * common.opt: Introduce -fpatchable-function-entry
634 command line option, and its variables function_entry_patch_area_size
635 and function_entry_patch_area_start.
636 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
637 including a two-value parser.
638 * target.def (print_patchable_function_entry): New target hook.
639 * targhooks.h (default_print_patchable_function_entry): New function.
640 * targhooks.c (default_print_patchable_function_entry): Likewise.
641 * toplev.c (process_options): Switch off IPA-RA if
642 patchable function entries are being generated.
643 * varasm.c (assemble_start_function): Look at the
644 patchable-function-entry command line switch and current
645 function attributes and maybe generate NOP instructions by
646 calling the print_patchable_function_entry hook.
647 * doc/extend.texi: Document patchable_function_entry attribute.
648 * doc/invoke.texi: Document -fpatchable_function_entry
650 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
652 * doc/tm.texi: Re-generate.
654 2017-07-25 Jakub Jelinek <jakub@redhat.com>
657 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
658 TARGET_AVX512DQ rather than TARGET_AVX512BW.
660 2017-07-25 Tamar Christina <tamar.christina@arm.com>
662 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
664 2017-07-25 Richard Biener <rguenther@suse.de>
666 PR tree-optimization/81455
667 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
668 not walk in cycles when looking for guards.
670 2017-07-25 Richard Biener <rguenther@suse.de>
672 PR tree-optimization/81529
673 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
674 when optimizing backedge uses.
676 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
678 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
680 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
681 to dl_section_ref. On AIX, append an expression to subtract
682 the size of the section length to dl_section_ref.
684 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
686 * configure.ac: If any of the config.* scripts fail, exit 1.
687 * configure: Regenerate.
689 2017-07-25 Richard Biener <rguenther@suse.de>
692 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
693 of immediate uses, be more verbose on errors.
695 2017-07-25 Richard Biener <rguenther@suse.de>
697 PR tree-optimization/81510
698 * tree-vect-loop.c (vect_is_simple_reduction): When the
699 reduction stmt is not inside the loop bail out.
701 2017-07-25 Richard Biener <rguenther@suse.de>
703 PR tree-optimization/81303
704 * tree-vect-loop-manip.c (vect_loop_versioning): Build
705 profitability check against LOOP_VINFO_NITERSM1.
707 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
709 * domwalk.c (cmp_bb_postorder): Simplify.
710 (sort_bbs_postorder): New function. Use it...
711 (dom_walker::walk): ...here to optimize common cases.
713 2017-07-25 Martin Liska <mliska@suse.cz>
716 * ipa-visibility.c (function_and_variable_visibility): Make the
717 redirection just on target that supports aliasing.
718 Fix GNU coding style.
720 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
723 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
725 * config/arm/rtems.h: Likewise.
726 * config/bfin/rtems.h: Likewise.
727 * config/i386/rtemself.h: Likewise.
728 * config/lm32/rtems.h: Likewise.
729 * config/m32c/rtems.h: Likewise.
730 * config/m68k/rtemself.h: Likewise.
731 * config/microblaze/rtems.h: Likewise.
732 * config/mips/rtems.h: Likewise.
733 * config/moxie/rtems.h: Likewise.
734 * config/nios2/rtems.h: Likewise.
735 * config/powerpcspe/rtems.h: Likewise.
736 * config/rs6000/rtems.h: Likewise.
737 * config/rtems.h: Likewise.
738 * config/sh/rtems.h: Likewise.
739 * config/sh/rtemself.h: Likewise.
740 * config/sparc/rtemself.h: Likewise.
742 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
745 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
746 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
747 * tree-ssa-structalias.c (alias_get_name): Same.
749 2017-07-25 Bin Cheng <bin.cheng@arm.com>
752 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
753 instructions if no du chain is found.
755 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
757 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
759 2017-07-25 Richard Biener <rguenther@suse.de>
762 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
765 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
767 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
769 (ISA_2_7_MASKS_SERVER): Likewise.
770 (ISA_3_0_MASKS_IEEE): Likewise.
771 (OTHER_P8_VECTOR_MASKS): Likewise.
772 (OTHER_VSX_VECTOR_MASKS): Likewise.
773 (POWERPC_MASKS): Likewise.
774 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
775 duplicate list of options.
776 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
777 explicit -mupper-regs options.
778 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
779 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
780 alias for -mupper-regs-df.
781 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
782 (rs6000_init_hard_regno_mode_ok): Likewise.
783 (rs6000_option_override_internal): Likewise.
784 (rs6000_opt_masks): Likewise.
785 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
786 options in terms of whether -mvsx or -mpower8-vector was used.
787 (TARGET_UPPER_REGS_DI): Likewise.
788 (TARGET_UPPER_REGS_SF): Likewise.
789 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
790 -mupper-regs-* options.
792 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
794 * passes.c (emergency_dump_function): Print some empty lines and a
795 header before the RTL dump.
797 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
799 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
801 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
804 * config/aarch64/aarch64.c (aarch64_classify_symbol):
805 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
807 2017-07-24 Carl Love <cel@us.ibm.com>
809 * config/rs6000/rs6000-c.c: Add support for built-in functions
810 vector float vec_extract_fp32_from_shorth (vector unsigned short);
811 vector float vec_extract_fp32_from_shortl (vector unsigned short);
812 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
813 vec_extract_fp_from_shortl): Add defines for the two builtins.
814 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
815 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
817 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
818 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
819 * doc/extend.texi: Update the built-in documentation file for the
820 new built-in function.
822 2017-07-24 Jakub Jelinek <jakub@redhat.com>
825 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
827 * doc/generic.texi: Likewise.
828 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
829 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
831 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
833 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
834 (aarch64_mls_elt_merge<mode>): Likewise.
836 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
838 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
841 2017-07-23 Michael Collison <michael.collison@arm.com>
843 * config/arm/arm.c (arm_option_override): Deprecate
844 use of -mstructure-size-boundary.
845 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
846 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
848 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
851 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
852 Reduce cost estimate for direct moves.
854 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
857 * config/i386/i386.c (ix86_option_override_internal): Disable
858 BMI, BMI2 and TBM instructions for -m16.
860 2017-07-21 Carl Love <cel@us.ibm.com>
862 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
863 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
864 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
865 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
866 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
867 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
868 VMULOSW): New enum "unspec" values.
869 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
870 altivec_vmulosw): New patterns.
871 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
872 VMULOSW): Add definitions.
874 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
876 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
878 * config/aarch64/aarch64-options-extensions.def (rdma); New.
879 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
880 (AARCH64_FL_V8_1): Renumber.
881 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
882 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
883 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
884 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
885 rdma to feature modifiers list.
887 2017-07-21 Yury Gribov <tetra2005@gmail.com>
890 * ipa-visibility (function_and_variable_visibility): Convert
891 recursive PLT call to direct call if appropriate.
893 2017-07-21 Andrew Pinski <apinski@cavium.com>
895 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
896 operand 1 to see if the types precision matches.
897 * fold-const.c (operand_equal_p): Likewise.
899 2017-07-21 Richard Biener <rguenther@suse.de>
901 PR tree-optimization/81303
902 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
903 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
904 (vect_peeling_hash_get_lowest_cost): Adjust.
905 (vect_enhance_data_refs_alignment): Likewise. Use
906 vect_get_peeling_costs_all_drs to compute the penalty for no
907 peeling to match up costs.
909 2017-07-21 Richard Biener <rguenther@suse.de>
911 PR tree-optimization/81500
912 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
913 we didn't identify a reduction path.
915 2017-07-21 Tom de Vries <tom@codesourcery.com>
916 Cesar Philippidis <cesar@codesourcery.com>
918 PR gcov-profile/81442
919 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
922 2017-07-21 Tom de Vries <tom@codesourcery.com>
925 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
927 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
928 nvptx_override_options_after_change.
930 2017-07-21 Ulrich Drepper <drepper@redhat.com>
932 * dwarf2out.c (output_file_names): Avoid double testing for
935 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
937 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
939 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
941 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
943 (try_crossjump_to_edge): Do not punt on partitioned functions.
945 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
947 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
948 Put all BBs reachable only via paths crossing cold region to cold
950 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
952 2016-07-21 Richard Biener <rguenther@suse.de>
954 PR tree-optimization/81303
955 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
956 into account prologue and epilogue iterations when raising
957 min_profitable_iters to sth at least covering one vector iteration.
959 2017-07-21 Tamar Christina <tamar.christina@arm.com>
961 * config/arm/arm.c (arm_test_cpu_arch_dat):
964 2017-07-20 Nathan Sidwell <nathan@acm.org>
967 * tree.h (TYPE_METHODS): Delete.
968 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
969 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
970 (dbxout_type_methods): Scan TYPE_FIELDS.
971 (dbxout_type): Don't check TYPE_METHODS here.
972 * function.c (use_register_for_decl): Always ignore register for
973 class types when not optimizing.
974 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
975 * tree.c (free_lang_data_in_type): Stitch out member functions and
976 templates from TYPE_FIELDS.
977 (build_distinct_type_copy, verify_type_variant,
978 verify_type): Member fns are on TYPE_FIELDS.
979 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
980 * tree-pretty-print.c (dump_generic_node): Likewise.
982 2017-07-20 Jakub Jelinek <jakub@redhat.com>
985 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
986 V2TImode and V4TImode.
987 (ix86_expand_vector_extract): Likewise.
988 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
989 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
990 (ssescalarmode): Handle V4TImode and V2TImode.
991 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
992 (*vec_extractv2ti, *vec_extractv4ti): New insns.
993 (VEXTRACTI128_MODE): New mode iterator.
994 (splitter for *vec_extractv?ti first element): New.
995 (VEC_INIT_MODE): New mode iterator.
996 (vec_init<mode>): Consolidate 3 expanders into one using
997 VEC_INIT_MODE mode iterator.
999 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
1001 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
1002 non_spilled_static_chain_regno_p.
1004 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
1006 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
1008 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
1010 * bb-reorder.c (connect_traces): Allow copying of blocks within
1013 2017-07-20 Richard Biener <rguenther@suse.de>
1015 * gimple.h (gimple_phi_result): Add gphi * overload.
1016 (gimple_phi_result_ptr): Likewise.
1017 (gimple_phi_arg): Likewise. Adjust index assert to only
1018 allow actual argument accesses rather than all slots available
1020 (gimple_phi_arg_def): Add gphi * overload.
1021 * tree-phinodes.c (make_phi_node): Initialize only actual
1023 (resize_phi_node): Clear memory not covered by old node,
1024 do not initialize excess argument slots.
1025 (reserve_phi_args_for_new_edge): Initialize new argument slot
1028 2017-07-20 Bin Cheng <bin.cheng@arm.com>
1030 PR tree-optimization/81388
1032 2016-07-21 Bin Cheng <bin.cheng@arm.com>
1034 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
1035 by removing computation of may_be_zero.
1037 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1038 Tom de Vries <tom@codesourcery.com>
1041 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
1042 when gimple level profile disagrees with what RTL expander did.
1044 2017-07-20 Richard Biener <rguenther@suse.de>
1046 PR tree-optimization/61171
1047 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
1048 (vect_analyze_stmt): Add slp instance parameter.
1049 (vectorizable_reduction): Likewise.
1050 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
1051 (vect_is_simple_reduction): Deal with chains not detected
1052 as SLP reduction chain, specifically not properly associated
1053 chains containing a mix of plus/minus.
1054 (get_reduction_op): Remove.
1055 (get_initial_defs_for_reduction): Simplify, pass in whether
1056 this is a reduction chain, pass in the SLP node for the PHIs.
1057 (vect_create_epilog_for_reduction): Get the SLP instance as
1059 (vectorizable_reduction): Get the SLP instance as arg.
1060 During analysis remember the SLP node with the PHIs in the
1061 instance. Simplify getting at the vectorized reduction PHIs.
1062 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
1063 through SLP instance.
1064 (vect_slp_analyze_operations): Likewise.
1065 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
1066 (vect_transform_stmt): Likewise.
1068 2017-07-20 Tom de Vries <tom@codesourcery.com>
1070 PR tree-optimization/81489
1071 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
1072 read of phi arg location to before loop that modifies phi.
1074 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
1076 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
1079 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1082 * except.c (execute): Fix ordering issue.
1084 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
1086 PR rtl-optimization/81423
1087 * combine.c (make_compound_operation_int): Don't try to optimize
1088 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
1090 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
1092 PR rtl-optimization/81423
1093 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
1094 with a constant that is -1 in the truncated to mode.
1096 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1098 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
1099 (determine_unlikely_bbs): ... here.
1100 * predict.h (propagate_unlikely_bbs_forward): Declare.
1101 * cfgexpand.c (pass_expand::execute): Use it.
1102 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
1104 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
1105 propagate_unlikely_bbs_forward.
1107 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1110 * except.c (maybe_add_nop_after_section_switch): New function.
1113 2017-07-19 Tom de Vries <tom@codesourcery.com>
1115 * gimple.h (gimple_phi_set_arg): Make assert more strict.
1117 2017-07-19 Tom de Vries <tom@codesourcery.com>
1119 * gimple.h (gimple_phi_arg): Make assert more strict.
1121 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
1123 * config.gcc (powerpc*-*-*): Add mmintrin.h.
1124 * config/rs6000/mmintrin.h: New file.
1125 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
1127 2017-07-19 Jakub Jelinek <jakub@redhat.com>
1129 PR tree-optimization/81346
1130 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
1132 2017-07-19 Tom de Vries <tom@codesourcery.com>
1134 * config/nvptx/nvptx.md (VECIM): Add V2DI.
1136 2017-07-19 Tom de Vries <tom@codesourcery.com>
1138 * config/nvptx/nvptx-modes.def: Add V2DImode.
1139 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
1140 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
1141 (nvptx_output_mov_insn): Handle lack of mov.b128.
1142 (nvptx_print_operand): Handle 'H' and 'L' codes.
1143 (nvptx_vector_mode_supported): Allow V2DImode.
1144 (nvptx_preferred_simd_mode): New function.
1145 (nvptx_data_alignment): New function.
1146 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
1147 nvptx_preferred_simd_mode.
1148 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
1150 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
1152 2017-07-19 Tom de Vries <tom@codesourcery.com>
1154 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
1155 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
1156 (nvptx_vector_mode_supported): New function. Allow V2SImode.
1157 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
1158 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
1159 (mov<VECIM>_insn): New define_insn.
1160 (define_expand "mov<VECIM>): New define_expand.
1162 2017-07-19 Tom de Vries <tom@codesourcery.com>
1164 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
1166 2017-07-19 Jakub Jelinek <jakub@redhat.com>
1168 PR tree-optimization/81346
1169 * fold-const.h (fold_div_compare, range_check_type): Declare.
1170 * fold-const.c (range_check_type): New function.
1171 (build_range_check): Use range_check_type.
1172 (fold_div_compare): No longer static, rewritten into
1173 a match.pd helper function.
1174 (fold_comparison): Don't call fold_div_compare here.
1175 * match.pd (X / C1 op C2): New optimization using fold_div_compare
1178 2017-07-19 Nathan Sidwell <nathan@acm.org>
1180 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
1181 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
1182 * tree.c (find_decls_types_r, verify_type): Use
1183 TYPE_{MIN,MAX}_VALUE_RAW.
1184 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
1185 (hash_tree): Likewise.
1186 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
1188 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
1191 2017-07-18 Tom de Vries <tom@codesourcery.com>
1194 * omp-expand.c (expand_omp_for_static_chunk): Handle
1195 equal-argument loop exit phi.
1197 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
1200 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
1201 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
1202 operand 2 predicate.
1203 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
1204 operand 2 predicate.
1205 (ror,rol -> rorx splitters): Use const_int_operand as
1206 operand 2 predicate.
1208 2017-06-18 Richard Biener <rguenther@suse.de>
1210 PR tree-optimization/81410
1211 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
1212 the gap in the ! slp_perm SLP case after each group.
1214 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1217 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
1220 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1223 * predict.c (set_even_probabilities): Cleanup; do not affect
1224 probabilities that are already known.
1225 (combine_predictions_for_bb): Call even when count is set.
1227 2017-07-18 Nathan Sidwell <nathan@acm.org>
1229 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
1232 2017-07-18 Bin Cheng <bin.cheng@arm.com>
1235 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
1236 optimization for loop niter analysis.
1238 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
1241 * config/avr/avr.c (avr_optimize_casesi): Don't use
1242 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
1244 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1246 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
1247 body_cost_vec from _vect_peel_extended_info.
1248 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
1249 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
1252 2017-07-18 Bin Cheng <bin.cheng@arm.com>
1254 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
1256 2017-07-18 Richard Biener <rguenther@suse.de>
1258 PR tree-optimization/80620
1259 PR tree-optimization/81403
1260 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
1261 info when re-using a VN table entry.
1263 2017-07-18 Richard Biener <rguenther@suse.de>
1265 PR tree-optimization/81418
1266 * tree-vect-loop.c (vectorizable_reduction): Properly compute
1267 vectype_in. Verify that with lane-reducing reduction operations
1268 we have a single def-use cycle.
1270 2017-07-17 Carl Love <cel@us.ibm.com>
1272 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
1274 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1275 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
1276 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
1277 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
1278 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
1279 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
1280 VMULOSW): New enum "unspec" values.
1281 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
1282 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
1283 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
1284 altivec_vmulosw): New patterns.
1285 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
1286 VMULOSW): Add definitions.
1287 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
1289 * config/alpha/alpha.c: Include predict.h.
1291 2017-07-17 Yury Gribov <tetra2005@gmail.com>
1293 * tree-vrp.c (compare_assert_loc): Fix comparison function
1294 to return predictable results.
1296 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1298 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
1301 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
1302 * doc/invoke.texi (mexpand-adddi): Update text.
1304 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1306 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
1307 that also clobbers the CC register. The old expand code is moved
1309 (*arc_clzsi2): ... here.
1310 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
1311 the CC register. The old expand code is moved to ...
1312 (arc_ctzsi2): ... here.
1314 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1316 * config/arc/arc.opt (mindexed-loads): Use initial value
1317 TARGET_INDEXED_LOADS_DEFAULT.
1318 (mauto-modify-reg): Use initial value
1319 TARGET_AUTO_MODIFY_REG_DEFAULT.
1320 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
1321 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
1322 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
1323 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
1325 2017-07-17 Martin Liska <mliska@suse.cz>
1328 * opts.c (finish_options): Do not allow -fgnu-tm
1329 w/ -fsanitize={kernel-,}address. Say sorry.
1331 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1334 * tree-loop-distribution.c (classify_partition): Only assert on
1335 numer of iterations.
1336 (merge_dep_scc_partitions): Delete prameter. Update function call.
1337 (distribute_loop): Remove code handling loop with unknown niters.
1338 (pass_loop_distribution::execute): Skip loop with unknown niters.
1340 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1343 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
1344 function sort_partitions_by_post_order.
1346 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1348 PR tree-optimization/81374
1349 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
1350 the max index of basic blocks, rather than number of basic blocks.
1352 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1354 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
1356 (arc_legitimate_pic_operand_p): Likewise.
1357 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
1359 (arc_needs_pcl_p): Likewise.
1360 (arc_legitimate_pc_offset_p): Likewise.
1361 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
1362 function is also used in constrains.md.
1363 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
1364 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
1365 PLUS. Only return true/false in known cases, otherwise assert.
1366 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
1367 is already called in arc_legitimate_constant_p.
1368 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
1370 (LEGITIMATE_PIC_OPERAND_P): Use
1371 arc_raw_symbolic_reference_mentioned_p function.
1372 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
1377 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1378 Andrew Burgess <andrew.burgess@embecosm.com>
1380 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
1381 (arc_return_address_register): New function.
1382 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
1383 (arc_handle_fndecl_attribute): Add naked attribute.
1384 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
1385 (TARGET_WARN_FUNC_RETURN): Likewise.
1386 (arc_allocate_stack_slots_for_args): New function.
1387 (arc_warn_func_return): Likewise.
1388 (machine_function): Change type fn_type.
1389 (arc_compute_function_type): Consider new naked function type,
1390 change function return type.
1391 (arc_must_save_register): Adapt to handle new
1392 arc_compute_function_type's return type.
1393 (arc_expand_prologue): Likewise.
1394 (arc_expand_epilogue): Likewise.
1395 (arc_return_address_regs): Delete.
1396 (arc_return_address_register): New function.
1397 (arc_epilogue_uses): Use above function.
1398 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
1399 (arc_function_type): Change encoding, add naked type.
1400 (ARC_INTERRUPT_P): Change to handle the new encoding.
1401 (ARC_FAST_INTERRUPT_P): Likewise.
1402 (ARC_NORMAL_P): Define.
1403 (ARC_NAKED_P): Likewise.
1404 (arc_compute_function_type): Delete prototype.
1405 * config/arc/arc.md (in_ret_delay_slot): Use
1406 arc_return_address_register function.
1407 (simple_return): Likewise.
1408 (p_return_i): Likewise.
1410 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1412 PR tree-optimization/81428
1413 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
1414 can't be built for those types.
1416 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
1418 Remove stuff dead since r239246.
1420 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
1421 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
1422 (avr_inform_devices): Remove dead stuff.
1424 2017-07-17 Tamar Christina <tamar.christina@arm.com>
1426 * config/arm/arm_neon.h: Fix softp typo.
1428 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1430 PR tree-optimization/81365
1431 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
1432 aggregate moves onto bb predecessor edges, make sure there are no
1433 loads that could alias the lhs in between the start of bb and the
1436 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
1439 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
1440 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
1441 [LSHIFTRT, outer_code = TRUNCATE]: Same.
1443 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1445 PR tree-optimization/81396
1446 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
1447 (init_symbolic_number): Initialize it to 1.
1448 (perform_symbolic_merge): Add n_ops from both operands into the new
1450 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
1451 without base_addr as useless if they need more than one operation.
1452 (bswap_replace): Handle !bswap case for NULL base_addr.
1454 2017-07-17 Tom de Vries <tom@codesourcery.com>
1457 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
1460 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
1462 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
1463 conditional builtin define __FIX_LEON3FT_B2BST.
1465 2017-07-17 Daniel Cederman <cederman@gaisler.com>
1467 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
1468 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
1471 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1473 PR rtl-optimization/81424
1474 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
1475 to remove potential trapping from operands if -fnon-call-exceptions.
1477 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1479 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
1480 profile_proability for scalling.
1481 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
1483 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1485 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
1487 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1489 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
1490 fixpoint arithmetics.
1492 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1494 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
1495 fixpoint arithmetics.
1497 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1499 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
1500 fixpoint arithmetics.
1502 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1504 * profile-count.h (profile_probability::from_reg_br_prob_note,
1505 profile_probability::to_reg_br_prob_note): New functions.
1506 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
1507 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
1508 * predict.c (probability_reliable_p): Update.
1509 (edge_probability_reliable_p): Update.
1510 (br_prob_note_reliable_p): Update.
1511 (invert_br_probabilities): Update.
1512 (add_reg_br_prob_note): New function.
1513 (combine_predictions_for_insn): Update.
1514 * asan.c (asan_clear_shadow): Update.
1515 * cfgbuild.c (compute_outgoing_frequencies): Update.
1516 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
1517 (update_br_prob_note): Update.
1518 (rtl_verify_edges): Update.
1519 (purge_dead_edges): Update.
1520 (fixup_reorder_chain): Update.
1521 * emit-rtl.c (try_split): Update.
1522 * ifcvt.c (cond_exec_process_insns): Update.
1523 (cond_exec_process_if_block): Update.
1524 (dead_or_predicable): Update.
1525 * internal-fn.c (expand_addsub_overflow): Update.
1526 (expand_neg_overflow): Update.
1527 (expand_mul_overflow): Update.
1528 * loop-doloop.c (doloop_modify): Update.
1529 * loop-unroll.c (compare_and_jump_seq): Update.
1530 * optabs.c (emit_cmp_and_jump_insn_1): Update.
1531 * predict.h: Update.
1532 * reorg.c (mostly_true_jump): Update.
1534 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
1535 * config/alpha/alpha.c (emit_unlikely_jump): Update.
1536 * config/arc/arc.c: (emit_unlikely_jump): Update.
1537 * config/arm/arm.c: (emit_unlikely_jump): Update.
1538 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
1539 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
1540 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
1541 (ix86_print_operand): Update.
1542 (ix86_split_fp_branch): Update.
1543 (predict_jump): Update.
1544 * config/ia64/ia64.c (ia64_print_operand): Update.
1545 * config/mmix/mmix.c (mmix_print_operand): Update.
1546 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
1547 (rs6000_expand_split_stack_prologue): Update.
1548 * config/rs6000/rs6000.c: Update.
1549 * config/s390/s390.c (s390_expand_vec_strlen): Update.
1550 (s390_expand_vec_movstr): Update.
1551 (s390_expand_cs_tdsi): Update.
1552 (s390_expand_split_stack_prologue): Update.
1553 * config/sh/sh.c (sh_print_operand): Update.
1554 (expand_cbranchsi4): Update.
1555 (expand_cbranchdi4): Update.
1556 * config/sparc/sparc.c (output_v9branch): Update.
1557 * config/spu/spu.c (get_branch_target): Update.
1558 (ea_load_store_inline): Update.
1559 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
1560 * config/tilepro/tilepro.c: Update.
1562 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1564 * gimplify.c (mostly_copy_tree_r): Revert latest change.
1565 (gimplify_save_expr): Likewise.
1567 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1569 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
1571 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1573 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
1575 * timevar.def (TV_IPA_FNSUMMARY): Define.
1577 2017-07-16 Daniel Cederman <cederman@gaisler.com>
1579 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
1580 to back store errata sensitive sequence from being generated.
1581 (sqrtdf2_fix): Likewise.
1583 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1585 * tree-ssa-threadupdate.c (compute_path_counts,
1586 update_joiner_offpath_counts): Use profile_probability.
1588 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
1591 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1593 * config/arm/arm-c.c (arm_cpu_builtins): Define
1594 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1596 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
1598 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1599 array entries to represent __ieee128 versions of the
1600 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
1601 scalar_extract_sig, and scalar_insert_exp built-in functions.
1602 (altivec_resolve_overloaded_builtin): Add special case handling
1603 for the __builtin_scalar_insert_exp function, as represented by
1604 the P9V_BUILTIN_VEC_VSIEDP constant.
1605 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
1606 exponent support for __ieee128 argument.
1607 (VSESQP): Add scalar extract signature support for __ieee128
1609 (VSTDCNQP): Add scalar test negative support for __ieee128
1611 (VSIEQP): Add scalar insert exponent support for __int128 argument
1612 with __ieee128 result.
1613 (VSIEQPF): Add scalar insert exponent support for __ieee128
1614 argument with __ieee128 result.
1615 (VSTDCQP): Add scalar test data class support for __ieee128
1617 (VSTDCNQP): Add overload support for scalar test negative with
1619 (VSTDCQP): Add overload support for scalar test data class
1621 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
1623 (UNSPEC_VSX_SIEXPQP): New constant.
1624 (xsxexpqp): New insn for VSX scalar extract exponent quad
1626 (xsxsigqp): New insn for VSX scalar extract significand quad
1628 (xsiexpqpf): New insn for VSX scalar insert exponent quad
1629 precision with floating point argument.
1630 (xststdcqp): New expand for VSX scalar test data class quad
1632 (xststdcnegqp): New expand for VSX scalar test negative quad
1634 (xststdcqp): New insn to match expansions for VSX scalar test data
1635 class quad precision and VSX scalar test negative quad precision.
1636 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
1637 special case operand checking to enforce that second operand of
1638 VSX scalar test data class with quad precision argument is a 7-bit
1640 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
1641 prototypes and descriptions of __ieee128 versions of
1642 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
1643 scalar_test_data_class, and scalar_test_neg built-in functions.
1645 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1647 PR tree-optimization/81162
1648 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
1649 replace a negate with an add.
1651 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
1653 * doc/invoke.texi (arm/-mcpu): Document +crypto.
1655 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1657 * config/arm/arm-c.c (arm_cpu_builtins): Define
1658 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1660 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1662 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
1663 (armv8-r): Set ARM Cortex-R52 as default CPU.
1664 * config/arm/arm-tables.opt: Regenerate.
1665 * config/arm/arm-tune.md: Regenerate.
1666 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
1668 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
1669 extension for -mcpu=cortex-r52.
1671 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1673 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
1674 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
1675 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
1676 (fp-armv8): Define it as FP_ARMv8 only.
1677 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
1678 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
1680 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
1681 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
1682 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
1683 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
1684 than TARGET_FPU_ARMV8.
1685 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
1686 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
1687 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
1689 * config/arm/neon.md (neon_vrint): Likewise.
1690 (neon_vcvt): Likewise.
1691 (neon_<fmaxmin_op><mode>): Likewise.
1692 (<fmaxmin><mode>3): Likewise.
1693 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
1694 * config/arm/predicates.md (arm_cond_move_operator): Check against
1695 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
1697 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
1699 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
1702 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1704 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
1705 loop in comment with memset.
1707 2017-07-14 Martin Liska <mliska@suse.cz>
1709 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
1710 * dwarf2out.c (is_java): Remove the function.
1711 (output_pubname): Remove usage of the function.
1712 (lower_bound_default): Remove usage of DW_LANG_Java.
1713 (gen_compile_unit_die): Likewise.
1714 * gcc.c: Remove compiler defaults for .java and .zip files.
1715 * gimple-expr.c (remove_suffix): Change as there's no longer
1716 extension than 4-letter one.
1717 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
1718 (gimplify_save_expr): Likewise.
1719 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
1720 as it's possible even for other languages than Java.
1721 * langhooks.h (struct lang_hooks): Remove Java from a comment.
1722 * lto-opts.c (lto_write_options): Remove reference to Java.
1723 * opts.c (strip_off_ending): Update file extension handling.
1724 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
1725 * tree-eh.c (lower_resx): Likewise.
1726 * tree.c (free_lang_data_in_type): Remove dead code.
1727 (find_decls_types_r): Likewise.
1728 (build_common_builtin_nodes): Remove Java from a comment.
1729 (verify_type): Remove dead code.
1730 * varasm.c (assemble_external): Remove Java from a comment.
1732 2017-07-14 Martin Liska <mliska@suse.cz>
1734 * opts.c (finish_options): Add quotes.
1735 (common_handle_option): Likewise.
1737 2017-07-14 Martin Liska <mliska@suse.cz>
1739 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
1740 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
1742 * dwarf2out.c (lower_bound_default): Do not handle
1744 (gen_compile_unit_die): Likewise.
1745 * gcc.c: Remove default extension binding for GNU Pascal.
1746 * stmt.c: Remove Pascal language from a comment.
1747 * xcoffout.c: Likewise.
1749 2017-07-13 David Malcolm <dmalcolm@redhat.com>
1752 * diagnostic-show-locus.c (fixit_cmp): New function.
1753 (layout::layout): Sort m_fixit_hints.
1754 (column_range::column_range): Assert that the values are valid.
1755 (struct char_span): New struct.
1756 (correction::overwrite): New method.
1757 (struct source_line): New struct.
1758 (line_corrections::add_hint): Add assertions. Reimplement memcpy
1759 calls in terms of classes source_line and char_span, and
1760 correction::overwrite.
1761 (selftest::test_overlapped_fixit_printing_2): New function.
1762 (selftest::diagnostic_show_locus_c_tests): Call it.
1764 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
1766 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
1767 early if there is no lhs.
1769 2017-07-13 Martin Liska <mliska@suse.cz>
1771 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
1772 (gen_reference_type_die): Likewise.
1773 * stor-layout.c: Remove Pascal-related comment.
1775 2017-07-13 Martin Liska <mliska@suse.cz>
1777 * opts.c (finish_options): Add quotes to error messages.
1778 (parse_sanitizer_options): Likewise.
1780 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
1782 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
1784 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
1786 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
1788 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
1790 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
1792 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
1794 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1797 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
1798 provides the hardware capability bits, define the macro
1799 __BUILTIN_CPU_SUPPORTS__.
1800 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
1801 if GLIBC does not provide the hardware capability bits. Add a
1802 gcc_unreachable call if the built-in cpu function is neither
1803 __builtin_cpu_is nor __builtin_cpu_supports.
1804 (rs6000_get_function_versions_dispatcher): Change the warning
1805 that an old GLIBC is used which does not export the capability
1806 bits to be an error.
1807 * doc/extend.texi (target_clones attribute): Document the
1808 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
1809 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
1810 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
1811 the macros defined by GCC if the newer GLIBC is available.
1813 2017-07-12 Jeff Law <law@redhat.com>
1815 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
1816 remaining includes slightly.
1817 * config/riscv/riscv-builtins.c: Include profile-count.h.
1819 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1822 * config/avr/avr.c (avr_set_current_function): In diagnostic
1823 messages: Quote keywords and (parts of) identifiers.
1824 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
1827 2017-07-12 Carl Love <cel@us.ibm.com>
1829 * config/rs6000/rs6000-c.c: Add support for built-in functions
1830 vector bool char vec_revb (vector bool char);
1831 vector bool short vec_revb (vector short char);
1832 vector bool int vec_revb (vector bool int);
1833 vector bool long long vec_revb (vector bool long long);
1834 * doc/extend.texi: Update the built-in documentation file for the
1835 new built-in functions.
1837 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1839 * config/s390/s390.md: Remove movcc splitter.
1841 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1843 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
1844 load/store on condition.
1846 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1849 * config/avr/avr.c (avr_encode_section_info)
1850 [progmem && !TREE_READONLY]: Error if progmem object needs
1853 2017-07-11 Michael Collison <michael.collison@arm.com>
1855 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
1858 2017-07-11 Carl Love <cel@us.ibm.com>
1860 * config/rs6000/rs6000-c.c: Add support for builtins
1861 vector unsigned int vec_parity_lsbb (vector signed int);
1862 vector unsigned int vec_parity_lsbb (vector unsigned int);
1863 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
1864 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
1865 vector unsigned long long vec_parity_lsbb (vector signed long long);
1866 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
1867 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
1868 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
1869 * doc/extend.texi: Update the built-in documentation file for the
1870 new built-in functions.
1872 2017-07-11 David Malcolm <dmalcolm@redhat.com>
1874 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
1875 (layout::m_primary_loc): New field.
1876 (layout::layout): Initialize new field. Move location filtering
1877 logic from here to...
1878 (layout::maybe_add_location_range): ...this new method. Add
1879 support for filtering to just the lines already specified by other
1881 (layout::will_show_line_p): New method.
1882 (gcc_rich_location::add_location_if_nearby): New method.
1883 (selftest::test_add_location_if_nearby): New test function.
1884 (selftest::diagnostic_show_locus_c_tests): Call it.
1885 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
1888 2017-07-11 Tom de Vries <tom@codesourcery.com>
1890 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
1891 (bb_first_real_insn): New function.
1892 (nvptx_single): Add extra initialization of broadcasted condition
1895 2017-07-11 Nathan Sidwell <nathan@acm.org>
1897 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
1899 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
1901 * doc/extend.texi (AVR Function Attributes): Remove weblink to
1902 Binutils doc as TEXI will mess them up.
1903 * doc/invoke.texi (AVR Options): Same here.
1905 2017-07-11 Daniel Cederman <cederman@gaisler.com>
1907 * config/sparc/sparc.opt (mfix-ut700): New option.
1908 (mfix-gr712rc): Likewise.
1909 (sparc_fix_b2bst): New variable.
1910 * doc/invoke.texi (SPARC options): Document them.
1911 (ARM options): Fix warnings.
1912 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
1913 instructions to prevent sequences that can trigger the store-store
1914 errata for certain LEON3FT processors.
1915 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
1916 (sparc_option_override): Set sparc_fix_b2bst appropriately.
1917 * config/sparc/sparc.md (fix_b2bst): New attribute.
1918 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
1920 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
1923 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
1925 (*rsqrtsf2_sse): Ditto.
1927 (div<mode>3): Macroize insn from divdf3 and divsf3
1928 using MODEF mode iterator.
1930 2017-07-10 Martin Sebor <msebor@redhat.com>
1932 PR tree-optimization/80397
1933 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
1934 instead of testing for equality to INTEGER_TYPE.
1936 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
1938 * config.gcc: Remove uclibc from arc target spec.
1940 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
1942 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
1944 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1947 * lto-wrapper.c (remove_option): New function.
1948 (merge_and_complain): Merge PIC/PIE options more realistically.
1950 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
1952 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
1956 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
1957 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
1958 * config.in: Regenerate.
1959 * configure: Regenerate.
1960 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
1961 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
1962 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
1963 (TARGET_GASISR_PROLOGUES): ...target mask.
1964 * common/config/avr/avr-common.c
1965 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
1966 Set -mgas-isr-prologues.
1967 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
1968 INSERT_PASS_BEFORE for it.
1969 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
1970 * config/avr/avr.c (avr_option_override)
1971 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
1972 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
1973 (avr_attribute_table) <no_gccisr>: Add new function attribute.
1974 (avr_set_current_function) <is_no_gccisr>: Init machine field.
1975 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
1977 (make_avr_pass_pre_proep): New function.
1978 (emit_push_sfr) <treg>: Add argument to function and use it
1980 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
1981 and set machine->gasisr.yes.
1982 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
1983 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
1984 __gcc_isr.n_pushed to .L__stack_usage.
1985 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
1986 (avr_asm_final_postscan_insn): ...this new static function.
1987 * config/avr/avr.h (machine_function)
1988 <is_no_gccisr, use_L__stack_usage>: New fields.
1989 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
1990 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
1991 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
1992 (gasisr, *gasisr): New expander and insn.
1993 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
1994 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
1995 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
1997 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
1999 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
2002 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
2004 Move jump-tables out of .text again.
2007 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
2008 (ASM_OUTPUT_ADDR_VEC): New function.
2009 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
2010 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
2011 INSN_ADDRESSes as asm comment.
2012 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
2013 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
2014 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
2015 * config/avr/avr.md (*tablejump): Adjust comment.
2016 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
2017 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
2019 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
2020 (avr_output_addr_vec): New proto.
2021 (avr_log_t) <insn_addresses>: New field.
2023 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
2026 * config/i386/i386.c (ix86_function_arg_advance): Set
2027 outgoing_args_on_stack to true if there are outgoing arguments
2029 (ix86_function_arg): Likewise.
2030 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
2031 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
2032 * config/i386/i386.h (machine_function): Add
2033 outgoing_args_on_stack.
2035 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
2037 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
2039 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
2041 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
2043 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
2044 (REAL_H): Remove $(MACHMODE_H).
2045 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
2047 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
2048 $(MACHMODE_H) and double-int.h.
2049 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
2051 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
2054 2017-07-07 Andrew Pinski <apinski@cavium.com>
2056 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
2057 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
2059 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
2061 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
2062 Add warning if GCC was not configured to link against a GLIBC that
2063 exports the hardware capability bits.
2064 (make_resolver_func): Make resolver function private and not a
2065 COMDAT function. Create the name with clone_function_name instead
2066 of make_unique_name.
2069 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
2070 correct operand in doing the split.
2072 2017-07-07 Carl Love <cel@us.ibm.com>
2074 * config/rs6000/rs6000-c: Add support for built-in function
2075 vector unsigned short vec_pack_to_short_fp32 (vector float,
2077 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
2078 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
2079 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
2080 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
2081 (convert_4f32_8i16): Add define_expand.
2082 * doc/extend.texi: Update the built-in documentation file for the
2083 new built-in function.
2085 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2087 * config/sparc/m8.md: New file.
2088 * config/sparc/sparc.md: Include m8.md.
2090 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2092 * config/sparc/sparc.opt: New option -mvis4b.
2093 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
2094 (sparc_option_override): Handle VIS4B.
2095 (enum sparc_builtins): Define
2096 SPARC_BUILTIN_DICTUNPACK{8,16,32},
2097 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
2098 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
2099 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
2100 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
2101 (check_constant_argument): New function.
2102 (sparc_vis_init_builtins): Define builtins
2103 __builtin_vis_dictunpack{8,16,32},
2104 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
2105 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
2106 __builtin_vis_fpcmpde{8,16,32}shl and
2107 __builtin_vis_fpcmpur{8,16,32}shl.
2108 (sparc_expand_builtin): Check that the constant operands to
2109 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
2110 constant and in range.
2111 * config/sparc/sparc-c.c (sparc_target_macros): Handle
2113 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
2114 (SPARC_IMM5_P): Likewise.
2115 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
2116 (enabled): Handle vis4b.
2117 (UNSPEC_DICTUNPACK): New unspec.
2118 (UNSPEC_FPCMPSHL): Likewise.
2119 (UNSPEC_FPUCMPSHL): Likewise.
2120 (UNSPEC_FPCMPDESHL): Likewise.
2121 (UNSPEC_FPCMPURSHL): Likewise.
2122 (cpu_feature): New CPU feature `vis4b'.
2123 (dictunpack{8,16,32}): New insns.
2124 (FPCSMODE): New mode iterator.
2125 (fpcscond): New code iterator.
2126 (fpcsucond): Likewise.
2127 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
2128 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
2129 (fpcmpde{8,16,32}{si,di}shl): Likewise.
2130 (fpcmpur{8,16,32}{si,di}shl): Likewise.
2131 * config/sparc/constraints.md: Define constraints `q' for unsigned
2132 2-bit integer constants and `t' for unsigned 5-bit integer
2134 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
2136 (imm5_operand_dictunpack16): Likewise.
2137 (imm5_operand_dictunpack32): Likewise.
2138 (imm2_operand): Likewise.
2139 * doc/invoke.texi (SPARC Options): Document -mvis4b.
2140 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
2141 ditunpack* and fpcmp*shl builtins.
2143 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2145 * config.gcc: Handle m8 in --with-{cpu,tune} options.
2146 * config.in: Add HAVE_AS_SPARC6 define.
2147 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
2149 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
2151 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
2152 (CPP_CPU_SPEC): Handle m8.
2153 (ASM_CPU_SPEC): Likewise.
2154 * config/sparc/sparc-opts.h (enum processor_type): Add
2156 * config/sparc/sparc.c (m8_costs): New struct.
2157 (sparc_option_override): Handle TARGET_CPU_m8.
2158 (sparc32_initialize_trampoline): Likewise.
2159 (sparc64_initialize_trampoline): Likewise.
2160 (sparc_issue_rate): Likewise.
2161 (sparc_register_move_cost): Likewise.
2162 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
2163 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
2164 (ASM_CPU64_DEFAULT_SPEC): Likewise.
2165 (CPP_CPU_SPEC): Handle M8.
2166 (ASM_CPU_SPEC): Likewise.
2167 (AS_M8_FLAG): Define.
2168 * config/sparc/sparc.md: Add m8 to the cpu attribute.
2169 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
2170 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
2172 * configure: Regenerate.
2173 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
2176 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2178 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
2180 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
2181 ("*movdi_insn_sp32"): Do not set v3pipe.
2182 ("*movsi_insn"): Likewise.
2183 ("*movdi_insn_sp64"): Likewise.
2184 ("*movsf_insn"): Likewise.
2185 ("*movdf_insn_sp32"): Likewise.
2186 ("*movdf_insn_sp64"): Likewise.
2187 ("*zero_extendsidi2_insn_sp64"): Likewise.
2188 ("*sign_extendsidi2_insn"): Likewise.
2189 ("*mov<VM32:mode>_insn"): Likewise.
2190 ("*mov<VM64:mode>_insn_sp64"): Likewise.
2191 ("*mov<VM64:mode>_insn_sp32"): Likewise.
2192 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
2193 ("<vlop:code><VL:mode>3"): Likewise.
2194 ("*not_<vlop:code><VL:mode>3"): Likewise.
2195 ("*nand<VL:mode>_vis"): Likewise.
2196 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
2197 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
2198 ("one_cmpl<VL:mode>2"): Likewise.
2199 ("faligndata<VM64:mode>_vis"): Likewise.
2200 ("alignaddrsi_vis"): Likewise.
2201 ("alignaddrdi_vis"): Likweise.
2202 ("alignaddrlsi_vis"): Likewise.
2203 ("alignaddrldi_vis"): Likewise.
2204 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
2205 ("bmaskdi_vis"): Likewise.
2206 ("bmasksi_vis"): Likewise.
2207 ("bshuffle<VM64:mode>_vis"): Likewise.
2208 ("cmask8<P:mode>_vis"): Likewise.
2209 ("cmask16<P:mode>_vis"): Likewise.
2210 ("cmask32<P:mode>_vis"): Likewise.
2211 ("pdistn<P:mode>_vis"): Likewise.
2212 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
2214 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2216 * config/sparc/sparc.md ("subtype"): New insn attribute.
2217 ("*wrgsr_sp64"): Set insn subtype.
2218 ("*rdgsr_sp64"): Likewise.
2219 ("alignaddrsi_vis"): Likewise.
2220 ("alignaddrdi_vis"): Likewise.
2221 ("alignaddrlsi_vis"): Likewise.
2222 ("alignaddrldi_vis"): Likewise.
2223 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
2224 ("fexpand_vis"): Likewise.
2225 ("fpmerge_vis"): Likewise.
2226 ("faligndata<VM64:mode>_vis"): Likewise.
2227 ("bshuffle<VM64:mode>_vis"): Likewise.
2228 ("cmask8<P:mode>_vis"): Likewise.
2229 ("cmask16<P:mode>_vis"): Likewise.
2230 ("cmask32<P:mode>_vis"): Likewise.
2231 ("fchksm16_vis"): Likewise.
2232 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
2233 ("fmean16_vis"): Likewise.
2234 ("fp<plusminus_insn>64_vis"): Likewise.
2235 ("<plusminus_insn>v8qi3"): Likewise.
2236 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
2237 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
2238 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
2239 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
2240 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
2241 ("*movqi_insn"): Likewise.
2242 ("*movhi_insn"): Likewise.
2243 ("*movsi_insn"): Likewise.
2244 ("movsi_pic_gotdata_op"): Likewise.
2245 ("*movdi_insn_sp32"): Likewise.
2246 ("*movdi_insn_sp64"): Likewise.
2247 ("movdi_pic_gotdata_op"): Likewise.
2248 ("*movsf_insn"): Likewise.
2249 ("*movdf_insn_sp32"): Likewise.
2250 ("*movdf_insn_sp64"): Likewise.
2251 ("*zero_extendhisi2_insn"): Likewise.
2252 ("*zero_extendqihi2_insn"): Likewise.
2253 ("*zero_extendqisi2_insn"): Likewise.
2254 ("*zero_extendqidi2_insn"): Likewise.
2255 ("*zero_extendhidi2_insn"): Likewise.
2256 ("*zero_extendsidi2_insn_sp64"): Likewise.
2257 ("ldfsr"): Likewise.
2258 ("prefetch_64"): Likewise.
2259 ("prefetch_32"): Likewise.
2260 ("tie_ld32"): Likewise.
2261 ("tie_ld64"): Likewise.
2262 ("*tldo_ldub_sp32"): Likewise.
2263 ("*tldo_ldub1_sp32"): Likewise.
2264 ("*tldo_ldub2_sp32"): Likewise.
2265 ("*tldo_ldub_sp64"): Likewise.
2266 ("*tldo_ldub1_sp64"): Likewise.
2267 ("*tldo_ldub2_sp64"): Likewise.
2268 ("*tldo_ldub3_sp64"): Likewise.
2269 ("*tldo_lduh_sp32"): Likewise.
2270 ("*tldo_lduh1_sp32"): Likewise.
2271 ("*tldo_lduh_sp64"): Likewise.
2272 ("*tldo_lduh1_sp64"): Likewise.
2273 ("*tldo_lduh2_sp64"): Likewise.
2274 ("*tldo_lduw_sp32"): Likewise.
2275 ("*tldo_lduw_sp64"): Likewise.
2276 ("*tldo_lduw1_sp64"): Likewise.
2277 ("*tldo_ldx_sp64"): Likewise.
2278 ("*mov<VM32:mode>_insn"): Likewise.
2279 ("*mov<VM64:mode>_insn_sp64"): Likewise.
2280 ("*mov<VM64:mode>_insn_sp32"): Likewise.
2282 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2284 * config/sparc/sparc.md ("type"): New insn type viscmp.
2285 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
2287 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
2288 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
2289 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
2290 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
2292 ("n7_vis_logical_11cycle"): Likewise.
2293 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
2294 * config/sparc/niagara2.md ("niag3_vis": Likewise.
2295 * config/sparc/niagara.md ("niag_vis"): Likewise.
2296 * config/sparc/ultra3.md ("us3_fga"): Likewise.
2297 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
2299 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2301 * config/sparc/sparc.md: New instruction type `bmask'.
2302 (bmaskdi_vis): Use the `bmask' type.
2303 (bmasksi_vis): Likewise.
2304 * config/sparc/ultra3.md (us3_array): Likewise.
2305 * config/sparc/niagara7.md (n7_array): Likewise.
2306 * config/sparc/niagara4.md (n4_array): Likewise.
2307 * config/sparc/niagara2.md (niag2_vis): Likewise.
2308 (niag3_vis): Likewise.
2309 * config/sparc/niagara.md (niag_vis): Likewise.
2311 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2313 * ipa-comdats.c: Remove optimize check from gate.
2314 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
2315 for functions not optimized.
2316 (ipa_fn_summary_read): Skip optimize check.
2317 (ipa_fn_summary_write): Likewise.
2318 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
2320 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
2322 (can_inline_edge_p): Check flag_pcc_struct_return for match.
2323 (check_callers): Give up on caller which is not optimized.
2324 (inline_small_functions): Likewise.
2325 (ipa_inline): Do not give up when not optimizing.
2326 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
2327 away unoptimizes cdtors.
2328 (whole_program_function_and_variable_visibility): Do
2329 ipa_discover_readonly_nonaddressable_vars in LTO mode.
2330 * ipa.c (process_references): Do not check optimize.
2331 (symbol_table::remove_unreachable_nodes): Update optimize check.
2332 (set_writeonly_bit): Update optimize check.
2333 (pass_ipa_cdtor_merge::gate): Do not check optimize.
2334 (pass_ipa_single_use::gate): Remove.
2336 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2338 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
2339 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
2340 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
2341 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
2342 permute_load, permute_store, adjust_extract, adjust_splat,
2343 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
2344 replace_swap_with_copy, dump_swap_insn_table,
2345 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
2346 recombine_lvx_pattern, recombine_stvx_pattern,
2347 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
2348 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
2349 to file rs6000-p8swap.c.
2350 * config/rs6000/rs6000-p8swap.c: New file.
2351 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
2352 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
2353 and rs6000*-*-* targets.
2355 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2357 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
2359 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2361 * lto-wrapper.c (merge_and_complain): Do not merge
2362 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
2363 fsigned_zeros, ftrapping_math, fwrapv.
2364 (append_compiler_options): Do not track these options.
2365 (append_linker_options): Likewie
2367 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2369 * cgraphunit.c (cgraph_node::finalize_function): When
2370 !flag_toplevel_reorde set no_reorder flag.
2371 (varpool_node::finalize_decl): Likewise.
2372 (symbol_table::compile): Drop no toplevel reorder path.
2374 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2376 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
2377 edges; zero probability is not better than uninitialized.
2379 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
2381 * asan.h (asan_sanitize_allocas_p): Declare.
2382 * asan.c (asan_sanitize_allocas_p): New function.
2383 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
2384 (handle_builtin_alloca): Likewise.
2385 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
2386 if !asan_sanitize_allocas_p.
2387 * params.def (asan-instrument-allocas): Add new option.
2388 * params.h (ASAN_PROTECT_ALLOCAS): Define.
2389 * opts.c (common_handle_option): Disable allocas sanitization for
2392 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
2394 * asan.c: Include gimple-fold.h.
2395 (get_last_alloca_addr): New function.
2396 (handle_builtin_stackrestore): Likewise.
2397 (handle_builtin_alloca): Likewise.
2398 (asan_emit_allocas_unpoison): Likewise.
2399 (get_mem_refs_of_builtin_call): Add new parameter, remove const
2400 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
2401 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
2402 (instrument_builtin_call): Pass gimple iterator to
2403 get_mem_refs_of_builtin_call.
2404 (last_alloca_addr): New global.
2405 * asan.h (asan_emit_allocas_unpoison): Declare.
2406 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
2407 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
2408 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
2409 if function calls alloca.
2410 * gimple-fold.c (replace_call_with_value): Remove static keyword.
2411 * gimple-fold.h (replace_call_with_value): Declare.
2412 * internal-fn.c: Include asan.h.
2413 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
2414 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
2416 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2418 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
2419 (C_SELFTEST_FLAGS): New.
2420 (CPP_SELFTEST_FLAGS): New.
2421 (SELFTEST_DEPS): New, from deps of s-selftest.
2422 (C_SELFTEST_DEPS): New, from deps of s-selftest.
2423 (CPP_SELFTEST_DEPS): New.
2424 (selftest): Add dependency on s-selftest-c++.
2425 (s-selftest): Rename to...
2426 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
2427 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
2428 than SELFTEST_FLAGS.
2429 (selftest-gdb): Rename to...
2430 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
2432 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
2433 (selftest-valgrind): Rename to...
2434 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
2436 (selftest-valgrind): Reintroduce as an alias for
2437 selftest-c-valgrind.
2438 (s-selftest-c++): New.
2439 (selftest-c++-gdb): New.
2440 (selftest-c++-valgrind): New.
2442 2017-07-06 Olivier Hainque <hainque@adacore.com>
2444 * gcc.c (process_command): When deciding if undefined variables
2445 should be ignored when processing specs, accept "gcc -v" as well.
2447 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2449 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
2450 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
2452 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
2454 * config/arm/arm-cpus.in (armv8-r): Add new entry.
2455 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
2456 * config/arm/arm-tables.opt: Regenerate.
2457 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
2459 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
2461 2017-07-06 Carl Love <cel@us.ibm.com>
2463 * ChangeLog: Clean up from mid air collision
2465 2017-07-06 Carl Love <cel@us.ibm.com>
2467 * config/rs6000/rs6000-c.c: Add support for built-in functions
2468 vector signed int vec_subc (vector signed int, vector signed int);
2469 vector signed __int128 vec_subc (vector signed __int128,
2470 vector signed __int128);
2471 vector unsigned __int128 vec_subc (vector unsigned __int128,
2472 vector unsigned __int128);
2473 vector signed int vec_sube (vector signed int, vector signed int,
2475 vector unsigned int vec_sube (vector unsigned int,
2476 vector unsigned int,
2477 vector unsigned int);
2478 vector signed __int128 vec_sube (vector signed __int128,
2479 vector signed __int128,
2480 vector signed__int128);
2481 vector unsigned __int128 vec_sube (vector unsigned __int128,
2482 vector unsigned __int128,
2483 vector unsigned __int128);
2484 vector signed int vec_subec (vector signed int, vector signed int,
2486 vector unsigned int vec_subec (vector unsigned int,
2487 vector unsigned int,
2488 vector unsigned int);
2489 vector signed __int128 vec_subec (vector signed __int128,
2490 vector signed __int128,
2491 vector signed__int128);
2492 vector unsigned __int128 vec_subec (vector unsigned __int128,
2493 vector unsigned __int128,
2494 vector unsigned __int128);
2495 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
2496 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
2497 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
2498 BU_ALTIVEC_OVERLOAD_X definitions.
2499 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
2500 * doc/extend.texi: Update the built-in documentation file for the new
2503 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2506 * diagnostic-show-locus.c (layout::layout): Use start and finish
2507 spelling location for the start and finish of each range.
2508 * genmatch.c (linemap_client_expand_location_to_spelling_point):
2509 Add unused aspect param.
2510 * input.c (expand_location_1): Add "aspect" param, and use it
2511 to access the correct part of the location.
2512 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
2514 (expand_location_to_spelling_point): Likewise.
2515 (linemap_client_expand_location_to_spelling_point): Add "aspect"
2516 param, and pass it to expand_location_1.
2518 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
2520 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
2521 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
2522 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
2523 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
2524 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
2525 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
2526 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
2527 _mm_maskz_getmant_ss): New intrinsics.
2528 (__builtin_ia32_getexpss128_mask): Changed to ...
2529 __builtin_ia32_getexpss128_round ... this.
2530 (__builtin_ia32_getexpsd128_mask): Changed to ...
2531 __builtin_ia32_getexpsd128_round ... this.
2532 * config/i386/i386-builtin-types.def
2533 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
2534 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
2535 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
2536 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
2537 __builtin_ia32_getmantss_mask_round): New builtins.
2538 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
2539 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
2540 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
2541 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
2542 * config/i386/sse.md
2543 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
2544 avx512f_sgetexp<mode><mask_scalar_name>
2545 <round_saeonly_scalar_name> ... this.
2546 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
2547 %0, %1, %2<round_saeonly_op3>}): Changed to ...
2548 vgetexp<ssescalarmodesuffix>
2549 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2550 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
2551 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
2552 avx512f_vgetmant<mode><mask_scalar_name>
2553 <round_saeonly_scalar_name> ... this.
2554 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
2555 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
2556 vgetmant<ssescalarmodesuffix>
2557 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
2558 %0<mask_scalar_operand4>, %1, %2
2559 <round_saeonly_scalar_mask_op4>, %3} ... this.
2560 * config/i386/subst.md (mask_scalar_operand4,
2561 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
2562 round_saeonly_scalar_nimm_predicate): New subst attributes.
2564 2017-07-06 Julia Koval <julia.koval@intel.com>
2566 * config/i386/i386.c (ix86_erase_embedded_rounding):
2567 Remove code for old rounding pattern.
2569 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
2571 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
2573 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
2575 * doc/sourcebuild.texi (Test Directives, Variants of
2576 dg-require-support): Add documentation for dg-require-stack-check.
2578 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
2580 * config/i386/subst.md (mask_scalar, round_scalar,
2581 round_saeonly_scalar): New meta-templates.
2582 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
2583 round_scalar_mask_operand3, round_scalar_mask_op3,
2584 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
2585 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
2586 round_saeonly_scalar_constraint,
2587 round_saeonly_scalar_prefix): New subst attribute.
2588 * config/i386/sse.md
2589 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
2590 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
2591 <round_scalar_name> ... this.
2592 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
2593 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
2594 <round_scalar_name> ... this.
2595 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
2596 <sse>_vm<code><mode>3<mask_scalar_name>
2597 <round_saeonly_scalar_name> ... this.
2598 (v<plusminus_mnemonic><ssescalarmodesuffix>
2599 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2600 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2601 v<plusminus_mnemonic><ssescalarmodesuffix>
2602 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2603 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2604 (v<multdiv_mnemonic><ssescalarmodesuffix>
2605 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2606 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2607 v<multdiv_mnemonic><ssescalarmodesuffix>
2608 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2609 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2610 (v<maxmin_float><ssescalarmodesuffix>
2611 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
2612 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
2613 v<maxmin_float><ssescalarmodesuffix>
2614 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2615 %0<mask_scalar_operand3>, %1, %<iptr>2
2616 <round_saeonly_scalar_mask_op3>} ... this.
2618 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
2620 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
2621 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
2623 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
2624 Alan Hayward <alan.hayward@arm.com>
2625 David Sherwood <david.sherwood@arm.com>
2627 * combine.c (simplify_if_then_else): Remove "enum" before
2629 * compare-elim.c (can_eliminate_compare): Likewise.
2630 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
2632 (aarch64_lookup_simd_builtin_type): Likewise.
2633 (aarch64_simd_builtin_type): Likewise.
2634 (aarch64_init_simd_builtin_types): Likewise.
2635 (aarch64_simd_expand_args): Likewise.
2636 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
2638 (aarch64_reverse_mask): Likewise.
2639 (aarch64_simd_emit_reg_reg_move): Likewise.
2640 (aarch64_gen_adjusted_ldpstp): Likewise.
2641 (aarch64_ccmp_mode_to_code): Likewise.
2642 (aarch64_operands_ok_for_ldpstp): Likewise.
2643 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2644 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
2646 (aarch64_min_divisions_for_recip_mul): Likewise.
2647 (aarch64_reassociation_width): Likewise.
2648 (aarch64_get_condition_code_1): Likewise.
2649 (aarch64_simd_emit_reg_reg_move): Likewise.
2650 (aarch64_simd_attr_length_rglist): Likewise.
2651 (aarch64_reverse_mask): Likewise.
2652 (aarch64_operands_ok_for_ldpstp): Likewise.
2653 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2654 (aarch64_gen_adjusted_ldpstp): Likewise.
2655 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
2657 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
2658 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
2659 (arm_lookup_simd_builtin_type): Likewise.
2660 (arm_simd_builtin_type): Likewise.
2661 (arm_init_simd_builtin_types): Likewise.
2662 (arm_expand_builtin_args): Likewise.
2663 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
2664 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
2665 (ft32_setup_incoming_varargs): Likewise.
2666 (ft32_function_arg): Likewise.
2667 (ft32_function_arg_advance): Likewise.
2668 (ft32_pass_by_reference): Likewise.
2669 (ft32_arg_partial_bytes): Likewise.
2670 (ft32_valid_pointer_mode): Likewise.
2671 (ft32_addr_space_pointer_mode): Likewise.
2672 (ft32_addr_space_legitimate_address_p): Likewise.
2673 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
2675 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
2676 (ix86_emit_outlined_ms2sysv_restore): Likewise.
2677 (iamcu_alignment): Likewise.
2678 (canonicalize_vector_int_perm): Likewise.
2679 (ix86_noce_conversion_profitable_p): Likewise.
2680 (ix86_mpx_bound_mode): Likewise.
2681 (ix86_operands_ok_for_move_multiple): Likewise.
2682 * config/microblaze/microblaze-protos.h
2683 (microblaze_expand_conditional_branch_reg): Likewise.
2684 * config/microblaze/microblaze.c
2685 (microblaze_expand_conditional_branch_reg): Likewise.
2686 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
2688 (rs6000_reassociation_width): Likewise.
2689 (rs6000_invalid_binary_op): Likewise.
2690 (fusion_p9_p): Likewise.
2691 (emit_fusion_p9_load): Likewise.
2692 (emit_fusion_p9_store): Likewise.
2693 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
2695 (riscv_hard_regno_mode_ok_p): Likewise.
2696 (riscv_address_insns): Likewise.
2697 (riscv_split_symbol): Likewise.
2698 (riscv_legitimize_move): Likewise.
2699 (riscv_function_value): Likewise.
2700 (riscv_hard_regno_nregs): Likewise.
2701 (riscv_expand_builtin): Likewise.
2702 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
2703 (riscv_build_integer): Likewise.
2704 (riscv_split_integer): Likewise.
2705 (riscv_legitimate_constant_p): Likewise.
2706 (riscv_cannot_force_const_mem): Likewise.
2707 (riscv_regno_mode_ok_for_base_p): Likewise.
2708 (riscv_valid_base_register_p): Likewise.
2709 (riscv_valid_offset_p): Likewise.
2710 (riscv_valid_lo_sum_p): Likewise.
2711 (riscv_classify_address): Likewise.
2712 (riscv_legitimate_address_p): Likewise.
2713 (riscv_address_insns): Likewise.
2714 (riscv_load_store_insns): Likewise.
2715 (riscv_force_binary): Likewise.
2716 (riscv_split_symbol): Likewise.
2717 (riscv_force_address): Likewise.
2718 (riscv_legitimize_address): Likewise.
2719 (riscv_move_integer): Likewise.
2720 (riscv_legitimize_const_move): Likewise.
2721 (riscv_legitimize_move): Likewise.
2722 (riscv_address_cost): Likewise.
2723 (riscv_subword): Likewise.
2724 (riscv_output_move): Likewise.
2725 (riscv_canonicalize_int_order_test): Likewise.
2726 (riscv_emit_int_order_test): Likewise.
2727 (riscv_function_arg_boundary): Likewise.
2728 (riscv_pass_mode_in_fpr_p): Likewise.
2729 (riscv_pass_fpr_single): Likewise.
2730 (riscv_pass_fpr_pair): Likewise.
2731 (riscv_get_arg_info): Likewise.
2732 (riscv_function_arg): Likewise.
2733 (riscv_function_arg_advance): Likewise.
2734 (riscv_arg_partial_bytes): Likewise.
2735 (riscv_function_value): Likewise.
2736 (riscv_pass_by_reference): Likewise.
2737 (riscv_setup_incoming_varargs): Likewise.
2738 (riscv_print_operand): Likewise.
2739 (riscv_elf_select_rtx_section): Likewise.
2740 (riscv_save_restore_reg): Likewise.
2741 (riscv_for_each_saved_reg): Likewise.
2742 (riscv_register_move_cost): Likewise.
2743 (riscv_hard_regno_mode_ok_p): Likewise.
2744 (riscv_hard_regno_nregs): Likewise.
2745 (riscv_class_max_nregs): Likewise.
2746 (riscv_memory_move_cost): Likewise.
2747 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
2748 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
2749 (rl78_addr_space_address_mode): Likewise.
2750 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2752 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2753 (rs6000_reassociation_width): Likewise.
2754 (rs6000_invalid_binary_op): Likewise.
2755 (fusion_p9_p): Likewise.
2756 (emit_fusion_p9_load): Likewise.
2757 (emit_fusion_p9_store): Likewise.
2758 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
2759 (ok_for_simple_move_operands): Likewise.
2760 (ok_for_simple_move_strict_operands): Likewise.
2761 (ok_for_simple_arith_logic_operands): Likewise.
2762 (visium_legitimize_reload_address): Likewise.
2763 (visium_select_cc_mode): Likewise.
2764 (output_cbranch): Likewise.
2765 (visium_split_double_move): Likewise.
2766 (visium_expand_copysign): Likewise.
2767 (visium_expand_int_cstore): Likewise.
2768 (visium_expand_fp_cstore): Likewise.
2769 * config/visium/visium.c (visium_pass_by_reference): Likewise.
2770 (visium_function_arg): Likewise.
2771 (visium_function_arg_advance): Likewise.
2772 (visium_libcall_value): Likewise.
2773 (visium_setup_incoming_varargs): Likewise.
2774 (visium_legitimate_constant_p): Likewise.
2775 (visium_legitimate_address_p): Likewise.
2776 (visium_legitimize_address): Likewise.
2777 (visium_secondary_reload): Likewise.
2778 (visium_register_move_cost): Likewise.
2779 (visium_memory_move_cost): Likewise.
2780 (prepare_move_operands): Likewise.
2781 (ok_for_simple_move_operands): Likewise.
2782 (ok_for_simple_move_strict_operands): Likewise.
2783 (ok_for_simple_arith_logic_operands): Likewise.
2784 (visium_function_value_1): Likewise.
2785 (rtx_ok_for_offset_p): Likewise.
2786 (visium_legitimize_reload_address): Likewise.
2787 (visium_split_double_move): Likewise.
2788 (visium_expand_copysign): Likewise.
2789 (visium_expand_int_cstore): Likewise.
2790 (visium_expand_fp_cstore): Likewise.
2791 (visium_split_cstore): Likewise.
2792 (visium_select_cc_mode): Likewise.
2793 (visium_split_cbranch): Likewise.
2794 (output_cbranch): Likewise.
2795 (visium_print_operand_address): Likewise.
2796 * expmed.c (flip_storage_order): Likewise.
2797 * expmed.h (emit_cstore): Likewise.
2798 (flip_storage_order): Likewise.
2799 * genrecog.c (validate_pattern): Likewise.
2800 * hsa-gen.c (gen_hsa_addr): Likewise.
2801 * internal-fn.c (expand_arith_overflow): Likewise.
2802 * ira-color.c (allocno_copy_cost_saving): Likewise.
2803 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2804 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
2805 (process_invariant_for_inheritance): Likewise.
2806 * lra-eliminations.c (move_plus_up): Likewise.
2807 * omp-low.c (lower_oacc_reductions): Likewise.
2808 * simplify-rtx.c (simplify_subreg): Likewise.
2809 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
2810 (TARGET_CHKP_BOUND_MODE): Likewise..
2811 * targhooks.c (default_chkp_bound_mode): Likewise.
2812 (default_setup_incoming_vararg_bounds): Likewise.
2813 * targhooks.h (default_chkp_bound_mode): Likewise.
2814 (default_setup_incoming_vararg_bounds): Likewise.
2815 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
2816 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
2817 (have_whole_vector_shift): Likewise.
2818 * tree-vect-stmts.c (vectorizable_load): Likewise.
2819 * doc/tm.texi: Regenerate.
2821 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2823 Graceful degrade if Binutils PR21472 is not available.
2826 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
2827 .rodata in flash test fails.
2828 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
2829 * confgure: Regenerate.
2830 * config.in: Regenerate.
2831 * config/avr/avr.c (avr_asm_named_section)
2832 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
2833 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
2834 (avr_asm_init_sections): Same.
2836 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2838 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
2839 (fma<VH:mode>4_intrinsic): Likewise.
2840 (*fmsub<VCVTF:mode>4): Likewise.
2841 (*fmsub<VH:mode>4_intrinsic): Likewise.
2843 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2846 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
2847 Don't depend on "optimize > 0".
2848 (out_movhi_r_mr, out_movqi_mr_r): Same.
2849 (out_movhi_mr_r, out_movqi_r_mr): Same.
2850 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
2851 io_address_operand on "optimize > 0".
2853 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2855 * tree-loop-distribution.c: Add general explanantion on the pass.
2856 (generate_loops_for_partition): Mark distributed loop.
2857 (pg_add_dependence_edges): New parameter. Handle alias data
2858 dependence specially and record it in the parameter if asked.
2859 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
2860 (init_partition_graph_vertices, add_partition_graph_edge): New.
2861 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
2862 (free_partition_graph_vdata, build_partition_graph): New.
2863 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
2864 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
2865 (data_ref_segment_size, latch_dominated_by_data_ref): New.
2866 (compute_alias_check_pairs, version_loop_by_alias_check): New.
2867 (version_for_distribution_p, finalize_partitions): New.
2868 (distribute_loop): Handle alias data dependence specially. Factor
2869 out loop fusion code as functions and call these functions.
2871 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2873 * tree-loop-distribution.c (classify_partition): New parameter and
2874 better handle reduction statement.
2875 (rdg_build_partitions): Revise comment.
2876 (distribute_loop): Compute statements in all partitions and pass it
2877 to classify_partition.
2879 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2881 * tree-loop-distribution.c (enum partition_type): New.
2882 (struct partition): New field type.
2883 (partition_merge_into): Add parameter. Update partition type.
2884 (data_dep_in_cycle_p, update_type_for_merge): New functions.
2885 (build_rdg_partition_for_vertex): Compute partition type.
2886 (rdg_build_partitions): Dump partition type.
2887 (distribute_loop): Update calls to partition_merge_into.
2889 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2891 * tree-loop-distribution.c (struct ddr_hasher): New.
2892 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
2894 (classify_partition): Call get_data_dependence.
2895 (pg_add_dependence_edges): Ditto.
2896 (distribute_loop): Release data dependence hash table.
2898 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2900 * tree-loop-distribution.c (ref_base_address): Delete.
2901 (similar_memory_accesses): Rename ...
2902 (share_memory_accesses): ... to this. Check if partitions access
2903 the same memory reference.
2904 (distribute_loop): Call share_memory_accesses.
2906 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2908 * tree-loop-distribution.c (struct partition): New field recording
2910 (partition_alloc, partition_free): Init and release data refs.
2911 (partition_merge_into): Merge data refs.
2912 (build_rdg_partition_for_vertex): Collect data refs for partition.
2913 (pg_add_dependence_edges): Change parameters from vector to bitmap.
2915 (distribute_loop): Remve data refs from vertice data of partition
2918 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2920 * tree-loop-distribution.c (params.h): Include header file.
2921 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
2922 (datarefs_vec): New global var.
2923 (create_rdg_vertices): Use datarefs_vec directly.
2924 (free_rdg): Don't free data references.
2925 (build_rdg): Update use. Don't free data references.
2926 (distribute_loop): Compute global variable for data references.
2927 Bail out if there are too many data references.
2929 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2931 * tree-loop-distribution.c (loop_nest): New global var.
2932 (build_rdg): Use loop directly, rather than loop nest.
2933 (pg_add_dependence_edges): Remove loop nest parameter. Use global
2935 (distribute_loop): Compute global variable loop nest. Update use.
2937 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2939 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
2940 (partition_merge_into): New parameter. Dump reason for fusion.
2941 (distribute_loop): Update use of partition_merge_into.
2943 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2945 * tree-loop-distribution.c (bb_top_order_index): New.
2946 (bb_top_order_index_size, bb_top_order_cmp): New.
2947 (stmts_from_loop): Use topological order.
2948 (pass_loop_distribution::execute): Compute and release topological
2949 order for basic blocks.
2951 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2953 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
2956 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2958 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
2959 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
2960 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
2961 * internal-fn.def (LOOP_DIST_ALIAS): New.
2962 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
2963 (fold_loop_internal_call): ... this.
2964 (vect_loop_dist_alias_call): New function.
2965 (set_uid_loop_bbs): Call fold_loop_internal_call.
2966 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
2969 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
2972 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
2973 Require dead FLAGS_REG at the beginning of a peephole.
2975 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
2978 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
2979 arguments in the call to __builtin_ia32_sbb_u32.
2980 (_subborrow_u64): Swap _X and _Y arguments in the call to
2981 __builtin_ia32_sbb_u64.
2983 2017-07-04 Jakub Jelinek <jakub@redhat.com>
2986 * tree-vrp.c (compare_assert_loc): Turn into a function template
2987 with stable template parameter. Only test if a->e is NULL,
2988 !a->e == !b->e has been verified already. Use e == NULL or
2989 e != NULL instead of e or ! e tests. If stable is true, don't use
2990 iterative_hash_expr, on the other side allow a or b or both NULL
2991 and sort the NULLs last.
2992 (process_assert_insertions): Sort using compare_assert_loc<false>
2993 instead of compare_assert_loc, later sort using
2994 compare_assert_loc<true> before calling process_assert_insertions_for
2995 in a loop. Use break instead of continue once seen NULL pointer.
2997 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
2999 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
3000 Cortex-R7 and Cortex-R8 processors.
3002 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
3004 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
3005 uninitialized while src is not.
3007 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
3009 * common/config/arm/arm-common.c: Adjust include path for
3011 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
3012 (arm-cpu.h): Create in build directory. Adjust dependency rules.
3013 (arm-cpu-data.h): Likewise.
3014 (arm-cpu-cdata.h): Likewise.
3015 * config/arm/arm-cpu.h: Delete.
3016 * config/arm/arm-cpu-cdata.h: Delete.
3017 * config/arm/arm-cpu-data.h: Delete.
3019 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
3021 * config/arm/arm-cpus.in (cortex-a55): New.
3022 (cortex-a75): Likewise.
3023 (cortex-a75.cortex-a55): Likewise.
3024 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
3026 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
3027 * config/arm/arm-cpu-cdata.h: Regenerate.
3028 * config/arm/arm-cpu-data.h: Regenerate.
3029 * config/arm/arm-cpu.h: Regenerate.
3030 * config/arm/arm-tables.opt: Regenerate.
3031 * config/arm/arm-tune.md: Regenerate.
3033 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
3035 * haifa-sched.c (sched_create_recovery_edges): Update profile.
3037 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
3039 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
3042 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
3044 PR tree-optimization/81292
3045 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
3046 full_string_p, also call adjust_related_strinfos if the adjustment
3047 is simple, otherwise invalidate related strinfos.
3049 2017-07-04 Martin Liska <mliska@suse.cz>
3052 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
3053 newly created variable as DECL_IGNORED_P.
3055 2017-07-04 Martin Liska <mliska@suse.cz>
3058 * ipa-inline.c (inline_small_functions):
3059 Use xstrdup_for_dump.
3061 2017-07-04 Tom de Vries <tom@codesourcery.com>
3063 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
3065 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
3068 * config/darwin.c (darwin_function_switched_text_sections):
3071 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
3073 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
3075 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
3077 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
3079 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3081 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
3082 min_profitable_iters, and th as inclusive lower bounds.
3083 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
3084 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
3085 for min_profitable_iters and min_profitable_estimate.
3086 (vect_transform_loop): Treat th as an inclusive lower bound.
3087 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
3089 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
3092 * config/darwin.c (darwin_function_switched_text_sections):
3093 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
3094 in two pieces, and suppress the use of buf.
3096 2017-07-03 Nathan Sidwell <nathan@acm.org>
3098 * hash-table.h (hash_table_mod1): Fix indentation.
3100 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3103 * predict.c (force_edge_cold): Be more careful about propagation
3105 * profile-count.h (profile_probability::guessed,
3106 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
3108 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
3110 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
3112 * doc/invoke.texi (rcpc architecture extension): Document it.
3114 2017-07-03 Richard Biener <rguenther@suse.de>
3116 PR tree-optimization/60510
3117 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
3118 the scalar reduction PHI and use it.
3119 (vectorizable_reduction): Properly guard the single_defuse_cycle
3120 path for non-SLP reduction chains where we cannot use it.
3121 Rework reduc_def/index and vector type deduction. Rework
3122 vector operand gathering during reduction op code-gen.
3123 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
3124 chains dissolve the chain and leave it to non-SLP reduction
3127 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3129 * tree-data-ref.h (dr_alignment): Declare.
3130 * tree-data-ref.c (dr_alignment): New function.
3131 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
3132 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
3134 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
3136 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3138 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
3139 and base_misalignment fields.
3140 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
3141 * tree-data-ref.c: Include builtins.h.
3142 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
3143 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
3144 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
3145 * tree-vect-data-refs.c: Include tree-cfg.h.
3146 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
3147 fields instead of calculating an alignment here.
3148 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
3149 innermost_loop_behavior fields.
3151 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3153 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
3155 (DR_STEP_ALIGNMENT): New macro.
3156 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
3157 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
3158 (create_data_ref): Print it.
3159 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
3160 to tell whether the step preserves vector (mis)alignment.
3161 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
3162 Move the check for an integer step and generalise to all INTEGER_CST.
3163 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
3164 Print the outer step alignment.
3166 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3168 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
3169 with offset_alignment.
3170 (DR_ALIGNED_TO): Delete.
3171 (DR_OFFSET_ALIGNMENT): New macro.
3172 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
3173 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
3174 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
3175 (create_data_ref): Likewise.
3176 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
3177 (vect_analyze_data_refs): Likewise.
3178 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
3179 creating dummy innermost behavior.
3181 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3183 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
3184 with a "innermost_loop_behavior *" and refeence tree.
3185 * tree-data-ref.c (dr_analyze_innermost): Likewise.
3186 (create_data_ref): Update call accordingly.
3187 * tree-predcom.c (find_looparound_phi): Likewise.
3189 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3191 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
3192 fields with dr_wrt_vec_loop.
3193 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
3194 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
3195 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
3196 (vect_dr_behavior): New function.
3197 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
3198 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
3199 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
3200 track whether the step preserves the misalignment.
3201 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
3202 Use vect_dr_behavior.
3203 (vect_setup_realignment): Update call accordingly.
3204 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
3205 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
3206 call to vect_create_addr_base_for_vector_ref.
3207 (vect_create_cond_for_align_checks): Likewise.
3208 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
3209 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
3210 (vect_recog_mask_conversion_pattern): Likewise.
3211 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
3212 (new_stmt_vec_info): Remove redundant zeroing.
3214 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
3216 * common/config/arm/arm-common.c (arm_be8_option): New function.
3217 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
3218 (ISA_ARMv6): Add isa_bit_be8.
3219 * config/arm/arm.h (arm_be8_option): Add prototype.
3220 (BE8_SPEC_FUNCTION): New define.
3221 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
3222 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
3223 (mlittle-endian): Similarly.
3224 (mbe8, mbe32): New options.
3225 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
3226 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
3228 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3230 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
3232 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3234 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
3235 (cleanup_tree_cfg_bb): Use it.
3236 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
3238 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
3240 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3243 * loop-doloop.c (add_test): Update profile.
3245 2017-07-03 Martin Liska <mliska@suse.cz>
3248 * sanopt.c (rewrite_usage_of_param): New function.
3249 (sanitize_rewrite_addressable_params): Likewise.
3250 (pass_sanopt::execute): Call rewrite_usage_of_param.
3252 2017-07-03 Richard Biener <rguenther@suse.de>
3254 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
3255 back to using VIEW_CONVERT_EXPR.
3257 2017-07-03 Martin Liska <mliska@suse.cz>
3260 * doc/extend.texi: Document when a resolver function is
3261 generated for target_clones.
3263 2017-07-03 Martin Liska <mliska@suse.cz>
3265 * asan.c (asan_emit_stack_protection): Unpoison just red zones
3266 and shadow memory of auto variables which are subject of
3267 use-after-scope sanitization.
3268 (asan_expand_mark_ifn): Add do set only when is_poison.
3270 2016-07-03 Richard Biener <rguenther@suse.de>
3272 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
3274 (vect_force_simple_reduction): Record reduction def -> phi mapping.
3275 (vectorizable_reduction): Perform reduction PHI creation when
3276 visiting a reduction PHI and adjust and simplify code generation
3277 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
3278 (vect_transform_loop): Visit reduction PHIs.
3279 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
3280 defs into the SLP tree.
3281 (vect_build_slp_tree): Reduction defs terminate the recursion.
3282 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
3284 (vect_get_vec_defs_for_stmt_copy): Export.
3285 (vect_get_vec_defs): Likewise.
3286 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
3288 (vect_get_vec_defs_for_stmt_copy): Declare.
3289 (vect_get_vec_defs): Likewise.
3291 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3293 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
3294 parameter with a "loop" parameter and use it instead of the
3295 loop containing DR_STMT. Don't check simple_iv when doing
3296 BB analysis. Describe the two analysis modes in the comment.
3298 2017-07-03 Tom de Vries <tom@codesourcery.com>
3300 PR tree-optimization/69468
3301 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
3302 (find_same_succ_bb): Handle ignore_edge_flags.
3304 2017-07-03 Tom de Vries <tom@codesourcery.com>
3306 PR tree-optimization/81192
3307 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
3309 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
3311 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
3313 2017-07-03 Tom de Vries <tom@codesourcery.com>
3315 PR tree-optimization/81192
3316 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
3317 BB_SAME_SUCC (bb) == NULL.
3319 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3321 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
3324 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3326 * dumpfile.c: Include profile-count.h
3327 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
3329 (insert_cond_bb): Update profile.
3330 * tree-cfg.h (insert_cond_bb): Update prototype.
3331 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
3332 * tree-dump.c: Do not include tree-cfg.
3334 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3336 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
3338 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3340 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
3343 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3345 * tree-complex.c (expand_complex_div_wide): update profile.
3347 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3348 Alan Hayward <alan.hayward@arm.com>
3349 David Sherwood <david.sherwood@arm.com>
3351 * Makefile.in (MACHMODE_H): Remove insn-modes.h
3352 (CORETYPES_H): New define.
3353 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
3354 (insn-modes-inline.h, s-modes-inline-h): New rules.
3355 (generated_files): Add insn-modes-inline.h.
3356 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
3357 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
3358 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
3359 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
3360 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
3361 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
3362 (build/gencodes.o, build/genconditions.o): Likewise.
3363 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
3364 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
3365 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
3366 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
3367 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
3368 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
3369 * coretypes.h: Include everything up to real.h for generators.
3370 Include insn-modes.h first. Include wide-int-print.h after
3371 wide-int.h. Include insn-modes-inline.h and then machmode.h.
3372 * machmode.h: Don't include insn-modes.h here.
3373 * function-tests.c: Remove includes of signop.h, machmode.h,
3374 double-int.h and wide-int.h.
3376 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
3378 * optc-save-gen.awk: Likewise.
3379 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
3380 * godump.c: Remove include of wide-int-print.h.
3381 * pretty-print.h: Likewise.
3382 * wide-int-print.cc: Likewise.
3383 * wide-int.cc: Likewise.
3384 * hash-map-tests.c: Remove include of signop.h.
3385 * hash-set-tests.c: Likewise.
3386 * rtl-tests.c: Likewise.
3387 * mkconfig.sh: Remove include of machmode.h.
3388 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
3390 (emit_insn_modes_inline_h): ...this new function. Emit the code
3391 into an insn-modes-inline.h header file, adding appropriate
3392 include guards and end comments.
3393 (emit_insn_modes_c_header): Remove include of machmode.h.
3394 (emit_min_insn_modes_c_header): Include coretypes.h rather than
3396 (main): Handle -i flag and call emit_insn_modes_inline_h when
3399 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3401 * tree-ssa-strlen.c (strinfo): Rename the length field to
3402 nonzero_chars. Add a full_string_p field.
3403 (compare_nonzero_chars, zero_length_string_p): New functions.
3404 (get_addr_stridx): Add an offset_out parameter.
3405 Use compare_nonzero_chars.
3406 (get_stridx): Update accordingly. Use compare_nonzero_chars.
3407 (new_strinfo): Update after above changes to strinfo.
3408 (set_endptr_and_length): Set full_string_p.
3409 (get_string_length): Update after above changes to strinfo.
3410 (unshare_strinfo): Update call to new_strinfo.
3411 (maybe_invalidate): Likewise.
3412 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
3413 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
3414 as a uhwi instead of an shwi. Update after above changes to
3415 strinfo and new_strinfo.
3416 (zero_length_string): Assert that chainsi contains full strings.
3417 Use zero_length_string_p. Update call to new_strinfo.
3418 (adjust_related_strinfos): Update after above changes to strinfo.
3419 Copy full_string_p from origsi.
3420 (adjust_last_stmt): Use zero_length_string_p.
3421 (handle_builtin_strlen): Update after above changes to strinfo and
3422 new_strinfo. Install the lhs as the string length if the previous
3423 entry didn't describe a full string.
3424 (handle_builtin_strchr): Update after above changes to strinfo
3426 (handle_builtin_strcpy): Likewise.
3427 (handle_builtin_strcat): Likewise.
3428 (handle_builtin_malloc): Likewise.
3429 (handle_pointer_plus): Likewise.
3430 (handle_builtin_memcpy): Likewise. Track nonzero characters
3431 that aren't necessarily followed by a nul terminator.
3432 (handle_char_store): Likewise.
3434 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3436 PR tree-optimization/80769
3437 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
3438 for malloc and calloc. Document the new invariant that all related
3439 strinfos have delayed lengths or none do.
3440 (verify_related_strinfos): Move earlier in file.
3441 (set_endptr_and_length): New function, split out from...
3442 (get_string_length): ...here. Also set the lengths of related
3444 (zero_length_string): Assert that chainsi has known (rather than
3446 (adjust_related_strinfos): Likewise.
3448 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3450 PR tree-optimization/81136
3451 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
3452 assert that two references with the same misalignment have the same
3453 compile-time misalignment if those compile-time misalignments
3456 2017-07-01 Andi Kleen <ak@linux.intel.com>
3458 * print-tree.c (print_node): Print all attributes.
3460 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3462 * cfg.c (scale_bbs_frequencies): New function.
3463 * cfg.h (scale_bbs_frequencies): Declare it.
3464 * cfgloopanal.c (single_likely_exit): Cleanup.
3465 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
3467 (scale_loop_profile): Likewise.
3468 (loop_version): Likewise.
3469 (create_empty_loop_on_edge): Update.
3470 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
3471 scale_loop_frequencies, scale_loop_profile, loopify,
3472 loop_version): Update prototypes.
3473 * modulo-sched.c (sms_schedule): Update.
3474 * predict.c (unlikely_executed_edge_p): Also check probability.
3475 (probably_never_executed_edge_p): Fix typo.
3476 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3477 * tree-parloops.c (gen_parallel_loop): Update.
3478 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
3479 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3480 * tree-ssa-loop-split.c (split_loop): Update.
3481 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3482 * tree-vect-loop-manip.c (vect_do_peeling): Update.
3483 (vect_loop_versioning): Update.
3484 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3486 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3488 * trans-mem.c (split_bb_make_tm_edge): Update profile.
3490 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3492 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
3493 to keep profile consistent.
3495 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3497 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
3498 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
3499 * profile-count.h (max_safe_multiplier): Make unsigned.
3500 (profile_count::guessed_zero): New.
3502 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3504 * bb-reorder.c (fix_up_crossing_landing_pad,
3505 fix_crossing_conditional_branches): Use make_single_succ_edge
3506 to keep profile consistent.
3508 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3510 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
3513 2017-07-01 Jakub Jelinek <jakub@redhat.com>
3516 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
3517 the right scopes, make sure cond_jump isn't preserved between multiple
3518 iterations. Search for fallthru edge whenever there are 3+ edges and
3519 use find_fallthru_edge for it.
3521 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3523 Patch by Alexander Monakov <amonakov@ispras.ru>
3524 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
3525 probabilities consistently.
3527 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3529 * pa.c (pa_expand_compare_and_swap_loop): Update call of
3530 emit_cmp_and_jump_insns.
3532 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3535 * tree-inline.c (expand_call_inline): Combine profile statuses.
3537 2017-06-30 Andrew Pinski <apinski@cavium.com>
3539 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
3540 fold_stmt returned true.
3542 2017-06-30 Nathan Sidwell <nathan@acm.org>
3544 * ggc.h (empty_string): Delete.
3545 * cfgexpand.c (expand_asm_stmt): Use plain "".
3546 * optabs.c (expand_asm_memory_barrier): Likewise.
3547 * stringpool.c (empty_string): Delete.
3548 (digit_vector, digit_string): Delete.
3549 (ggc_alloc_string): Use plain "", don't optimize single digit
3550 strings. Use ggc_alloc_atomic.
3552 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
3554 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
3555 comparison set and one other set, use the cost of the non-comparison
3558 2017-06-30 Nathan Sidwell <nathan@acm.org>
3560 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
3563 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
3565 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
3566 loops. Remove now unneeded calls to gimple_switch_set_label() that
3567 just set removed labels to NULL_TREE.
3569 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
3571 * tree-ssanames.c (set_range_info_raw): Abstract from ...
3572 (set_range_info): ...here. Only call set_range_info_raw if domain
3574 (set_nonzero_bits): Call set_range_info_raw.
3575 * tree-ssanames.h (set_range_info_raw): New.
3577 2017-06-30 Jakub Jelinek <jakub@redhat.com>
3580 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
3581 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
3582 of nonimmediate_operand and <store_mask_constraint> instead of m
3583 for the input operand. For V8FI iterator, always split if input
3584 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
3585 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
3586 <store_mask_predicate> instead of register_operand and
3587 <store_mask_constraint> instead of v for the input operand. Make
3588 sure both operands aren't MEMs for if not <mask_applied>.
3590 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
3592 * lto-wrapper.c (copy_file) Close both file descriptors before
3595 2017-06-30 Martin Liska <mliska@suse.cz>
3598 * multiple_target.c (create_dispatcher_calls): Make ifunc
3599 also for function that don't have calls or are not referenced.
3601 2017-06-30 Richard Biener <rguenther@suse.de>
3603 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
3604 analyze the first scalar stmt. Move vector type computation
3605 for the BB case here from ...
3606 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
3607 live operation processing in the SLP case properly.
3609 2017-06-30 Richard Biener <rguenther@suse.de>
3611 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
3613 2017-06-30 Martin Liska <mliska@suse.cz>
3616 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
3617 before BUILT_IN_UNWIND_RESUME when ASAN is used.
3619 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
3621 * doc/invoke.texi (AArch64): Add missing options and remove redundant
3624 2017-06-30 Richard Biener <rguenther@suse.de>
3626 PR tree-optimization/81249
3627 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
3628 condition reduction result to original scalar type.
3630 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3632 * profile-count.h (enum profile_quality): Fix typos and whitespace
3635 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3637 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
3638 type for branch probabilities.
3640 2017-06-29 Julian Brown <julian@codesourcery.com>
3641 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3643 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
3644 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
3645 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
3646 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
3648 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3650 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
3651 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
3652 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
3653 CC usage from generic code to here.
3654 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
3655 CC usage into the target macros.
3657 2017-06-29 Maya Rashish <coypu@sdf.org>
3659 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
3662 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3664 * arm/arm-builtins.c: Include profile-count.h
3665 * except.c (sjlj_emit_function_enter): Use
3666 profile_probability::unlikely.
3668 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3670 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
3671 and tocrel_offset be pointer args rather than implicitly using
3673 (legitimate_constant_pool_address_p, rs6000_emit_move,
3674 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
3675 tocrel_offset and use in toc_relative_expr_p call.
3676 (print_operand, print_operand_address): Use static tocrel_base_oac
3677 and tocrel_offset_oac.
3678 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
3681 2017-06-29 Maya Rashish <coypu@sdf.org>
3683 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
3685 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
3687 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
3688 objects, take into account only the alignment of 'op0' and 'mode1' if
3691 2017-06-29 Steve Ellcey <sellcey@cavium.com>
3693 * ccmp.c (ccmp_tree_comparison_p): New function.
3694 (ccmp_candidate_p): Update to use above function.
3695 (get_compare_parts): New function.
3696 (expand_ccmp_next): Update to use new functions.
3697 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
3699 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
3700 take mode as argument.
3701 * ccmp.h (expand_ccmp_expr): Add mode as argument.
3702 * expr.c (expand_expr_real_1): Pass mode as argument.
3704 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
3706 * combine.c (combine_instructions): Print insns to dump_file, together
3709 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3711 * asan.c (asan_emit_stack_protection): Update.
3712 (create_cond_insert_point): Update.
3713 * auto-profile.c (afdo_propagate_circuit): Update.
3714 * basic-block.h (struct edge_def): Turn probability to
3715 profile_probability.
3716 (EDGE_FREQUENCY): Update.
3717 * bb-reorder.c (find_traces_1_round): Update.
3718 (better_edge_p): Update.
3719 (sanitize_hot_paths): Update.
3720 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
3721 (make_single_succ_edge): Update.
3722 (check_bb_profile): Update.
3723 (dump_edge_info): Update.
3724 (update_bb_profile_for_threading): Update.
3725 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
3726 probabilitycount to 0.
3727 * cfgbuild.c (compute_outgoing_frequencies): Update.
3728 * cfgcleanup.c (try_forward_edges): Update.
3729 (outgoing_edges_match): Update.
3730 (try_crossjump_to_edge): Update.
3731 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
3732 (expand_gimple_tailcall): Update.
3733 (construct_init_block): Use make_single_succ_edge.
3734 (construct_exit_block): Use make_single_succ_edge.
3735 * cfghooks.c (verify_flow_info): Update.
3736 (redirect_edge_succ_nodup): Update.
3737 (split_edge): Update.
3738 (account_profile_record): Update.
3739 * cfgloopanal.c (single_likely_exit): Update.
3740 * cfgloopmanip.c (scale_loop_profile): Update.
3741 (set_zero_probability): Remove.
3742 (duplicate_loop_to_header_edge): Update.
3743 * cfgloopmanip.h (loop_version): Update prototype.
3744 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3745 (force_nonfallthru_and_redirect): Update.
3746 (update_br_prob_note): Update.
3747 (rtl_verify_edges): Update.
3748 (purge_dead_edges): Update.
3749 (rtl_lv_add_condition_to_bb): Update.
3750 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
3751 * cgraphunit.c (init_lowered_empty_function): Update.
3752 (cgraph_node::expand_thunk): Update.
3753 * cilk-common.c: Include profile-count.h
3754 * dojump.c (inv): Remove.
3755 (jumpifnot): Update.
3756 (jumpifnot_1): Update.
3757 (do_jump_1): Update.
3759 (do_jump_by_parts_greater_rtx): Update.
3760 (do_compare_rtx_and_jump): Update.
3761 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
3762 do_jump_1. do_compare_rtx_and_jump): Update prototype.
3763 * dwarf2cfi.c: Include profile-count.h
3764 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
3765 (sjlj_emit_dispatch_table): Likewise.
3766 * explow.c: Include profile-count.h
3767 * expmed.c (emit_store_flag_force): Update.
3768 (do_cmp_and_jump): Update.
3769 * expr.c (compare_by_pieces_d::generate): Update.
3770 (compare_by_pieces_d::finish_mode): Update.
3771 (emit_block_move_via_loop): Update.
3772 (store_expr_with_bounds): Update.
3773 (store_constructor): Update.
3774 (expand_expr_real_2): Update.
3775 (expand_expr_real_1): Update.
3776 * expr.h (try_casesi, try_tablejump): Update prototypes.
3777 * gimple-pretty-print.c (dump_probability): Update.
3778 (dump_profile): New.
3779 (dump_gimple_label): Update.
3780 (dump_gimple_bb_header): Update.
3781 * graph.c (draw_cfg_node_succ_edges): Update.
3782 * hsa-gen.c (convert_switch_statements): Update.
3783 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
3784 (find_if_case_1): Update.
3785 (find_if_case_2): Update.
3786 * internal-fn.c (expand_arith_overflow_result_store): Update.
3787 (expand_addsub_overflow): Update.
3788 (expand_neg_overflow): Update.
3789 (expand_mul_overflow): Update.
3790 (expand_vector_ubsan_overflow): Update.
3791 * ipa-cp.c (good_cloning_opportunity_p): Update.
3792 * ipa-split.c (split_function): Use make_single_succ_edge.
3793 * ipa-utils.c (ipa_merge_profiles): Update.
3794 * loop-doloop.c (add_test): Update.
3795 (doloop_modify): Update.
3796 * loop-unroll.c (compare_and_jump_seq): Update.
3797 (unroll_loop_runtime_iterations): Update.
3798 * lra-constraints.c (lra_inheritance): Update.
3799 * lto-streamer-in.c (input_cfg): Update.
3800 * lto-streamer-out.c (output_cfg): Update.
3801 * mcf.c (adjust_cfg_counts): Update.
3802 * modulo-sched.c (sms_schedule): Update.
3803 * omp-expand.c (expand_omp_for_init_counts): Update.
3804 (extract_omp_for_update_vars): Update.
3805 (expand_omp_ordered_sink): Update.
3806 (expand_omp_for_ordered_loops): Update.
3807 (expand_omp_for_generic): Update.
3808 (expand_omp_for_static_nochunk): Update.
3809 (expand_omp_for_static_chunk): Update.
3810 (expand_cilk_for): Update.
3811 (expand_omp_simd): Update.
3812 (expand_omp_taskloop_for_outer): Update.
3813 (expand_omp_taskloop_for_inner): Update.
3814 * omp-simd-clone.c (simd_clone_adjust): Update.
3815 * optabs.c (expand_doubleword_shift): Update.
3816 (expand_abs): Update.
3817 (emit_cmp_and_jump_insn_1): Update.
3818 (expand_compare_and_swap_loop): Update.
3819 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
3820 * predict.c (predictable_edge_p): Update.
3821 (edge_probability_reliable_p): Update.
3822 (set_even_probabilities): Update.
3823 (combine_predictions_for_insn): Update.
3824 (combine_predictions_for_bb): Update.
3825 (propagate_freq): Update.
3826 (estimate_bb_frequencies): Update.
3827 (force_edge_cold): Update.
3828 * profile-count.c (profile_count::dump): Add missing space into dump.
3829 (profile_count::debug): Add newline.
3830 (profile_count::differs_from_p): Explicitly convert to unsigned.
3831 (profile_count::stream_in): Update.
3832 (profile_probability::dump): New member function.
3833 (profile_probability::debug): New member function.
3834 (profile_probability::differs_from_p): New member function.
3835 (profile_probability::differs_lot_from_p): New member function.
3836 (profile_probability::stream_in): New member function.
3837 (profile_probability::stream_out): New member function.
3838 * profile-count.h (profile_count_quality): Rename to ...
3839 (profile_quality): ... this one.
3840 (profile_probability): New.
3841 (profile_count): Update.
3842 * profile.c (compute_branch_probabilities): Update.
3843 * recog.c (peep2_attempt): Update.
3844 * sched-ebb.c (schedule_ebbs): Update.
3845 * sched-rgn.c (find_single_block_region): Update.
3846 (compute_dom_prob_ps): Update.
3847 (schedule_region): Update.
3848 * sel-sched-ir.c (compute_succs_info): Update.
3849 * stmt.c (struct case_node): Update.
3850 (do_jump_if_equal): Update.
3851 (get_outgoing_edge_probs): Update.
3852 (conditional_probability): Update.
3853 (emit_case_dispatch_table): Update.
3854 (expand_case): Update.
3855 (expand_sjlj_dispatch_table): Update.
3856 (emit_case_nodes): Update.
3857 * targhooks.c: Update.
3858 * tracer.c (better_p): Update.
3859 (find_best_successor): Update.
3860 * trans-mem.c (expand_transaction): Update.
3861 * tree-call-cdce.c: Update.
3862 * tree-cfg.c (gimple_split_edge): Upate.
3863 (move_sese_region_to_fn): Upate.
3864 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
3865 * tree-eh.c (lower_resx): Upate.
3866 (cleanup_empty_eh_move_lp): Upate.
3867 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3868 * tree-inline.c (copy_edges_for_bb): Update.
3869 (copy_cfg_body): Update.
3870 * tree-parloops.c (gen_parallel_loop): Update.
3871 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
3872 (gimple_gen_time_profiler): Update.
3873 * tree-ssa-dce.c (remove_dead_stmt): Update.
3874 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3875 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3876 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3877 (unloop_loops): Update.
3878 (try_peel_loop): Update.
3879 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3880 * tree-ssa-loop-split.c (connect_loops): Update.
3881 (split_loop): Update.
3882 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3883 (hoist_guard): Update.
3884 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3885 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3886 (value_replacement): Update.
3887 * tree-ssa-reassoc.c (branch_fixup): Update.
3888 * tree-ssa-tail-merge.c (replace_block_by): Update.
3889 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3890 (create_edge_and_update_destination_phis): Update.
3891 (compute_path_counts): Update.
3892 (recompute_probabilities): Update.
3893 (update_joiner_offpath_counts): Update.
3894 (freqs_to_counts_path): Update.
3895 (duplicate_thread_path): Update.
3896 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3897 (struct switch_conv_info): Update.
3898 (gen_inbound_check): Update.
3899 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3900 (vect_do_peeling): Update.
3901 (vect_loop_versioning): Update.
3902 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3903 (optimize_mask_stores): Update.
3904 * ubsan.c (ubsan_expand_null_ifn): Update.
3905 * value-prof.c (gimple_divmod_fixed_value): Update.
3906 (gimple_divmod_fixed_value_transform): Update.
3907 (gimple_mod_pow2): Update.
3908 (gimple_mod_pow2_value_transform): Update.
3909 (gimple_mod_subtract): Update.
3910 (gimple_mod_subtract_transform): Update.
3911 (gimple_ic): Update.
3912 (gimple_stringop_fixed_value): Update.
3913 (gimple_stringops_transform): Update.
3914 * value-prof.h: Update.
3916 2017-06-29 Carl Love <cel@us.ibm.com>
3918 * config/rs6000/rs6000-c.c: Add support for built-in functions
3919 vector signed int vec_signed (vector float);
3920 vector signed long long vec_signed (vector double);
3921 vector signed int vec_signed2 (vector double, vector double);
3922 vector signed int vec_signede (vector double);
3923 vector signed int vec_signedo (vector double);
3924 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
3925 instruction generator.
3926 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3927 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
3928 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
3930 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
3931 vunsignede_v2df): Add define_expands.
3932 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
3933 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
3934 VEC_UNSIGNEDO): Add definitions.
3935 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3936 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
3937 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
3938 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
3939 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
3940 * config/rs6000/altivec.h (vec_signed, vec_signed2,
3941 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
3942 vec_unsignede, vec_unsignedo): Add builtin defines.
3943 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
3945 * doc/extend.texi: Update the built-in documentation file for the
3946 new built-in functions.
3948 2017-06-29 Richard Biener <rguenther@suse.de>
3950 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
3951 reduction chains to LOOP_VINFO_REDUCTIONS.
3952 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
3953 SLP reductions after processing reduction chains.
3955 2017-06-29 Nathan Sidwell <nathan@acm.org>
3957 * builtins.c (fold_builtin_FUNCTION): Use
3958 lang_hooks.decl_printable_name.
3960 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
3963 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
3964 with only one label.
3965 * stmt.c (expand_case): Assert NCASES is greater than one.
3967 2017-06-29 Richard Biener <rguenther@suse.de>
3969 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
3971 (group_case_labels): Likewise.
3972 (find_taken_edge): Push sanity checking on val to workers...
3973 (find_taken_edge_cond_expr): ... here
3974 (find_taken_edge_switch_expr): ... and here, handle cases
3975 with just a default label.
3976 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
3977 (group_case_labels): Likewise.
3978 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
3979 group_case_labels does anything cleanup the CFG again.
3981 2017-06-29 Bin Cheng <bin.cheng@arm.com>
3983 PR tree-optimization/81196
3984 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
3985 exit condition comparing two IVs.
3987 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
3989 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
3990 profile to the dummy entry at the end of the list of architectures.
3991 * config/arm/arm-cpu-cdata.h: Regenerated.
3993 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3994 Michael Collison <michael.collison@arm.com>
3997 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
3999 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
4000 (*aarch64_reg_<mode>3_minus_mask): New pattern.
4001 (*aarch64_<optab>_reg_di3_mask2): New pattern.
4002 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
4003 of shift when the shift amount is masked with constant equal to
4004 the size of the mode.
4005 * config/aarch64/predicates.md (subreg_lowpart_operator): New
4008 2017-06-29 Martin Liska <mliska@suse.cz>
4010 * config/i386/i386.opt: Change range from [1,5] to [0,5].
4012 2017-06-29 Yury Gribov <tetra2005@gmail.com>
4015 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
4018 (edge_growth_cache_entry::edge_growth_cache_entry): New
4020 (reset_edge_growth_cache): Update to use constructor.
4022 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4024 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
4025 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
4026 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
4028 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
4030 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
4031 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
4033 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
4035 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
4036 (*-linux-uclibc*): Add t-uclibc tmake_file.
4037 * config/t-musl: New.
4038 * config/t-uclibc: New.
4040 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
4042 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
4044 (gen_comm_data): Emit architectural setting of arch_prof.
4045 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
4047 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
4048 (armv8-m.base, armv8-m.main): Likewise.
4049 * arm-protos.h (arm_build_target): Add profile field.
4050 (arch_option): Likewise.
4051 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
4053 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
4054 arm_active_target.profile.
4056 2017-06-28 Richard Biener <rguenther@suse.de>
4059 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
4060 TYPE_OVERFLOW_WRAPS.
4061 * match.pd (negate_expr_p): Likewise.
4062 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
4063 fold_build2, not fold_binary.
4065 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
4067 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
4068 Convert memory address to Pmode.
4069 (aarch64_print_operand): Assert MEM operands are always Pmode.
4071 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
4074 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
4075 Remove redundant if.
4076 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
4077 * config/arm/aarch-common-protos.h
4078 (aarch_forward_to_shift_is_not_shifted_re): Remove.
4079 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
4081 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
4084 * multiple_target.c (create_dispatcher_calls): Set the default
4085 clone to be static, not public.
4087 2017-06-28 Richard Biener <rguenther@suse.de>
4089 * tree-vect-loop.c (vectorizable_reduction): Move special
4090 cond reduction IV var creation ...
4091 (vect_create_epilog_for_reduction): ... here. Remove induction_index
4092 parameter. Use STMT_VINFO_VECTYPE.
4093 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
4096 2017-06-28 Martin Liska <mliska@suse.cz>
4099 * ipa-visibility.c (non_local_p): Handle visibility.
4101 2017-06-28 Martin Liska <mliska@suse.cz>
4104 * common.opt: Add IntegerRange to various options.
4105 * opt-functions.awk (integer_range_info): New function.
4106 * optc-gen.awk: Add integer_range_info to cl_options struct.
4107 * opts-common.c (decode_cmdline_option): Handle
4108 CL_ERR_INT_RANGE_ARG.
4109 (cmdline_handle_error): Likewise.
4110 * opts.c (print_filtered_help): Show valid interval in
4111 when --help is provided.
4112 * opts.h (struct cl_option): Add range_min and range_max fields.
4113 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
4115 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
4117 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
4118 (x * C EQ/NE y * C): New transformation.
4120 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
4122 * genmultilib (combination_space): Accept '+' in option names.
4124 2017-06-28 Martin Liska <mliska@suse.cz>
4127 * asan.c (instrument_derefs): Bail out inner references
4128 that are hard register variables.
4130 2017-06-28 Jakub Jelinek <jakub@redhat.com>
4133 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
4134 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
4136 2017-06-28 Richard Biener <rguenther@suse.de>
4138 * tree-vectorizer.h (vect_get_vec_defs): Remove.
4139 (vect_get_slp_defs): Adjust.
4140 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
4142 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
4144 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4145 get_initial_defs_for_reduction instead of vect_get_vec_defs.
4146 (vectorizable_reduction): Adjust.
4147 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
4149 (vect_get_slp_defs): Likewise.
4150 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
4151 (vectorizable_bswap): Adjust.
4152 (vectorizable_call): Likewise.
4153 (vectorizable_conversion): Likewise.
4154 (vectorizable_assignment): Likewise.
4155 (vectorizable_shift): Likewise.
4156 (vectorizable_operation): Likewise.
4157 (vectorizable_store): Likewise.
4158 (vectorizable_condition): Likewise.
4159 (vectorizable_comparison): Likewise.
4161 2017-06-28 Michael Collison <michael.collison@arm.com>
4164 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
4166 (arm_gen_movmemqi): Removed unused variable 'i'.
4167 Convert 'for' loop into 'while' loop.
4168 (arm_expand_prologue): Remove last unnecessary set of insn.
4169 (thumb_pop): Remove unused variable 'pushed_words'.
4170 (thumb_exit): Remove last unnecessary set of regs_to_pop.
4172 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4174 * config/s390/predicates.md: Use s390_rel_address_ok_p.
4175 * config/s390/s390-protos.h: Add prototype of
4176 s390_rel_address_ok_p.
4177 * config/s390/s390.c (s390_got_symbol): New function.
4178 (s390_rel_address_ok_p): New function.
4179 (legitimize_pic_address): Use s390_rel_address_ok_p.
4180 (s390_load_got): Use s390_got_symbol.
4181 (s390_option_override): Issue error if
4182 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
4183 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
4185 * config/s390/s390.opt: New option mpic-data-is-text-relative.
4187 2017-06-27 Andrew Pinski <apinski@cavium.com>
4189 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
4190 (X * copysign (1.0, X)): New pattern.
4191 (X * copysign (1.0, -X)): New pattern.
4192 (copysign (-1.0, CST)): New pattern.
4194 2017-06-27 Joseph Myers <joseph@codesourcery.com>
4196 * genmultilib (combination_space): Remove variable.
4197 Validate reuse rules against regular expression for any sequence
4198 of multilib options in any order.
4200 2017-06-27 Michael Collison <michael.collison@arm.com>
4202 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
4203 call aarch64_split_simd_combine.
4204 * (aarch64_combine_internal<mode>): Delete pattern.
4205 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
4206 Allow register and subreg operands.
4208 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4210 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
4211 specific need, just fallback on defaults.
4212 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
4214 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4215 Olivier Hainque <hainque@adacore.com>
4217 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
4219 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
4220 targets. Pick a default if no particular attempt applied.
4221 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
4224 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4226 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
4227 (x86_64-wrs-vxworks7): Likewise.
4229 2017-06-27 Marek Polacek <polacek@redhat.com>
4232 * ubsan.c (instrument_null): Check get_base_address's result for null.
4234 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
4236 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
4238 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
4240 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
4241 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
4242 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
4243 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
4245 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
4246 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
4247 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
4248 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
4249 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
4250 BUILT_IN_FEUPDATEENV): New builtins.
4251 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
4252 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
4253 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
4254 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
4256 (builtin_structptr_types): Adjust size.
4257 * tree.c (builtin_structptr_types): Add four entries.
4259 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4260 Olivier Hainque <hainque@adacore.com>
4262 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
4263 (TLS_SYM): New local macro, forcing reference to __tls__ on
4264 link command lines for VxWorks 7 RTPs, triggering initialization
4266 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
4267 OS features TLS support, true for RTPs on VxWorks 7.
4268 * config/vxworks.c (vxworks_override_options): Setup emutls
4271 2017-06-27 Jakub Jelinek <jakub@redhat.com>
4273 * predict.c (test_prediction_value_range): Use -1U instead of -1
4274 to avoid narrowing conversion warning.
4275 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
4276 to avoid narrowing conversion warning.
4277 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
4279 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
4281 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4283 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
4284 64bit configurations.
4285 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
4286 (SIZE_TYPE): Likewise.
4287 * config/vxworks.c (vxworks_emutls_var_fields): Use
4288 long_unsigned_type_node instead of unsigned_type_node as the offset
4289 field type, which is "pointer" mode in emutls.c.
4291 2017-06-27 Jakub Jelinek <jakub@redhat.com>
4294 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
4297 * gimple-fold.c (replace_call_with_call_and_fold): Handle
4298 gimple_vuse copying separately from gimple_vdef copying.
4300 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4302 * value-prof.c (free_hist): Remove call to memset and the enclosing if
4305 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
4306 Olivier Hainque <hainque@adacore.com>
4308 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
4309 for all vxworks7 targets.
4310 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
4311 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
4312 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
4313 variations for VX6/VX7 and 32/64bits later on in ...
4314 (VXWORKS_LIB_SPEC): Leverage new macros.
4315 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
4316 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
4318 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
4320 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
4321 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
4323 2017-06-26 Carl Love <cel@us.ibm.com>
4325 * config/rs6000/rs6000-c.c: Add support for built-in functions
4326 vector bool char vec_reve (vector bool char);
4327 vector signed char vec_reve (vector signed char);
4328 vector unsigned char vec_reve (vector unsigned char);
4329 vector bool int vec_reve (vector bool int);
4330 vector signed int vec_reve (vector signed int);
4331 vector unsigned int vec_reve (vector unsigned int);
4332 vector bool long long vec_reve (vector bool long long);
4333 vector signed long long vec_reve (vector signed long long);
4334 vector unsigned long long vec_reve (vector unsigned long long);
4335 vector bool short vec_reve (vector bool short);
4336 vector signed short vec_reve (vector signed short);
4337 vector double vec_reve (vector double);
4338 vector float vec_reve (vector float);
4339 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
4340 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
4341 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
4342 (altivec_vreve): New pattern.
4343 * config/rs6000/altivec.h (vec_reve): New define.
4344 * doc/extend.texi (vec_rev): Update the built-in documentation file
4345 for the new built-in functions.
4347 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4349 PR tree-optimization/71815
4350 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
4352 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
4354 (slsr_process_phi): Likewise.
4355 (replace_uncond_cands_and_profitable_phis): Don't replace a
4356 multiply candidate with a stride of 1 (copy or cast).
4357 (phi_incr_cost): Call uses_consumed_by_stmt rather than
4359 (lowest_cost_path): Likewise.
4360 (total_savings): Likewise.
4362 2017-06-26 Richard Biener <rguenther@suse.de>
4365 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
4366 Use def_builtin_pure for all gather builtins.
4368 2017-06-26 Richard Biener <rguenther@suse.de>
4370 PR tree-optimization/81203
4371 * tree-tailcall.c (find_tail_calls): Do not move stmts into
4374 2017-06-26 Marek Polacek <polacek@redhat.com>
4377 * doc/invoke.texi: Document -Wmultistatement-macros.
4379 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
4381 * doc/sourcebuild.texi (ARM-specific attributes): Document new
4382 arm_neon_ok_no_float_abi effective target.
4384 2017-06-26 Richard Biener <rguenther@suse.de>
4386 PR tree-optimization/80928
4387 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
4388 (copy_bbs): Set BB_DUPLICATED flag early.
4389 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
4391 (execute_on_shrinking_pred): Likewise.
4392 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
4393 BB_DUPLICATED blocks.
4394 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
4395 iterate over all PHIs considering removal of *gsi.
4397 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
4399 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
4402 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4404 * config/rs6000/rs6000-string.c: (expand_block_clear,
4405 do_load_for_compare, select_block_compare_mode,
4406 compute_current_alignment, expand_block_compare,
4407 expand_strncmp_align_check, expand_strn_compare,
4408 expand_block_move, rs6000_output_load_multiple)
4409 Move functions related to string/block move/compare
4411 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
4412 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
4413 for this function which is now used in two files.
4414 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
4415 * config.gcc: Add rs6000-string.o to extra_objs for
4416 targets powerpc*-*-* and rs6000*-*-*.
4418 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4421 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
4422 32-bit, since indexed is not valid for DImode.
4423 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
4424 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
4425 (define_peephole2 for Altivec d-form load): Add 32-bit support.
4426 (define_peephole2 for Altivec d-form store): Likewise.
4429 * multiple_target.c (create_dispatcher_calls): Only create the
4430 dispatcher call if the function is the default clone of a
4433 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
4436 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
4437 a call, force the call to not be a tail call.
4439 2017-06-23 Jeff Law <law@redhat.com>
4441 * doc/contrib.texi: Add entry for Steven Pemberton's work on
4444 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
4446 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
4447 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
4448 handling for early expansion of vector shifts (sl,sr,sra,rl).
4449 (builtin_function_type): Add vector shift right instructions
4450 to the unsigned argument list.
4452 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
4454 rtl-optimizatoin/79286
4455 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
4456 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
4457 trap. PIC register plus a const unspec without offset can never trap.
4459 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
4461 * tree.h (builtin_structptr_type): New type.
4462 (builtin_structptr_types): Declare new array.
4463 * tree.c (builtin_structptr_types): New array.
4464 (free_lang_data, build_common_tree_nodes): Use it.
4466 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
4469 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
4472 2017-06-22 Matt Turner <mattst88@gmail.com>
4474 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
4475 Lake models to skylake case. Assume skylake for unknown
4476 models with clflushopt.
4478 2017-06-22 Jeff Law <law@redhat.com>
4480 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
4481 frame sizes that do not satisfy aarch64_uimm12_shift.
4483 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
4485 * profile-count.h (apply_probability,
4486 apply_scale, probability_in): Fix checks for zero.
4488 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4490 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
4491 * doc/cppdiropts.texi (-I @var{dir}): Document it.
4493 2016-06-22 Richard Biener <rguenther@suse.de>
4495 * tree-vect-loop.c (vect_model_reduction_cost): Handle
4496 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
4497 REDUC_MAX_EXPR support.
4498 (vectorizable_reduction): Likewise.
4499 (vect_create_epilog_for_reduction): Likewise.
4501 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
4503 * match.pd (A / (1 << B) -> A >> B): New.
4504 * generic-match-head.c: Include optabs-tree.h.
4505 * gimple-match-head.c: Likewise.
4506 * optabs-tree.h (target_supports_op_p): New.
4507 * optabs-tree.c (target_supports_op_p): New.
4509 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4511 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
4512 $gcc_cv_ld --help output.
4513 (gcc_cv_ld_demangle): Likewise.
4514 (gcc_cv_ld_eh_frame_hdr): Likewise.
4515 (gcc_cv_ld_pie): Likewise.
4516 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
4517 (gcc_cv_ld_buildid): Likewise.
4518 (gcc_cv_ld_sysroot): Likewise.
4519 (ld_bndplt_support): Likewise.
4520 (ld_pushpopstate_support): Likewise.
4521 * configure: Regenerate.
4522 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
4524 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4527 * config/i386/sse.md (round<mode>2): Renumber match_dup and
4528 operands indexes to avoid gap between operands and match_dups.
4530 2017-06-21 Andrew Pinski <apinski@cavium.com>
4532 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
4533 Increment Arith_shift and Arith_shift_reg by 1.
4534 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
4536 * config/aarch64/aarch64.c (thunderx_tunings): Enable
4537 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
4538 (aarch64_strip_extend): Add new argument and test for it.
4539 (aarch64_cheap_mult_shift_p): New function.
4540 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
4541 add a cost if it is true.
4542 Update calls to aarch64_strip_extend.
4543 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
4545 2017-06-21 Andrew Pinski <apinski@cavium.com>
4547 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
4549 (thunderxt88): Likewise.
4550 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
4551 (thunderx_prefetch_tune): New variable.
4552 (thunderx2t99_prefetch_tune): Update for the correct values.
4553 (thunderxt88_tunings): New variable.
4554 (thunderx_tunings): Use thunderx_prefetch_tune instead of
4555 generic_prefetch_tune.
4556 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
4558 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4560 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
4561 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
4562 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
4563 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
4564 (aarch64_atomic_cas<mode>, GPI): Likewise.
4566 2017-06-21 Martin Liska <mliska@suse.cz>
4568 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
4569 statements on cold and hot labels.
4570 * predict.c (tree_estimate_probability_bb): Remove the
4571 prediction from this place.
4573 2017-06-21 Martin Liska <mliska@suse.cz>
4575 PR tree-optimization/79489
4576 * gimplify.c (maybe_add_early_return_predict_stmt): New
4578 (gimplify_return_expr): Call the function.
4579 * predict.c (tree_estimate_probability_bb): Remove handling
4581 * predict.def: Update comment about early return predictor.
4582 * gimple-predict.h (is_gimple_predict): New function.
4583 * predict.def: Change default value of early return to 66.
4584 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
4586 * passes.def: Put pass_strip_predict_hints to the beginning of
4589 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
4591 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
4592 FUNCTION_DECL declarations.
4593 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
4595 (dwaf2out_decl): Likewise.
4596 * godump.c (go_early_global_decl): Skip call to the real debug hook
4597 for FUNCTION_DECL declarations.
4598 * passes.c (rest_of_decl_compilation): Skip call to the
4599 early_global_decl debug hook for FUNCTION_DECL declarations, unless
4600 -fdump-go-spec is passed.
4602 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4604 * config/i386/i386.c (struct builtin_isa): New field pure_p.
4605 Reorder for compactness.
4606 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
4607 (def_builtin_pure, def_builtin_pure2): New functions.
4608 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
4610 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4612 * match.pd (nop_convert): New predicate.
4613 ((A +- CST1) +- CST2): Allow some NOP conversions.
4615 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4618 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
4619 with ctors/dtors if GOVD_SHARED is set.
4621 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4623 * config/aarch64/aarch64.md (movti_aarch64):
4624 Emit mov rather than orr.
4625 (movtf_aarch64): Likewise.
4626 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
4627 Emit mov rather than orr.
4629 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4631 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
4632 Swap alternatives, make integer dup more expensive.
4634 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4636 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
4637 Return true for non-tls symbols.
4639 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
4641 * config/aarch64/aarch64-cores.def (cortex-a55): New.
4642 (cortex-a75): Likewise.
4643 (cortex-a75.cortex-a55): Likewise.
4644 * config/aarch64/aarch64-tune.md: Regenerate.
4645 * doc/invoke.texi (-mtune): Document new values for -mtune.
4647 2017-06-21 Tom de Vries <tom@codesourcery.com>
4649 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
4651 (Effective-Target Keywords, Other attributes): Suggest using
4652 dg-add-options stack_size feature to get stack limit in stack_size
4653 effective target documentation.
4655 2017-06-21 Julian Brown <julian@codesourcery.com>
4656 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4658 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
4659 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
4660 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
4662 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
4663 attribute type list for neon_multiply.
4664 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
4665 attribute type list for neon_multiply.
4666 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
4667 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
4668 attribute type list for neon_multiply.
4669 * config/arm/types.md (crypto_pmull): Add.
4670 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
4671 attribute type list.
4673 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
4675 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
4678 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4680 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
4681 to make sure not to dereference a NULL cost_classes_ptr pointer.
4683 2017-06-20 Carl Love <cel@us.ibm.com>
4685 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4686 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4687 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4688 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4689 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4690 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4691 VMULOSW): New enum "unspec" values.
4692 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
4693 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
4694 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4695 altivec_vmulosw): New patterns.
4696 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4697 VMULOSW): Add definitions.
4699 2017-06-20 Julia Koval <julia.koval@intel.com>
4701 * config/i386/i386.c: Fix rounding expand for new pattern.
4702 * config/i386/subst.md: Fix pattern (parallel -> unspec).
4704 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4706 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
4707 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
4709 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4711 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
4714 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4716 * config/aarch64/aarch64-cores.def: Rearrange to sort by
4717 architecture, then by implementer ID.
4718 * config/aarch64/aarch64-tune.md: Regenerate.
4720 2017-06-20 Richard Biener <rguenther@suse.de>
4723 * fold-const.c (split_tree): Fold to type before negating.
4725 2017-06-20 David Malcolm <dmalcolm@redhat.com>
4727 * diagnostic-show-locus.c
4728 (selftest::test_fixit_deletion_affecting_newline): New function.
4729 (selftest::diagnostic_show_locus_c_tests): Call it.
4731 2017-06-20 Andreas Schwab <schwab@suse.de>
4734 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
4737 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
4739 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
4740 __ARM_FEATURE_COPROC according to support.
4742 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4744 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
4745 Rewritten to avoid overflow for > 32-bit pointers.
4748 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
4749 by removing enum keyword.
4750 (ubsan_type_descriptor): Likewise. Formatting fix.
4753 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
4754 splitter): Require TARGET_SSE2 in the condition.
4756 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
4759 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
4760 for doing vector set of SFmode on ISA 3.0.
4761 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
4762 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
4764 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
4765 SFmode value into a V4SF variable that was extracted from another
4766 V4SF variable without converting the element to double precision
4767 and back to single precision vector format.
4768 (vsx_insert_extract_v4sf_p9_2): Likewise.
4770 2017-06-19 Jakub Jelinek <jakub@redhat.com>
4772 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
4773 in UWHI to avoid undefined overflow.
4776 * ubsan.h (enum ubsan_encode_value_phase): New.
4777 (ubsan_encode_value): Change second argument to
4778 enum ubsan_encode_value_phase with default value of
4779 UBSAN_ENCODE_VALUE_GENERIC.
4780 * ubsan.c (ubsan_encode_value): Change second argument to
4781 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
4782 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
4783 create_tmp_var_raw instead of create_tmp_var and use a
4785 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
4786 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
4787 ubsan_encode_value callers.
4790 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
4791 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
4792 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
4794 2017-06-19 Richard Biener <rguenther@suse.de>
4797 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
4798 estimates if we changed anything.
4800 2017-06-19 Richard Biener <rguenther@suse.de>
4802 PR tree-optimization/80887
4803 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
4804 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
4805 simplified lookups, then reset mprts_hook.
4806 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
4808 (try_to_simplify): Likewise.
4810 2017-06-19 Martin Liska <mliska@suse.cz>
4813 * gimplify.c (gimplify_switch_expr):
4814 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
4816 2017-06-19 Martin Liska <mliska@suse.cz>
4818 * doc/install.texi: Document that PGO runs in 4 stages.
4820 2017-06-19 Martin Liska <mliska@suse.cz>
4823 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
4824 to dispatcher function name.
4825 * multiple_target.c (replace_function_decl): New function.
4826 (create_dispatcher_calls): Redirect both edges and references.
4828 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
4830 * profile-count.c (profile_count::dump): Dump quality.
4831 (profile_count::differs_from_p): Update for unsigned val.
4832 * profile-count.h (profile_count_quality): New enum.
4833 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
4835 2017-06-19 Richard Biener <rguenther@suse.de>
4837 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
4838 struct function as arg.
4839 (estimate_numbers_of_iterations): Export overload with loop arg.
4840 (free_numbers_of_iterations_estimates_loop): Use an overload of
4841 free_numbers_of_iterations_estimates instead.
4842 * tree-cfg.c (remove_bb): Adjust.
4843 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
4844 * tree-parloops.c (gen_parallel_loop): Likewise.
4845 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
4847 (tree_unroll_loops_completely): Likewise.
4848 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
4849 Use an overload instead and export.
4850 (estimated_loop_iterations): Adjust.
4851 (max_loop_iterations): Likewise.
4852 (likely_max_loop_iterations): Likewise.
4853 (estimate_numbers_of_iterations): Take struct function as arg
4855 (loop_exits_before_overflow): Adjust.
4856 (free_numbers_of_iterations_estimates_loop): Use an overload.
4857 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
4858 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
4860 2017-06-19 Richard Biener <rguenther@suse.de>
4863 * ipa-prop.c (find_constructor_constant_at_offset): Handle
4864 RANGE_EXPR conservatively.
4866 2017-06-16 Carl Love <cel@us.ibm.com>
4868 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4869 definitions for vec_float, vec_float2, vec_floato,
4870 vec_floate built-ins.
4871 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
4872 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
4874 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
4875 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
4876 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
4877 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
4878 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
4879 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
4880 vec_floato): Add builtin defines.
4881 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
4882 Update the built-in documentation file for the new built-in
4885 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4887 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
4888 (mthumb): Mark as the negative of -marm.
4890 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4892 * doc/invoke.texi (ARM Options, -mcpu): Document supported
4894 (ARM Options, -mtune): Document that this accepts the same
4895 extension options as -mcpu.
4896 (ARM Options, -mfpu): Document addition of -mfpu=auto.
4898 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4900 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
4901 permitted extensions.
4903 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4905 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
4906 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
4907 (armv8-m.main): Add option +nodsp.
4908 * config/arm/arm-cpu-cdata.h: Regenerated.
4910 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4912 * config/arm/t-fuchsia: New file.
4913 * config.gcc (arm*-*-fuchsia*): Use it.
4915 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4917 * config/arm/t-symbian: Rewrite for new option infrastructure.
4919 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4921 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
4922 (MULTILIB_REQUIRED): Likewise.
4924 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4926 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
4927 (MULTILIB_RESUE): Likewise.
4928 (MULTILIB_MATCHES): Likewise.
4929 (MULTLIB_REQUIRED): Likewise.
4931 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4933 * config/arm/t-rtems: Rewrite for new option framework.
4935 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4937 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
4938 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
4939 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
4940 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
4941 * config/arm/t-multilib: ... here.
4942 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
4943 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
4944 armv7-a and armv8*-a when A-profile libraries have not been built.
4945 * config/arm/t-rmprofile: Rewrite.
4947 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4949 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
4950 with a backslash. Remove the backslash after substituting unescaped
4952 * doc/fragments.texi (MULTILIB_REUSE): Document it.
4954 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4956 * config.gcc: (arm*-*-*): When building a-profile libraries, force
4957 the driver to pass through the default setting of -mfloat-abi.
4958 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
4960 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
4961 (all_feat_combs): New rule.
4962 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
4964 * config/arm/t-aprofile: Rewrite.
4966 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4968 * config/arm/arm.h (FPUTYPE_AUTO): Define.
4969 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
4970 fpu is not specified by the user/command-line.
4971 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
4972 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
4973 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
4974 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
4975 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
4976 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
4978 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4980 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
4981 * config/arm/t-arm-elf: Rewritten.
4983 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4985 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
4986 have some floating-point instructions.
4987 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
4988 (TARGET_MAYBE_HARD_FLOAT): New macro.
4989 * config/arm/arm-builtins.c (arm_init_builtins): Use
4990 TARGET_MAYBE_HARD_FLOAT.
4991 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
4993 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4995 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
4996 (configargs.h): Include it.
4997 (arm_print_hint_for_fpu_option): New function.
4998 (arm_parse_fpu_option): New function.
4999 (candidate_extension): New class.
5000 (arm_canon_for_multilib): New function.
5001 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
5002 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
5003 (ARCH_CANONICAL_SPECS): New macro.
5004 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
5006 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5008 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
5009 are set after handling multilib fragments. Set target_cpu_default2
5012 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5014 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
5016 (arm*-*-*): Set target_cpu_default2 to a quoted string.
5017 * config/arm/parsecpu.awk (check_cpu): Validate any extension
5019 (check_arch): Likewise.
5020 * config/arm/arm.c (arm_configure_build_target): Handle
5021 TARGET_CPU_DEFAULT being a string constant. Scan any feature
5022 options in the default.
5024 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5026 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
5027 when an option is an alias of another.
5028 * config/arm/parsecpu.awk (optalias): New parser token.
5029 (gen_comm_data): Mark non-alias options as such. Emit entries
5030 for extension aliases.
5031 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
5032 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
5033 (armv6kz, armv6zk, armv6t2): Likewise.
5034 (armv7): Make vfpv3-d16 an alias.
5035 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
5037 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
5038 Sort in canonical order.
5039 (armv8-a): Sort in canonical order.
5040 (armv8.1-a, armv8.2-a): Likewise.
5041 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
5043 (cortex-a9): Sort in canonical order.
5044 * config/arm/arm.c (selftests.h): Include it.
5045 (arm_test_cpu_arch_data): New function.
5046 (arm_run_self_tests): New function.
5047 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
5048 (targetm): Move declaration to the end of the file.
5049 * arm-cpu-cdata.h: Regenerated.
5051 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5053 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
5054 call to target_mode_check describing the type of option passed.
5055 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
5056 (arm_target_thumb_only): Use arm_parse_arch_option_name or
5057 arm_parse_cpu_option_name to match parameters against list of
5059 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
5060 arm_arch_core_flags data structure.
5061 * config/arm/arm-cpu_cdata.h: Regenerated.
5063 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5065 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
5067 (arm_print_hint_for_cpu_option): Likewise.
5068 (arm_print_hint_for_arch_option): Likewise.
5069 (arm_parse_cpu_option_name): Likewise.
5070 (arm_parse_arch_option_name): Likewise.
5071 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
5072 of entries in the all_fpus list.
5073 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
5074 (arm_parse_cpu_option_name): Declare.
5075 (arm_parse_arch_option_name): Declare.
5076 (arm_parse_option_features): Declare.
5077 (arm_intialize_isa): Declare.
5078 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
5080 (gen_comm_data): ... here. Make definitions non-static.
5081 * config/arm/arm-cpu-data.h: Regenerated.
5082 * config/arm/arm-cpu-cdata.h: Regenerated.
5084 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5086 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
5087 (cpu_arch_extension): New structure.
5088 (cpu_arch_option, arch_option, cpu_option): New structures.
5089 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
5091 (gen_data): Generate new format data tables.
5092 * config/arm/arm.c (cpu_tune): New structure.
5093 (cpu_option, processors): Delete.
5094 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
5095 (arm_print_hint_for_cpu_option): ... this and ...
5096 (arm_print_hint_for_arch_option): ... this.
5097 (arm_parse_arch_cpu_name): Delete. Replace with ...
5098 (arm_parse_cpu_option_name): ... this and ...
5099 (arm_parse_arch_option_name): ... this.
5100 (arm_unrecognized_feature): Change type of target parameter to
5102 (arm_parse_arch_cpu_features): Delete. Replace with ...
5103 (arm_parse_option_features): ... this.
5104 (arm_configure_build_target): Rework to use new configuration data
5106 (arm_print_tune_info): Rework for new configuration data tables.
5107 * config/arm/arm-cpu-data.h: Regenerated.
5108 * config/arm/arm-cpu.h: Regenerated.
5110 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5112 * Makefile.in (OBJS): Move sbitmap.o from here ...
5113 (OBJS-libcommon): ... to here.
5115 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5117 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
5118 (ISA_ALL_CRYPTO): New macro.
5119 (ISA_ALL_SIMD): New macro
5120 (ISA_ALL_FP): New macro.
5121 * config/arm/arm.c (fpu_bitlist): Update initializer.
5122 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
5124 (arm9e): Add fpu. Add option for nofp
5125 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
5126 (arm926ej-s, arm1026ej-s): Likewise.
5127 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
5128 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
5129 neon-fp16, neon-vfpv4, nofp and nosimd.
5130 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
5131 (cortex-a8): Add fpu. Add option for nofp.
5132 (cortex-a9): Add fpu. Add options for nosimd and nofp.
5133 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
5134 (cortex-r4f): Add fpu.
5135 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
5136 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
5138 (cortex-r8): Likewise.
5139 (cortex-m4): Add fpu. Add option for nofp.
5140 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
5141 (cortex-a17.cortex-a7): Likewise.
5142 (cortex-a32): Add fpu. Add options for crypto and nofp.
5143 (cortex-a35, cortex-a53): Likewise.
5144 (cortex-a57): Add fpu. Add option for crypto.
5145 (cortex-a72, cortex-a73): Likewise.
5146 (exynos-m1): Likewise.
5147 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
5148 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
5149 (cortex-m33): Add fpu. Add option for nofp.
5150 * config/arm/arm-cpu-cdata.h: Regenerated
5151 * config/arm/arm-cpu-data.h: Regenerated.
5153 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5155 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
5156 (armv5te, armv5tej): Likewise.
5157 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
5158 (armv7): Add options fp and vfpv3-d16.
5159 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
5160 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
5162 (armv7ve): Likewise.
5163 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
5164 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
5165 (armv8-a): Add nocrypto option.
5166 (armv8.1-a, armv8.2-a): Likewise.
5167 (armv8-m.main): add options fp, fp.dp and nofp.
5169 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5171 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
5173 (armv8-a+crc): Delete.
5174 (armv8.1-a): Add options simd, crypto and nofp.
5175 (armv8.2-a): Add options fp16, simd, crypto and nofp.
5176 (armv8.2-a+fp16): Delete.
5177 (armv8-m.main): Add option dsp.
5178 (armv8-m.main+dsp): Delete.
5179 (cortex-a8): Add fpu. Add nofp option.
5180 (cortex-a9): Add fpu. Add nofp and nosimd options.
5181 * config/arm/parsecpu.awk (gen_data): Generate option tables and
5182 link to main cpu and architecture data structures.
5183 (gen_comm_data): Only put isa attributes from the main architecture
5185 (option): New statement for architecture and CPU entries.
5186 * arm.c (struct cpu_option): New structure.
5187 (struct processors): Add entry for options.
5188 (arm_unrecognized_feature): New function.
5189 (arm_parse_arch_cpu_name): Ignore any characters after the first
5191 (arm_parse_arch_cpu_feature): New function.
5192 (arm_configure_build_target): Separate out any CPU and architecture
5193 features and parse separately. Don't error out if -mfpu=auto is
5194 used with only an architecture string.
5195 (arm_print_asm_arch_directives): New function.
5196 (arm_file_start): Call it.
5197 * config/arm/arm-cpu-cdata.h: Regenerated.
5198 * config/arm/arm-cpu-data.h: Likewise.
5199 * config/arm/arm-tables.opt: Likewise.
5201 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5203 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
5204 assembler when it is not -mfpu=auto.
5206 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5208 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
5209 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
5210 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
5211 (ASM_CPU_SPEC): Rewrite.
5212 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
5213 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
5214 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
5215 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
5216 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
5217 copied string is NUL-terminated. Also strip any characters prefixed
5219 (arm_rewrite_selected_arch): New function.
5220 (arm_rewrite_march): New function.
5222 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5224 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
5225 (x_arm_cpu_string, x_arm_tune_string): Likewise.
5226 (march, mcpu, mtune): Convert to string-based options.
5227 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
5228 (arm_parse_arch_cpu_name): New function.
5229 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
5230 identify selected architecture or CPU.
5231 (arm_option_save): New function.
5232 (TARGET_OPTION_SAVE): Redefine.
5233 (arm_option_restore): Restore string options.
5234 (arm_option_print): Print string options.
5236 2017-06-16 Martin Sebor <msebor@redhat.com>
5238 PR tree-optimization/80933
5239 PR tree-optimization/80934
5240 * builtins.c (fold_builtin_3): Do not handle bcmp here.
5241 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
5242 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
5243 (gimple_fold_builtin): Call them.
5245 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5247 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
5248 as unlikely; update profile.
5250 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5252 * predict.c (force_edge_cold): Handle declaring edges impossible
5255 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5257 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
5259 (try_unroll_loop_completely): Fix reporting.
5261 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5263 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
5265 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
5268 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
5269 if given a non-constant argument for an intrinsic which requires a
5272 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5274 * profile.c (compare_freqs): New function.
5275 (branch_prob): Sort edge list.
5276 (find_spanning_tree): Assume that the list is priority sorted.
5278 2017-06-16 Richard Biener <rguenther@suse.de>
5280 PR tree-optimization/81090
5281 * passes.def (pass_record_bounds): Remove.
5282 * tree-pass.h (make_pass_record_bounds): Likewise.
5283 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
5284 make_pass_record_bounds): Likewise.
5285 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
5286 not free niter estimates at the beginning but at the end.
5287 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
5289 2017-06-16 Richard Biener <rguenther@suse.de>
5291 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
5292 initializer to workaround ICE in host GCC 4.8.
5294 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5296 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
5298 (clone_inlined_nodes): Update.
5300 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5302 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
5303 prefetch settings, and enable prefetching by default at -O3.
5305 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5307 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5308 Set flag_prefetch_loop_arrays according to tuning data.
5310 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5312 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
5314 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
5315 [Unrelated to main purpose of the patch] Place the pointer field last
5316 to enable type checking errors when tune structure are wrongly merged.
5317 * config/aarch64/aarch64.c (generic_prefetch_tune,)
5318 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
5319 (thunderx2t99_prefetch_tune): New tune constants.
5320 (tune_params *_tunings): Update all tunings (no functional change).
5321 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
5322 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
5323 from tunings structures.
5325 2017-06-16 Jakub Jelinek <jakub@redhat.com>
5328 * ubsan.c (instrument_null): Add T argument, use it instead
5329 of computing it based on IS_LHS.
5330 (instrument_object_size): Likewise.
5331 (pass_ubsan::execute): Adjust instrument_null and
5332 instrument_object_size callers to pass gimple_get_lhs or
5333 gimple_assign_rhs1 result to it. Use instrument_null instead of
5334 calling get_base_address and instrument_mem_ref. Handle
5335 aggregate call arguments for object-size sanitization.
5337 2017-06-16 Yury Gribov <tetra2005@gmail.com>
5339 PR tree-optimization/81089
5340 * tree-vrp.c (is_masked_range_test): Validate operands of
5343 2017-06-15 Martin Sebor <msebor@redhat.com>
5346 * dumpfile.c (dump_register): Avoid calling memset to initialize
5347 a class with a default ctor.
5348 * gcc.c (struct compiler): Remove const qualification.
5349 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
5350 * hash-table.h: Ditto.
5351 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
5353 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
5354 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
5356 * params.h (struct param_info): Make struct members non-const.
5357 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
5358 with default initialization.
5359 * vec.h (vec_copy_construct, vec_default_construct): New helper
5361 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
5362 with vec_copy_construct.
5363 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
5364 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
5365 * doc/invoke.texi (-Wclass-memaccess): Document.
5367 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5369 * emit-rtl.h (is_leaf): Update comment about local
5372 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5375 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
5376 for a variable to have a section before checking if the section has a
5378 Set section to.persistent if persistent attribute is set.
5379 Warn if .persistent attribute is used on an automatic variable.
5381 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
5383 PR rtl-optimization/80474
5384 * reorg.c (update_block): Do not ignore instructions in a delay slot.
5386 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
5388 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
5391 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
5393 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
5394 (casesi): Emit bounds checking as RTL.
5395 (casesi_internal_mips16_<mode>): Remove bounds checking.
5397 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
5399 * config/xtensa/xtensa.c (xtensa_option_override): Append
5400 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
5401 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
5402 xtensa_doloop_hooks): Define unconditionally.
5403 (xtensa_reorg_loops): Only call reorg_loops in the presence of
5405 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
5406 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
5407 for it in xtensa_option_override.
5408 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
5409 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
5411 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
5413 * doc/cppopts.texi: Document '-' special value to -MF.
5415 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
5417 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
5418 (cortex_a53_fconst): Likewise.
5419 (cortex_a53_fpmul): Likewise.
5420 (cortex_a53_f_load_64): Likewise.
5421 (cortex_a53_f_load_many): Likewise.
5422 (cortex_a53_advsimd_alu): Likewise.
5423 (cortex_a53_advsimd_alu_q): Likewise.
5424 (cortex_a53_advsimd_mul): Likewise.
5425 (cortex_a53_advsimd_mul_q): Likewise.
5426 (fpmac bypass): Add new bypass for fpmac-fpmac case.
5427 Add missing fmul, r2f_cvt and fconst cases.
5429 2017-06-14 Richard Biener <rguenther@suse.de>
5432 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
5434 (fold_binary_loc): When associating do not treat pre-existing
5435 TREE_OVERFLOW on literal constants as a reason to allow
5436 TREE_OVERFLOW on associated literal constants.
5438 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5440 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
5441 (MASK_FEATURES): New macro.
5442 * config/sparc/sparc.c (sparc_option_override): Remove the special
5443 handling of -mfpu and generalize it to all MASK_FEATURES switches.
5445 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5447 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
5448 a division of 0 if non-call exceptions are enabled.
5450 2017-06-14 Andrew Pinski <apinski@cavium.com>
5451 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5454 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
5455 Improve vector initialization code gen for only variable case.
5457 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5459 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
5461 2017-06-14 Richard Biener <rguenther@suse.de>
5463 PR tree-optimization/81083
5464 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
5467 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5469 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
5470 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
5471 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
5472 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
5473 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
5474 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
5476 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5478 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
5479 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
5481 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5483 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
5485 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5487 * config/rs6000/t-rtems: Don't handle SPE.
5489 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5491 * config/rs6000/t-linux: Don't handle SPE.
5493 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5495 * config/rs6000/eabispe.h: Delete file.
5497 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5499 * config/rs6000/t-spe: Delete file.
5501 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5503 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
5504 (rs6000_legitimate_offset_address_p): Return false for anything in
5505 V2SImode or V2SFmode.
5507 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5509 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
5510 except V2SF and V2SI. Rearrange the vector modes, and add comments.
5511 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
5513 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
5514 (rs6000_legitimate_offset_address_p): Ditto.
5515 (rs6000_emit_move): Ditto.
5516 (rs6000_init_builtins): Remove V4HI_type_node.
5518 2017-06-13 Martin Liska <mliska@suse.cz>
5521 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
5522 (gate_asan): Likewise.
5523 * asan.h (asan_no_sanitize_address_p): Remove the function.
5524 (sanitize_flags_p): New function.
5525 * builtins.def: Fix coding style.
5526 * common.opt: Use renamed enum value.
5527 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
5528 * doc/extend.texi: Document no_sanitize attribute.
5529 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
5530 to SANITIZE_UNDEFINED_NONDEFAULT.
5531 * gcc.c (sanitize_spec_function): Use the renamed enum value.
5532 * gimple-fold.c (optimize_atomic_compare_exchange_p):
5533 Use sanitize_flags_p.
5534 * gimplify.c (gimplify_function_tree): Likewise.
5535 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
5536 * opts.c (parse_no_sanitize_attribute): New function.
5537 (common_handle_option): Use renamed enum value.
5538 * opts.h (parse_no_sanitize_attribute): Declare.
5539 * tree.c (sanitize_flags_p): New function.
5540 * tree.h: Declared here.
5541 * tsan.c: Use sanitize_flags_p.
5542 * ubsan.c (ubsan_expand_null_ifn): Likewise.
5543 (instrument_mem_ref): Likewise.
5544 (instrument_bool_enum_load): Likewise.
5545 (do_ubsan_in_current_function): Remove the function.
5546 (pass_ubsan::execute): Use sanitize_flags_p.
5547 * ubsan.h: Remove do_ubsan_in_current_function
5548 * tree-cfg.c (print_no_sanitize_attr_value): New function.
5549 (dump_function_to_file): Use it here.
5551 2017-06-13 Martin Jambor <mjambor@suse.cz>
5553 PR tree-optimization/80803
5554 PR tree-optimization/81063
5555 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
5556 (propagate_subaccesses_across_link): Enqueue subtree whenever
5557 necessary instead of relying on the caller.
5559 2017-06-13 Martin Jambor <mjambor@suse.cz>
5561 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
5562 that have a first_link.
5563 (sort_and_splice_var_accesses): Do not check first_link before
5565 (subtree_mark_written_and_enqueue): Likewise.
5566 (propagate_all_subaccesses): Likewise and do not stop at first
5567 parent with a first_link.
5569 2017-06-13 Martin Jambor <mjambor@suse.cz>
5571 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
5574 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5576 * match.pd: New pattern.
5578 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5580 * tree-vrp.c (is_masked_range_test): New function.
5581 (register_edge_assert_for): Determine ranges for
5584 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5586 PR tree-optimization/67328
5587 * fold-const.c (maskable_range_p): New function.
5588 (build_range_check): Generate bittests if possible.
5590 2017-06-13 Martin Liska <mliska@suse.cz>
5592 * gimple-pretty-print.c (dump_probability): Add new argument.
5593 (dump_edge_probability): Dump both probability and count.
5594 (dump_gimple_label): Likewise.
5595 (dump_gimple_bb_header): Likewise.
5597 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
5600 * config/avr/avr-devices.c: Fix indentation.
5601 * config/avr/gen-avr-mmcu-specs.c: Dito.
5603 2017-06-13 Richard Biener <rguenther@suse.de>
5605 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
5606 instead get vector type from stmt_info.
5607 (vectorizable_reduction): Adjust. Remove dead code.
5609 2017-06-13 Richard Biener <rguenther@suse.de>
5612 * fold-const.c (extract_muldiv_1): Remove bogus distribution
5613 case of C * (x * C2 + C3).
5614 (fold_addr_of_array_ref_difference): Properly fold index difference.
5616 2017-06-12 David S. Miller <davem@davemloft.net>
5619 * config/sparc/sparc.md (return expander): Emit frame blockage if
5620 function uses alloca.
5622 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
5624 * combine.c (make_field_assignment): Check len rather than the mode
5625 precision when calling force_to_mode.
5627 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
5629 Support multilibs and devices that see flash in RAM address range.
5632 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
5633 (avr_mcu_t) <flash_pm_offset>: New field.
5634 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
5635 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
5636 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
5637 (AVR_TINY_PM_OFFSET): Remove macro.
5638 * config/avr/avr.opt (-mshort-calls): New option.
5639 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
5640 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
5641 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
5642 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
5643 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
5644 instead of avr_arch->have_jmp_call.
5645 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
5646 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
5647 avr_arch->flash_pm_offset to define.
5648 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
5649 new field flash_pm_offset. Add entry for avrxmega3.
5650 (avr_texinfo): Add entry for avrxmega3.
5651 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
5652 attiny212, attiny214,
5653 attiny412, attiny414, attiny416, attiny417,
5654 attiny814, attiny816, attiny817,
5655 attiny1614, attiny1616, attiny1617,
5656 attiny3214, attiny3216, attiny3217.
5657 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
5658 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
5659 (avr_print_operand_address) [AVR_TINY]: Same.
5660 (avr_asm_init_sections) <readonly_data_section>: Only patch
5661 callback if avr_arch->flash_pm_offset = 0.
5662 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
5663 for rodata if avr_arch->flash_pm_offset != 0.
5664 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
5665 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
5666 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
5667 (m_options): Append opt_rcall.
5668 (m_dirnames): Append dir_rcall.
5669 * config/avr/t-multilib: Regenerate.
5671 * configure.ac [target=avr]: Check whether avrxmega3 default
5672 linker description file works as needed.
5673 * configure: Regenerate.
5674 * doc/avr-mmcu.texi: Regenerate.
5675 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
5676 <__AVR_ARCH__>: Document avrxmega3 and 103.
5677 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
5678 <__AVR_SHORT_CALLS__>: Document it.
5679 <__AVR_PM_BASE_ADDRESS__>: Document it.
5680 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
5681 (AVR Variable Attributes) <progmem>: Document this is
5682 not needed for avrxmega3.
5683 (AVR Named Address Spaces) <__flash>: Dito.
5685 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
5687 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
5689 2017-06-12 Doug Rupp <rupp@adacore.com>
5691 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
5692 Append vxworks-stdint.h to the tm_file list.
5693 * config/vxworks-stdint.h: New file.
5695 2017-06-12 Martin Liska <mliska@suse.cz>
5697 PR tree-optimization/81041
5698 * tree-profile.c (gimple_gen_ic_func_profiler):
5699 Create an extra BB in profile-generate
5700 (gimple_gen_time_profiler): Likewise.
5702 2017-06-12 Jakub Jelinek <jakub@redhat.com>
5704 PR tree-optimization/81003
5705 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
5706 (update_range_test): Use it instead of force_gimple_operand_gsi.
5708 2017-06-12 Richard Biener <rguenther@suse.de>
5710 PR tree-optimization/81053
5711 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
5712 with backedge value not defined in loop. Simplify def stmt
5715 2017-06-11 Tom de Vries <tom@codesourcery.com>
5718 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
5720 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
5721 nvptx_cannot_force_const_mem.
5723 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5725 * opts.c (finish_options): Move test for flag_split_stack after
5726 it has been initialized.
5728 2017-06-11 Jason Merrill <jason@redhat.com>
5730 * tree.h (id_equal): New.
5731 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
5732 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
5733 instead of strcmp of IDENTIFIER_POINTER.
5735 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5737 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
5738 (mark_all_inlined_calls_cdtor): Fix formating.
5739 (inline_transform): Rescale profile before inlining.
5741 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5743 * cgraph.h (cgraph_edge::clone): Update prototype.
5744 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
5745 (cgraph_node::create_clone): Update.
5746 (cgraph_node::create_version_clone): Update.
5747 * tree-inline.c (copy_bb): Update.
5748 (expand_call_inline): Update.
5750 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5752 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
5753 factored out from ...
5754 (rs6000_emit_prologue): ... here.
5756 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5758 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
5759 factored out from ...
5760 (rs6000_emit_prologue): ... here.
5762 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5764 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
5766 (handle_missing_profiles): Fix computation of tp_first_run.
5767 (counts_to_freqs): Do not touch freqs when count is 0.
5769 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5771 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
5774 2017-06-10 Tom de Vries <tom@codesourcery.com>
5776 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
5777 attributes): Document signal effective target.
5779 2017-06-10 Tom de Vries <tom@codesourcery.com>
5781 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
5782 Document effective target stack_size.
5784 2017-06-09 David Malcolm <dmalcolm@redhat.com>
5786 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
5787 to the edit_context if they can be auto-applied.
5789 2017-06-9 Ian Lance Taylor <iant@golang.org>
5791 * opts.c (finish_options): If -fsplit-stack, disable implicit
5792 -forder-blocks-and-partition.
5793 * doc/invoke.texi (Optimize Options): Document that when using
5794 -fsplit-stack -forder-blocks-and-partition is not implicitly
5797 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5799 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
5800 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
5801 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
5802 * builtins.def (abort, trap, unreachable): Declare cold.
5803 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
5804 * tree-core.h (ECF_COLD): New.
5805 * tree.c (set_call_expr_flags): Handle ECF_COLD.
5806 (build_common_builtin_nodes): Mark unreachable and abort as cold.
5808 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5810 * predict.c (unlikely_executed_stmt_p): Cleanup.
5812 2017-06-09 Richard Biener <rguenther@suse.de>
5814 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
5815 model if the ref is always written to.
5817 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5819 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
5821 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5823 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
5826 2017-06-09 Tom de Vries <tom@codesourcery.com>
5829 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
5830 "target cannot support label values" when encountering LABEL_REF.
5832 2017-06-09 Martin Liska <mliska@suse.cz>
5834 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
5835 (gimple_gen_ic_func_profiler): Emit direct comparison
5836 of __gcov_indirect_call_callee with NULL.
5837 (gimple_gen_time_profiler): Change probability from
5838 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
5840 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5842 * profile.c (edge_gcov_counts): Turn to pointer.
5843 (compute_branch_probabilities, compute_branch_probabilities): Update.
5844 (branch_prob): Do not clear edge_gcov_count.
5845 * profile.h (edge_gcov_counts): Turn to pointer.
5846 (edge_gcov_count): Update.
5848 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5850 * gimple.h (gimple_check_failed): Mark cold.
5852 2017-06-09 Richard Biener <rguenther@suse.de>
5854 PR tree-optimization/66623
5855 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
5856 refactor check_reduction into two parts, properly computing
5857 whether we have to check reduction validity for outer loop
5860 2017-06-09 Richard Biener <rguenther@suse.de>
5862 PR tree-optimization/79483
5863 * graphite-scop-detection.c (order): New global.
5864 (get_order): Compute bb to order mapping that satisfies code
5865 generation constraints.
5866 (cmp_pbbs): New helper.
5867 (build_scops): Start domwalk at entry block, sort generated
5870 2017-06-09 Richard Biener <rguenther@suse.de>
5873 * ipa-polymorphic-call.c
5874 (ipa_polymorphic_call_context::restrict_to_inner_class):
5875 Skip FIELD_DECLs with error_mark_node type.
5876 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
5879 2017-06-09 Martin Liska <mliska@suse.cz>
5881 * predict.c (struct branch_predictor): New struct.
5882 (test_prediction_value_range): New test.
5883 (predict_c_tests): New function.
5884 * selftest-run-tests.c (selftest::run_tests): Run the function.
5885 * selftest.h: Declare new tests.
5887 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
5890 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
5891 gen_add3_insn did not fail.
5892 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
5893 r0, construct that number in a temporary reg and add that reg to r0.
5894 If asked to put the result in r0 as well, fail.
5896 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
5898 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5899 for early expansion of vec_eqv.
5901 2017-06-08 Jakub Jelinek <jakub@redhat.com>
5904 * ubsan.c (instrument_null): Avoid pointless code temporary.
5905 (pass_ubsan::execute): Instrument aggregate arguments of calls.
5907 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
5911 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
5914 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
5917 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5919 * predict.c (unlikely_executed_edge_p): Move ahead.
5920 (probably_never_executed_edge_p): Use it.
5922 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5925 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
5926 gimple_call_builtin_p call.
5928 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5930 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
5931 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
5932 rtl_check_failed_type2, rtl_check_failed_code1,
5933 rtl_check_failed_code2, rtl_check_failed_code_mode,
5934 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
5935 rtvec_check_failed_bounds, rtl_check_failed_flag,
5936 _fatal_insn_not_found, _fatal_insn): Likewise.
5937 * tree.h (tree_contains_struct_check_failed,
5938 tree_check_failed, tree_not_check_failed,
5939 tree_class_check_failed, tree_range_check_failed,
5940 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
5941 tree_vec_elt_check_failed, phi_node_elt_check_failed,
5942 tree_operand_check_failed, omp_clause_check_failed,
5943 omp_clause_operand_check_failed, omp_clause_range_check_failed):
5946 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5948 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
5949 flag_branch_probabilities.
5950 * ipa-inline.c (edge_badness): Likewise.
5951 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
5952 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
5953 * predict.c (maybe_hot_frequency_p): Likewise.
5954 (probably_never_executed): Likewise.
5955 * sched-ebb.c (schedule_ebbs): Likewise.
5956 * sched-rgn.c (find_single_block_region): Likewise.
5957 * tracer.c (tail_duplicate): Likewise.
5959 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5961 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
5962 longer requires x_flag_profile_use.
5964 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5966 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
5967 instead of flag_reorder_blocks_and_partition.
5968 * dbxout.c (dbxout_function_end): Likewise.
5969 * dwarf2out.c (gen_subprogram_die): Likewise.
5970 * haifa-sched.c (sched_create_recovery_edges): Likewise.
5971 * hw-doloop.c (reorg_loops): Likewise.
5972 * varasm.c (assemble_start_function,
5973 assemble_end_function): Likewise.
5974 (decide_function_section): Do not check for
5975 flag_reorder_blocks_and_partition.
5977 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5979 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
5981 (chkp_get_hard_register_fake_addr_expr): Ditto.
5982 (chkp_build_addr_expr): Add check for hard reg case.
5983 (chkp_parse_array_and_component_ref): Ditto.
5984 (chkp_find_bounds_1): Ditto.
5985 (chkp_process_stmt): Don't generate bounds store for
5988 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5990 * predict.c (maybe_hot_bb_p): Do not check profile status.
5991 (maybe_hot_edge_p): Likewise.
5992 (probably_never_executed): Check for zero counts even if profile
5994 (unlikely_executed_edge_p): New function.
5995 (unlikely_executed_stmt_p): New function.
5996 (unlikely_executed_bb_p): New function.
5997 (set_even_probabilities): Use unlikely predicates.
5998 (combine_predictions_for_bb): Likewise.
5999 (predict_paths_for_bb): Likewise.
6000 (predict_paths_leading_to_edge): Likewise.
6001 (determine_unlikely_bbs): New function.
6002 (estimate_bb_frequencies): Use it.
6003 (compute_function_frequency): Use zero counts even if profile is
6005 * profile-count.h: Fix typo.
6007 2017-08-08 Julia Koval <julia.koval@intel.com>
6009 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
6010 _mm512_mask_cvtsepi16_storeu_epi8,
6011 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
6012 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
6013 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
6014 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
6015 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
6016 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
6017 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
6018 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
6019 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
6020 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
6021 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
6022 __builtin_ia32_pmovuswb256mem_mask,
6023 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
6024 __builtin_ia32_pmovwb512mem_mask): New builtins.
6026 2017-08-08 Julia Koval <julia.koval@intel.com>
6028 PR target/73350,80862
6029 * config/i386/subst.md (round): Fix round pattern.
6030 * config/i386/i386.c (ix86_erase_embedded_rounding):
6031 Fix erasing rounding for the fixed pattern.
6033 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6035 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
6037 2017-06-08 Martin Liska <mliska@suse.cz>
6039 PR gcov-profile/80911
6040 * gcov.c (block_info::block_info): New constructor.
6042 2017-06-07 Carl Love <cel@us.ibm.com>
6044 * config/rs6000/rs6000-c: The return type of the following
6045 built-in functions was implemented as int not long long. Fix sign
6046 of return value for the unsigned version of vec_mulo and vec_mule.
6047 vector unsigned long long vec_bperm (vector unsigned long long,
6048 vector unsigned char)
6049 vector signed long long vec_mule (vector signed int,
6051 vector unsigned long long vec_mule (vector unsigned int,
6052 vector unsigned int)
6053 vector signed long long vec_mulo (vector signed int,
6055 vector unsigned long long vec_mulo (vector unsigned int,
6056 vector unsigned int)
6057 * doc/extend.texi: Fix the documentation for the built-in
6060 2017-06-07 Carl Love <cel@us.ibm.com>
6063 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
6066 2017-06-07 Carl Love <cel@us.ibm.com>
6068 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
6069 support, Generate doublehv for signed int/float for BE case only.
6071 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
6073 * doc/invoke.texi (mcx16): Rewrite.
6075 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6077 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
6078 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
6079 *mov<mode>_softfloat, and an anonymous splitter): Use
6080 nonimmediate_operand instead of rs6000_nonimmediate_operand.
6082 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6084 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
6086 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
6087 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
6088 (rs6000_debug_reg_global): Adjust.
6089 (rs6000_init_hard_regno_mode_ok): Adjust.
6090 (rs6000_dbx_register_number): Adjust.
6091 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
6092 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
6093 Remove SPE_ACC and SPEFSCR.
6094 (REG_ALLOC_ORDER): Ditto.
6095 (FRAME_POINTER_REGNUM): Change to 111.
6096 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
6097 (REG_CLASS_NAMES): Ditto.
6098 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
6099 (REGISTER_NAMES): Ditto.
6100 (ADDITIONAL_REG_NAMES): Ditto.
6101 (rs6000_reg_names): Ditto.
6102 * config/rs6000/rs6000.md: Renumber some register number
6105 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6107 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
6109 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
6110 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
6112 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
6113 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
6114 Delete the SPE high registers.
6115 (REG_ALLOC_ORDER): Ditto.
6116 (enum reg_class): Remove SPE_HIGH_REGS.
6117 (REG_CLASS_NAMES): Ditto.
6118 (REG_CLASS_CONTENTS): Delete the SPE high registers.
6119 (REGISTER_NAMES): Ditto.
6120 (rs6000_reg_names): Ditto.
6121 * doc/tm.texi.in: Remove SPE as example.
6122 * doc/tm.texi: Regenerate.
6124 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6126 * config/rs6000/8540.md (ppc8540_brinc): Delete.
6127 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
6128 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
6129 * config/rs6000/rs6000.md (type): Remove "brinc".
6131 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6133 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
6134 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
6135 * config/rs6000/linuxspe.h: Delete file.
6136 * config/rs6000/rs6000.md: Don't include spe.md.
6137 * config/rs6000/spe.h: Delete file.
6138 * config/rs6000/spe.md: Delete file.
6139 * config/rs6000/t-rs6000: Remove spe.md.
6141 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6143 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
6144 (reg_or_none500mem_operand): Delete.
6145 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
6146 instead of reg_or_none500mem_operand.
6148 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6150 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
6151 handling of SPE flags.
6152 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
6154 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6156 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
6158 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
6159 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
6160 paired_divv2sf3): Similar.
6161 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
6162 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
6163 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
6165 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
6166 Rename the paired_* instruction patterns.
6167 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
6169 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
6170 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
6171 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
6172 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
6173 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
6175 (struct machine_function): Delete field spe_insn_chain_scanned_p.
6176 (spe_func_has_64bit_regs_p): Delete.
6177 (spe_expand_predicate_builtin): Delete.
6178 (spe_expand_evsel_builtin): Delete.
6179 (TARGET_DWARF_REGISTER_SPAN): Do not define.
6180 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
6181 (invalid_e500_subreg): Delete.
6182 (rs6000_legitimize_address): Always force_reg op2 as well, for
6183 paired single memory accesses.
6184 (rs6000_member_type_forces_blk): Delete.
6185 (rs6000_spe_function_arg): Delete.
6186 (rs6000_expand_unop_builtin): Delete SPE handling.
6187 (rs6000_expand_binop_builtin): Ditto.
6188 (spe_expand_stv_builtin): Delete.
6189 (bdesc_2arg_spe): Delete.
6190 (spe_expand_builtin): Delete.
6191 (spe_expand_predicate_builtin): Delete.
6192 (spe_expand_evsel_builtin): Delete.
6193 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
6194 (spe_init_builtins): Delete.
6195 (spe_func_has_64bit_regs_p): Delete.
6196 (savres_routine_name): Delete "info" parameter. Adjust callers.
6197 (rs6000_emit_stack_reset): Ditto.
6198 (rs6000_dwarf_register_span): Delete.
6199 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
6200 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
6201 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
6203 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
6205 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
6206 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
6207 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
6208 mulv2sf3, divv2sf3): Delete expanders.
6210 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6212 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
6214 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6216 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
6217 * config/rs6000/rs6000.c: Ditto.
6219 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6221 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
6222 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
6223 comparison_operator.
6225 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6227 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
6228 * config/rs6000/rs6000.opt: Ditto.
6229 * config/rs6000/t-rtems: Ditto.
6231 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6233 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
6234 TARGET_E500_SINGLE by 0, simplify.
6235 * config/rs6000/rs6000.c: Ditto.
6236 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
6237 (spe_build_register_parallel): Delete.
6238 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
6239 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
6240 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
6241 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
6242 (E500_CONVERT): Delete.
6243 * config/rs6000/spe.md: Remove many patterns and all define_constants.
6245 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6247 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
6248 * config/rs6000/dfp.md: Ditto.
6249 (negdd2, *negdd2_fpr): Merge.
6250 (absdd2, *absdd2_fpr): Merge.
6251 (negtd2, *negtd2_fpr): Merge.
6252 (abstd2, *abstd2_fpr): Merge.
6253 * config/rs6000/e500.h: Delete file.
6254 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
6255 TARGET_FPRS by 1 and simplify.
6256 * config/rs6000/rs6000-c.c: Ditto.
6257 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
6259 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
6261 * config/rs6000/rs6000.md: Ditto.
6262 (floatdidf2, *floatdidf2_fpr): Merge.
6263 (move_from_CR_gt_bit): Delete.
6264 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
6265 (E500_CR_IOR_COMPARE): Delete.
6266 (All patterns that require !TARGET_FPRS): Delete.
6267 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
6269 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6271 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
6273 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6275 * graphds.c (add_edge): Intitialize edge's attached data.
6276 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
6277 pointer parameter. Call pointed function on each edge during
6278 graph traversing. Skip traversing the edge when the function
6280 (graphds_dfs, graphds_scc): Ditto.
6281 (for_each_edge): New parameter. Pass the new parameter to callback
6283 * graphds.h (skip_edge_callback): New function pointer type.
6284 (graphds_dfs, graphds_scc): New function pointer parameter.
6285 (graphds_edge_callback, for_each_edge): New parameter.
6287 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6289 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
6290 out code checking if runtime alias check is possible to below ...
6291 Call the new function.
6292 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
6293 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
6295 2017-06-07 Marek Polacek <polacek@redhat.com>
6298 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
6299 TYPE_OVERFLOW_WRAPS check.
6301 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6303 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
6304 if versioning is required.
6305 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
6306 peeling with the check for versioning.
6308 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6310 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
6311 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
6312 Set true to new parameter if new ssa variable is defined.
6313 (vect_gen_vector_loop_niters): Refactor. Set range information
6314 for the new vector loop bound variable.
6315 (vect_do_peeling): Ditto.
6317 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6319 * tree-affine.c (ssa.h): Include header file.
6320 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
6321 has wrapping overflow behavior.
6323 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6325 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
6327 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6329 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
6330 (tree_to_aff_combination): ... here.
6332 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6334 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
6335 reg_pressure model function.
6336 (ivopts_global_cost_for_size): Delete.
6337 (determine_set_costs, iv_ca_recount_cost): Call new model function
6338 ivopts_estimate_reg_pressure.
6340 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6342 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
6343 expensive than udiv. Remove floating point cases from mod.
6345 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6347 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
6350 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6352 * config/aarch64/aarch64.md
6353 (copysignsf3): Fix mask generation.
6355 2017-06-07 Jakub Jelinek <jakub@redhat.com>
6357 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
6359 (class dump_manager): Add register_dumps method.
6360 * dumpfile.c: Include langhooks.h.
6361 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
6362 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
6363 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
6364 (dump_manager::dump_register): Start with 512 entries instead of 32.
6365 (dump_manager::register_dumps): New method.
6366 * toplev.c (general_init): Instead of invoking register_dumps
6367 langhook, invoke register_dumps method on the dump manager.
6368 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
6371 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
6373 * doc/md.texi: Clarify the restrictions on a define_insn condition.
6374 Say that # requires an associated define_split to exist, and that
6375 the define_split must be suitable for use after register allocation.
6377 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6379 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
6380 (compute_outgoing_frequencies): Also initialize zero counts.
6381 (find_many_sub_basic_blocks): Do not produce uninitialized profile
6382 around loops; preserve more of profile when nothing changes.
6384 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
6386 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
6388 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
6389 * config/arm/arm-cpu-cdata.h: Regenerate.
6390 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
6391 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
6392 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
6393 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
6394 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
6396 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
6398 * config/arm/t-rmprofile: Likewise.
6399 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
6401 2017-06-06 David S. Miller <davem@davemloft.net>
6404 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
6405 blockage if function uses alloca.
6407 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
6409 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
6410 New "uid" fields to hold pretty-print IDs of group and ref.
6411 Memory references are now identified as <group_id>:<ref_id>
6412 instead of using [random] addresses.
6413 (dump_mem_details): Simplify, no functional change.
6414 (dump_mem_ref): Simplify and make output more concise.
6415 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
6416 (find_or_create_group): Initialize group uid.
6417 (record_ref): Initialize ref uid. Improve debug output.
6418 (prune_group_by_reuse, should_issue_prefetch_p,)
6419 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
6420 (mark_nontemporal_store, determine_loop_nest_reuse):
6421 Improve debug output.
6423 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
6425 * dbgcnt.def (prefetch): New debug counter.
6426 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
6427 (schedule_prefetches): Stop issueing prefetches if debug counter
6430 2017-06-06 Tom de Vries <tom@codesourcery.com>
6432 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
6433 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
6435 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6437 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
6438 Use aarch64_reg_or_zero predicate for operand 4.
6439 (aarch64_compare_and_swap<mode> define_insn_and_split):
6440 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
6441 (aarch64_store_exclusive<mode>): Likewise for operand 2.
6443 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
6445 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
6446 (arm_compute_save_core_reg_mask): This.
6447 (thumb1_compute_save_reg_mask): Rename into ...
6448 (thumb1_compute_save_core_reg_mask): This.
6449 (arm_compute_save_reg0_reg12_mask): Adapt comment.
6450 (arm_compute_frame_layout): Likewise.
6452 2017-06-06 Richard Biener <rguenther@suse.de>
6454 PR tree-optimization/80974
6455 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
6456 keep or clear leaders SSA info.
6458 2017-06-06 Tom de Vries <tom@codesourcery.com>
6460 * config/nvptx/nvptx.c (split_mode_p): New function.
6461 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
6463 2017-06-06 Tom de Vries <tom@codesourcery.com>
6465 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
6467 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6470 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
6473 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6475 * shrink-wrap.c (handle_simple_exit): Update profile.
6476 (try_shrink_wrapping): Upate profile.
6478 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6480 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
6481 (tree_guess_outgoing_edge_probabilities): New.
6482 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
6483 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
6485 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6487 * ipa-split.c (split_function): Initialize return bb profile.
6489 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6491 * profile.c (compute_branch_probabilities): Also initialize
6494 2017-06-06 Richard Biener <rguenther@suse.de>
6496 PR tree-optimization/80928
6497 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
6498 (vect_analyze_loop_operations): Properly guard analysis for
6500 (vect_transform_loop): Likewise.
6501 (vect_analyze_loop_2): Also reset SLP type on PHIs.
6502 (vect_model_induction_cost): Do not cost for pure SLP.
6503 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
6504 of induction in inner loop vectorization.
6505 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
6506 (vect_get_and_check_slp_defs): Handle vect_induction_def.
6507 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
6509 (vect_analyze_slp_cost_1): Cost induction.
6510 (vect_detect_hybrid_slp_stmts): Handle PHIs.
6511 (vect_get_slp_vect_defs): Likewise.
6512 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
6513 (vect_transform_stmt): Handle SLP reductions.
6514 * tree-vectorizer.h (vectorizable_induction): Adjust.
6516 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6518 * config/rs6000/rs6000.c (make_resolver_func): Update
6519 init_lowered_empty_function call.
6521 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
6523 * doc/invoke.texi: Document the -fprofile-abs-path option.
6524 * common.opt (fprofile-abs-path): New option.
6525 * gcov-io.h (gcov_write_filename): Declare.
6526 * gcov-io.c (gcov_write_filename): New function.
6527 * coverage.c (coverage_begin_function): Use gcov_write_filename.
6528 * profile.c (output_location): Likewise.
6530 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6532 * shring-wrap.c: Revert accidental commit.
6534 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
6536 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
6538 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6540 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
6542 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
6543 profile in callgraph edge.
6544 * profile-count.h (apply_probability): If THIS is 0, then result is 0
6545 (apply_scale): Likewise.
6546 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
6547 Also scale profile when inlining function with zero profile.
6548 (initialize_cfun): Update exit block profile even when it is zero.
6549 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
6550 when profile is read.
6552 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6554 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
6555 (CLONE_*): New constants to define the processors we can generate
6556 code for with the target_clone attribute.
6557 (rs6000_clone_map): New array to identify which clone processors
6558 the current program is running on.
6559 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
6560 target_clone attribute.
6561 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
6562 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
6563 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
6564 (cpu_expand_builtin): Add support for target_clone attribute.
6565 (rs6000_valid_attribute_p): Allow "default" attribute.
6566 (get_decl_name): New debug function to simplify printing the
6567 current function name in debugging statements.
6568 (rs6000_clone_priority): New functions to support the target_clone
6569 attribute, and be able to generate code to switch between ISA 2.05
6570 through ISA 3.0 (power6 through power9).
6571 (rs6000_compare_version_priority): Likewise.
6572 (rs6000_get_function_versions_dispatcher): Likewise.
6573 (make_resolver_func): Likewise.
6574 (add_condition_to_bb): Likewise.
6575 (dispatch_function_versions): Likewise.
6576 (rs6000_generate_version_dispatcher_body): Likewise.
6577 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
6578 (fusion_gpr_load_p): Fix a spacing issue.
6579 * doc/extend.texi (Common Function Attributes): Document that the
6580 PowerPC supports the target_clone attribute.
6582 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
6584 * config/arm/arm.h: explain F symbol found in description of ARM
6585 register allocation in its legend.
6587 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6589 * config/mips/frame-header-opt.c: Include profile-count.h.
6590 * config/riscv/riscv.c: Include profile-count.h
6592 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6594 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
6596 (sm_set_flag_if_changed): Add bbs field.
6597 (execute_sm_if_changed_flag_set): Pass BBS.
6598 (execute_sm): Update.
6600 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6602 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6605 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6607 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
6608 (peephole2): New peephole2 to emit the above.
6609 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
6611 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6613 * config/aarch64/aarch64.c (define_peephole2 above
6614 *sub_<shift>_<mode>): New peephole.
6616 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
6618 * config/i386/i386.c (make_resolver_func): Update.
6619 * Makefile.in: Add profile-count.h and profile-count.o
6620 * auto-profile.c (afdo_indirect_call): Update to new API.
6621 (afdo_set_bb_count): Update.
6622 (afdo_propagate_edge): Update.
6623 (afdo_propagate_circuit): Update.
6624 (afdo_calculate_branch_prob): Update.
6625 (afdo_annotate_cfg): Update.
6626 * basic-block.h: Include profile-count.h
6627 (struct edge_def): Turn count to profile_count.
6628 (struct basic_block_def): Likewie.
6629 (REG_BR_PROB_BASE): Move to profile-count.h
6630 (RDIV): Move to profile-count.h
6631 * bb-reorder.c (max_entry_count): Turn to profile_count.
6632 (find_traces): Update.
6633 (rotate_loop):Update.
6634 (connect_traces):Update.
6635 (sanitize_hot_paths):Update.
6636 * bt-load.c (migrate_btr_defs): Update.
6637 * cfg.c (RDIV): Remove.
6638 (init_flow): Use alloc_block.
6639 (alloc_block): Uninitialize count.
6640 (unchecked_make_edge): Uninitialize count.
6641 (check_bb_profile): Update.
6642 (dump_edge_info): Update.
6643 (dump_bb_info): Update.
6644 (update_bb_profile_for_threading): Update.
6645 (scale_bbs_frequencies_int): Update.
6646 (scale_bbs_frequencies_gcov_type): Update.
6647 (scale_bbs_frequencies_profile_count): New.
6648 * cfg.h (update_bb_profile_for_threading): Update.
6649 (scale_bbs_frequencies_profile_count): Declare.
6650 * cfgbuild.c (compute_outgoing_frequencies): Update.
6651 (find_many_sub_basic_blocks): Update.
6652 * cfgcleanup.c (try_forward_edges): Update.
6653 (try_crossjump_to_edge): Update.
6654 * cfgexpand.c (expand_gimple_tailcall): Update.
6655 (construct_exit_block): Update.
6656 * cfghooks.c (verify_flow_info): Update.
6657 (dump_bb_for_graph): Update.
6658 (split_edge): Update.
6659 (make_forwarder_block): Update.
6660 (duplicate_block): Update.
6661 (account_profile_record): Update.
6662 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
6663 (get_estimated_loop_iterations): Update.
6664 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
6665 (single_likely_exit): Update.
6666 * cfgloopmanip.c (scale_loop_profile): Update.
6668 (set_zero_probability): Update.
6669 (lv_adjust_loop_entry_edge): Update.
6670 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
6671 (purge_dead_edges): Update.
6672 (rtl_account_profile_record): Update.
6673 * cgraph.c (cgraph_node::create): Uninitialize count.
6674 (symbol_table::create_edge): Uninitialize count.
6675 (cgraph_update_edges_for_call_stmt_node): Update.
6676 (cgraph_edge::dump_edge_flags): Update.
6677 (cgraph_node::dump): Update.
6678 (cgraph_edge::maybe_hot_p): Update.
6679 * cgraph.h: Include profile-count.h
6680 (create_clone), create_edge, create_indirect_edge): Update.
6681 (cgraph_node): Turn count to profile_count.
6682 (cgraph_edge0: Likewise.
6683 (make_speculative, clone): Update.
6684 (create_edge): Update.
6685 (init_lowered_empty_function): Update.
6686 * cgraphclones.c (cgraph_edge::clone): Update.
6687 (duplicate_thunk_for_node): Update.
6688 (cgraph_node::create_clone): Update.
6689 * cgraphunit.c (cgraph_node::analyze): Update.
6690 (cgraph_node::expand_thunk): Update.
6691 * final.c (dump_basic_block_info): Update.
6692 * gimple-streamer-in.c (input_bb): Update.
6693 * gimple-streamer-out.c (output_bb): Update.
6694 * graphite.c (print_global_statistics): Update.
6695 (print_graphite_scop_statistics): Update.
6696 * hsa-brig.c: Include basic-block.h.
6697 * hsa-dump.c: Include basic-block.h.
6698 * hsa-gen.c (T sum_slice): Update.
6699 (convert_switch_statements):Update.
6700 * hsa-regalloc.c: Include basic-block.h.
6701 * ipa-chkp.c (chkp_produce_thunks): Update.
6702 * ipa-cp.c (struct caller_statistics): Update.
6703 (init_caller_stats): Update.
6704 (gather_caller_stats): Update.
6705 (ipcp_cloning_candidate_p): Update.
6706 (good_cloning_opportunity_p): Update.
6707 (get_info_about_necessary_edges): Update.
6708 (dump_profile_updates): Update.
6709 (update_profiling_info): Update.
6710 (update_specialized_profile): Update.
6711 (perhaps_add_new_callers): Update.
6712 (decide_about_value): Update.
6713 (ipa_cp_c_finalize): Update.
6714 * ipa-devirt.c (struct odr_type_warn_count): Update.
6715 (struct decl_warn_count): Update.
6716 (struct final_warning_record): Update.
6717 (possible_polymorphic_call_targets): Update.
6718 (ipa_devirt): Update.
6719 * ipa-fnsummary.c (redirect_to_unreachable): Update.
6720 * ipa-icf.c (sem_function::merge): Update.
6721 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
6722 * ipa-inline.c (compute_uninlined_call_time): Update.
6723 (compute_inlined_call_time): Update.
6724 (want_inline_small_function_p): Update.
6725 (want_inline_self_recursive_call_p): Update.
6726 (edge_badness): Update.
6727 (lookup_recursive_calls): Update.
6728 (recursive_inlining): Update.
6729 (inline_small_functions): Update.
6730 (dump_overall_stats): Update.
6731 (dump_inline_stats): Update.
6732 * ipa-profile.c (ipa_profile_generate_summary): Update.
6733 (ipa_propagate_frequency): Update.
6734 (ipa_profile): Update.
6735 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
6736 * ipa-utils.c (ipa_merge_profiles): Update.
6737 * loop-doloop.c (doloop_modify): Update.
6738 * loop-unroll.c (report_unroll): Update.
6739 (unroll_loop_runtime_iterations): Update.
6740 * lto-cgraph.c (lto_output_edge): Update.
6741 (lto_output_node): Update.
6742 (input_node): Update.
6743 (input_edge): Update.
6744 (merge_profile_summaries): Update.
6745 * lto-streamer-in.c (input_cfg): Update.
6746 * lto-streamer-out.c (output_cfg): Update.
6747 * mcf.c (create_fixup_graph): Update.
6748 (adjust_cfg_counts): Update.
6749 (sum_edge_counts): Update.
6750 * modulo-sched.c (sms_schedule): Update.
6751 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
6752 * predict.c (maybe_hot_count_p): Update.
6753 (probably_never_executed): Update.
6754 (dump_prediction): Update.
6755 (combine_predictions_for_bb): Update.
6756 (propagate_freq): Update.
6757 (handle_missing_profiles): Update.
6758 (counts_to_freqs): Update.
6759 (rebuild_frequencies): Update.
6760 (force_edge_cold): Update.
6761 * predict.h: Include profile-count.h
6762 (maybe_hot_count_p, counts_to_freqs): UPdate.
6763 * print-rtl-function.c: Do not include cfg.h
6764 * print-rtl.c: Include basic-block.h
6765 * profile-count.c: New file.
6766 * profile-count.h: New file.
6767 * profile.c (is_edge_inconsistent): Update.
6768 (correct_negative_edge_counts): Update.
6769 (is_inconsistent): Update.
6770 (set_bb_counts): Update.
6771 (read_profile_edge_counts): Update.
6772 (compute_frequency_overlap): Update.
6773 (compute_branch_probabilities): Update; Initialize and deinitialize
6775 (branch_prob): Update.
6776 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
6777 (edge_gcov_count): New.
6778 (bb_gcov_count): New.
6779 * shrink-wrap.c (try_shrink_wrapping): Update.
6780 * tracer.c (better_p): Update.
6781 * trans-mem.c (expand_transaction): Update.
6782 (ipa_tm_insert_irr_call): Update.
6783 (ipa_tm_insert_gettmclone_call): Update.
6784 * tree-call-cdce.c: Update.
6785 * tree-cfg.c (gimple_duplicate_sese_region): Update.
6786 (gimple_duplicate_sese_tail): Update.
6787 (gimple_account_profile_record): Update.
6788 (execute_fixup_cfg): Update.
6789 * tree-inline.c (copy_bb): Update.
6790 (copy_edges_for_bb): Update.
6791 (initialize_cfun): Update.
6792 (freqs_to_counts): Update.
6793 (copy_cfg_body): Update.
6794 (expand_call_inline): Update.
6795 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
6796 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
6797 (try_unroll_loop_completely): Update.
6798 (try_peel_loop): Update.
6799 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6800 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
6801 * tree-ssa-loop-split.c (connect_loops): Update.
6802 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
6803 * tree-ssa-reassoc.c (branch_fixup): Update.
6804 * tree-ssa-tail-merge.c (replace_block_by): Update.
6805 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
6806 (compute_path_counts): Update.
6807 (update_profile): Update.
6808 (recompute_probabilities): Update.
6809 (update_joiner_offpath_counts): Update.
6810 (estimated_freqs_path): Update.
6811 (freqs_to_counts_path): Update.
6812 (clear_counts_path): Update.
6813 (ssa_fix_duplicate_block_edges): Update.
6814 (duplicate_thread_path): Update.
6815 * tree-switch-conversion.c (case_bit_test_cmp): Update.
6816 (struct switch_conv_info): Update.
6817 * tree-tailcall.c (decrease_profile): Update.
6818 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
6819 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6820 * value-prof.c (check_counter): Update.
6821 (gimple_divmod_fixed_value): Update.
6822 (gimple_mod_pow2): Update.
6823 (gimple_mod_subtract): Update.
6824 (gimple_ic_transform): Update.
6825 (gimple_stringop_fixed_value): Update.
6826 * value-prof.h (gimple_ic): Update.
6828 2017-06-02 Carl Love <cel@us.ibm.com>
6830 * config/rs6000/rs6000-c: Add support for built-in functions
6831 vector double vec_doublee (vector signed int);
6832 vector double vec_doublee (vector unsigned int);
6833 vector double vec_doublee (vector float);
6834 vector double vec_doubleh (vector signed int);
6835 vector double vec_doubleh (vector unsigned int);
6836 vector double vec_doubleh (vector float);
6837 vector double vec_doublel (vector signed int);
6838 vector double vec_doublel (vector unsigned int);
6839 vector double vec_doublel (vector float);
6840 vector double vec_doubleo (vector signed int);
6841 vector double vec_doubleo (vector unsigned int);
6842 vector double vec_doubleo (vector float);.
6843 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
6844 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
6846 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
6847 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
6848 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
6850 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
6851 vec_doublel, vec_doubleh.
6852 * doc/extend.texi: Update the built-in documentation file for the
6853 new built-in functions.
6855 2017-06-02 David Malcolm <dmalcolm@redhat.com>
6858 * ipa-inline-analysis.c (free_growth_caches): Set
6859 edge_removal_hook_holder to NULL after removing it.
6861 2017-06-02 Sudakshina Das <sudi.das@arm.com>
6863 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
6864 comparision with zero.
6866 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6867 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6868 for early expansion of vec_min and vec_max builtins.
6869 (builtin_function_type): Add min/max unsigned variants to those
6870 identified as having unsigned arguments.
6872 2017-06-02 Olivier Hainque <hainque@adacore.com>
6874 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
6876 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6878 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
6879 Use VALL_F16 iterator rather than VALL.
6881 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6883 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
6884 Emit CBNZ inside loop when doing a strong exchange and comparing
6885 against zero. Generate the CC flags after the loop.
6887 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
6889 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
6890 (dl_section_ref): New.
6891 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
6892 On AIX, append an expression to subtract the size of the
6893 section length to dl_section_ref.
6895 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6897 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6898 for early expansion of vector absolute builtins.
6900 2017-06-02 Richard Biener <rguenther@suse.de>
6902 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
6903 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
6905 2017-06-02 Richard Biener <rguenther@suse.de>
6907 PR tree-optimization/80948
6908 * tree-tailcall.c (find_tail_calls): Track stmts to move in
6911 2017-06-02 Richard Biener <rguenther@suse.de>
6913 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
6915 * tree-vect-stmts.c (process_use): Do not mark backedge defs
6916 for inductions as relevant.
6918 2017-06-02 Richard Biener <rguenther@suse.de>
6920 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
6921 (vectorizable_induction): ... this. Remove dead code.
6923 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
6925 * builtins. (expand_builtin_alloca): Remove second parameter and
6926 infer its value from the first parameter instead.
6927 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
6929 2017-06-02 Jakub Jelinek <jakub@redhat.com>
6931 PR rtl-optimization/80903
6932 * loop-doloop.c (add_test): Unshare sequence.
6934 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
6936 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
6938 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
6940 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
6942 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
6943 xlogue_layout::get_instance, logue_layout::xlogue_layout,
6944 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
6945 (xlogue_layout::get_stub_rtx): Make static.
6946 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
6947 (xlogue_layout::compute_stub_managed_regs): Rename to...
6948 (xlogue_layout::count_stub_managed_regs): ...this.
6949 (xlogue_layout::is_stub_managed_reg): New function.
6950 (xlogue_layout::m_stub_names): Rename to...
6951 (xlogue_layout::s_stub_names): ...this, make static.
6952 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
6953 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
6954 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
6955 xlogue_layout::s_stub_names): Instantiate statics.
6956 (stub_managed_regs): Remove.
6957 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
6958 (disable_call_ms2sysv_xlogues): Rename to...
6959 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
6960 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
6962 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
6963 change after reload_completed.
6964 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
6966 (ix86_expand_prologue): Likewise.
6967 (ix86_expand_epilogue): Likewise.
6968 (ix86_expand_split_stack_prologue): Likewise.
6969 (ix86_compute_frame_layout): Remove frame parameter ...
6970 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
6971 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
6973 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
6974 (ix86_frame): Move from here ...
6975 * config/i386/i386.h (ix86_frame): ... to here.
6976 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
6977 complete ix86_frame data structure instead. Remove some_ld_name.
6979 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
6981 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
6982 symbols that hold a DECL_VALUE_EXPR.
6984 2017-06-01 Martin Jambor <mjambor@suse.cz>
6986 PR tree-optimization/80898
6987 * tree-sra.c (process_subtree_disqualification): Removed.
6988 (disqualify_candidate): Do not acll
6989 process_subtree_disqualification.
6990 (subtree_mark_written_and_enqueue): New function.
6991 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
6992 RHS has been disqualified and re-queue LHS if necessary. Apart
6993 from that, ignore disqualified RHS.
6995 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6997 * config/s390/s390.c (s390_emit_epilogue): Disable early return
6998 address fetch for z10 or later.
7000 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7002 * config/arc/arc.md (tst_movb): Add guard when splitting.
7004 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7006 * config/arc/arc.c (arc_can_eliminate): Test against
7007 arc_frame_pointer_needed.
7009 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7011 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
7012 to prevent store reordering.
7013 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
7014 (type): Add block type.
7015 (stack_tie): Define special instruction to be used in
7018 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7020 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
7021 constraint. It is not valid for the pattern.
7022 (noncommutative_binary_comparison): Likewise.
7024 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7026 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
7029 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7031 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
7032 be used by the reg-alloc.
7034 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7036 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
7037 reg-alloc when having mul64 or mul32x16 instructions.
7038 (mulsidi3): Likewise.
7039 (umulsidi3): Likewise.
7040 (mulsi32x16): New pattern.
7041 (mulsi64): Likewise.
7042 (mulsidi64): Likewise.
7043 (umulsidi64): Likewise.
7044 (MUL32x16_REG): Define.
7045 (mul64_600): Use MUL32x16_REG.
7046 (mac64_600): Likewise.
7047 (umul64_600): Likewise.
7048 (umac64_600): Likewise.
7050 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7052 * config/arc/arc.md (mulsi3_700): Make it commutative.
7054 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
7056 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
7058 (*sign_extendsidi2_insn): Likewise for movstosw.
7060 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
7062 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
7063 the type of the input discriminant value. Convert the
7064 discriminant value of signedness vary.
7066 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
7068 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
7069 Add to -Wall section.
7071 2017-06-01 Richard Biener <rguenther@suse.de>
7074 * fold-const.c (fold_plusminus_mult_expr): If the factored
7075 factor may be zero use a wrapping type for the inner operation.
7076 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
7077 and handle moved defs.
7078 (process_assignment): Properly guard the unary op case. Return a
7079 tri-state indicating that moving the stmt before the call may allow
7080 to continue. Pass through to_move.
7081 (find_tail_calls): Handle moving unrelated defs before
7084 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
7087 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
7088 splitter result in the canonical way.
7090 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
7092 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
7093 also for 32bit target. Update insn attributes.
7094 (zero-extendsidi2 splitter): Allow all registers for operand 1.
7096 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
7098 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
7099 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
7100 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
7101 (_mm_maskz_min_ss): New intrinsics.
7103 2017-05-31 Martin Liska <mliska@suse.cz>
7105 * tree-vect-loop.c (vect_create_epilog_for_reduction):
7106 Change comment style to one we normally use.
7107 (vectorizable_reduction): Likewise.
7108 (vectorizable_induction): Likewise.
7109 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7110 (vectorizable_call): Likewise.
7111 (vectorizable_simd_clone_call): Likewise.
7112 (vectorizable_conversion): Likewise.
7113 (vectorizable_assignment): Likewise.
7114 (vectorizable_shift): Likewise.
7115 (vectorizable_operation): Likewise.
7116 (vectorizable_store): Likewise.
7117 (vectorizable_load): Likewise.
7118 * tree-vectorizer.h: Likewise.
7120 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
7122 * passes.c (emergency_dump_function): New.
7123 * tree-pass.h (emergency_dump_function): Declare.
7124 * plugin.c (plugins_internal_error_function): Remove.
7125 * plugin.h (plugins_internal_error_function): Remove declaration.
7126 * toplev.c (internal_error_function): New static function. Use it...
7127 (general_init): ...here.
7129 2017-05-31 Graham Markall <graham.markall@embecosm.com>
7131 * config/arc/arc.c (arc_print_operand): Handle constant operands.
7132 (arc_rtx_costs): Add costs for new patterns.
7133 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
7134 * config/arc/predicates.md: Add _1_2_3_operand predicate.
7136 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
7138 * tree-ssa-strlen.c (get_next_strinfo): New function.
7139 (get_stridx_plus_constant): Use it.
7140 (zero_length_string): Likewise.
7141 (adjust_related_strinfos): Likewise.
7142 (adjust_last_stmt): Likewise.
7144 2017-05-31 Richard Biener <rguenther@suse.de>
7147 * config/i386/i386.c (ix86_expand_builtin): Remove assert
7148 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
7150 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
7152 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
7153 loop_vinfo argument and use of dependence distance vectors.
7154 Check instead whether the two references differ only in their
7155 initial value and assume that they have the same alignment if the
7156 difference is a multiple of the vector alignment.
7157 (vect_analyze_data_refs_alignment): Update call accordingly.
7159 2017-05-31 Martin Liska <mliska@suse.cz>
7162 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
7164 2017-05-31 Bin Cheng <bin.cheng@arm.com>
7166 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
7167 (create_intersect_range_checks): Move from ...
7168 * tree-data-ref.c (create_intersect_range_checks_index)
7169 (create_intersect_range_checks): ... to here.
7170 (create_runtime_alias_checks): New function factored from ...
7171 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
7172 here. Call above function.
7173 * tree-data-ref.h (create_runtime_alias_checks): New function.
7175 2017-05-31 Bin Cheng <bin.cheng@arm.com>
7177 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
7178 segment length for dr_b and compute it in wide_int.
7180 2017-05-31 Richard Biener <rguenther@suse.de>
7182 PR tree-optimization/80906
7183 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
7184 and pass through iv_map.
7185 (copy_bb_and_scalar_dependences): Adjust.
7186 (translate_pending_phi_nodes): Likewise.
7187 (copy_loop_close_phi_args): Handle code-generating IVs instead
7190 2017-05-30 David Malcolm <dmalcolm@redhat.com>
7192 * diagnostic-color.c (color_dict): Add "type-diff".
7193 (parse_gcc_colors): Update comment.
7194 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
7195 -fdiagnostics-show-template-tree and -fno-elide-type.
7196 (GCC_COLORS): Add type-diff to example.
7198 (-fdiagnostics-show-template-tree): New.
7199 (-fno-elide-type): New.
7200 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
7201 the pp_format_decoder callback. Call any m_format_postprocessor's
7203 (pretty_printer::pretty_printer): Initialize
7204 m_format_postprocessor.
7205 (pretty_printer::~pretty_printer): Delete any
7206 m_format_postprocessor.
7207 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
7208 (class format_postprocessor): New class.
7209 (struct pretty_printer::format_decoder): Document the new parameters.
7210 (struct pretty_printer::m_format_postprocessor): New field.
7211 * tree-diagnostic.c (default_tree_printer): Update for new
7212 bool and const char ** params.
7213 * tree-diagnostic.h (default_tree_printer): Likewise.
7215 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
7217 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
7218 (lwa_operand): Delete rs6000_gen_cell_microcode test.
7219 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
7220 rs6000_gen_cell_microcode code.
7221 (rs6000_final_prescan_insn): Delete.
7222 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
7223 "warn-cell-microcode" entries.
7224 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
7225 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
7226 throughout. Change cc_reg_not_micro_cr0_operand to
7227 cc_reg_not_cr0_operand throughout.
7228 (*extendhi<mode>2_noload): Delete.
7229 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
7230 (mwarn-cell-microcode): Delete.
7231 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
7232 -mgen-cell-microcode and -mwarn-cell-microcode.
7234 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
7237 * config/i386/constraints.md (Yd): New constraint.
7239 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
7240 and (?Yd, r) alternatives. Update insn attributes.
7241 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
7242 and (?*Yd, r) alternatives. Update insn attributes.
7243 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
7245 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
7247 * gimplify.c (gimplify_modify_expr): Don't create a
7248 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
7251 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
7253 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
7255 2017-05-30 Richard Biener <rguenther@suse.de>
7257 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
7258 and reduc_def fields.
7259 (STMT_VINFO_REDUC_TYPE): New define.
7260 (STMT_VINFO_REDUC_DEF): Likewise.
7261 (vect_force_simple_reduction): Adjust prototype.
7262 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
7263 (vect_is_simple_reduction): Remove check_reduction argument.
7264 (vect_force_simple_reduction): Adjust and set
7265 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
7266 (vectorizable_reduction): Do not re-do reduction analysis
7267 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
7268 * tree-parloops.c (gather_scalar_reductions): Adjust.
7270 2017-05-30 Richard Biener <rguenther@suse.de>
7273 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
7276 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
7278 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
7279 Introduce unknown_misalignment parameter and remove vf.
7280 (vect_peeling_hash_get_lowest_cost):
7281 Pass unknown_misalignment parameter.
7282 (vect_enhance_data_refs_alignment):
7283 Fix unsupportable data ref treatment.
7285 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7287 * tree-vect-data-refs.c (vect_get_data_access_cost):
7288 Workaround for SLP handling.
7289 (vect_enhance_data_refs_alignment):
7290 Compute costs for doing no peeling at all, compare to the best
7291 peeling costs so far and avoid peeling if cheaper.
7293 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7295 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
7296 Return peeling info and set costs to zero for unlimited cost
7298 (vect_enhance_data_refs_alignment): Also inspect all datarefs
7299 with unknown misalignment. Compute and costs for unknown
7300 misalignment, compare them to the costs for known misalignment
7301 and choose the cheapest for peeling.
7303 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7305 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
7306 (vect_get_peeling_costs_all_drs): Create function.
7307 (vect_peeling_hash_get_lowest_cost):
7308 Use vect_get_peeling_costs_all_drs.
7309 (vect_peeling_supportable): Create function.
7310 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
7312 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7314 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
7315 DR_HAS_NEGATIVE_STEP.
7316 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
7317 (vect_enhance_data_refs_alignment): Use.
7318 (vect_duplicate_ssa_name_ptr_info): Use.
7319 * tree-vectorizer.h (dr_misalignment): Use.
7320 (known_alignment_for_access_p): Use.
7322 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7325 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
7327 (has_section_name): New function.
7329 2017-05-30 Martin Liska <mliska@suse.cz>
7332 * auto-profile.c (get_function_decl_from_block): Fix
7335 2017-05-30 Richard Biener <rguenther@suse.de>
7338 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
7340 2017-05-30 Martin Liska <mliska@suse.cz>
7342 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
7343 * dumpfile.h (struct dump_file_info): Remove ctors.
7345 2017-05-30 Martin Liska <mliska@suse.cz>
7347 * predict.def: Fix GNU coding style.
7349 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
7351 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
7352 Mark 'to' argument with ATTRIBUTE_UNUSED.
7354 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
7356 * config/xtensa/xtensa.c (xtensa_emit_call): Use
7357 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
7358 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
7361 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
7363 * doc/install.texi (Options specification): Restore entry of
7364 --enable-sjlj-exceptions.
7366 2017-05-27 Michael Eager <eager@eagercon.com>
7369 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
7371 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
7373 * config/microblaze/microblaze.h
7374 (FIXED_REGISTERS): Update in macro.
7375 (CALL_USED_REGISTERS): Update in macro.
7377 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
7379 * doc/install.texi: Add links to macOS binary distributions.
7381 2017-05-27 Jakub Jelinek <jakub@redhat.com>
7385 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7387 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
7389 2017-05-26 Martin Liska <mliska@suse.cz>
7391 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
7393 2017-05-26 Martin Liska <mliska@suse.cz>
7395 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
7396 always leading ';; '.
7397 (dump_bb_info): Likewise.
7398 (brief_dump_cfg): Likewise.
7399 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
7400 * dumpfile.c: Remove usage of TDF_VERBOSE.
7401 * dumpfile.h (enum dump_kind): Likewise.
7402 (dump_gimple_bb_header): Do not use TDF_COMMENT.
7403 * print-tree.c (debug_verbose): Remove.
7404 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
7405 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
7406 * tree-diagnostic.c (default_tree_printer): Replace
7407 TDF_DIAGNOSTIC with TDF_SLIM.
7409 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7411 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
7412 in parameter loop, rather than loop_vinfo.
7413 (create_intersect_range_checks): Ditto.
7414 (vect_create_cond_for_alias_checks): Update call to above functions.
7416 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7418 PR tree-optimization/80815
7419 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
7420 for merging runtime alias checks. Handle negative DR_STEPs.
7422 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7424 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
7426 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
7427 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
7428 out code pruning runtime alias checks.
7429 * tree-data-ref.c (prune_runtime_alias_test_list): New function
7430 factored out from above.
7431 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
7433 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
7435 (prune_runtime_alias_test_list): New decalaration.
7437 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7439 * tree-vect-data-refs.c (compare_tree): Rename and move ...
7440 * tree-data-ref.c (data_ref_compare_tree): ... to here.
7441 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
7442 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
7443 (operator==, comp_dr_with_seg_len_pair): Ditto.
7444 (vect_prune_runtime_alias_test_list): Ditto.
7446 2017-05-26 Martin Liska <mliska@suse.cz>
7449 * params.def: Bound partial-inlining-entry-probability param.
7451 2017-05-26 Marek Polacek <polacek@redhat.com>
7454 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
7457 2017-05-26 Richard Biener <rguenther@suse.de>
7459 PR tree-optimization/80842
7460 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
7463 2017-05-26 Richard Biener <rguenther@suse.de>
7465 PR tree-optimization/80844
7466 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
7468 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
7470 * doc/md.texi (Machine Constraints): Update x86 family
7471 machine constraints section to match 'config/i386/constraints.md'.
7473 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
7475 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
7477 2017-05-25 Nathan Sidwell <nathan@acm.org>
7479 * doc/invoke.texi (--enable-languages): Update documentation.
7481 2017-05-25 Martin Liska <mliska@suse.cz>
7483 * dumpfile.c: Add TDF_FOLDING.
7484 * dumpfile.h (enum dump_kind): Likewise.
7485 * genmatch.c (dt_simplify::gen_1): Use it.
7487 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7489 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
7491 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7493 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
7494 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
7496 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7498 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
7499 * match.pd (X == C): Rewrite it here.
7500 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
7501 with_certain_nonzero_bits2): New predicates.
7502 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
7504 2017-05-24 Nathan Sidwell <nathan@acm.org>
7506 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
7509 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
7512 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
7514 * config/powerpcspe: New port. Files are copied from the rs6000
7515 port, with "rs6000" in filenames replaced by "powerpcspe".
7517 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
7519 PR rtl-optimization/80754
7520 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
7522 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
7524 * config/sparc/sparc.md (length): Return the correct value for -mflat
7525 sibcalls to match output_sibcall.
7527 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
7531 * config/rs6000/rs6000.c (struct machine_function): Add new field
7533 (rs6000_get_separate_components): Init that field, use it.
7534 (rs6000_components_for_bb): Use the field.
7536 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
7538 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
7540 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
7543 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
7545 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7548 * config/s390/s390.c (s390_check_qrst_address): Check incoming
7549 address against address_operand predicate.
7550 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
7552 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
7554 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
7555 parameters passed indirectly.
7557 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
7559 * config/i386/i386.md (*movdi_internal): Remove SSE4
7560 alternative 18 (?r, *v). Update insn attributes.
7561 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
7562 Update insn attributes.
7563 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
7564 Update insn attributes.
7565 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
7566 alternative 1 (r, v). Remove isa attribute.
7567 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
7568 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
7569 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
7571 2017-05-23 Tom de Vries <tom@codesourcery.com>
7573 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
7576 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7578 * cgraphunit.c (symbol_table::process_new_functions): Update.
7579 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
7580 (inline_generate_summary): Rename to ...
7581 (ipa_fn_summary_generate): ... this one.
7582 (inline_read_summary): Rename to ...
7583 (ipa_fn_summary_read): ... this one.
7584 (inline_write_summary): Rename to ...
7585 (ipa_fn_summary_write): ... this one.
7586 (inline_free_summary): Rename to ...
7587 (ipa_free_fn_summary): ... this one.
7588 (pass_data_local_fn_summary, pass_local_fn_summary,
7589 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
7590 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
7591 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
7592 make_pass_ipa_fn_summary): New.
7593 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
7594 inline_write_summary, inline_free_summary): Remove.
7595 (ipa_free_fn_summary) : New.
7596 * ipa-inline.c (ipa_inline): Update.
7597 (pass_ipa_inline): Do not generate summaries.
7598 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
7600 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
7601 and add pass_ipa_fn_summary.
7602 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
7604 (make_pass_inline_parameters): Remove.
7606 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
7608 * omp-low.c (struct omp_context): Remove "default_kind" member.
7611 * omp-offload.c (execute_oacc_device_lower): Remove the
7612 parallelism dimensions function attributes for unparallelized
7613 OpenACC kernels constructs.
7615 2017-05-23 Martin Liska <mliska@suse.cz>
7617 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
7619 (cgraph_edge::make_speculative): Likewise.
7620 (cgraph_edge::resolve_speculation): Likewise.
7621 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
7622 (cgraph_node::dump): Likewise.
7623 * cgraph.h: Likewise.
7624 * cgraphunit.c (analyze_functions): Likewise.
7625 (symbol_table::compile): Likewise.
7626 * ipa-cp.c (print_all_lattices): Likewise.
7627 (determine_versionability): Likewise.
7628 (initialize_node_lattices): Likewise.
7629 (ipcp_verify_propagated_values): Likewise.
7630 (estimate_local_effects): Likewise.
7631 (update_profiling_info): Likewise.
7632 (create_specialized_node): Likewise.
7633 (perhaps_add_new_callers): Likewise.
7634 (decide_about_value): Likewise.
7635 (decide_whether_version_node): Likewise.
7636 (identify_dead_nodes): Likewise.
7637 (ipcp_store_bits_results): Likewise.
7638 * ipa-devirt.c (dump_targets): Likewise.
7639 (ipa_devirt): Likewise.
7640 * ipa-icf.c (sem_item::dump): Likewise.
7641 (sem_function::equals): Likewise.
7642 (sem_variable::equals): Likewise.
7643 (sem_item_optimizer::read_section): Likewise.
7644 (sem_item_optimizer::execute): Likewise.
7645 (congruence_class::dump): Likewise.
7646 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
7647 (dump_inline_summary): Likewise.
7648 (estimate_node_size_and_time): Likewise.
7649 (inline_analyze_function): Likewise.
7650 * ipa-inline-transform.c (inline_call): Likewise.
7651 * ipa-inline.c (report_inline_failed_reason): Likewise.
7652 (want_early_inline_function_p): Likewise.
7653 (edge_badness): Likewise.
7654 (update_edge_key): Likewise.
7655 (inline_small_functions): Likewise.
7656 * ipa-profile.c (ipa_profile): Likewise.
7657 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
7658 (ipa_make_edge_direct_to_target): Likewise.
7659 (remove_described_reference): Likewise.
7660 (ipa_impossible_devirt_target): Likewise.
7661 (propagate_controlled_uses): Likewise.
7662 (ipa_print_node_params): Likewise.
7663 (ipcp_transform_function): Likewise.
7664 * ipa-pure-const.c (pure_const_read_summary): Likewise.
7665 (propagate_pure_const): Likewise.
7666 * ipa-reference.c (generate_summary): Likewise.
7667 (read_write_all_from_decl): Likewise.
7668 (propagate): Likewise.
7669 (ipa_reference_read_optimization_summary): Likewise.
7670 * ipa-utils.c (ipa_merge_profiles): Likewise.
7671 * ipa.c (walk_polymorphic_call_targets): Likewise.
7672 (symbol_table::remove_unreachable_nodes): Likewise.
7673 (ipa_single_use): Likewise.
7674 * passes.c (execute_todo): Likewise.
7675 * predict.c (drop_profile): Likewise.
7676 * symtab.c (symtab_node::get_dump_name): New function.
7677 (symtab_node::dump_name): Likewise.
7678 (symtab_node::dump_asm_name): Likewise.
7679 (symtab_node::dump_references): Likewise.
7680 (symtab_node::dump_referring): Likewise.
7681 (symtab_node::dump_base): Likewise.
7682 (symtab_node::debug_symtab): Likewise.
7683 * tree-sra.c (convert_callers_for_node): Likewise.
7684 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
7685 * value-prof.c (init_node_map): Likewise.
7687 2017-05-23 Martin Liska <mliska@suse.cz>
7689 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
7690 and symtab_node::debug_symtab to symbol_table::debug.
7691 * cgraphunit.c (analyze_functions): Use the renamed function.
7692 (symbol_table::compile): Likewise.
7693 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
7694 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
7695 * passes.c (execute_todo): Likewise.
7696 * symtab.c (symbol_table::dump): New function.
7697 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
7699 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7701 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
7702 that nonconst implies exec.
7704 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7706 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7707 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7708 (inline_edge_summary_vec): Turn into ...
7709 (ipa_call_summaries): ... this one.
7710 (redirect_to_unreachable, edge_set_predicate,
7711 evaluate_properties_for_edge, inline_summary_alloc,
7712 reset_ipa_call_summary, reset_inline_summary,
7713 inline_summary_t::duplicate): Update.
7714 (inline_edge_duplication_hook): Turn to ...
7715 (ipa_call_summary_t::duplicate): ... this one.
7716 (inline_edge_removal_hook): Turn to ...
7717 (ipa_call_summary_t::remove): ... this one.
7718 (dump_inline_edge_summary): Turn to ...
7719 (dump_ipa_call_summary): ... this one.
7720 (estimate_function_body_sizes): Update.
7721 (inline_update_callee_summaries): Update.
7722 (remap_edge_change_prob): Update.
7723 (remap_edge_summaries): Update.
7724 (inline_merge_summary): Update.
7725 (do_estimate_edge_time): Update.
7726 (inline_generate_summary): Update.
7727 (inline_read_section): Update.
7728 (inline_read_summary): Update.
7729 (inline_free_summary): Update.
7730 * ipa-inline.c (can_inline_edge_p): Update.
7731 (compute_inlined_call_time): Update.
7732 (want_inline_small_function_p): Update.
7733 (edge_badness): Update.
7734 (early_inliner): Update.
7735 * ipa-inline.h (inline_edge_summary): Turn to ...
7736 (ipa_call_summary): ... this one.
7737 (ipa_call_summary_t): New class.
7738 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
7739 (ipa_call_summaries): New.
7740 (inline_edge_summary): Remove.
7741 (estimate_edge_growth): Update.
7742 * ipa-profile.c (ipa_propagate_frequency_1): Update.
7743 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7744 * ipa-split.c (execute_split_functions): Update.
7745 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
7747 2017-05-23 Tom de Vries <tom@codesourcery.com>
7749 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7750 attributes): Document rdrand effective target.
7752 2017-05-23 Tom de Vries <tom@codesourcery.com>
7754 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7755 attributes): Sort alphabetically.
7757 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
7759 * config/avr/genmultilib.awk: Use gsub instead of gensub.
7761 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
7764 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
7765 V2DF/V2DI splat into two separate patterns, one that handles
7766 registers, and the other that only handles memory. Drop support
7767 for splatting from a GPR on ISA 2.07 and then splitting the
7768 splat into direct move and splat.
7769 (vsx_splat_<mode>_reg): Likewise.
7770 (vsx_splat_<mode>_mem): Likewise.
7772 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
7774 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
7776 2017-05-22 Jakub Jelinek <jakub@redhat.com>
7779 * omp-low.c (finish_taskreg_remap): New function.
7780 (finish_taskreg_scan): If unit size of ctx->record_type
7781 is non-constant, unshare the size expression and replace
7782 decls in it with possible outer var refs.
7785 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
7786 GOVD_SHARED rather than GOVD_PRIVATE with it.
7787 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
7788 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
7791 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
7792 as last argument to build_outer_var_ref for pointer bases of array
7795 2017-05-19 Martin Sebor <msebor@redhat.com>
7797 * print-tree.c (print_node): Print DECL_READ_P flag.
7799 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7801 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
7802 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
7803 * cgraph.c: Likewise.
7804 * cgraphunit.c: Likewise.
7805 * gengtype.c: Likewise.
7806 * ipa-cp.c: Likewise.
7807 * ipa-devirt.c: Likewise.
7808 * ipa-icf.c: Likewise.
7809 * ipa-predicate.c: Likewise.
7810 * ipa-profile.c: Likewise.
7811 * ipa-prop.c: Likewise.
7812 * ipa-split.c: Likewise.
7814 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
7815 edge_predicate_pool, dump_inline_hints,
7816 inline_summary::account_size_time, redirect_to_unreachable,
7817 edge_set_predicate, set_hint_predicate,
7818 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
7819 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
7820 inline_summary_t::remove, remap_hint_predicate_after_duplication,
7821 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
7822 ipa_call_summary_t::remove, initialize_growth_caches,
7823 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
7824 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
7825 mark_modified, unmodified_parm_1, unmodified_parm,
7826 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
7827 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
7828 compute_bb_predicates, will_be_nonconstant_expr_predicate,
7829 will_be_nonconstant_predicate, record_modified_bb_info,
7830 get_minimal_bb, record_modified, param_change_prob,
7831 phi_result_unknown_predicate, predicate_for_phi_result,
7832 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
7833 estimate_function_body_sizes, compute_inline_parameters,
7834 compute_inline_parameters_for_curren, pass_data_inline_parameters,
7835 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
7836 inline_update_callee_summaries, remap_edge_change_prob,
7837 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
7838 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
7839 inline_analyze_function, inline_summary_t::insert,
7840 inline_generate_summary, read_ipa_call_summary, inline_read_section,
7841 inline_read_summary, write_ipa_call_summary, inline_write_summary,
7842 inline_free_summary): Move to ipa-fnsummary.h
7843 (predicate_t): Remove.
7844 * ipa-fnsummary.c: New file.
7845 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
7846 (enum inline_hints_vals, inline_hints, agg_position_info,
7847 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
7848 inline_summaries, ipa_call_summary, ipa_call_summary_t,
7849 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
7850 dump_inline_summary, dump_inline_hints, inline_generate_summary,
7851 inline_read_summary, inline_write_summary, inline_free_summary,
7852 inline_analyze_function, initialize_inline_failed,
7853 inline_merge_summary, inline_update_overall_summary,
7854 compute_inline_parameters): Move to ipa-fnsummary.h
7855 * ipa-fnsummary.h: New file.
7856 * ipa-inline-transform.h: Include ipa-inline.h.
7857 * ipa-inline.c: LIkewise.
7859 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7861 * ipa-inline.c (edge_badness): Use inlined_time instead of
7862 inline_summaries->get.
7864 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7866 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
7868 2017-05-22 Nathan Sidwell <nathan@acm.org>
7870 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
7871 (fdump-lang): Document 'raw' option.
7872 * dumpfile.h (TDI_tu): Delete.
7873 * dumpfile.c (dump_files): Remove translation-unit.
7874 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
7876 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
7878 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
7879 command option from $(AWK) call.
7880 * config/avr/genmultilib.awk: Simplify and rewrite so that it
7881 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
7882 [FORMAT]: Remove handling of variable.
7883 * config/avr/t-multilib: Regenerate.
7885 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7887 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
7889 (dump_inline_summary): Do not print self_time.
7890 (estimate_function_body_sizes): Do not set self_time.
7891 (compute_inline_parameters): Likewise.
7892 (inline_read_section, inline_write_summary): Do not stream self_time.
7893 * ipa-inline.h (inline_summary): Drop self_time.
7895 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7897 * ipa-inline-analysis.c (account_size_time): Rename to ...
7898 (inline_summary::account_size_time): ... this one.
7899 (reset_ipa_call_summary): Turn to ...
7900 (ipa_call_summary::reset): ... this one.
7901 (reset_inline_summary): Turn to ...
7902 (inline_summary::reset): ... this one.
7903 (inline_summary_t::remove): Update.
7904 (inline_summary_t::duplicate): Update.
7905 (ipa_call_summary_t::remove): Update.
7906 (dump_inline_summary): Update.
7907 (estimate_function_body_sizes): Update.
7908 (compute_inline_parameters): Update.
7909 (estimate_node_size_and_time): Update.
7910 (inline_merge_summary): Update.
7911 (inline_update_overall_summary): Update.
7912 (inline_read_section): Update.
7913 (inline_write_summary): Update.
7914 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
7915 add account_size_time and reset member functions.
7916 (ipa_call_summary): Add reset function.
7917 * ipa-predicate.h (predicate::operator &): Constify.
7919 2017-05-22 Richard Biener <rguenther@suse.de>
7921 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
7923 2017-05-19 Jason Merrill <jason@redhat.com>
7925 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
7927 2017-05-19 Marek Polacek <polacek@redhat.com>
7930 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
7931 TYPE_OVERFLOW_WRAPS checks.
7933 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
7935 * tree-core.h (enum omp_clause_default_kind): Add
7936 "OMP_CLAUSE_DEFAULT_PRESENT".
7937 * tree-pretty-print.c (dump_omp_clause): Handle it.
7938 * gimplify.c (enum gimplify_omp_var_data): Add
7939 "GOVD_MAP_FORCE_PRESENT".
7940 (gimplify_adjust_omp_clauses_1): Map it to
7941 "GOMP_MAP_FORCE_PRESENT".
7942 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
7944 * gimplify.c (oacc_default_clause): Clarify.
7946 2017-05-19 Nathan Sidwell <nathan@acm.org>
7948 LANG_HOOK_REGISTER_DUMPS
7949 * toplev.c (general_init): Call register dump lang hook.
7950 * doc/invoke.texi: Document -fdump-lang option family.
7951 * dumpfile.c (dump_files): Remove class dump here.
7952 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
7953 * dumpfile.h (tree_dump_index): Remove TDI_class.
7954 * langhooks-def.h (lhd_register_dumps): Declare.
7955 (LANG_HOOKS_REGISTER_DUMPS): Define.
7956 (LANG_HOOKS_INITIALIZER): Add it.
7957 * langhooks.c (lhd_register_dumps): Define.
7958 * langhooks.h (struct lang_hooks): Add register_dumps.
7960 2017-05-19 Nathan Sidwell <nathan@acm.org>
7962 * context.h (context::set_passes): New.
7963 * context.c (context::context): Do not create pass manager.
7964 * toplev.c (general_init): Create pass manager here.
7966 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
7968 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
7969 use this splitter if two add or or instructions would also work for
7970 the constant we want to generate.
7972 2017-05-19 Richard Biener <rguenther@suse.de>
7975 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
7976 predicate evaluation.
7978 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
7980 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
7982 * ipa-inline.c (want_inline_small_function_p): Do not cast to
7985 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
7987 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7988 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7989 (inline_edge_summary_vec): Turn into ...
7990 (ipa_call_summaries): ... this one.
7991 (redirect_to_unreachable, edge_set_predicate,
7992 evaluate_properties_for_edge, inline_summary_alloc,
7993 reset_ipa_call_summary, reset_inline_summary,
7994 inline_summary_t::duplicate): Update.
7995 (inline_edge_duplication_hook): Turn to ...
7996 (ipa_call_summary_t::duplicate): ... this one.
7997 (inline_edge_removal_hook): Turn to ...
7998 (ipa_call_summary_t::remove): ... this one.
7999 (dump_inline_edge_summary): Turn to ...
8000 (dump_ipa_call_summary): ... this one.
8001 (estimate_function_body_sizes): Update.
8002 (inline_update_callee_summaries): Update.
8003 (remap_edge_change_prob): Update.
8004 (remap_edge_summaries): Update.
8005 (inline_merge_summary): Update.
8006 (do_estimate_edge_time): Update.
8007 (inline_generate_summary): Update.
8008 (inline_read_section): Update.
8009 (inline_read_summary): Update.
8010 (inline_free_summary): Update.
8011 * ipa-inline.c (can_inline_edge_p): Update.
8012 (compute_inlined_call_time): Update.
8013 (want_inline_small_function_p): Update.
8014 (edge_badness): Update.
8015 (early_inliner): Update.
8016 * ipa-inline.h (inline_edge_summary): Turn to ...
8017 (ipa_call_summary): ... this one.
8018 (ipa_call_summary_t): New class.
8019 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
8020 (ipa_call_summaries): New.
8021 (inline_edge_summary): Remove.
8022 (estimate_edge_growth): Update.
8023 * ipa-profile.c (ipa_propagate_frequency_1): Update.
8024 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
8025 * ipa-split.c (execute_split_functions): Update.
8026 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
8028 2017-05-19 Richard Biener <rguenther@suse.de>
8031 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
8033 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
8035 * config/rs6000/rs6000.c (struct machine_function): Add field
8036 fpr_is_wrapped_separately.
8037 (rs6000_get_separate_components): Use 64 components. Handle the
8039 (rs6000_components_for_bb): Handle the FPR components.
8040 (rs6000_emit_prologue_components): Handle the FPR components.
8041 (rs6000_emit_epilogue_components): Handle the FPR components.
8042 (rs6000_set_handled_components): Handle the FPR components.
8043 (rs6000_emit_prologue): Don't output prologue code for those FPRs
8044 that are already separately shrink-wrapped.
8045 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
8046 that are already separately shrink-wrapped.
8048 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
8051 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
8054 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
8055 (define_peephole2 for Altivec d-form load): Add peepholes to catch
8056 cases where the register allocator uses a move and an offsettable
8057 memory operation to/from a FPR register on ISA 2.06/2.07.
8058 (define_peephole2 for Altivec d-form store): Likewise.
8060 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
8063 * config/i386/mmx.md (*mov<mode>_internal): Enable
8064 alternatives 11, 12, 13 and 14 also for 32bit targets.
8065 Remove alternatives 15, 16, 17 and 18.
8066 * config/i386/sse.md (vec_concatv2di): Change
8067 alternative (!x, *y) to (x, ?!*Yn).
8069 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
8071 * dumpfile.h (enum dump_kind): Remove stray comma.
8073 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
8075 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
8076 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
8077 predicate::num_conditions
8078 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
8079 (CHANGED): turn into predicate::changed.
8080 (agg_position_info): Move to ipa-predicate.h
8081 (add_condition, predicate::add_clause, predicate::operator &=,
8082 predicate::or_with, predicate::evaluate, predicate::probability,
8083 dump_condition, dump_clause, predicate::dump,
8084 predicate::remap_after_duplication, predicate::remap_after_inlining,
8085 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
8086 (evaluate_conditions_for_known_args): Update.
8087 (set_cond_stmt_execution_predicate): Update.
8088 * ipa-inline.h: Include ipa-predicate.h
8089 (condition, inline_param_summary, conditions, agg_position_info,
8090 predicate): Move to ipa-predicate.h
8091 * ipa-predicate.c: New file.
8092 * ipa-predicate.h: New file.
8094 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
8096 * final.c (leaf_function_p): Check we are not in a sequence.
8098 2017-05-18 Martin Liska <mliska@suse.cz>
8100 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
8101 * dumpfile.c (dump_register): Use new enum dump_kind.
8102 (get_dump_file_name): Likewise.
8103 (dump_enable_all): Likewise.
8104 (dump_switch_p_1): Likewise.
8105 (enable_rtl_dump_file): Remove usage of TDF_RTL.
8106 * dumpfile.h (enum dump_kind): New enum type.
8107 (struct dump_file_info): Create constructor and
8108 format fields and comments.
8109 * passes.c (pass_manager::register_one_dump_file):
8111 * statistics.c (statistics_early_init): Likewise.
8112 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
8113 TDF_TREE with TDF_SLIM.
8114 (gather_memory_references_ref): Likewise.
8116 2017-05-18 Martin Liska <mliska@suse.cz>
8118 * vec.h (struct vnull): Use it.
8120 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
8122 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
8123 (true_predicate, false_predicate, true_predicate_p,
8124 false_predicate_p): Remove.
8125 (single_cond_predicate, not_inlined_predicate): Turn to member function
8127 (add_condition): Update.
8128 (add_clause): Turn to...
8129 (predicate::add_clause): ... this one; update; allow passing NULL
8131 (and_predicates): Turn to ...
8132 (predicate::operator &=): ... this one.
8133 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
8134 (or_predicates): Turn to ...
8135 (predicate::or_with): ... this one.
8136 (evaluate_predicate): Turn to ...
8137 (predicate::evaluate): ... this one.
8138 (predicate_probability): Turn to ...
8139 (predicate::probability): ... this one.
8140 (dump_condition): Update.
8141 (dump_predicate): Turn to ...
8142 (predicate::dump): ... this one.
8143 (account_size_time): Update.
8144 (edge_set_predicate): Update.
8145 (set_hint_predicate): UPdate.
8146 (evaluate_conditions_for_known_args): Update.
8147 (evaluate_properties_for_edge): Update.
8148 (remap_predicate_after_duplication): Turn to...
8149 (predicate::remap_after_duplication): ... this one.
8150 (remap_hint_predicate_after_duplication): Update.
8151 (inline_summary_t::duplicate): UPdate.
8152 (dump_inline_edge_summary): Update.
8153 (dump_inline_summary): Update.
8154 (set_cond_stmt_execution_predicate): Update.
8155 (set_switch_stmt_execution_predicate): Update.
8156 (compute_bb_predicates): Update.
8157 (will_be_nonconstant_expr_predicate): Update.
8158 (will_be_nonconstant_predicate): Update.
8159 (phi_result_unknown_predicate): Update.
8160 (predicate_for_phi_result): Update.
8161 (array_index_predicate): Update.
8162 (estimate_function_body_sizes): Update.
8163 (estimate_node_size_and_time): Update.
8164 (estimate_ipcp_clone_size_and_time): Update.
8165 (remap_predicate): Rename to ...
8166 (predicate::remap_after_inlining): ... this one.
8167 (remap_hint_predicate): Update.
8168 (inline_merge_summary): Update.
8169 (inline_update_overall_summary): Update.
8170 (estimate_size_after_inlining): Update.
8171 (read_predicate): Rename to ...
8172 (predicate::stream_in): ... this one.
8173 (read_inline_edge_summary): Update.
8174 (write_predicate): Rename to ...
8175 (predicate::stream_out): ... this one.
8176 (write_inline_edge_summary): Update.
8177 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
8178 (clause_t): Turn to uint32_t
8179 (predicate): Turn to class; implement constructor and operators
8181 (size_time_entry): Update.
8182 (inline_summary): Update.
8183 (inline_edge_summary): Update.
8185 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
8187 * fold-const.c (fold_binary_loc): Move transformation...
8188 * match.pd (C - X CMP X): ... here.
8190 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
8192 * config/sparc/sparc.c (sparc_option_override): Set function
8193 alignment for -mcpu=niagara7 to 64 to match the I$ line.
8194 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
8196 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
8198 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
8200 2017-05-18 Marek Polacek <polacek@redhat.com>
8203 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
8204 (pass_ubsan::execute): Call gimple_assign_single_p instead of
8205 gimple_assign_load_p.
8207 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
8210 * real.c (do_compare): Give decimal_do_compare preference over
8211 comparing just the signs.
8213 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
8215 * doc/md.texi (Canonicalization of Instructions): Describe the
8216 canonical form of instructions that inherently set a condition
8219 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
8222 * tree-cfg.c: Move deletion of unreachable case statements to after
8223 the merging of consecutive case labels.
8225 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
8227 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
8228 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
8229 restoring of callee-saved registers.
8231 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
8233 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
8234 * config/visium/visium.c (single_set_and_flags): Likewise.
8235 * config/visium/visium.md (Substitutions): Likewise.
8237 2017-05-17 Martin Liska <mliska@suse.cz>
8239 * cfg.c: Introduce dump_flags_t type and
8240 use it instead of int type.
8242 * cfghooks.c: Likewise.
8243 * cfghooks.h (struct cfg_hooks): Likewise.
8244 * cfgrtl.c: Likewise.
8245 * cfgrtl.h: Likewise.
8246 * cgraph.c (cgraph_node::get_body): Likewise.
8247 * coretypes.h: Likewise.
8248 * domwalk.c: Likewise.
8249 * domwalk.h: Likewise.
8250 * dumpfile.c (struct dump_option_value_info): Likewise.
8251 (dump_enable_all): Likewise.
8252 (dump_switch_p_1): Likewise.
8253 (opt_info_switch_p): Likewise.
8254 * dumpfile.h (enum tree_dump_index): Likewise.
8255 (struct dump_file_info): Likewise.
8256 * genemit.c: Likewise.
8257 * generic-match-head.c: Likewise.
8258 * gengtype.c (open_base_files): Likewise.
8259 * gimple-pretty-print.c: Likewise.
8260 * gimple-pretty-print.h: Likewise.
8261 * graph.c (print_graph_cfg): Likewise.
8262 * graphite-scop-detection.c (dot_all_sese): Likewise.
8263 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
8264 * loop-unroll.c (report_unroll): Likewise.
8265 * passes.c (pass_manager::register_one_dump_file): Likewise.
8266 * print-tree.c: Likewise.
8267 * statistics.c: Likewise.
8268 * tree-cfg.c: Likewise.
8269 * tree-cfg.h: Likewise.
8270 * tree-dfa.c: Likewise.
8271 * tree-dfa.h: Likewise.
8272 * tree-dump.c (dump_function): Likewise.
8273 * tree-dump.h (struct dump_info): Likewise.
8274 * tree-pretty-print.c: Likewise.
8275 * tree-pretty-print.h: Likewise.
8276 * tree-ssa-live.c: Likewise.
8277 * tree-ssa-live.h: Likewise.
8278 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
8279 * tree-vect-loop.c: Likewise.
8280 * tree-vect-slp.c: Likewise.
8282 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
8283 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8285 PR tree-optimization/80457
8286 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
8287 of all arguments to a statement as scalar_to_vec operations.
8288 (vectorizable_call): Adjust call to vect_model_simple_cost for
8290 (vectorizable_conversion): Likewise.
8291 (vectorizable_assignment): Likewise.
8292 (vectorizable_shift): Likewise.
8293 (vectorizable_operation): Likewise.
8294 (vectorizable_comparison): Likewise.
8295 (vect_is_simple_cond): Record the def types for operands.
8296 (vectorizable_condition): Likewise, call vect_model_simple_cost.
8297 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
8298 for statement argument count.
8300 2017-05-16 Carl Love <cel@us.ibm.com>
8302 * config/rs6000/rs6000-c: Add support for built-in functions
8303 vector unsigned long long vec_bperm (vector unsigned long long,
8304 vector unsigned char)
8305 vector signed long long vec_mule (vector signed int,
8307 vector unsigned long long vec_mule (vector unsigned int,
8308 vector unsigned int)
8309 vector signed long long vec_mulo (vector signed int,
8311 vector unsigned long long vec_mulo (vector unsigned int,
8312 vector unsigned int)
8313 vector signed char vec_sldw (vector signed char,
8316 vector unsigned char vec_sldw (vector unsigned char,
8317 vector unsigned char,
8319 vector signed short vec_sldw (vector signed short,
8320 vector signed short,
8322 vector unsigned short vec_sldw (vector unsigned short,
8323 vector unsigned short,
8325 vector signed int vec_sldw (vector signed int,
8328 vector unsigned int vec_sldw (vector unsigned int,
8329 vector unsigned int,
8331 vector signed long long vec_sldw (vector signed long long,
8332 vector signed long long,
8334 vector unsigned long long vec_sldw (vector unsigned long long,
8335 vector unsigned long long,
8337 * config/rs6000/rs6000-c: Add support for built-in functions
8338 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
8339 * config/rs6000/altivec.h: Add defintion for vec_sldw.
8340 * doc/extend.texi: Update the built-in documentation for the
8341 new built-in functions.
8343 2017-05-16 Marek Polacek <polacek@redhat.com>
8347 * tree.c (save_expr): Don't fold the expression.
8349 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
8351 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
8352 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
8353 and (?*y,m). Update insn attributes.
8355 2017-05-16 Martin Liska <mliska@suse.cz>
8357 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
8358 flags argument of print_gimple_stmt, print_gimple_expr,
8359 print_generic_stmt and print_generic_expr.
8360 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
8361 * coretypes.h: Likewise.
8362 * except.c (dump_eh_tree): Likewise.
8363 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8364 * gimple-pretty-print.h: Likewise.
8365 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
8366 (backprop::push_to_worklist): Likewise.
8367 (backprop::pop_from_worklist): Likewise.
8368 (backprop::process_use): Likewise.
8369 (backprop::intersect_uses): Likewise.
8370 (note_replacement): Likewise.
8371 * gimple-ssa-store-merging.c
8372 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
8373 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
8374 (pass_store_merging::execute): Likewise.
8375 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
8376 (ssa_base_cand_dump_callback): Likewise.
8377 (dump_incr_vec): Likewise.
8378 (replace_refs): Likewise.
8379 (replace_mult_candidate): Likewise.
8380 (create_add_on_incoming_edge): Likewise.
8381 (create_phi_basis): Likewise.
8382 (insert_initializers): Likewise.
8383 (all_phi_incrs_profitable): Likewise.
8384 (introduce_cast_before_cand): Likewise.
8385 (replace_one_candidate): Likewise.
8386 * gimplify.c (gimplify_expr): Likewise.
8387 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
8388 (set_rename): Likewise.
8389 (rename_uses): Likewise.
8390 (copy_loop_phi_nodes): Likewise.
8391 (add_close_phis_to_merge_points): Likewise.
8392 (copy_loop_close_phi_args): Likewise.
8393 (copy_cond_phi_args): Likewise.
8394 (graphite_copy_stmts_from_block): Likewise.
8395 (translate_pending_phi_nodes): Likewise.
8396 * graphite-poly.c (print_pdr): Likewise.
8397 (dump_gbb_cases): Likewise.
8398 (dump_gbb_conditions): Likewise.
8399 (print_scop_params): Likewise.
8400 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
8401 (build_cross_bb_scalars_use): Likewise.
8402 (gather_bbs::before_dom_children): Likewise.
8403 * hsa-dump.c (dump_hsa_immed): Likewise.
8404 * ipa-cp.c (print_ipcp_constant_value): Likewise.
8405 (get_replacement_map): Likewise.
8406 * ipa-inline-analysis.c (dump_condition): Likewise.
8407 (estimate_function_body_sizes): Likewise.
8408 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
8409 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
8410 * ipa-prop.c (ipa_dump_param): Likewise.
8411 (ipa_print_node_jump_functions_for_edge): Likewise.
8412 (ipa_modify_call_arguments): Likewise.
8413 (ipa_modify_expr): Likewise.
8414 (ipa_dump_param_adjustments): Likewise.
8415 (ipa_dump_agg_replacement_values): Likewise.
8416 (ipcp_modif_dom_walker::before_dom_children): Likewise.
8417 * ipa-pure-const.c (check_stmt): Likewise.
8418 (pass_nothrow::execute): Likewise.
8419 * ipa-split.c (execute_split_functions): Likewise.
8420 * omp-offload.c (dump_oacc_loop_part): Likewise.
8421 (dump_oacc_loop): Likewise.
8422 * trans-mem.c (tm_log_emit): Likewise.
8423 (tm_memopt_accumulate_memops): Likewise.
8424 (dump_tm_memopt_set): Likewise.
8425 (dump_tm_memopt_transform): Likewise.
8426 * tree-cfg.c (gimple_verify_flow_info): Likewise.
8427 (print_loop): Likewise.
8428 * tree-chkp-opt.c (chkp_print_addr): Likewise.
8429 (chkp_gather_checks_info): Likewise.
8430 (chkp_get_check_result): Likewise.
8431 (chkp_remove_check_if_pass): Likewise.
8432 (chkp_use_outer_bounds_if_possible): Likewise.
8433 (chkp_reduce_bounds_lifetime): Likewise.
8434 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
8435 (chkp_mark_completed_bounds): Likewise.
8436 (chkp_register_incomplete_bounds): Likewise.
8437 (chkp_mark_invalid_bounds): Likewise.
8438 (chkp_maybe_copy_and_register_bounds): Likewise.
8439 (chkp_build_returned_bound): Likewise.
8440 (chkp_get_bound_for_parm): Likewise.
8441 (chkp_build_bndldx): Likewise.
8442 (chkp_get_bounds_by_definition): Likewise.
8443 (chkp_generate_extern_var_bounds): Likewise.
8444 (chkp_get_bounds_for_decl_addr): Likewise.
8445 * tree-chrec.c (chrec_apply): Likewise.
8446 * tree-data-ref.c (dump_data_reference): Likewise.
8447 (dump_subscript): Likewise.
8448 (dump_data_dependence_relation): Likewise.
8449 (analyze_overlapping_iterations): Likewise.
8450 * tree-inline.c (expand_call_inline): Likewise.
8451 (tree_function_versioning): Likewise.
8452 * tree-into-ssa.c (dump_defs_stack): Likewise.
8453 (dump_currdefs): Likewise.
8454 (dump_names_replaced_by): Likewise.
8455 (dump_update_ssa): Likewise.
8456 (update_ssa): Likewise.
8457 * tree-object-size.c (pass_object_sizes::execute): Likewise.
8458 * tree-parloops.c (build_new_reduction): Likewise.
8459 (try_create_reduction_list): Likewise.
8460 (ref_conflicts_with_region): Likewise.
8461 (oacc_entry_exit_ok_1): Likewise.
8462 (oacc_entry_exit_single_gang): Likewise.
8463 * tree-pretty-print.h: Likewise.
8464 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
8465 (get_scalar_evolution): Likewise.
8466 (add_to_evolution): Likewise.
8467 (get_loop_exit_condition): Likewise.
8468 (analyze_evolution_in_loop): Likewise.
8469 (analyze_initial_condition): Likewise.
8470 (analyze_scalar_evolution): Likewise.
8471 (instantiate_scev): Likewise.
8472 (number_of_latch_executions): Likewise.
8473 (gather_chrec_stats): Likewise.
8474 (final_value_replacement_loop): Likewise.
8475 (scev_const_prop): Likewise.
8476 * tree-sra.c (dump_access): Likewise.
8477 (disqualify_candidate): Likewise.
8478 (create_access): Likewise.
8480 (maybe_add_sra_candidate): Likewise.
8481 (create_access_replacement): Likewise.
8482 (analyze_access_subtree): Likewise.
8483 (analyze_all_variable_accesses): Likewise.
8484 (sra_modify_assign): Likewise.
8485 (initialize_constant_pool_replacements): Likewise.
8486 (find_param_candidates): Likewise.
8487 (decide_one_param_reduction): Likewise.
8488 (replace_removed_params_ssa_names): Likewise.
8489 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
8490 * tree-ssa-copy.c (dump_copy_of): Likewise.
8491 (copy_prop_visit_cond_stmt): Likewise.
8492 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
8493 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
8494 (record_equivalences_from_stmt): Likewise.
8495 * tree-ssa-dse.c (compute_trims): Likewise.
8496 (delete_dead_call): Likewise.
8497 (delete_dead_assignment): Likewise.
8498 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
8499 (forward_propagate_into_cond): Likewise.
8500 (pass_forwprop::execute): Likewise.
8501 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8502 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
8504 (move_computations_worker): Likewise.
8505 (execute_sm): Likewise.
8506 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
8507 (remove_exits_and_undefined_stmts): Likewise.
8508 (remove_redundant_iv_tests): Likewise.
8509 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
8510 (adjust_iv_update_pos): Likewise.
8511 * tree-ssa-math-opts.c (bswap_replace): Likewise.
8512 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
8513 (value_replacement): Likewise.
8514 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
8515 * tree-ssa-pre.c (print_pre_expr): Likewise.
8516 (get_representative_for): Likewise.
8517 (create_expression_by_pieces): Likewise.
8518 (insert_into_preds_of_block): Likewise.
8519 (eliminate_insert): Likewise.
8520 (eliminate_dom_walker::before_dom_children): Likewise.
8521 (eliminate): Likewise.
8522 (remove_dead_inserted_code): Likewise.
8523 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
8524 * tree-ssa-reassoc.c (get_rank): Likewise.
8525 (eliminate_duplicate_pair): Likewise.
8526 (eliminate_plus_minus_pair): Likewise.
8527 (eliminate_not_pairs): Likewise.
8528 (undistribute_ops_list): Likewise.
8529 (eliminate_redundant_comparison): Likewise.
8530 (update_range_test): Likewise.
8531 (optimize_range_tests_var_bound): Likewise.
8532 (optimize_vec_cond_expr): Likewise.
8533 (rewrite_expr_tree): Likewise.
8534 (rewrite_expr_tree_parallel): Likewise.
8535 (linearize_expr): Likewise.
8536 (break_up_subtract): Likewise.
8537 (linearize_expr_tree): Likewise.
8538 (attempt_builtin_powi): Likewise.
8539 (attempt_builtin_copysign): Likewise.
8540 (transform_stmt_to_copy): Likewise.
8541 (transform_stmt_to_multiply): Likewise.
8542 (dump_ops_vector): Likewise.
8543 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
8544 (print_scc): Likewise.
8545 (set_ssa_val_to): Likewise.
8546 (visit_reference_op_store): Likewise.
8547 (visit_use): Likewise.
8548 (sccvn_dom_walker::before_dom_children): Likewise.
8549 (run_scc_vn): Likewise.
8550 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
8552 (expr_hash_elt::print): Likewise.
8553 (const_and_copies::pop_to_marker): Likewise.
8554 (const_and_copies::record_const_or_copy_raw): Likewise.
8555 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
8556 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
8557 (dump_predicates): Likewise.
8558 (find_uninit_use): Likewise.
8559 (warn_uninitialized_phi): Likewise.
8560 (pass_late_warn_uninitialized::execute): Likewise.
8561 * tree-ssa.c (verify_vssa): Likewise.
8562 (verify_ssa): Likewise.
8563 (maybe_optimize_var): Likewise.
8564 * tree-vrp.c (dump_value_range): Likewise.
8565 (dump_all_value_ranges): Likewise.
8566 (dump_asserts_for): Likewise.
8567 (register_edge_assert_for_2): Likewise.
8568 (vrp_visit_cond_stmt): Likewise.
8569 (vrp_visit_switch_stmt): Likewise.
8570 (vrp_visit_stmt): Likewise.
8571 (vrp_visit_phi_node): Likewise.
8572 (simplify_cond_using_ranges_1): Likewise.
8573 (fold_predicate_in): Likewise.
8574 (evrp_dom_walker::before_dom_children): Likewise.
8575 (evrp_dom_walker::push_value_range): Likewise.
8576 (evrp_dom_walker::pop_value_range): Likewise.
8577 (execute_early_vrp): Likewise.
8579 2017-05-16 Richard Biener <rguenther@suse.de>
8581 * dwarf2out.c (loc_list_from_tree_1): Do not create
8582 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
8584 2017-05-16 Richard Biener <rguenther@suse.de>
8586 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
8588 (note_variable_value_in_expr): If we resolved the decl ref
8589 do not push to the stack.
8591 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
8593 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
8594 operations in fast-math mode.
8595 (vaddq_f16): Likewise.
8596 (vmul_f16): Likewise.
8597 (vmulq_f16): Likewise.
8598 (vsub_f16): Likewise.
8599 (vsubq_f16): Likewise.
8600 * config/arm/neon.md (add<mode>3): New.
8602 (fma:<VH:mode>3): New. Also remove outdated comment.
8605 2017-05-16 Martin Liska <mliska@suse.cz>
8609 * ipa-devirt.c (warn_types_mismatch): Fix typo.
8610 (odr_types_equivalent_p): Likewise.
8612 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
8614 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
8616 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
8619 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
8620 non-interunit SSE move alternatives with '?'.
8621 (zero-extendsidi peephole2): New peephole to skip intermediate
8622 general register in SSE zero-extend sequence.
8624 2017-05-15 Jeff Law <law@redhat.com>
8626 * reorg.c (relax_delay_slots): Create a new variable to hold
8627 the temporary target rather than clobbering TARGET_LABEL.
8629 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
8630 missing argument to extract_bit_field call.
8631 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
8633 2017-05-15 Martin Liska <mliska@suse.cz>
8636 * gcc.c (process_command): Do not allow empty argument of -o option.
8638 2017-05-15 Renlin Li <renlin.li@arm.com>
8640 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
8641 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
8642 * config/aarch64/constraints.md (Usf): Add long call check.
8643 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
8644 (call_value): Likewise.
8645 (sibcall): Likewise.
8646 (sibcall_value): Likewise.
8648 (call_value_insn): New.
8649 (sibcall_insn): Update rtx pattern.
8650 (sibcall_value_insn): Likewise.
8651 (call_internal): Remove.
8652 (call_value_internal): Likewise.
8653 (sibcall_internal): Likewise.
8654 (sibcall_value_internal): Likewise.
8655 (call_reg): Likewise.
8656 (call_symbol): Likewise.
8657 (call_value_reg): Likewise.
8658 (call_value_symbol): Likewise.
8660 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
8663 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
8665 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
8667 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
8668 compatible with CCGOCmode and with CCZmode.
8670 2017-05-14 Martin Sebor <msebor@redhat.com>
8673 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
8674 (gimple_fold_builtin_snprintf): Same.
8675 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
8676 (gimple_fold_builtin_snprintf): Same.
8677 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
8679 (is_call_safe): New function.
8680 (try_substitute_return_value): Call it.
8681 (try_simplify_call): New function.
8682 (pass_sprintf_length::handle_gimple_call): Call it.
8684 2017-05-14 Martin Sebor <msebor@redhat.com>
8687 * builtins.c (expand_builtin_stpncpy): Simplify.
8689 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
8691 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
8692 * config/i386/i386.h
8693 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
8694 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
8695 (struct machine_function): Add new members call_ms2sysv,
8696 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
8697 (struct machine_frame_state): New fields sp_realigned and
8698 sp_realigned_offset.
8699 * config/i386/i386.c
8700 (enum xlogue_stub): New enum.
8701 (enum xlogue_stub_sets): New enum.
8702 (class xlogue_layout): New class.
8703 (struct ix86_frame): New fields stack_realign_allocate_offset,
8704 stack_realign_offset and outlined_save_offset. Modify comments to
8705 detail stack layout when using out-of-line stubs.
8706 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
8707 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
8708 -mcall-ms2sysv-xlogues.
8709 (stub_managed_regs): New static variable.
8710 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
8711 registers managed by out-of-line stub.
8712 (disable_call_ms2sysv_xlogues): New function.
8713 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
8714 m->call_ms2sysv when appropriate and compute frame layout for
8716 (sp_valid_at, fp_valid_at): New inline functions.
8717 (choose_basereg): New function.
8718 (choose_baseaddr): Add align parameter, use choose_basereg and modify
8720 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
8721 Use align parameter of choose_baseaddr to generated aligned SSE movs
8723 (pro_epilogue_adjust_stack): Modify to track
8724 machine_frame_state::sp_realigned.
8725 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
8726 (ix86_nsaved_sseregs): Likewise.
8727 (ix86_emit_save_regs): Likewise.
8728 (ix86_emit_save_regs_using_mov): Likewise.
8729 (ix86_emit_save_sse_regs_using_mov): Likewise.
8730 (get_scratch_register_on_entry): Likewise.
8731 (gen_frame_set): New function.
8732 (gen_frame_load): Likewise.
8733 (gen_frame_store): Likewise.
8734 (emit_outlined_ms2sysv_save): Likewise.
8735 (emit_outlined_ms2sysv_restore): Likewise.
8736 (ix86_expand_prologue): Modify stack re-alignment code and call
8737 emit_outlined_ms2sysv_save when appropriate.
8738 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
8739 parameter rtx_insn *insn, which allows the function to be used to only
8741 (ix86_expand_epilogue): Modify validity checks of frame and stack
8742 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
8743 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
8744 * config/i386/predicates.md
8745 (save_multiple): New predicate.
8746 (restore_multiple): Likewise.
8747 * config/i386/sse.md
8748 (save_multiple<mode>): New pattern.
8749 (save_multiple_realign<mode>): Likewise.
8750 (restore_multiple<mode>): Likewise.
8751 (restore_multiple_and_return<mode>): Likewise.
8752 (restore_multiple_leave_return<mode>): Likewise.
8753 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
8755 2017-05-14 Julia Koval <julia.koval@intel.com>
8757 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
8758 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
8759 (__builtin_ia32_xsetbv): New builtins.
8760 * config/i386/i386.c (ix86_expand_special_args_builtin):
8762 (ix86_expand_builtin): Special expand for new intrinsics.
8763 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
8764 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
8765 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
8767 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8769 * cfganal.c (inverted_post_order_compute): Change argument type
8771 * cfganal.h (inverted_post_order_compute): Adjust prototype.
8772 * df-core.c (rest_of_handle_df_initialize): Adjust.
8773 (rest_of_handle_df_finish): Likewise.
8774 (df_analyze_1): Likewise.
8775 (df_analyze): Likewise.
8776 (loop_inverted_post_order_compute): Change argument to be a vec *.
8777 (df_analyze_loop): Adjust.
8778 (df_get_n_blocks): Likewise.
8779 (df_get_postorder): Likewise.
8780 * df.h (struct df_d): Change field to be a vec.
8781 * lcm.c (compute_laterin): Adjust.
8782 (compute_available): Likewise.
8783 * lra-lives.c (lra_create_live_ranges_1): Likewise.
8784 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
8785 * tree-ssa-pre.c (compute_antic): Likewise.
8787 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8789 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
8790 (depth_first_search::depth_first_search): Change structure init
8791 function to this constructor.
8792 (depth_first_search::add_bb): Rename function to this member.
8793 (depth_first_search::execute): Likewise.
8794 (flow_dfs_compute_reverse_finish): Adjust.
8796 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8798 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
8799 (longest_simple_path): Likewise.
8800 * shrink-wrap.c (spread_components): Likewise.
8801 (disqualify_problematic_components): Likewise.
8802 (emit_common_heads_for_components): Likewise.
8803 (emit_common_tails_for_components): Likewise.
8804 (insert_prologue_epilogue_for_components): Likewise.
8806 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8808 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
8811 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8813 * df-core.c (df_set_blocks): Start using auto_bitmap.
8814 (df_compact_blocks): Likewise.
8815 * df-problems.c (df_rd_confluence_n): Likewise.
8816 * df-scan.c (df_insn_rescan_all): Likewise.
8817 (df_process_deferred_rescans): Likewise.
8818 (df_update_entry_block_defs): Likewise.
8819 (df_update_exit_block_uses): Likewise.
8820 (df_entry_block_bitmap_verify): Likewise.
8821 (df_exit_block_bitmap_verify): Likewise.
8822 (df_scan_verify): Likewise.
8823 * lra-constraints.c (lra_constraints): Likewise.
8824 (undo_optional_reloads): Likewise.
8825 (lra_undo_inheritance): Likewise.
8826 * lra-remat.c (calculate_gen_cands): Likewise.
8827 (do_remat): Likewise.
8828 * lra-spills.c (assign_spill_hard_regs): Likewise.
8829 (spill_pseudos): Likewise.
8830 * tree-ssa-pre.c (bitmap_set_and): Likewise.
8831 (bitmap_set_subtract_values): Likewise.
8833 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8835 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
8836 management with auto_bitmap.
8837 (fix_inter_tick): Likewise.
8838 (fix_recovery_deps): Likewise.
8839 * ira.c (add_store_equivs): Likewise.
8840 (find_moveable_pseudos): Likewise.
8841 (split_live_ranges_for_shrink_wrap): Likewise.
8842 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
8843 (rtx_reuse_manager::seen_def_p): Likewise.
8844 (rtx_reuse_manager::set_seen_def): Likewise.
8845 * print-rtl.h (class rtx_reuse_manager): Likewise.
8847 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8849 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
8851 (migrate_btr_def): Likewise.
8852 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
8853 * df-core.c (loop_post_order_compute): Likewise.
8854 (loop_inverted_post_order_compute): Likewise.
8855 * hsa-common.h: Likewise.
8856 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
8857 * init-regs.c (initialize_uninitialized_regs): Likewise.
8858 * ipa-inline.c (resolve_noninline_speculation): Likewise.
8859 (inline_small_functions): Likewise.
8860 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
8861 * ira.c (combine_and_move_insns): Likewise.
8862 (build_insn_chain): Likewise.
8863 * loop-invariant.c (find_invariants): Likewise.
8864 * lower-subreg.c (propagate_pseudo_copies): Likewise.
8865 * predict.c (tree_predict_by_opcode): Likewise.
8866 (predict_paths_leading_to): Likewise.
8867 (predict_paths_leading_to_edge): Likewise.
8868 (estimate_loops_at_level): Likewise.
8869 (estimate_loops): Likewise.
8870 * shrink-wrap.c (try_shrink_wrapping): Likewise.
8871 (spread_components): Likewise.
8872 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
8873 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
8874 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
8875 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
8876 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
8877 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
8878 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
8879 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
8880 (mark_threaded_blocks): Likewise.
8881 (thread_through_all_blocks): Likewise.
8882 * tree-ssa.c (verify_ssa): Likewise.
8883 (execute_update_addresses_taken): Likewise.
8884 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
8886 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8888 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
8890 (post_order_compute): Likewise.
8891 (inverted_post_order_compute): Likewise.
8892 (pre_and_rev_post_order_compute_fn): Likewise.
8894 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8896 * genrecog.c (int_set::int_set): Explicitly construct our
8897 auto_vec base class.
8898 * vec.h (auto_vec::auto_vec): New constructor.
8900 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8902 * bitmap.h (class auto_bitmap): New constructor taking
8903 bitmap_obstack * argument.
8905 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8907 * bitmap.h (class auto_bitmap): Change type of m_bits to
8908 bitmap_head, and adjust ctor / dtor and member operators.
8910 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
8912 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
8913 when returned register mode doesn't match original mode.
8915 2017-05-12 Jeff Law <law@redhat.com>
8916 Jakub Jelinek <jakub@redhat.com>
8918 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
8919 we look for cc setter after the compare-elim changes.
8920 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
8921 within the vector to match what compare-elim now expects.
8922 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
8923 (xorsi3_flags, one_cmplsi2_flags): Likewise.
8925 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
8926 after the compare-elim changes.
8927 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
8928 the vector to match what compare-elim now expects.
8929 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
8930 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
8931 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
8932 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
8933 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
8935 * config/visium/visium.c (single_set_and_flags): Fix where
8936 we look for cc setter after the compare-elim changes.
8937 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
8938 with the vector to match what compare-elim now expects.
8939 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
8940 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
8941 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
8942 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
8943 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
8944 (neg<mode>2_insn_set_overflow): Likewise.
8946 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
8949 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
8950 maybe_expand_insn call, set ops[0].target. If still set after call,
8951 set alt_rtl. Add extra arg to recursive calls.
8952 (extract_bit_field): Add alt_rtl argument. Pass to
8954 * expmed.h (extract_bit_field): Fix prototype.
8955 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
8956 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
8957 to extract_bit_field_calls.
8958 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
8959 Pass alt_rtl to extract_bit_field calls.
8960 * calls.c (store_unaligned_arguments_into_psuedos)
8961 load_register_parameters): Pass extra NULL to extract_bit_field calls.
8962 * optabs.c (maybe_legitimize_operand): Clear op->target when call
8964 * optabs.h (struct expand_operand): Add target bitfield.
8966 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
8968 * compare-elim.c (try_eliminate_compare): Canonicalize
8969 operation with embedded compare to
8970 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
8971 (set (reg) (operation)].
8973 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
8975 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
8978 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
8979 cost of adding a carry flag for ADC instruction.
8980 [case MINUS]: Ignore the cost of subtracting a carry flag
8981 for SBB instruction.
8983 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
8985 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
8987 * config/rs6000/bmiintrin.h: New file.
8988 * config/rs6000/bmi2intrin.h: New file.
8989 * config/rs6000/x86intrin.h: New file.
8991 2017-05-12 Jeff Law <law@redhat.com>
8993 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
8996 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
8999 * tree-cfg.c: Remove cfg edges of unreachable case statements.
9001 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9003 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
9004 early expansion of vector divide builtins.
9005 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
9006 builtins identified as having unsigned arguments.
9008 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9010 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
9011 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
9012 expansion of vector logical operations (and, andc, or, xor,
9015 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9017 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
9018 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
9020 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9022 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
9023 early GIMPLE expansion of vector multiplies.
9025 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9027 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
9028 TARGET_HAVE_MOVT conditional.
9029 (movt splitter): Likewise.
9031 2017-05-12 Richard Biener <rguenther@suse.de>
9033 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
9034 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9035 Fold all stmts not inplace.
9037 2017-05-12 Richard Biener <rguenther@suse.de>
9039 PR tree-optimization/80713
9040 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
9041 inserted_exprs bit for not removed stmts.
9043 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
9046 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
9047 parallelized" attribute for parallelized OpenACC kernels.
9048 * omp-offload.c (execute_oacc_device_lower): Use it.
9050 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
9051 Set "oacc kernels" attribute.
9052 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
9053 parameter. Adjust all users.
9054 (oacc_fn_attrib_kernels_p): Remove function.
9055 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
9056 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
9057 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
9058 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
9059 assert "oacc kernels" attribute is set.
9061 2017-05-11 Carl Love <cel@us.ibm.com>
9063 * config/rs6000/rs6000-c: Add support for built-in functions
9064 vector unsigned char vec_popcnt (vector signed char)
9065 vector unsigned char vec_popcnt (vector unsigned char)
9066 vector unsigned short vec_popcnt (vector signed short)
9067 vector unsigned short vec_popcnt (vector unsigned short)
9068 vector unsigned int vec_popcnt (vector signed int)
9069 vector unsigned int vec_popcnt (vector unsigned int)
9070 vector unsigned long long vec_popcnt (vector signed long long)
9071 vector unsigned long long vec_popcnt (vector unsigned long long)
9072 vector signed long long vec_slo (vector signed long long,
9074 vector signed long long vec_slo (vector signed long long,
9075 vector unsigned char)
9076 vector unsigned long long vec_slo (vector unsigned long long,
9078 vector unsigned long long vec_slo (vector unsigned long long,
9079 vector unsigned char)
9080 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
9081 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
9082 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
9083 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
9084 * doc/extend.texi: Update the built-in documentation file for the
9085 new built-in functions.
9087 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
9089 * attribs.h (sorted_attr_string): Move machine independent
9090 functions for target clone support from the i386 port to common
9091 code. Rename ix86_function_versions to common_function_versions.
9092 Rename make_name to make_unique_name.
9093 (common_function_versions): Likewise.
9094 (make_unique_name): Likewise.
9095 (make_dispatcher_decl): Likewise.
9096 (is_function_default_version): Likewise.
9097 * attribs.c (attr_strcmp): Likewise.
9098 (sorted_attr_string): Likewise.
9099 (common_function_versions): Likewise.
9100 (make_unique_name): Likewise.
9101 (make_dispatcher_decl): Likewise.
9102 (is_function_default_version): Likewise.
9103 * config/i386/i386.c (attr_strcmp): Likewise.
9104 (sorted_attr_string): Likewise.
9105 (ix86_function_versions): Likewise.
9106 (make_name): Likewise.
9107 (make_dispatcher_decl): Likewise.
9108 (is_function_default_version): Likewise.
9109 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
9111 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9114 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
9115 Account for direct move costs for vec_construct of integer
9118 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
9121 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
9122 (UNSPEC_STX_ATOMIC): Ditto.
9123 (loaddi_via_sse): New insn.
9124 (storedi_via_sse): Ditto.
9125 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
9126 Update corresponding peephole2 patterns.
9127 (atomic_storedi_fpu): Ditto.
9129 2017-05-11 Julia Koval <julia.koval@intel.com>
9131 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
9132 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
9134 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
9135 (__builtin_ia32_rsqrt14ss_mask): New builtins.
9136 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
9138 2017-05-11 Nathan Sidwell <nathan@acm.org>
9140 * graphite-poly.c: Include dumpfile.h.
9142 * dumpfle.h (dump_function): Declare here ...
9143 * tree-dump.h (dump_function): ... not here.
9144 * dumpfile.c: #include tree-cfg.h.
9145 (dump_function): Move here from ...
9146 * tree-dump.c (dump_function): ... here.
9147 * gimplify.c: #include splay-tree.h, not tree-dump.h.
9148 * graphite-poly.c: Don't include tree-dump.h.
9149 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
9150 * print-tree.c: Likewise.
9151 * stor-layout.c: Likewise.
9152 * tree-nested.c: Likewise.
9154 * dumpfile.c (dump_start): Use TDF_FLAGS.
9155 (dump_enable_all): Fix TDF_KIND check thinko.
9157 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
9159 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9160 array entries to represent two legal parameterizations of the
9161 overloaded __builtin_cmpb function, as represented by the
9162 P6_OV_BUILTIN_CMPB constant.
9163 (altivec_resolve_overloaded_builtin): Add special case handling
9164 for the __builtin_cmpb function, as represented by the
9165 P6_OV_BUILTIN_CMPB constant.
9166 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
9167 (BU_P6_64BIT_2): New macro.
9168 (BU_P6_OVERLOAD_2): New macro
9169 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
9170 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
9171 (CMPB): Add overload support to represent both 32-bit and 64-bit
9172 compare-bytes function.
9173 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
9174 support for TARGET_CMPB.
9175 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
9176 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
9177 documentation of the __builtin_cmpb overloaded built-in function.
9179 2017-05-11 Richard Biener <rguenther@suse.de>
9181 PR tree-optimization/80705
9182 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
9183 bases are not vectorizable.
9185 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9187 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
9188 when counting register pressure.
9190 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9192 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
9193 (iv_ca_more_deps): Renamed to ...
9194 (iv_ca_compare_deps): ... this.
9195 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
9197 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9199 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
9201 (determine_group_iv_costs): ... here.
9202 (find_inv_vars_cb): Record inv var if it's not recorded before.
9204 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9206 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
9207 (get_shiftadd_cost): Ditto.
9209 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9211 * tree-ssa-address.c: Include header file.
9212 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
9214 (add_to_parts): Refactor.
9215 (addr_to_parts): New parameter. Update use of move_hint_to_base.
9216 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
9219 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9221 PR tree-optimization/53090
9222 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
9224 (extract_cond_operands): Detect condition with IV on both sides
9225 and return COMP_IV_EXPR_2.
9226 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
9227 (rewrite_use_compare): Simplify by removing call to function
9228 extract_cond_operands.
9230 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9232 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
9233 (extract_cond_operands): Detect condition comparing against non-
9234 invariant bound and return appropriate enum value.
9235 (find_interesting_uses_cond): Update use of extract_cond_operands.
9236 Handle its return value accordingly.
9237 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
9239 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9241 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
9242 nonlinear iv_use computation in loop invariant sensitive way.
9244 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9246 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
9247 (find_iv_candidates): Call relate_compare_use_with_all_cands.
9249 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9251 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
9252 (dump_cand): Support iv_cand.inv_exprs.
9253 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
9255 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
9258 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9260 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
9262 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
9263 as local function. Include necessary header files.
9264 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
9266 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9268 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
9270 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9272 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
9273 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
9274 RSHIFT_EXPR and BIT_NOT_EXPR.
9276 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9278 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
9279 (adjust_setup_cost): New parameter supporting round up adjustment.
9280 (struct address_cost_data): Delete.
9281 (force_expr_to_var_cost): Don't bound cost with spill_cost.
9282 (split_address_cost, ptr_difference_cost): Delete.
9283 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
9284 (struct ainc_cost_data): New struct.
9285 (get_address_cost_ainc): New function.
9286 (get_address_cost, get_computation_cost): Reimplement.
9287 (determine_group_iv_cost_address): Record inv_expr for all uses of
9289 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
9290 (iv_ca_has_deps): Reimplemented to ...
9291 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
9293 (iv_ca_extend): Call iv_ca_more_deps.
9295 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9297 * tree-ssa-address.c (struct mem_address): Move to header file.
9298 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
9299 * tree-ssa-address.h (struct mem_address): Move from C file.
9300 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
9302 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9304 * tree-affine.h (aff_combination_type): New interface.
9305 (aff_combination_zero_p): Remove static.
9306 (aff_combination_const_p): New interface.
9307 (aff_combination_singleton_var_p): New interfaces.
9309 2017-05-11 Richard Biener <rguenther@suse.de>
9311 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9312 Skip unreachable blocks and destinations.
9313 (eliminate): Move stmt removal and fixup ...
9314 (fini_eliminate): ... here. Skip inserted exprs.
9315 (pass_pre::execute): Move fini_pre after fini_eliminate.
9316 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
9317 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
9318 PRE to get rid of dead code that has invalid SSA form and
9319 split critical edges again.
9321 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9323 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
9325 2017-05-11 Richard Biener <rguenther@suse.de>
9327 * passes.c (execute_function_todo): Verify loops if they are
9328 said to be up-to-date.
9329 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
9330 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
9332 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
9335 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
9336 handle calling assemble_external ourself.
9339 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
9340 modes with zero size. Enhance comment.
9342 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9344 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
9345 built-ins for vec_xl and vec_xst with short and char pointer
9348 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
9350 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
9351 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
9352 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
9353 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
9354 (_mm_maskz_min_round_ss): New intrinsics.
9355 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
9356 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
9357 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
9358 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
9359 (__builtin_ia32_minss_mask_round): New builtins.
9360 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9361 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9362 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
9364 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
9365 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
9367 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
9370 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
9372 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
9373 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
9374 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
9375 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
9376 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
9377 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
9378 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
9379 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9380 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
9381 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
9382 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
9383 (__builtin_ia32_mulss_mask_round): New builtins.
9384 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9385 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9386 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
9388 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
9389 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
9391 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
9394 2017-05-10 Julia Koval <julia.koval@intel.com>
9396 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
9397 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
9398 (_mm256_setr_m128i): New intrinsics.
9400 2017-05-10 Julia Koval <julia.koval@intel.com>
9402 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
9403 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
9404 (_mm_maskz_rcp14_ss): New intrinsics.
9405 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
9406 (__builtin_ia32_rcp14ss_mask): New builtins.
9407 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
9409 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
9411 PR tree-optimization/51513
9412 * tree-cfg.c (gimple_seq_unreachable_p): New function.
9413 (assert_unreachable_fallthru_edge_p): Use it.
9414 (group_case_labels_stmt): Likewise.
9415 * tree-cfg.h: Prototype it.
9416 * stmt.c: Include cfghooks.h and tree-cfg.h.
9417 (emit_case_dispatch_table) <gap_label>: New local variable.
9418 Use it to fill dispatch table gaps.
9419 Test for default_label before updating probabilities.
9420 (expand_case) <default_label>: Remove unneeded initialization.
9421 Test for unreachable default case statement and remove its edge.
9422 Set default_label accordingly.
9423 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
9425 2017-05-10 Carl Love <cel@us.ibm.com>
9427 * config/rs6000/rs6000-c: Add support for built-in functions
9428 vector signed char vec_neg (vector signed char)
9429 vector signed short int vec_neg (vector short int)
9430 vector signed int vec_neg (vector signed int)
9431 vector signed long long vec_neg (vector signed long long)
9432 vector float vec_neg (vector float)
9433 vector double vec_neg (vector double)
9434 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
9436 * config/rs6000/altivec.h: Add define for vec_neg
9437 * doc/extend.texi: Update the built-in documentation for the
9438 new built-in functions.
9440 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9442 PR tree-optimization/77644
9443 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
9445 2017-05-10 Nathan Sidwell <nathan@acm.org>
9447 * dumpfile.h (TDI_lang_all): New.
9448 (TDF_KIND): New. Renumber others
9449 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
9451 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
9453 (get_dump_file_name): Adjust suffix generation.
9454 (dump_enable_all): Use TDF_KIND.
9455 * doc/invoke.texi (-fdump-lang-all): Document.
9457 * dumpfile.h: Tabify.
9459 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
9462 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
9463 Move member access before delete.
9465 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
9467 * tree-inline.c (expand_call_inline): Split block at stmt
9470 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9473 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
9474 are now unused after splitting mov{sf,sd}_hardfloat.
9475 (f32_lr2): Likewise.
9477 (f32_lm2): Likewise.
9479 (f32_li2): Likewise.
9482 (f32_sr2): Likewise.
9484 (f32_sm2): Likewise.
9486 (f32_si2): Likewise.
9489 (f32_vsx): Likewise.
9491 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
9492 For movsf, order stores so the VSX stores occur before the GPR
9493 store which encourages the register allocator to use a traditional
9494 FPR instead of a GPR. For movsd, order the stores so that the GPR
9495 store comes before the VSX stores to allow the power6 to work.
9496 This is due to the power6 not having a 32-bit integer store
9497 instruction from a FPR.
9498 (movsf_hardfloat): Likewise.
9499 (movsd_hardfloat): Likewise.
9501 2017-05-09 Martin Sebor <msebor@redhat.com>
9503 PR translation/80280
9504 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
9507 PR translation/80280
9508 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
9509 data member added in r247778.
9510 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
9512 2017-05-09 Nathan Sidwell <nathan@acm.org>
9514 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
9516 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
9519 2017-05-09 Marek Polacek <polacek@redhat.com>
9521 * doc/invoke.texi: Fix typo.
9523 2017-05-09 Richard Biener <rguenther@suse.de>
9525 * tree-vrp.c (vrp_val_is_max): Adjust comment.
9526 (vrp_val_is_min): Likewise.
9527 (set_value_range_to_value): Likewise.
9528 (set_value_range_to_nonnegative): Likewise.
9529 (gimple_assign_nonzero_p): Likewise.
9530 (gimple_stmt_nonzero_p): Likewise.
9531 (vrp_int_const_binop): Likewise. Remove unreachable case.
9532 (adjust_range_with_scev): Adjust comments.
9533 (compare_range_with_value): Likewise.
9534 (extract_range_from_phi_node): Likewise.
9535 (test_for_singularity): Likewise.
9537 2017-05-09 Richard Biener <rguenther@suse.de>
9539 * tree-vrp.c (get_single_symbol): Add assert that we don't
9540 get overflowed constants as invariant part.
9541 (compare_values_warnv): Add comment before the TREE_NO_WARNING
9542 checks. Use wi::cmp instead of recursing for integer constants.
9543 (compare_values): Just ignore whether we assumed undefined
9544 overflow instead of failing the compare.
9545 (extract_range_for_var_from_comparison_expr): Add comment before the
9546 TREE_NO_WARNING sets.
9547 (test_for_singularity): Likewise.
9548 (extract_range_from_comparison): Do not disable optimization
9549 when we assumed undefined overflow.
9550 (extract_range_basic): Remove init of unused var.
9552 2017-05-09 Richard Biener <rguenther@suse.de>
9554 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
9555 (extract_range_from_multiplicative_op_1): Adjust.
9556 (extract_range_from_binary_expr_1): Use int_const_binop.
9558 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
9561 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
9562 rs6000_store_data_bypass_p in seven define_bypass directives and
9563 in several comments.
9564 * config/rs6000/rs6000-protos.h: Add prototype for
9565 rs6000_store_data_bypass_p function.
9566 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
9567 function implements slightly different (rs6000-specific) semantics
9568 than store_data_bypass_p, returning false rather than aborting
9569 with assertion error when arguments do not satisfy the
9570 requirements of store data bypass.
9571 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
9572 rs6000_store_data_bypass_p.
9574 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
9576 * config/xtensa/xtensa-protos.h
9577 (xtensa_initial_elimination_offset): New declaration.
9578 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
9579 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
9580 macro definition, add case for FRAME_POINTER_REGNUM when
9581 FRAME_GROWS_DOWNWARD.
9582 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
9583 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
9584 xtensa_initial_elimination_offset.
9586 2017-05-08 Nathan Sidwell <nathan@acm.org>
9588 * doc/invoke.texi: Alphabetize -fdump options.
9590 2017-05-08 Martin Sebor <msebor@redhat.com>
9592 PR translation/80280
9593 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
9595 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
9597 * target.def (compute_frame_layout): New optional target hook.
9598 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
9599 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
9600 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
9602 * reload1.c (verify_initial_elim_offsets): Likewise.
9603 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
9604 (use_simple_return_p): Call arm_compute_frame_layout if needed.
9605 (arm_get_frame_offsets): Split up into this ...
9606 (arm_compute_frame_layout): ... and this function.
9608 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
9610 * config/aarch64/constraints.md (Usa): New constraint.
9611 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
9613 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
9615 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
9616 with_multilib_list after it has been checked.
9618 2017-05-08 Richard Biener <rguenther@suse.de>
9620 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
9621 (bitmap_set_subtract_values): Likewise.
9623 2017-05-08 Richard Biener <rguenther@suse.de>
9625 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
9626 (gimple_assign_nonzero): ... this and remove strict_overflow_p
9628 (gimple_stmt_nonzero_warnv_p): Rename to ...
9629 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
9631 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
9632 (extract_range_basic): Adjust, do not disable propagation on
9633 strict overflow sensitive simplification.
9634 (vrp_visit_cond_stmt): Likewise.
9636 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
9638 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
9639 body size unconditionally.
9641 2017-05-07 Jeff Law <law@redhat.com>
9644 2017-05-06 Jeff Law <law@redhat.com>
9645 PR tree-optimization/78496
9646 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9649 PR tree-optimization/78496
9650 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9651 (simplify_stmt_using_ranges): Call it.
9652 (vrp_dom_walker::before_dom_children): Extract equivalences
9653 from an ASSERT_EXPR with an equality comparison against a
9656 2017-05-06 Jeff Law <law@redhat.com>
9658 PR tree-optimization/78496
9659 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9662 PR tree-optimization/78496
9663 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9664 (simplify_stmt_using_ranges): Call it.
9665 (vrp_dom_walker::before_dom_children): Extract equivalences
9666 from an ASSERT_EXPR with an equality comparison against a
9669 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9671 * lra-constraints.c (lra_copy_reg_equiv): New function.
9672 (split_reg): Use it to copy equivalence information from the
9673 original register to the spill register.
9675 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9677 PR rtl-optimization/75964
9678 * simplify-rtx.c (simplify_const_relational_operation): Remove
9679 invalid handling of comparisons of integer ABS.
9681 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
9683 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
9685 (init_regs): Remove declaration.
9686 (function_arg_advance_32): Initialize error_p as boolean variable.
9688 2017-05-05 Nathan Sidwell <nathan@acm.org>
9690 * store-motion.c (remove_reachable_equiv_notes): Reformat long
9691 lines. Use for (;;).
9693 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9695 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
9696 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
9697 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
9698 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
9699 VF=2 that require versioning.
9701 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9703 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
9706 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9708 * diagnostic.h (diagnostic_override_option_index): Convert from
9709 macro to inline function.
9711 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9713 * diagnostic.c (last_module_changed_p): New function.
9714 (set_last_module): New function.
9715 (diagnostic_report_current_module): Convert macro usage to
9716 the above functions.
9717 * diagnostic.h (diagnostic_context::last_module): Strengthen
9718 from const line_map * to const line_map_ordinary *.
9719 (diagnostic_last_module_changed): Delete macro.
9720 (diagnostic_set_last_module): Delete macro.
9722 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9724 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
9725 with diagnostic_report_diagnostic.
9726 (diagnostic_n_impl_richloc): Likewise.
9727 * diagnostic.h (report_diagnostic): Delete macro.
9728 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
9729 with diagnostic_report_diagnostic.
9730 * substring-locations.c (format_warning_va): Likewise.
9732 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9734 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
9735 save/restor of format_spec. Move option-printing code to...
9736 (print_option_information): ...this new function, and
9737 reimplement by simply printing to the pretty_printer,
9738 rather than appending to the format string.
9740 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9742 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
9743 handling logic into...
9744 (update_effective_level_from_pragmas): ...this new function.
9746 2017-05-04 Andrew Waterman <andrew@sifive.com>
9748 * config/riscv/riscv.opt (mstrict-align): New option.
9749 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
9750 (SLOW_UNALIGNED_ACCESS): Define.
9751 (riscv_slow_unaligned_access): Declare.
9752 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
9754 (riscv_slow_unaligned_access): New variable.
9755 (rocket_tune_info): Set slow_unaligned_access to true.
9756 (optimize_size_tune_info): Set slow_unaligned_access to false.
9757 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
9758 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
9759 (riscv_option_override): Set riscv_slow_unaligned_access.
9760 * doc/invoke.texi: Add -mstrict-align to RISC-V.
9762 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
9764 * config/riscv/riscv.md: Unify indentation.
9766 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
9771 * config/rs6000/rs6000.md (u code attribute): Add FIX and
9773 (extendsi<mode>2): Add support for doing sign extension via
9774 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
9775 don't have ISA 3.0 instructions.
9776 (extendsi<mode>2 splitter): Likewise.
9777 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
9778 generate the normal insns since SImode can now go in vector
9779 registers. Disallow the special UNSPECs needed for previous
9780 machines to hide SImode being used. Add new insns
9781 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
9782 (fix_trunc<mode>si2_stfiwx): Likewise.
9783 (fix_trunc<mode>si2_internal): Likewise.
9784 (fixuns_trunc<mode>si2): Likewise.
9785 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9786 (fctiw<u>z_<mode>_smallint): Likewise.
9787 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
9788 of floating point to 32-bit integer from doing a direct move to
9789 the GPR registers to do a store.
9790 (fctiwz_<mode>): Break long line.
9792 2017-05-05 Bin Cheng <bin.cheng@arm.com>
9794 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
9795 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
9796 (addr_list, addr_offset_valid_p): New.
9797 (split_address_groups): Check offset validity with above function.
9798 (gt-tree-ssa-loop-ivopts.h): Include header file.
9800 2017-05-05 Nathan Sidwell <nathan@acm.org>
9802 * config.gcc (arm*-*-*): Add missing 'fi'.
9804 2017-05-05 Steve Ellcey <sellcey@cavium.com>
9806 * doc/invoke.texi (-fopt-info): Explicitly say order of options
9807 included in -fopt-info does not matter.
9808 * doc/optinfo.texi (-fopt-info): Fix description of default
9809 behavour. Explicitly say order of options included in -fopt-info
9812 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
9814 * config.gcc: Allow combinations of aprofile and rmprofile values for
9815 --with-multilib-list.
9816 * config/arm/t-multilib: New file.
9817 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
9818 variables. Remove setting of ISA and floating-point ABI in
9819 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
9820 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
9821 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
9822 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
9824 * config/arm/t-rmprofile: Likewise except for the matches changes.
9825 * doc/install.texi (--with-multilib-list): Document the combination of
9826 aprofile and rmprofile values and warn about pitfalls in doing that.
9828 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9830 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
9831 (movdi_aarch64): Likewise.
9833 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9835 PR tree-optimization/80632
9836 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
9838 (build_arrays): Initialize it for virtual phis.
9839 (fix_phi_nodes): Use it for virtual phis.
9841 PR tree-optimization/80558
9842 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
9843 [x, y] op z into [x op, y op z] for op & or | if conditions
9846 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
9847 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9850 * config/arm/arm.md (use_literal_pool): Remove.
9851 (64-bit immediate split): No longer takes cost into consideration
9852 if arm_disable_literal_pool is enabled.
9853 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
9854 used when arm_disable_literal_pool is enabled.
9855 (arm_max_const_double_inline_cost): Remove use of
9856 arm_disable_literal_pool.
9857 (push_minipool_fix): Add assert.
9858 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
9859 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
9860 (no_literal_pool_sf_immediate): New.
9862 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9864 PR tree-optimization/80613
9865 * tree-ssa-dce.c (propagate_necessity): Remove cases for
9866 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
9868 2017-05-05 Richard Biener <rguenther@suse.de>
9870 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
9872 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
9874 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
9875 of this flag from insn conditions due to removal from r247495.
9877 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9879 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
9881 (arm_early_store_addr_dep_ptr): Likewise.
9882 * config/arm/aarch-common-protos.h
9883 (arm_early_load_addr_dep_ptr): Add prototype.
9884 (arm_early_store_addr_dep_ptr): Likewise.
9885 * config/arm/cortex-a53.md: Add new bypasses.
9887 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9889 * tree.c (next_type_uid): Change type to unsigned.
9890 (type_hash_canon): Decrement back next_type_uid if
9891 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
9892 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
9895 2017-05-04 Martin Sebor <msebor@redhat.com>
9897 * builtins.c: Fix a trivial typo in a comment.
9900 * builtins.c (check_sizes): Adjust to handle reading past the end.
9901 Avoid printing excessive upper bound of ranges. Use %E to print
9902 tree nodes instead of converting them to %wu.
9903 (expand_builtin_memchr): New function.
9904 (compute_dest_size): Rename...
9905 (compute_objsize): ...to this.
9906 (expand_builtin_memcpy): Adjust.
9907 (expand_builtin_mempcpy): Adjust.
9908 (expand_builtin_strcat): Adjust.
9909 (expand_builtin_strcpy): Adjust.
9910 (check_strncat_sizes): Adjust.
9911 (expand_builtin_strncat): Adjust.
9912 (expand_builtin_strncpy): Adjust and simplify.
9913 (expand_builtin_memset): Adjust.
9914 (expand_builtin_bzero): Adjust.
9915 (expand_builtin_memcmp): Adjust.
9916 (expand_builtin): Handle memcmp.
9917 (maybe_emit_chk_warning): Check strncat just once.
9919 2017-05-04 Martin Sebor <msebor@redhat.com>
9921 PR preprocessor/79214
9924 * builtins.c (check_sizes): Add inlining context and issue
9925 warnings even when -Wno-system-headers is set.
9926 (check_strncat_sizes): Same.
9927 (expand_builtin_strncat): Same.
9928 (expand_builtin_memmove): New function.
9929 (expand_builtin_stpncpy): Same.
9930 (expand_builtin): Handle memmove and stpncpy.
9932 2017-05-04 Bin Cheng <bin.cheng@arm.com>
9934 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
9935 which is not used any more.
9937 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9939 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
9941 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9943 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
9944 (cortexa53_tunings): Likewise.
9945 (cortexa57_tunings): Likewise.
9946 (cortexa72_tunings): Likewise.
9947 (cortexa73_tunings): Likewise.
9949 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9951 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
9952 Set loop alignment to 8.
9954 2017-05-04 Martin Sebor <msebor@redhat.com>
9956 PR translation/80280
9957 * builtins.c (expand_builtin_object_size): Add missing quoting to
9958 %D and like directives.
9959 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
9960 (hsa_type_for_tree_type): Same.
9961 (verify_function_arguments): Same.
9962 * symtab.c (symbol_table::change_decl_assembler_name): Same.
9963 * varasm.c (get_section): Same.
9966 2017-05-04 Martin Sebor <msebor@redhat.com>
9968 PR translation/80280
9969 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
9971 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9973 * config/aarch64/aarch64.c (generic_addrcost_table):
9974 Change HI/TI mode setting.
9976 2017-05-04 Martin Jambor <mjambor@suse.cz>
9978 PR tree-optimization/80622
9979 * tree-sra.c (comes_initialized_p): New function.
9980 (build_accesses_from_assign): Only set write lazily when
9981 comes_initialized_p is false.
9982 (analyze_access_subtree): Use comes_initialized_p.
9983 (propagate_subaccesses_across_link): Assert !comes_initialized_p
9984 instead of testing for PARM_DECL.
9986 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9988 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
9989 constraint on operand 0 to allow more general addressing modes.
9990 Adjust output template.
9991 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
9993 * config/aarch64/aarch64-protos.h
9994 (aarch64_address_valid_for_prefetch_p): Declare prototype.
9995 * config/aarch64/constraints.md (Dp): New address constraint.
9996 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
9999 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
10001 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
10002 update use of estimate_ipcp_clone_size_and_time.
10003 (estimate_local_effects): Update use of
10004 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
10005 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
10006 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
10007 Return nonspecialized time.
10009 2017-05-04 Richard Biener <rguenther@suse.de>
10011 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
10012 for the last VUSE which def dominates the PHI. Directly call
10014 (get_continuation_for_phi_1): Remove.
10016 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
10018 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
10019 to explain the use of truncating division. Cap the number of
10020 iterations to the maximum given by nb_iterations_upper_bound,
10023 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
10025 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
10026 * configure: Regenerate.
10027 * config.in: Regenerate.
10028 * config/i386/driver-mingw32.c: new file.
10029 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
10030 * config.host: Link driver-mingw32.o on MinGW host.
10031 * doc/install.texi: Document new --enable-mingw-wildcard configure
10034 2017-05-04 Marek Polacek <polacek@redhat.com>
10036 PR tree-optimization/80612
10037 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
10039 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10040 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
10042 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
10043 (movt splitter): Likewise.
10044 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
10045 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
10046 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
10047 block for Thumb-1 with MOVT.
10048 (thumb2_legitimate_address_p): Move code block ...
10049 (can_avoid_literal_pool_for_label_p): ... into this new function.
10050 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
10052 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
10053 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
10054 "M-profile targets with the MOVT instruction".
10056 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10058 * config/arm/arm-builtins.c (arm_init_builtins): Rename
10059 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
10060 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
10062 2017-05-04 Martin Liska <mliska@suse.cz>
10064 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
10065 variable cond_code.
10067 2017-05-04 Richard Biener <rguenther@suse.de>
10069 * tree.c (array_at_struct_end_p): Handle arrays at struct
10070 end with flexarrays more conservatively. Refactor and treat
10071 arrays of arrays or aggregates more strict. Fix
10072 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
10073 * tree.c (array_at_struct_end_p): Adjust prototype.
10074 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
10075 * gimple-fold.c (get_range_strlen): Likewise.
10076 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
10078 2017-05-04 Richard Biener <rguenther@suse.de>
10080 PR tree-optimization/31130
10081 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
10083 (supports_overflow_infinity): Likewise.
10084 (is_negative_overflow_infinity): Likewise.
10085 (is_positive_overflow_infinity): Likewise.
10086 (is_overflow_infinity): Likewise.
10087 (stmt_overflow_infinity): Likewise.
10088 (overflow_infinity_range_p): Likewise.
10089 (usable_range_p): Remove as always returning true.
10090 (make_overflow_infinity): Remove.
10091 (negative_overflow_infinity): Likewise.
10092 (positive_overflow_infinity): Likewise.
10093 (avoid_overflow_infinity): Likewise.
10094 (set_value_range): Adjust accordingly.
10095 (set_value_range_to_nonnegative): Likewise, remove now unused
10096 overflow_infinity arg.
10097 (vrp_operand_equal_p): Adjust.
10098 (update_value_range): Likewise.
10099 (range_int_cst_singleton_p): Likewise.
10100 (operand_less_p): Likewise.
10101 (compare_values_warnv): Likewise.
10102 (extract_range_for_var_from_comparison_expr): Likewise.
10103 (vrp_int_const_binop): Likewise.
10104 (zero_nonzero_bits_from_vr): Likewise.
10105 (extract_range_from_multiplicative_op_1): Likewise.
10106 (extract_range_from_binary_expr_1): Likewise.
10107 (extract_range_from_unary_expr): Likewise.
10108 (extract_range_from_comparison): Likewise.
10109 (extract_range_basic): Likewise.
10110 (adjust_range_with_scev): Likewise.
10111 (compare_ranges): Likewise.
10112 (compare_range_with_value): Likewise.
10113 (dump_value_range): Likewise.
10114 (test_for_singularity): Likewise, remove strict_overflow_p parameter
10116 (simplify_cond_using_ranges): Adjust.
10118 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
10120 * brig-builtins.def: Added a builtin for class_f64.
10121 * builtin-types.def: Added a builtin type needed by class_f64.
10123 2017-05-03 Jason Merrill <jason@redhat.com>
10125 * timevar.def: Add TV_CONSTEXPR.
10127 2017-05-03 David Malcolm <dmalcolm@redhat.com>
10129 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
10131 2017-05-03 Martin Jambor <mjambor@suse.cz>
10133 * ipa-prop.c (ipa_update_after_lto_read): Removed.
10134 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
10135 * ipa-cp.c (ipcp_propagate_stage): Do not call
10136 ipa_update_after_lto_read.
10137 * ipa-inline.c (ipa_inline): Likewise.
10139 2017-05-03 Martin Jambor <mjambor@suse.cz>
10141 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
10142 tag. Added a default constructor and a destructor.
10143 (ipa_edge_args_sum_t): New class;
10144 (ipa_edge_args_sum): Declare.
10145 (ipa_edge_args_vector): Remove declaration.
10146 (IPA_EDGE_REF): Use ipa_edge_args_sum.
10147 (ipa_free_edge_args_substructures): Remove declaration.
10148 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
10149 (ipa_edge_args_info_available_for_edge_p): Likewise.
10150 * ipa-prop.c (ipa_edge_args_vector): Removed.
10151 (edge_removal_hook_holder): Likewise.
10152 (edge_duplication_hook_holder): Likewise.
10153 (ipa_edge_args_sum): New variable.
10154 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
10155 ipa_edge_args_vector.
10156 (ipa_free_edge_args_substructures): Likewise.
10157 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
10158 ipa_edge_args_vector.
10159 (ipa_edge_removal_hook): Turned into method
10160 ipa_edge_args_sum_t::remove.
10161 (ipa_edge_duplication_hook): Turned into method
10162 ipa_edge_args_sum_t::duplicate.
10163 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
10164 registering edge hooks.
10165 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
10166 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
10167 ipa_edge_args_sum instead of ipa_edge_args_vector.
10168 * ipa-profile.c (ipa_profile): Likewise.
10170 2017-05-03 Martin Jambor <mjambor@suse.cz>
10172 * symbol-summary.h (function_summary): New method exists.
10173 (function_summary::symtab_removal): Deallocate through release.
10174 (call_summary): New class.
10175 (gt_ggc_mx): New overload.
10176 (gt_pch_nx): Likewise.
10177 (gt_pch_nx): Likewise.
10179 2017-05-03 Jeff Law <law@redhat.com>
10181 PR tree-optimization/78496
10182 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
10183 from simplify_cond_using_ranges. Split off code to walk
10184 backwards through casts into ...
10185 (simplify_cond_using_ranges_2): New function.
10186 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
10187 (execute_vrp): After identifying jump threads, call
10188 simplify_cond_using_ranges_2.
10190 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
10193 * ipa-inline.h (inline_summary): Add ctor.
10194 (create_ggc): Do not use ggc_cleared_alloc.
10196 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
10197 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10199 * gcc.c (handle_braces): Support escaping in switch matching
10201 * doc/invoke.texi (Spec Files): Document it.
10202 Remove superfluous @code markup in items.
10204 2017-05-03 David Malcolm <dmalcolm@redhat.com>
10206 * diagnostic-show-locus.c (struct column_range): New struct.
10207 (get_affected_columns): New function.
10208 (get_printed_columns): New function.
10209 (struct correction): New struct.
10210 (correction::ensure_capacity): New function.
10211 (correction::ensure_terminated): New function.
10212 (struct line_corrections): New struct.
10213 (line_corrections::~line_corrections): New dtor.
10214 (line_corrections::add_hint): New function.
10215 (layout::print_trailing_fixits): Reimplement in terms of the new
10217 (selftest::test_overlapped_fixit_printing): New function.
10218 (selftest::diagnostic_show_locus_c_tests): Call it.
10220 2017-05-03 Nathan Sidwell <nathan@acm.org>
10222 Canonicalize canonical type hashing
10223 * tree.h (type_hash_canon_hash): Declare.
10224 * tree.c (type_hash_list, attribute_hash_list): Move into
10225 type_hash_canon_hash.
10226 (build_type_attribute_qual_variant): Break out hash code calc into
10227 type_hash_canon_hash.
10228 (type_hash_canon_hash): New. Generic type hash computation.
10229 (build_range_type_1, build_array_type_1, build_function_type,
10230 build_method_type_directly, build_offset_type, build_complex_type,
10231 make_vector_type): Call it.
10233 2017-05-03 Richard Biener <rguenther@suse.de>
10235 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
10236 When all DRs have unknown misaligned do not always peel
10237 when there is a store but apply the same costing model as if
10238 there were only loads.
10240 2017-05-03 Richard Biener <rguenther@suse.de>
10243 PR tree-optimization/80492
10244 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
10245 compare_base_decls returning dont-know properly.
10247 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
10249 * config/arm/iterators.md (CCSI): New mode iterator.
10250 (arch): New mode attribute.
10251 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
10252 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
10253 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
10254 code iterator for success result mode.
10255 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
10256 the corresponding new insn generators.
10258 2017-05-03 Bin Cheng <bin.cheng@arm.com>
10261 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10262 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
10264 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
10266 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
10267 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
10268 (DDR_A): Wrap DDR argument in brackets.
10269 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
10270 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
10271 (DDR_REVERSED_P): Likewise.
10273 2017-05-03 Jakub Jelinek <jakub@redhat.com>
10275 PR tree-optimization/79472
10276 * tree-switch-conversion.c (struct switch_conv_info): Add
10277 contiguous_range and default_case_nonstandard fields.
10278 (collect_switch_conv_info): Compute contiguous_range and
10279 default_case_nonstandard fields, don't clear final_bb if
10280 contiguous_range and only the default case doesn't have the required
10282 (check_all_empty_except_final): Set default_case_nonstandard instead
10283 of failing if contiguous_range and the default case doesn't have empty
10285 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
10286 and only the default case doesn't have the required constants. Skip
10288 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
10289 if default_case_nonstandard.
10290 (build_constructors): Build constant 1 just once. Assert that default
10291 values aren't inserted in between cases if contiguous_range. Skip
10293 (build_arrays): Skip virtual phis.
10294 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
10295 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
10296 Handle virtual phis.
10297 (gen_inbound_check): Handle default_case_nonstandard case.
10298 (process_switch): Adjust check_final_bb caller. Call
10299 gather_default_values with the first non-default case instead of
10300 default case if default_case_nonstandard.
10302 2017-05-02 Nathan Sidwell <nathan@acm.org>
10304 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
10305 check. Fix formatting.
10307 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
10309 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
10310 errors when comparing specialized and unspecialized times.
10312 2017-05-02 David Malcolm <dmalcolm@redhat.com>
10314 * diagnostic-show-locus.c
10315 (layout::should_print_annotation_line_p): Make private.
10316 (layout::print_annotation_line): Make private.
10317 (layout::annotation_line_showed_range_p): Make private.
10318 (layout::show_ruler): Make private.
10319 (layout::print_source_line): Make private. Pass in line and
10320 line_width, rather than calling location_get_source_line. Drop
10322 (layout::print_leading_fixits): New method.
10323 (layout::print_any_fixits): Rename to...
10324 (layout::print_trailing_fixits): ...this, and make private.
10325 Don't print newline fixits.
10326 (diagnostic_show_locus): Move logic for printing one row into...
10327 (layout::print_line): ...this new function. Move the
10328 location_get_source_line call and error-handling from
10329 print_source_line to here. Call print_leading_fixits, and rename
10330 print_any_fixits to print_trailing_fixits.
10331 (selftest::test_fixit_insert_containing_newline): Update now that
10332 newlines are partially supported.
10333 (selftest::test_fixit_insert_containing_newline_2): New test.
10334 (selftest::test_fixit_replace_containing_newline): Update comments.
10335 (selftest::diagnostic_show_locus_c_tests): Call the new test.
10336 * edit-context.c (class added_line): New class.
10337 (class edited_line): Describe newline handling in comment.
10338 (edited_line::actually_edited_p): New method.
10339 (edited_line::print_content): Delete redundant decl.
10340 (edited_line::m_predecessors): New field.
10341 (edited_file::print_content): Call edited_line::print_content.
10342 (edited_file::print_diff): Update to support newlines.
10343 (edited_file::print_diff_hunk): Likewise.
10344 (edited_file::print_run_of_changed_lines): New function.
10345 (edited_file::print_diff_line): Convert to...
10346 (print_diff_line): ...this.
10347 (edited_file::get_effective_line_count): New function.
10348 (edited_line::edited_line): Initialize new field m_predecessors.
10349 (edited_line::~edited_line): Clean up m_predecessors.
10350 (edited_line::apply_fixit): Handle newlines.
10351 (edited_line::get_effective_line_count): New function.
10352 (edited_line::print_content): New function.
10353 (edited_line::print_diff_lines): New function.
10354 (selftest::test_applying_fixits_insert_containing_newline): New
10356 (selftest::test_applying_fixits_replace_containing_newline): New
10358 (selftest::insert_line): New function.
10359 (selftest::test_applying_fixits_multiple_lines): Add example of
10361 (selftest::edit_context_c_tests): Call the new tests.
10363 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10365 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
10366 parameter cand. Update dump information.
10367 (get_computation_cost): Update uses.
10369 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10371 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
10372 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
10373 (get_computation_at, rewrite_use_address): Update use of
10374 get_computation_aff.
10376 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10378 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
10379 (get_computation): Delete.
10380 (get_computation_cost): Implement like get_computation_cost_at.
10381 Use get_computation_at.
10382 (get_computation_cost_at): Delete.
10383 (rewrite_use_nonlinear_expr): Use get_computation_at.
10384 (rewrite_use_compare, remove_unused_ivs): Ditto.
10386 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10388 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
10390 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10392 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
10393 (ivopts_global_cost_for_size): Rename parameter and update uses.
10394 (iv_ca_recount_cost): Update uses.
10395 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
10396 candidates seperately in n_invs and n_cands.
10397 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
10399 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10401 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
10402 (find_inv_vars_cb): New.
10403 (find_depends): Renamed to ...
10404 (find_inv_vars): ... this.
10405 (add_candidate_1, force_var_cost): Call find_inv_vars.
10406 (split_address_cost, determine_group_iv_cost_cond): Ditto.
10408 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10410 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
10411 inv_vars. Add inv_exprs.
10412 (struct iv_cand): Rename depends_on to inv_vars.
10413 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
10414 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
10415 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
10416 (dump_cand): Dump inv_vars.
10417 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
10418 (record_invariant, find_depends, add_candidate_1): Ditto.
10419 (set_group_iv_cost, force_var_cost): Ditto.
10420 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
10421 (get_computation_cost_at, get_computation_cost): Ditto.
10422 (determine_group_iv_cost_generic): Ditto.
10423 (determine_group_iv_cost_address): Ditto.
10424 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
10425 (determine_group_iv_costs): Ditto.
10426 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
10427 (iv_ca_set_remove_invariants): Renamed to ...
10428 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
10429 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
10430 (iv_ca_set_add_invariants): Renamed to ...
10431 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
10432 (iv_ca_set_cp): Use iv_ca_set_add_invs.
10433 (iv_ca_has_deps): Support inv_vars and inv_exprs.
10434 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
10435 (create_new_ivs): Remove useless dump.
10437 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10439 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
10441 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
10442 (iv_ca_set_no_cp, create_new_iv): Ditto.
10444 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10446 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
10448 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10450 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
10451 function tree_check2.
10453 2017-05-02 Martin Liska <mliska@suse.cz>
10455 * doc/gcov.texi: Add missing preposition.
10456 * gcov.c (function_info::function_info): Properly fill up
10457 all member variables.
10459 2017-05-02 Tamar Christina <tamar.christina@arm.com>
10461 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
10463 2017-05-02 Tamar Christina <tamar.christina@arm.com>
10465 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
10467 2017-05-02 Martin Liska <mliska@suse.cz>
10470 * lto-streamer-in.c (lto_read_tree_1): Remove
10471 LTO_STREAMER_DEBUG.
10472 * lto-streamer.c (struct tree_hash_entry): Likewise.
10473 (struct tree_entry_hasher): Likewise.
10474 (tree_entry_hasher::hash): Likewise.
10475 (tree_entry_hasher::equal): Likewise.
10476 (lto_streamer_init): Likewise.
10477 (lto_orig_address_map): Likewise.
10478 (lto_orig_address_get): Likewise.
10479 (lto_orig_address_remove): Likewise.
10480 * lto-streamer.h: Likewise.
10481 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
10482 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
10484 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
10486 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
10487 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
10488 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
10489 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
10490 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
10491 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
10492 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
10493 (mm_maskz_sub_ss): New intrinsics.
10494 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
10495 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
10496 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
10497 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
10498 (__builtin_ia32_subss_mask_round): New builtins.
10499 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
10500 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
10501 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
10503 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
10504 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
10506 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
10509 2017-05-02 Martin Jambor <mjambor@suse.cz>
10511 PR tree-optimization/78687
10512 * tree-sra.c (access): New field parent.
10513 (process_subtree_disqualification): New function.
10514 (disqualify_candidate): Call it.
10515 (build_accesses_from_assign): Reset write flag if creating an
10517 (build_access_subtree): Fill in parent field and also prpagate
10518 down grp_write flag.
10519 (create_artificial_child_access): New parameter set_grp_write, set
10520 grp_write to its value.
10521 (propagate_subaccesses_across_link): Also propagate grp_write flag
10523 (propagate_all_subaccesses): Push the closest parent back to work
10524 queue if add_access_to_work_queue returned true.
10526 2017-05-02 Richard Biener <rguenther@suse.de>
10528 * common.opt (fstrict-overflow): Alias negative to fwrapv.
10529 * doc/invoke.texi (fstrict-overflow): Remove all traces of
10530 -fstrict-overflow documentation.
10531 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
10532 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
10533 flag_strict_overflow.
10534 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
10535 * lto-opts.c (lto_write_options): Do not stream it.
10536 * lto-wrapper.c (merge_and_complain): Do not handle it.
10537 * opts.c (default_options_table): Do not set -fstrict-overflow.
10538 (finish_options): Likewise do not clear it when sanitizing.
10539 * simplify-rtx.c (simplify_const_relational_operation): Do not
10540 test flag_strict_overflow.
10542 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10544 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
10545 using enabled attribute.
10546 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
10547 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
10548 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
10549 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
10550 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
10551 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
10552 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
10553 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
10554 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
10555 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
10557 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10559 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
10561 2017-05-02 Richard Biener <rguenther@suse.de>
10563 PR tree-optimization/80591
10565 2017-04-10 Richard Biener <rguenther@suse.de>
10567 * tree-ssa-structalias.c (find_func_aliases): Properly handle
10570 2017-05-02 Richard Biener <rguenther@suse.de>
10572 PR tree-optimization/80549
10573 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
10574 (cleanup_tree_cfg_noloop): Create forwarders to known loop
10575 headers if they do not have a preheader.
10577 2017-05-02 Martin Liska <mliska@suse.cz>
10580 * common.opt: Fix typo.
10581 * doc/invoke.texi: Likewise.
10583 2017-05-01 Jan Beulich <jbeulich@suse.com>
10585 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
10586 swapping, add (x,x,m,x,n) alternative.
10588 2017-05-01 Nathan Sidwell <nathan@acm.org>
10590 * calls.c (combine_pending_stack_adjustment_and_call): Remove
10591 unnecessary unadjusted_alignment check.
10593 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10596 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
10598 * gimplify.c (gimplify_cilk_detach): New function.
10599 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
10600 * tree-core.h: Document EXPR_CILK_SPAWN.
10601 * tree.h (EXPR_CILK_SPAWN): Define.
10603 2017-05-01 David Malcolm <dmalcolm@redhat.com>
10605 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
10606 to use new fixit_hint representation, using the "replace" logic.
10607 (get_line_span_for_fixit_hint): Likewise.
10608 (layout::print_any_fixits): Likewise.
10609 (selftest::test_one_liner_many_fixits): Rename to...
10610 (selftest::test_one_liner_many_fixits_1): ...this, and update
10611 comment and expected output to reflect that the multiple fix-it
10612 hints are now consolidated into one insertion.
10613 (selftest::test_one_liner_many_fixits_2): New test.
10614 (selftest::test_diagnostic_show_locus_one_liner): Update for
10616 (selftest::test_fixit_consolidation): Update for fix-it API
10618 * diagnostic.c (print_parseable_fixits): Likewise.
10619 * edit-context.c (edited_line::m_line_events): Convert from
10620 auto_vec <line_event *> to auto_vec <line_event>.
10621 (class line_event): Convert from abstract base class to a concrete
10622 class, taking over the role of replace_event.
10623 (class insert_event): Delete.
10624 (class replace_event): Rename to class line_event. Convert to
10626 (edit_context::add_fixits): Reimplement.
10627 (edit_context::apply_insert): Delete.
10628 (edit_context::apply_replace): Rename to...
10629 (edit_context::apply_fixit): ...this. Convert to half-open range.
10630 (edited_file::apply_insert): Delete.
10631 (edited_file::apply_replace): Rename to...
10632 (edited_file::apply_fixit): ...this.
10633 (edited_line::~edited_line): Drop deletion of events.
10634 (edited_line::apply_insert): Delete.
10635 (edited_line::apply_replace): Rename to...
10636 (edited_line::apply_fixit): ...this. Convert to half-open range.
10637 Update for change to type of m_line_events.
10638 * edit-context.h (edit_context::apply_insert): Delete.
10639 (edit_context::apply_replace): Rename to...
10640 (edit_context::apply_fixit): ...this.
10642 2017-05-01 Martin Sebor <msebor@redhat.com>
10644 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
10647 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
10650 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
10651 __get_cpuid_max returns 0.
10652 (__get_cpuid_count): Ditto.
10654 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
10656 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
10657 replacement expression is another instance of one of its arguments.
10659 2017-05-01 Jakub Jelinek <jakub@redhat.com>
10662 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
10663 check for stack push/pop autoinc.
10664 * config/i386/i386.c (ix86_agi_dependent): Return false
10665 if the only reason why modified_in_p returned true is that
10666 addr is SP based and set_insn is a push or pop.
10668 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10670 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
10673 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10676 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
10677 (account_size_time): Use two predicates - exec_pred and
10679 (evaluate_conditions_for_known_args): Compute both clause and
10681 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
10682 (inline_summary_t::duplicate): Update.
10683 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
10685 (compute_inline_parameters): Likewise.
10686 (estimate_edge_size_and_time): Update caluclation of time.
10687 (estimate_node_size_and_time): Compute both time and nonspecialized
10689 (estimate_ipcp_clone_size_and_time): Update.
10690 (inline_merge_summary): Update.
10691 (do_estimate_edge_time): Update.
10692 (do_estimate_edge_size): Update.
10693 (do_estimate_edge_hints): Update.
10694 (inline_read_section, inline_write_summary): Stream both new predicates.
10695 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
10697 (compute_inlined_call_time): Cleanup.
10698 (big_speedup_p): Update.
10699 (edge_badness): Update.
10700 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
10701 (size_time_entry): Replace predicate by exec_predicate and
10702 nonconst_predicate.
10703 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
10704 (estimate_edge_time): Return also nonspec_time.
10705 (reset_edge_growth_cache): Update.
10707 2017-04-29 Jakub Jelinek <jakub@redhat.com>
10709 PR rtl-optimization/80491
10710 * ifcvt.c (noce_process_if_block): When looking for x setter
10711 with missing else_bb, don't check only the insn right before
10712 cond_earliest, but look for the last insn that x is modified in
10713 within the same bb.
10715 PR rtl-optimization/80491
10716 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
10718 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
10720 PR tree-optimization/80487
10721 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
10723 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10725 PR tree-optimization/79697
10726 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
10727 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
10728 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
10730 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
10731 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
10733 2017-04-28 Martin Sebor <msebor@redhat.com>
10735 PR tree-optimization/80523
10736 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
10737 (init_target_to_host_charmap, target_to_host, target_strtol10): New
10739 (maybe_warn, format_directive, parse_directive): Use new functions.
10740 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
10742 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
10744 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
10746 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
10748 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
10749 target_header_dir): Set correctly.
10750 * configure: Regenerated.
10751 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
10752 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
10753 instead of SYSTEM_HEADER_DIR.
10755 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
10757 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
10758 (estimate_local_effects): Likewise.
10759 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
10760 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
10761 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
10762 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
10763 do_estimate_edge_time, estimate_edge_time): Likewise.
10764 * ipa-inline-analysis.c (estimate_node_size_and_time,
10765 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
10766 (estimate_time_after_inlining): Remove.
10768 2017-04-28 Martin Liska <mliska@suse.cz>
10770 * doc/gcov.texi: Enhance documentation of gcov.
10772 2017-04-28 Martin Liska <mliska@suse.cz>
10774 * doc/gcov.texi: Sort options in alphabetic order.
10775 * doc/gcov-dump.texi: Likewise.
10776 * doc/gcov-tool.texi: Likewise.
10777 * gcov.c (print_usage): Likewise.
10778 * gcov-dump.c (print_usage): Likewise.
10779 * gcov-tool.c (print_merge_usage_message): Likewise.
10780 (print_rewrite_usage_message): Likewise.
10781 (print_overlap_usage_message): Likewise.
10783 2017-04-28 Martin Liska <mliska@suse.cz>
10785 PR gcov-profile/53915
10786 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
10788 2017-04-28 Martin Liska <mliska@suse.cz>
10790 PR gcov-profile/79891
10791 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
10792 is marked by compiler as living on a line.
10793 (get_cycles_count): Remove usage of the union.
10794 (output_intermediate_file): Likewise.
10795 (find_source): Fix GNU coding style.
10796 (accumulate_line_counts): Remove old non-all block mode.
10797 (output_lines): Remove usage of the union.
10798 * profile.c (output_location): Include all BBs, even if
10799 belonging to a same line (and file) as a previous BB.
10801 2017-04-28 Martin Liska <mliska@suse.cz>
10803 * gcov.c (process_args): Handle new argument 'w'.
10804 (read_graph_file): Assign ID to BBs.
10805 (output_branch_count): Display BB # if verbose flag is set.
10806 (output_lines): Likewise for arcs.
10807 (print_usage): Add '--verbose' option help.
10808 * doc/gcov.texi: Document --verbose (-w) option.
10810 2017-04-28 Martin Liska <mliska@suse.cz>
10812 * gcov.c (struct block_location_info): New struct.
10813 (process_file): Fill up the new structure.
10814 (read_graph_file): Replace usage of encoding by the newly added
10816 (add_line_counts): Likewise.
10817 (accumulate_line_counts): Remove usage of the union.
10818 (function_info::function_info): New function.
10819 (function_info::~function_info): Likewise.
10820 (process_file): Call delete instead of release_function.
10821 (release_function): Release the function.
10822 (release_structures): Call delete instead of release_function.
10823 (solve_flow_graph): Replace usage of num_blocks.
10824 (find_exception_blocks): Likewise.
10825 (output_lines): Fix GNU coding style.
10827 2017-04-28 Martin Liska <mliska@suse.cz>
10830 * coverage.c (coverage_remove_note_file): New function.
10831 * coverage.h: Declare the function.
10832 * toplev.c (finalize): Clean if an error has been seen.
10834 2017-04-28 Martin Liska <mliska@suse.cz>
10836 PR gcov-profile/80031
10837 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
10838 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
10839 * gcov.c (read_graph_file): Read just number of blocks.
10840 * profile.c (branch_prob): Do not stream 0 flags per a basic
10843 2017-04-28 Martin Liska <mliska@suse.cz>
10845 * gcov-dump.c (tag_*): Add new argument to declarations.
10846 (dump_gcov_file): Likewise.
10847 (tag_blocks): Add and use new argument depth.
10848 (tag_arcs): Likewise.
10849 (tag_lines): Likewise.
10850 (tag_counters): Likewise.
10851 (tag_summary): Likewise.
10852 (dump_working_sets): Use depth to do a proper indentation.
10854 2017-04-28 Jakub Jelinek <jakub@redhat.com>
10857 * cgraph.h (symtab_node::debug_symtab): No longer inline.
10858 * symtab.c (symtab_node::debug_symtab): Move definition here.
10860 2017-04-28 Richard Biener <rguenther@suse.de>
10862 * lto-streamer.h (LTO_major_version): Bump to 7.
10864 2017-04-28 Richard Biener <rguenther@suse.de>
10866 * tree-vrp.c (assert_info): New struct.
10867 (add_assert_info): New helper.
10868 (register_edge_assert_for_2): Refactor to add asserts to a vector
10870 (register_edge_assert_for_1): Likewise.
10871 (register_edge_assert_for): Likewise.
10872 (finish_register_edge_assert_for): New helper actually registering
10873 asserts where live on edge.
10874 (find_conditional_asserts): Adjust.
10875 (find_switch_asserts): Likewise.
10876 (evrp_dom_walker::try_find_new_range): Generalize.
10877 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
10879 2017-04-27 Marek Polacek <polacek@redhat.com>
10882 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
10883 arg10 and arg11 to itype.
10885 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
10887 * doc/extend.texi (Object Size Checking): Improve grammar.
10889 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
10892 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
10893 that the logic for permitting reciprocal estimates matches that
10896 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10899 * tree.c (type_cache_hasher::equal): Only compare
10900 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
10901 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
10902 non-aggregate element types.
10903 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
10904 about the flag on ARRAY_TYPEs in the comment, formatting fix.
10906 2017-04-27 Richard Biener <rguenther@suse.de>
10908 PR middle-end/80533
10909 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
10910 stripping ARRAY_REFs from MEM_EXPR make sure we're not
10911 keeping a reference to a trailing array.
10913 2017-04-27 Richard Biener <rguenther@suse.de>
10915 PR middle-end/80539
10916 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
10917 being in loop-closed SSA form conservatively.
10918 (chrec_fold_multiply_poly_poly): Likewise.
10920 2017-04-27 Tamar Christina <tamar.christina@arm.com>
10922 PR middle-end/79665
10923 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
10924 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
10926 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10929 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
10930 (aarch64_function_arg_alignment): Return unsigned int again, but still
10931 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
10932 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
10933 Don't emit -Wpsabi note.
10934 (aarch64_function_arg_boundary): Likewise.
10935 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
10938 2017-04-26 Nathan Sidwell <nathan@acm.org>
10940 * tree.h (crc32_unsigned_n): Declare.
10941 (crc32_unsigned, crc32_unsigned): Make inline.
10942 * tree.c (crc32_unsigned_bits): Replace with ...
10943 (crc32_unsigned_n): ... this.
10944 (crc32_unsigned, crc32_byte): Remove.
10945 (crc32_string): Remove unnecessary braces.
10947 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
10949 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
10950 * ipa-inline-analysis.c (MAX_TIME): Remove.
10951 (account_size_time): Use sreal for time.
10952 (dump_inline_summary): Update.
10953 (estimate_function_body_sizes): Update.
10954 (estimate_edge_size_and_time): Update.
10955 (estimate_calls_size_and_time): Update.
10956 (estimate_node_size_and_time): Update.
10957 (inline_merge_summary): Update.
10958 (inline_update_overall_summary): Update.
10959 (estimate_time_after_inlining): Update.
10960 (inline_read_section): Update.
10961 (inline_write_summary): Update.
10962 * ipa-inline.c (compute_uninlined_call_time): Update.
10963 (compute_inlined_call_time): Update.
10964 (recursive_inlining): Update.
10965 (inline_small_functions): Update.
10966 (dump_overall_stats): Update.
10967 * ipa-inline.h: Include sreal.h.
10968 (size_time_entry): Turn time to sreal.
10969 (inline_summary): Turn self_time nad time to sreal.
10971 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
10973 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
10975 (sreal::stream_out, sreal::stream_in): New.
10976 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
10978 2017-04-25 Jakub Jelinek <jakub@redhat.com>
10980 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
10983 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
10986 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
10988 (dimode_scalar_chain::compute_convert_gain): Ditto.
10989 (dimode_scalar_chain::make_vector_copies): Ditto.
10990 (dimode_scalar_chain::convert_reg): Ditto.
10991 (dimode_scalar_chain::convert_insn): Ditto.
10992 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
10993 (VI248_AVX512BW_1): New mode iterator.
10994 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
10995 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
10998 2017-04-25 Martin Sebor <msebor@redhat.com>
11000 PR tree-optimization/80497
11001 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
11002 constants are representable in HOST_WIDE_INT.
11003 (parse_directive): Ditto.
11005 2017-04-25 Martin Sebor <msebor@redhat.com>
11008 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
11009 (new_zero_array): Adjust signature.
11010 (dom_info::dom_init): Used unsigned rather that size_t.
11011 (dom_info::dom_info): Same.
11013 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11014 Jakub Jelinek <jakub@redhat.com>
11017 * config/arm/arm.c: Include gimple.h.
11018 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
11019 returns negative, increment ncrn only if it returned positive.
11020 (arm_needs_doubleword_align): Return int instead of bool,
11021 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
11022 members, but if there is any such non-FIELD_DECL
11023 > PARM_BOUNDARY aligned decl, return -1 instead of false.
11024 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
11025 returns negative, increment nregs only if it returned positive.
11026 (arm_setup_incoming_varargs): Likewise.
11027 (arm_function_arg_boundary): Emit -Wpsabi note if
11028 arm_needs_doubleword_align returns negative, return
11029 DOUBLEWORD_ALIGNMENT only if it returned positive.
11031 2017-04-25 Marek Polacek <polacek@redhat.com>
11034 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
11035 first argument to type.
11037 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
11040 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
11041 type checks to test for compatibility instead of equality.
11043 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11044 Jakub Jelinek <jakub@redhat.com>
11047 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
11049 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
11050 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
11051 the alignment computation, but return their maximum in warn_alignment.
11052 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
11053 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
11055 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
11056 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
11059 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11061 * config/arc/simdext.md (dmpyh): Fix typo.
11063 2017-04-25 Richard Biener <rguenther@suse.de>
11065 PR tree-optimization/80492
11066 * alias.c (compare_base_decls): Handle registers with asm
11067 specification conservatively.
11068 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
11069 compare_base_decls returning dont-know properly.
11071 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11073 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
11074 (legitimate_offset_address_p): New function.
11075 (arc_legitimate_address_p): Use above function.
11077 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11079 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
11081 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11083 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
11084 ACCH registers whenever they are available.
11086 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11088 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
11089 double regs fix when not used.
11091 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11093 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
11095 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
11096 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
11098 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11100 * config/arc/arc.c (arc_output_addsi): Check for h-register class
11101 when emitting short ADD instructions.
11103 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11105 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
11107 (cmpsi_cc_c_insn): Likewise.
11108 (cbranchsi4_scratch): Compute proper instruction length using
11109 compact_hreg_operand.
11110 * config/arc/predicates.md (compact_hreg_operand): New predicate.
11112 2017-04-25 Richard Biener <rguenther@suse.de>
11114 PR middle-end/80509
11115 * passes.c (pass_manager::pass_manager): Initialize
11116 m_name_to_pass_map.
11118 2017-04-25 Richard Biener <rguenther@suse.de>
11120 PR tree-optimization/79201
11121 * tree-ssa-sink.c (statement_sink_location): Handle calls.
11123 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11126 * config/s390/vector.md: Split MEM->GPR vector moves for
11127 non-s_operand addresses.
11129 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11132 * config/s390/predicates.md (reload_const_wide_int_operand): New
11134 * config/s390/s390.md ("movti"): Remove d/P alternative.
11135 ("movti_bigconst"): New pattern definition.
11137 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11140 * s390-protos.h (s390_expand_cs_hqi): Removed.
11141 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
11142 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
11143 modes as well as CCZ1mode and CCZmode.
11144 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
11145 signature of s390_emit_compare_and_swap.
11146 (s390_expand_cs_hqi): Likewise, make static.
11147 (s390_expand_cs_tdsi): Generate an explicit compare before trying
11148 compare-and-swap, in some cases.
11149 (s390_expand_cs): Wrapper function.
11150 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
11152 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
11153 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
11154 patterns for small and large integers. Forbid symref memory operands.
11155 Move expander to s390.c. Require cc register.
11156 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
11157 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
11158 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
11159 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
11160 symref memory operands. Remove CC mode and call s390_match_ccmode
11162 ("atomic_exchange<mode>"): Allow and implement all integer modes.
11164 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11166 * config/s390/s390.md (define_peephole2): New peephole to help
11167 combining the load-and-test pattern with volatile memory.
11169 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11171 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
11172 with CCZmode for TARGET_Z196.
11174 2017-04-25 Jakub Jelinek <jakub@redhat.com>
11176 PR rtl-optimization/80501
11177 * combine.c (make_compound_operation_int): Set subreg_code to SET
11178 even for AND with mask of the sign bit of mode.
11180 PR rtl-optimization/80500
11181 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
11182 sum's initial value.
11184 2017-04-25 Julian Brown <julian@codesourcery.com>
11185 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11187 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
11189 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
11191 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
11193 2017-04-25 Julian Brown <julian@codesourcery.com>
11194 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11196 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
11197 (thunderx2t99_sha): New Reservation.
11199 2017-04-25 Julian Brown <julian@codesourcery.com>
11200 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11202 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
11203 type for 1-element load.
11205 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
11207 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
11209 2017-04-24 Martin Jambor <mjambor@suse.cz>
11211 PR tree-optimization/80293
11212 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
11213 char arrays not totally scalarizable if it is false.
11214 (analyze_all_variable_accesses): Pass correct value in the new
11215 parameter. Add a statistics counter.
11217 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
11219 PR middle-end/79931
11220 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
11222 2017-04-24 Richard Biener <rguenther@suse.de>
11224 PR tree-optimization/80494
11225 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
11226 out for complex types.
11228 2017-04-24 Richard Biener <rguenther@suse.de>
11230 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
11231 * tree-ssa-sccvn.c (print_scc): Print SCC size.
11232 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
11233 (DFS): Adjust and never fail.
11234 (sccvn_dom_walker::fail): Remove.
11235 (sccvn_dom_walker::before_dom_children): Adjust.
11236 (run_scc_vn): Likewise and never fail.
11237 * tree-ssa-pre.c (pass_pre::execute): Adjust.
11238 (pass_fre::execute): Likewise.
11240 2017-04-24 Richard Biener <rguenther@suse.de>
11242 PR tree-optimization/79725
11243 * tree-ssa-sink.c (statement_sink_location): Return whether
11244 failure reason was zero uses. Move that check later.
11245 (sink_code_in_bb): Deal with zero uses by removing the stmt
11248 2017-04-24 Richard Biener <rguenther@suse.de>
11251 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
11252 pointer-based references.
11254 2017-04-24 Richard Biener <rguenther@suse.de>
11257 * pass_manager.h (pass_manager::operator new): Remove.
11258 (pass_manager::operator delete): Likewise.
11259 * passes.c (pass_manager::operator new): Remove.
11260 (pass_manager::operator delete): Likewise.
11261 (pass_manager::pass_manager): Zero individual pass members.
11263 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
11266 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
11267 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
11268 Check "XEXP (src, 1)" operand here.
11269 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
11270 Check "XEXP (src, 1)" operand here.
11271 (dimode_scalar_chain::make_vector_copies): Detect count register
11272 of a shift instruction. Zero extend count register from QImode
11273 to DImode to satisfy vector shift pattern count operand predicate.
11274 Substitute vector shift count operand with a DImode copy.
11275 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
11278 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
11280 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
11281 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
11282 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
11283 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
11284 (UNSPEC_NOREX_MEM): Remove definition.
11286 2017-04-21 Richard Biener <rguenther@suse.de>
11288 PR tree-optimization/79547
11289 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
11290 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
11291 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
11292 without any constraints.
11294 2017-04-21 Richard Biener <rguenther@suse.de>
11296 PR tree-optimization/78847
11297 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
11299 2017-04-21 Richard Biener <rguenther@suse.de>
11301 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
11302 (build_distinct_type_copy): Likewise.
11303 (build_variant_type_copy): Likewise.
11304 * tree.c (build_qualified_type): Pass down mem-stat info.
11305 (build_distinct_type_copy): Likewise.
11306 (build_variant_type_copy): Likewise.
11308 2017-04-21 Richard Biener <rguenther@suse.de>
11310 PR tree-optimization/80237
11311 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
11313 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
11314 for a simplified result.
11316 2016-04-21 Richard Biener <rguenther@suse.de>
11318 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
11319 sth as strict as a simple_iv but a chrec without symbols and an
11320 operand defined in the loop we are peeling (and not some subloop).
11321 (propagate_constants_for_unrolling): Propagate all constants.
11323 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
11326 * config/i386/i386.c (print_reg): Remove assert for disalowed
11327 regno values, call output_operand_lossage instead.
11329 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
11332 * config/i386/constraints.md (Yc): New register constraint.
11333 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
11334 Use Yc constraint for alternative 2 of operand 0. Remove
11335 preferred_for_speed attribute.
11337 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
11339 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
11340 lastprivate clauses in SIMT case.
11342 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
11344 * doc/invoke.texi (-Wextra-semi): Document new warning option.
11346 2017-04-20 Richard Biener <rguenther@suse.de>
11348 PR tree-optimization/57796
11349 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
11350 as N scalar stores.
11351 (vect_model_load_cost): Cost gathers as N scalar loads.
11353 2017-04-20 Richard Biener <rguenther@suse.de>
11355 * ggc-page.c (ggc_allocated_p): Rename to ...
11356 (safe_lookup_page_table_entry): ... this and return the lookup
11358 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
11360 2017-04-20 Richard Biener <rguenther@suse.de>
11362 PR tree-optimization/80453
11363 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
11364 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
11365 from the conditions.
11366 (vn_phi_eq): Pass them down.
11367 (vn_phi_lookup): Record them.
11368 (vn_phi_insert): Likewise.
11370 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
11372 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
11373 uninitialized variable warning to avoid buffer overrun.
11375 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
11378 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
11379 is suppressed for '{ 0 }' in C.
11381 2017-04-20 Jakub Jelinek <jakub@redhat.com>
11383 * BASE-VER: Set to 8.0.0.
11385 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
11387 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
11388 priority .init_array and .fini_array section with SECTION_NOTYPE
11391 2017-04-20 Jakub Jelinek <jakub@redhat.com>
11393 PR middle-end/80423
11394 * tree.h (build_array_type): Add typeless_storage default argument.
11395 * tree.c (type_cache_hasher::equal): Also compare
11396 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
11397 (build_array_type): Add typeless_storage argument, set
11398 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
11400 (build_nonshared_array_type): Adjust build_array_type_1 caller.
11401 (build_array_type): Likewise. Add typeless_storage argument.
11403 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
11404 Jakub Jelinek <jakub@redhat.com>
11406 PR tree-optimization/80426
11407 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
11408 operation on symbolic operands, also compute the overflow for the
11409 invariant part when the operation degenerates into a negation.
11411 2017-04-19 Jakub Jelinek <jakub@redhat.com>
11414 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
11415 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
11418 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
11420 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
11423 * config/avr/avr.c (tree.h): Include it.
11424 (cgraph.h): Include it.
11425 (avr_encode_section_info): Don't warn for uninitialized progmem
11426 variable if it's just an alias.
11428 2017-04-19 Richard Biener <rguenther@suse.de>
11431 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
11432 when needed by AutoPGO.
11434 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
11437 * doc/lto.texi: Remove an extra 'that'.
11439 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
11441 PR rtl-optimization/80429
11442 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
11443 are only used in debug insns.
11445 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
11446 Vladimir Makarov <vmakarov@redhat.com>
11448 * config/sparc/predicates.md (input_operand): Add comment. Return
11449 true for any memory operand when LRA is in progress.
11450 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
11452 2017-04-18 Jeff Law <law@redhat.com>
11455 * mips.md ({return,simple_return}_internal): Do not overwrite
11458 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11460 PR tree-optimization/80443
11461 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
11462 instead of adding 1, subtract -1 and similarly instead of subtracting
11465 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
11467 PR rtl-optimization/80357
11468 * haifa-sched.c (tmp_bitmap): New variable.
11469 (model_recompute): Handle duplicate use records.
11470 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
11471 (free_global_sched_pressure_data): Free it.
11473 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
11476 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
11477 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
11478 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
11479 instead of SYSTEM_HEADER_DIR.
11481 2017-04-18 Jeff Law <law@redhat.com>
11483 PR middle-end/80422
11484 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
11485 predecessors after walking up the insn chain.
11487 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11490 * dwarf2out.c (modified_type_die): Try harder not to emit internal
11491 sizetype type into debug info.
11493 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
11496 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
11497 unneeded test for TARGET_UPPER_REGS_SF.
11498 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
11500 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11503 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
11504 instead of gsi_after_labels.
11506 2017-04-18 Jeff Law <law@redhat.com>
11508 * regcprop.c (maybe_mode_change): Avoid creating copies of the
11512 2017-04-13 Jeff Law <law@redhat.com>
11513 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11514 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11516 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
11519 * config/avr/avr.c (intl.h): Include it.
11520 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
11522 2017-04-18 Martin Liska <mliska@suse.cz>
11524 PR gcov-profile/78783
11525 * gcov-tool.c (gcov_output_files): Validate that destination
11526 file is either removed by the tool or by a user.
11528 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
11529 Guy Benyei <guybe@mellanox.com>
11531 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
11532 block, and do not negate it, the stored id is already negative.
11534 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
11536 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
11538 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
11541 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
11542 masks of options that should be turned off if the VSX vector
11543 options are turned off.
11544 (OTHER_P8_VECTOR_MASKS): Likewise.
11545 (OTHER_VSX_VECTOR_MASKS): Likewise.
11546 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
11547 rs6000_disable_incompatible_switches to validate no type switches
11549 (rs6000_incompatible_switch): New function to disallow turning on
11550 other vector options if -mno-vsx, -mno-power8-vector, or
11551 -mno-power9-vector are specified.
11553 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11555 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
11557 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11559 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
11560 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
11561 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
11562 (ARG_POINTER_CFA_OFFSET): Likewise.
11564 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11566 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
11567 conditions to take advantage of various optimizations.
11569 2017-04-13 Jeff Law <law@redhat.com>
11571 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11572 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11573 (zero_extendsidi2_dext): Likewise.
11575 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11578 * fold-const.c (fold_ternary_loc): Revert
11579 use op0 instead of fold_convert_loc (loc, type, arg0) part of
11582 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
11584 PR rtl-optimization/80343
11585 * lra-remat.c (update_scratch_ops): Assign original hard reg to
11586 new scratch pseudo.
11588 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
11591 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
11592 to ubsan_encode_value.
11594 2017-04-13 Jeff Law <law@redhat.com>
11596 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
11597 appearing in DEBUG_INSNs.
11599 2017-04-13 Martin Liska <mliska@suse.cz>
11601 PR gcov-profile/80413
11602 * gcov-io.c (gcov_write_string): Copy to buffer just when
11603 allocated size is greater than zero.
11605 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11608 * dwarf2out.c (decls_for_scope): Ignore declarations of
11609 current_function_decl in BLOCK_NONLOCALIZED_VARS.
11611 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
11614 * ipa-visibility.c (non_local_p): Fix typos.
11615 (localize_node): When localizing symbol in same comdat group,
11616 dissolve the group only when we know external symbols are going
11618 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
11620 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11622 PR tree-optimization/79390
11623 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
11624 order does not result in usable sequence, retry with reversed operand
11630 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
11631 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
11632 op0 instead of fold_convert_loc (loc, type, arg0).
11634 2017-04-12 Jeff Law <law@redhat.com>
11636 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
11637 has a delay slot in the generated code.
11639 * config/cris/cris.md (cris_preferred_reload_class): Return
11640 GENNONACR_REGS rather than GENERAL_REGS.
11642 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11645 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
11646 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
11647 signedness of the result type.
11649 2017-04-12 Richard Biener <rguenther@suse.de>
11650 Jeff Law <law@redhat.com>
11652 PR tree-optimization/80359
11653 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
11654 trim stores to TARGET_MEM_REFs.
11656 2017-04-12 Richard Biener <rguenther@suse.de>
11658 PR tree-optimization/79390
11659 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
11660 threading case even more.
11662 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
11665 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
11666 for quad_address_p for TImode, instead of just not indexed_address.
11668 2017-04-12 Richard Biener <rguenther@suse.de>
11669 Bernd Edlinger <bernd.edlinger@hotmail.de>
11671 PR middle-end/79671
11672 * alias.c (component_uses_parent_alias_set_from): Handle
11673 TYPE_TYPELESS_STORAGE.
11674 (get_alias_set): Likewise.
11675 * tree-core.h (tree_type_common): Add typeless_storage flag.
11676 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
11677 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
11678 for types containing members with TYPE_TYPELESS_STORAGE.
11679 (place_field): Likewise.
11680 (layout_type): Likewise for ARRAY_TYPE.
11681 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
11682 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
11683 TYPE_TYPELESS_STORAGE.
11684 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
11686 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11689 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
11690 first argument to type.
11692 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11696 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
11697 CONST0_RTX (mode) rather than const0_rtx where appropriate.
11698 (rs6000_expand_binop_builtin): Likewise.
11699 (rs6000_expand_ternop_builtin): Likewise; also add missing
11700 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
11701 vshasigma built-ins.
11702 * doc/extend.texi: Document that vec_xxpermdi's third argument
11703 must be a constant.
11705 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
11707 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
11708 Use shift_const cost parameter when calculating gain of STV shifts.
11710 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
11712 PR rtl-optimization/70478
11713 * lra-constraints.c (process_alt_operands): Check memory for
11714 disfavoring memory insn operand.
11716 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11718 PR middle-end/80100
11719 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
11720 left shift in unsigned HOST_WIDE_INT type.
11722 PR rtl-optimization/80385
11723 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
11724 (not (neg X)) into (plus X -1) for complex or non-integral modes.
11727 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
11728 if they have any depend clauses.
11730 2017-04-11 Martin Liska <mliska@suse.cz>
11733 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
11734 * ipa-split.c (split_function): Create a local comdat symbol
11735 if caller is in a comdat group.
11737 2017-04-11 Martin Liska <mliska@suse.cz>
11740 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
11743 2017-04-11 Martin Sebor <msebor@redhat.com>
11745 PR middle-end/80364
11746 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
11747 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
11749 (directive::set_width, directive::set_precision, format_character):
11751 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
11754 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
11757 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
11758 conflict, set target->arch_name instead of target->cpu_name.
11760 2017-04-11 Richard Biener <rguenther@suse.de>
11762 PR tree-optimization/80374
11763 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
11764 build_zero_cst, remove fold_convertible_p check again.
11766 2017-04-11 Martin Liska <mliska@suse.cz>
11769 * ubsan.c (instrument_object_size): Do not instrument register
11772 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11775 * config/i386/i386-builtin-types.def
11776 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
11777 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
11778 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
11779 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
11780 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
11781 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
11782 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
11783 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
11784 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
11785 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
11786 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
11787 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
11788 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
11789 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
11790 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
11791 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
11792 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
11793 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
11794 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
11795 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
11796 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
11797 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
11798 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
11799 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
11800 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
11801 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
11802 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
11803 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
11804 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
11805 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
11806 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
11807 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
11808 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
11809 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
11810 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
11811 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
11812 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
11813 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
11814 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
11815 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
11816 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
11817 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
11818 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
11819 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
11820 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
11821 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
11823 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
11824 flag to second_arg_count, handle 4 argument function type _COUNT
11825 aliases, handle second_arg_count on second argument rather than last.
11827 2017-04-10 Jeff Law <law@redhat.com>
11829 PR tree-optimization/80374
11830 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
11831 record anything if we can not convert integer_zero_node to the
11834 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
11837 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11838 Enhance special handling given to the TARGET_P9_MINMAX option in
11839 relation to certain other options.
11841 2017-04-10 Bin Cheng <bin.cheng@arm.com>
11843 PR tree-optimization/80153
11844 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
11845 remove POINTER_PLUS_EXPR's base part directly, rather than through
11848 2017-04-10 Richard Biener <rguenther@suse.de>
11849 Bin Cheng <bin.cheng@arm.com>
11851 PR tree-optimization/80153
11852 * tree-affine.c (aff_combination_to_tree): Get base pointer from
11853 the first element of pointer type aff_tree. Build result expr in
11855 (add_elt_to_tree): Convert to type unconditionally. Remove other
11856 fold_convert calls.
11857 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
11858 (rewrite_use_nonlinear_expr): Check invariant using iv information.
11860 2017-04-10 Richard Biener <rguenther@suse.de>
11862 * tree-ssa-structalias.c (find_func_aliases): Properly handle
11865 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
11867 PR rtl-optimization/70478
11868 * lra-constraints.c (curr_small_class_check): New.
11869 (update_and_check_small_class_inputs): New.
11870 (process_alt_operands): Update curr_small_class_check. Disfavor
11871 alternative insn memory operands. Check available regs for small
11874 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
11877 * config/mips/mips.opt (-mvirt): Update description.
11878 * doc/invoke.texi (-mvirt): Likewise.
11880 2017-04-10 Richard Biener <rguenther@suse.de>
11882 PR middle-end/80362
11883 * fold-const.c (fold_binary_loc): Look at unstripped ops when
11884 looking for NEGATE_EXPR in -A / -B to A / B folding.
11886 2017-04-10 Martin Liska <mliska@suse.cz>
11888 PR gcov-profile/80224
11889 * gcov.c (print_usage): Fix usage string.
11890 (get_gcov_intermediate_filename): Remove.
11891 (output_gcov_file): Use both for normal and intermediate format.
11892 (generate_results): Do not initialize special file for
11893 intermediate format.
11895 2017-04-10 Richard Biener <rguenther@suse.de>
11897 PR tree-optimization/80304
11898 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
11901 2017-04-10 Nathan Sidwell <nathan@acm.org>
11904 * config/rs6000/rs6000.c (rs6000_vector_type): New.
11905 (rs6000_init_builtins): Use it.
11907 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11909 * config/arm/arm.md (<mrc>): Add mode to SET source.
11910 (<mrrc>): Likewise.
11912 2017-04-10 Richard Biener <rguenther@suse.de>
11914 PR middle-end/80344
11915 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
11917 2017-04-10 Jakub Jelinek <jakub@redhat.com>
11920 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
11921 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
11922 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
11923 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
11924 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
11925 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
11926 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
11927 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
11928 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
11929 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
11930 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
11931 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
11932 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
11933 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
11934 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
11935 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
11936 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
11937 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
11938 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
11939 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
11940 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
11941 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
11942 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
11944 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
11946 PR rtl-optimization/70478
11947 * lra-constraints.c: Reverse the last patch.
11949 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
11951 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
11952 Add comment for WCHAR_T.
11954 2017-04-08 Martin Liska <mliska@suse.cz>
11957 2017-04-07 Martin Liska <mliska@suse.cz>
11960 * ipa-split.c (split_function): Add function part to a same comdat
11963 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11966 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
11968 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
11970 * rs6000/rs6000.c (vec_load_pendulum): Rename...
11971 (vec_pairing): ...to this.
11972 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
11973 (rs6000_sched_init): Adjust for name change.
11974 (struct rs6000_sched_context): Likewise.
11975 (rs6000_init_sched_context): Likewise.
11976 (rs6000_set_sched_context): Likewise.
11978 2017-04-07 Jakub Jelinek <jakub@redhat.com>
11984 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
11986 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
11987 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
11988 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
11990 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
11992 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
11994 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
11996 PR rtl-optimization/70703
11997 * ira-color.c (update_conflict_hard_regno_costs): Use
11998 int64_t instead of HOST_WIDE_INT.
12000 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
12002 PR rtl-optimization/70478
12003 * lra-constraints.c (process_alt_operands): Disfavor alternative
12004 insn memory operands.
12006 2017-04-07 Jeff Law <law@redhat.com>
12008 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
12009 CALL and NOTE_INSN_CALL_ARG_LOCATION.
12011 2017-04-07 Martin Liska <mliska@suse.cz>
12014 * config/aarch64/aarch64.c (aarch64_process_target_attr):
12015 Show error message instead of an ICE.
12017 2017-04-07 Martin Liska <mliska@suse.cz>
12020 * ipa-split.c (split_function): Add function part to a same comdat
12023 2017-04-07 Richard Biener <rguenther@suse.de>
12025 PR middle-end/80341
12026 * tree.c (get_unwidened): Also handle ! for_type case for
12028 * convert.c (do_narrow): Split out from ...
12029 (convert_to_integer_1): ... here. Do not pass final truncation
12030 type to get_unwidened for TRUNC_DIV_EXPR.
12032 2017-04-07 Richard Biener <rguenther@suse.de>
12034 * tree-affine.c (wide_int_ext_for_comb): Take type rather
12036 (aff_combination_const): Adjust.
12037 (aff_combination_scale): Likewise.
12038 (aff_combination_add_elt): Likewise.
12039 (aff_combination_add_cst): Likewise.
12040 (aff_combination_convert): Likewise.
12041 (add_elt_to_tree): Likewise. Remove unused argument.
12042 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
12044 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
12046 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
12048 * config/arm/arm.c (arm_default_short_enums): Use
12049 ARM_DEFAULT_SHORT_ENUMS.
12050 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
12052 2017-04-06 Jakub Jelinek <jakub@redhat.com>
12055 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
12056 members with redundant out-of-class redeclaration.
12058 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
12061 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
12062 * config/i386/i386.md (*zero_extendsidi2):
12063 Add (?*x,*x) and (?*v,*v) alternatives.
12065 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
12068 * config/i386/i386.c (ix86_expand_builtin)
12069 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
12070 mode from insn data. Convert operands to insn operand mode.
12071 Copy operands that don't satisfy insn predicate to a register.
12073 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
12075 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
12078 2017-04-06 Richard Biener <rguenther@suse.de>
12080 PR tree-optimization/80334
12081 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
12082 preserve alignment of accesses.
12084 2017-04-06 Richard Biener <rguenther@suse.de>
12086 PR tree-optimization/80262
12087 * tree-sra.c (build_ref_for_offset): Preserve address-space
12089 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
12090 Drop useless address-space information on MEM_REF offsets.
12092 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
12094 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
12096 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
12098 PR rtl-optimization/70703
12099 * ira-color.c (update_conflict_hard_regno_costs): Use
12100 HOST_WIDE_INT instead of long.
12102 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
12105 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
12106 not defined for x86_64 target. Add -mmmx target option when __SSE2__
12108 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
12109 for x86_64 target. Handle -m3dnowa option.
12111 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
12113 PR rtl-optimization/70703
12114 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
12115 (update_conflict_hard_regno_costs): Use long instead of unsigned
12116 arithmetic for cost calculation.
12118 2017-04-05 Jakub Jelinek <jakub@redhat.com>
12119 Bernd Edlinger <bernd.edlinger@hotmail.de>
12122 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
12125 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
12128 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
12129 ptr_mode with Pmode throughout.
12130 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
12131 into probe_stack_range and use DImode.
12133 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
12136 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
12137 call_eh_return is true.
12139 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12141 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
12142 Initialize last_match_fntype_index.
12144 2017-04-05 Jakub Jelinek <jakub@redhat.com>
12147 * tree-nvr.c: Include internal-fn.h.
12148 (pass_return_slot::execute): Ignore internal calls without
12151 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12152 Richard Biener <rguenther@suse.de>
12155 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
12156 captures used multiple times, except for the last use.
12157 * generic-match-head.c: Include gimplify.h.
12159 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12161 PR tree-optimization/79390
12162 * target.h (struct noce_if_info): Declare.
12163 * targhooks.h (default_noce_conversion_profitable_p): Declare.
12164 * target.def (noce_conversion_profitable_p): New target hook.
12165 * ifcvt.h (struct noce_if_info): New type, moved from ...
12166 * ifcvt.c (struct noce_if_info): ... here.
12167 (noce_conversion_profitable_p): Renamed to ...
12168 (default_noce_conversion_profitable_p): ... this. No longer
12170 (noce_try_store_flag_constants, noce_try_addcc,
12171 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
12172 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
12173 instead of noce_conversion_profitable_p.
12174 * config/i386/i386.c: Include ifcvt.h.
12175 (ix86_option_override_internal): Don't override
12176 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
12177 (ix86_noce_conversion_profitable_p): New function.
12178 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
12179 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
12180 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
12181 * doc/tm.texi: Regenerated.
12183 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12185 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
12188 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
12191 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
12192 instructions for small multiply cores.
12194 2017-04-04 Jeff Law <law@redhat.com>
12196 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
12198 (mips_expand_vec_perm_const): Initialize elements in orig_perm
12199 that are not set by the loop over the elements.
12201 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12204 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
12205 int mode, convert_modes it to mode as unsigned, otherwise use
12206 lowpart_subreg to mode rather than SImode.
12207 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
12208 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
12209 Use DImode instead of SImode for the shift count operand.
12210 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
12213 2017-04-04 Richard Biener <rguenther@suse.de>
12215 PR middle-end/80281
12216 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
12217 arithmetic done for the negate or the plus. Simplify.
12218 (A - (-B) -> A + B): Likewise.
12219 * fold-const.c (split_tree): Make sure to not negate pointers.
12221 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
12223 PR rtl-optimization/60818
12224 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
12225 a compare of comparisons with the thing compared if this results
12226 in a different machine mode.
12228 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
12230 * alias.c (base_alias_check): Fix typo in comment.
12231 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
12232 * cgraphunit.c (symbol_table::compile): Likewise.
12233 * collect2.c (maybe_run_lto_and_relink): Likewise.
12234 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
12235 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
12236 * config/avr/avr.c (avr_map_op_t): Likewise.
12237 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
12238 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
12239 * config/epiphany/epiphany.md (movcc): Likewise.
12240 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
12241 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
12243 * config/mips/mips.c (mips_save_restore_reg): Likewise.
12244 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
12245 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
12246 * config/sh/sh.c (sh_rtx_costs): Likewise.
12247 * fold-const.c (fold_truth_andor): Likewise.
12248 * genautomata.c (collapse_flag): Likewise.
12249 * gengtype.h (struct type::u::s): Likewise.
12250 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
12251 * input.c (FORMAT_AMOUNT): Likewise.
12252 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
12253 (known_aggs_to_agg_replacement_list): Likewise.
12254 * ipa-inline-analysis.c: Likewise.
12255 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
12256 * ipa-polymorphic-call.c
12257 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
12258 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
12259 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
12261 * modulo-sched.c (apply_reg_moves): Likewise.
12262 * omp-expand.c (build_omp_regions_1): Likewise.
12263 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
12264 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
12265 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
12266 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
12267 * value-prof.c: Likewise.
12268 * var-tracking.c (val_reset): Likewise.
12270 2017-04-03 Richard Biener <rguenther@suse.de>
12272 PR tree-optimization/80275
12273 * fold-const.c (split_address_to_core_and_offset): Handle
12276 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
12278 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
12279 descriptors is at least equal to that of functions.
12281 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
12283 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
12285 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
12288 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
12289 (mov<IMOD4:mode>): New expander.
12290 (*mov<IMOD4:mode>_internal): New insn and split pattern.
12292 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
12294 PR rtl-optimization/79405
12295 * fwprop.c (propagations_left): New variable.
12296 (forward_propagate_into): Decrement it.
12297 (fwprop_init): Initialize it.
12298 (fw_prop): If the variable has reached zero, stop propagating.
12299 (fwprop_addr): Ditto.
12301 2017-03-31 Jakub Jelinek <jakub@redhat.com>
12304 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
12305 a FUNCTION_DECL, pass it as decl instead of origin to
12308 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
12310 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
12313 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
12316 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
12317 TARGET_VSX_SMALL_INTEGER.
12319 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12321 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
12322 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
12324 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
12326 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
12327 extraction from odd-numbered MSA register.
12329 2017-03-31 Jakub Jelinek <jakub@redhat.com>
12331 PR middle-end/80173
12332 * expmed.c (store_bit_field_1): Don't attempt to create
12333 a word subreg out of hard registers wider than word if they
12334 have HARD_REGNO_NREGS of 1 for their mode.
12336 PR middle-end/80163
12337 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
12338 conversions to integer types wider than word and pointer.
12341 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
12342 (rtx_equal_for_cselib_p): Pass 0 to it.
12343 * cselib.c (cselib_hasher::equal): Likewise.
12344 (rtx_equal_for_cselib_1): Add depth argument. If depth
12345 is 128, don't look up VALUE locs and punt. Increment
12346 depth in recursive calls when walking VALUE locs.
12348 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
12350 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
12351 (make_gcov_file_name): Use the canonical path name for generating
12353 (read_line): Fix handling of files with ascii null bytes.
12355 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
12357 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
12358 to initialise a vector register instead
12359 of using a const_int.
12361 2017-03-30 Jakub Jelinek <jakub@redhat.com>
12363 PR translation/80189
12364 * gimplify.c (omp_default_clause): Use %qs instead of %s in
12365 diagnostic messages.
12367 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
12370 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
12371 (dfp_diex_<mode>): Update mode of operand 1.
12372 * doc/extend.texi (dxex, dxexq): Document change to return type.
12373 (diex, diexq): Document change to argument type.
12375 2017-03-30 Martin Jambor <mjambor@suse.cz>
12378 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
12379 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
12380 it reflects the signature changes performed at the callee side.
12381 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
12382 to cgraph_build_function_type_skip_args.
12383 (build_function_decl_skip_args): Adjust call to the above function.
12385 2017-03-30 Jakub Jelinek <jakub@redhat.com>
12388 * config/i386/sse.md
12389 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
12390 register as dest whenever it is a MEM not rtx_equal_p to the
12391 corresponding dup operand, and when forcing into reg move the
12392 reg into the memory afterwards.
12393 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
12394 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
12395 for the force_reg mode.
12396 (avx512vl_vextractf128<mode>): Use register as dest either
12397 always when a MEM, or when it is a MEM not rtx_equal_p to the
12398 corresponding dup operand, or even not when it is a CONST_VECTOR
12399 depending on the mode and lo vs. hi.
12400 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
12402 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
12403 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
12404 Likewise. Require that operands[2] is even.
12405 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
12406 Remove extraneous parens. Require that operands[2] is a multiple
12408 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
12409 operands[0] is a MEM if <mask_applied>, the predicates/constraints
12410 disallow memory then.
12412 2017-03-30 Richard Biener <rguenther@suse.de>
12414 PR tree-optimization/77498
12415 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
12416 to non-constants over backedges.
12418 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
12420 PR rtl-optimization/80233
12421 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
12422 as last_combined_insn. Do not test for BARRIER_P separately.
12424 2017-03-29 Andreas Schwab <schwab@suse.de>
12427 * calls.c (prepare_call_address): Convert funexp to Pmode before
12428 copying to temp reg.
12430 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12432 PR tree-optimization/80158
12433 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
12434 Handle possible future case of more than one alternate
12436 (replace_rhs_if_not_dup): Likewise.
12437 (replace_one_candidate): Likewise.
12439 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
12441 PR rtl-optimization/80193
12442 * ira.c (ira): Do not check allocation for LRA.
12444 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
12446 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
12447 (nvptx_output_simt_exit): Declare.
12448 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
12449 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
12450 (init_softstack_frame): Move initialization of crtl->is_leaf to...
12451 (nvptx_declare_function_name): ...here. Emit declaration of local
12452 memory space buffer for omp_simt_enter insn.
12453 (nvptx_output_unisimt_switch): New.
12454 (nvptx_output_softstack_switch): New.
12455 (nvptx_output_simt_enter): New.
12456 (nvptx_output_simt_exit): New.
12457 * config/nvptx/nvptx.h (struct machine_function): New fields
12458 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
12459 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
12460 (UNSPECV_SIMT_EXIT): Ditto.
12461 (omp_simt_enter_insn): New insn.
12462 (omp_simt_enter): New expansion.
12463 (omp_simt_exit): New insn.
12464 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
12466 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
12467 (expand_GOMP_SIMT_ENTER_ALLOC): New.
12468 (expand_GOMP_SIMT_EXIT): New.
12469 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
12470 (GOMP_SIMT_ENTER_ALLOC): Ditto.
12471 (GOMP_SIMT_EXIT): Ditto.
12472 * target-insns.def (omp_simt_enter): New insn.
12473 (omp_simt_exit): Ditto.
12474 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
12476 (lower_rec_simd_input_clauses): Implement SIMT privatization.
12477 (lower_rec_input_clauses): Likewise.
12478 (lower_lastprivate_clauses): Handle SIMT privatization.
12480 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
12481 (ompdevlow_adjust_simt_enter): New.
12482 (find_simtpriv_var_op): New.
12483 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
12484 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
12486 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
12487 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
12488 (copy_decl_for_dup_finish): Ditto.
12490 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
12492 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
12495 * config/i386/i386.c (ix86_option_override_internal): Always
12496 allow -mpreferred-stack-boundary=3 for 64-bit targets.
12498 2017-03-28 Bin Cheng <bin.cheng@arm.com>
12500 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
12502 2017-03-28 Bin Cheng <bin.cheng@arm.com>
12504 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
12505 mark new edge's irreducible flag accordign to it.
12506 (vect_do_peeling): Check loop preheader edge's irreducible flag
12507 and pass it to function slpeel_add_loop_guard.
12509 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
12511 PR tree-optimization/80218
12512 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
12513 Update block frequencies and counts.
12515 2017-03-28 Richard Biener <rguenther@suse.de>
12517 PR tree-optimization/78644
12518 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
12519 of a simplification result we may not use it at all.
12521 2017-03-28 Richard Biener <rguenther@suse.de>
12524 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
12525 without arguments, generate default definition of a SSA name.
12527 2017-03-28 Richard Biener <rguenther@suse.de>
12529 PR middle-end/80222
12530 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
12531 TYPE_REF_CAN_ALIAS_ALL references.
12532 * fold-const.c (fold_indirect_ref_1): Likewise.
12534 2017-03-28 Martin Liska <mliska@suse.cz>
12537 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
12538 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
12540 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12541 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
12543 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
12544 (EXTRA_SPECS): Define.
12545 (SUBTARGET_EXTRA_SPECS): Likewise.
12546 (SUBTARGET_CPP_SPEC): Likewise.
12547 * config/arc/elf.h (EXTRA_SPECS): Renamed to
12548 SUBTARGET_EXTRA_SPECS.
12549 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
12551 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12553 * config/arc/simdext.md (vst64_insn): Update pattern.
12554 (vld32wh_insn): Likewise.
12555 (vld32wl_insn): Likewise.
12556 (vld64_insn): Likewise.
12557 (vld32_insn): Likewise.
12559 2017-03-28 Marek Polacek <polacek@redhat.com>
12562 * fold-const.c (fold_comparison): Use protected_set_expr_location
12563 instead of SET_EXPR_LOCATION.
12565 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
12567 * tree.c (add_expr): Avoid name lookup warning.
12569 2017-03-27 Jeff Law <law@redhat.com>
12571 PR tree-optimization/80216
12572 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
12573 function name. Limit recursion depth.
12574 (record_temporary_equivalences): Corresponding changes.
12576 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
12578 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
12581 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12584 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
12586 * cfgcleanup.c (reg_note_cfa_p): New array.
12587 (insns_have_identical_cfa_notes): New function.
12588 (old_insns_match_p): Don't cross-jump in between /f
12589 and non-/f instructions. If both i1 and i2 are frame related,
12590 verify all CFA notes, their order and content.
12592 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
12595 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
12596 HImode and SImode with zero extend to DImode to one insn.
12597 (bswap<mode>2_extenddi): Likewise.
12598 (bswapsi2_extenddi): Likewise.
12599 (bswaphi2_extendsi): Likewise.
12600 (bswaphi2): Combine bswap HImode and SImode into one insn.
12601 Separate memory insns from swapping register.
12602 (bswapsi2): Likewise.
12603 (bswap<mode>2): Likewise.
12604 (bswaphi2_internal): Delete, no longer used.
12605 (bswapsi2_internal): Likewise.
12606 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
12607 store, and gpr<-gpr swap insns.
12608 (bswap<mode>2_store): Likewise.
12609 (bswaphi2_reg): Register only splitter, combine with the splitter.
12610 (bswaphi2 splitter): Likewise.
12611 (bswapsi2_reg): Likewise.
12612 (bswapsi2 splitter): Likewise.
12613 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
12614 the insns into load, store, and register/register insns.
12615 (bswapdi2_ldbrx): Likewise.
12616 (bswapdi2_load): Likewise.
12617 (bswapdi2_store): Likewise.
12618 (bswapdi2_reg): Likewise.
12620 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
12622 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
12623 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
12625 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12628 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
12630 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
12631 special handling for target option conflicts between dform
12632 options (-mpower9-dform, -mpower9-dform-vector,
12633 -mpower9-dform-scalar) and -mno-direct-move.
12635 2017-03-27 Richard Biener <rguenther@suse.de>
12637 PR tree-optimization/80181
12638 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
12640 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12642 * config/arc/predicates.md (move_double_src_operand): Replace the
12643 call to move_double_src_operand with a call to address_operand.
12645 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12647 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
12648 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
12649 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
12651 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12653 * config/arc/predicates.md (long_immediate_loadstore_operand):
12654 Consider scaled addresses cases.
12656 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12658 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
12659 restored when in interrupt.
12660 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
12661 doesn't have delay slot.
12663 2017-03-27 Richard Biener <rguenther@suse.de>
12666 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
12667 inlined thunk clones.
12669 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12672 * asan.c (instrument_derefs): Copy over last operand from
12673 original COMPONENT_REF to the new COMPONENT_REF with
12674 DECL_BIT_FIELD_REPRESENTATIVE.
12675 * ubsan.c (instrument_object_size): Likewise.
12677 2017-03-27 Richard Biener <rguenther@suse.de>
12679 PR tree-optimization/80170
12680 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
12681 sure DR/SCEV didnt fold in constants we do not see when looking
12682 at the reference base alignment.
12684 2017-03-27 Richard Biener <rguenther@suse.de>
12686 PR middle-end/80171
12687 * gimple-fold.c (fold_ctor_reference): Properly guard against
12688 NULL return value from canonicalize_constructor_val.
12690 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
12693 * config/i386/i386.c (ix86_expand_builtin)
12694 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
12695 flags reg setting and flags reg using instructions.
12696 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
12697 clobbering instructions to zero extend op2.
12699 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
12701 * doc/install.texi (Configuration) <--with-aix-soname>:
12702 Update link to AIX ld.
12704 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
12706 PR rtl-optimization/80160
12707 PR rtl-optimization/80159
12708 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
12709 reg_alternate_class into account.
12711 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
12714 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
12715 to consider in curr_insn_transform.
12717 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12719 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
12720 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
12721 and emit_mode_inner.
12723 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12725 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
12726 argument to the overloaded builtin variants. Use the new flag to
12727 deprecate certain builtin variants.
12728 * config/s390/s390-builtin-types.def: Add new builtin types.
12729 * config/s390/s390-builtins.h: Support new flags field for
12730 overloaded builtins.
12731 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
12732 (s390_macro_to_expand): Enable vector float data type.
12733 (s390_cpu_cpp_builtins_internal): Indicate support of the new
12734 builtins by incrementing the __VEC__ version number.
12735 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
12737 (s390_resolve_overloaded_builtin): Emit error messages depending
12738 on the builtin flags.
12739 * config/s390/s390.c (s390_expand_builtin): Support additional
12740 flags argument. Change error message to match the messages
12741 emitted in s390-c.c.
12742 * config/s390/s390.md: New UNSPEC_* constants.
12743 (op_type): Add new instruction types.
12744 * config/s390/vecintrin.h: Add new builtins and test data class
12746 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
12747 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
12748 (VEC_INEXACT, VEC_NOINEXACT): New constants.
12749 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
12750 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
12751 ("vec_mergel<mode>"): V_HW -> VEC_HW.
12753 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
12754 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
12755 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
12756 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
12758 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
12759 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
12760 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
12761 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
12763 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
12764 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
12765 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
12766 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
12767 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
12768 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
12769 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
12771 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
12772 ("vec_scatter_element<V_HW_4:mode>_DI")
12773 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
12774 ("vec_fpint<mode>", "vflls")
12775 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
12776 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
12777 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
12778 ("*vec_cmphe<mode>_cc"): ... these.
12780 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
12781 mode constant instead of magic value.
12783 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12785 * config/s390/s390.c (s390_expand_vec_compare): Support other
12786 vector floating point modes than just V2DF.
12787 (s390_expand_vcond): Likewise.
12788 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
12789 (s390_cannot_change_mode_class): Prevent mode changes between TF
12790 and V1TF in vector registers.
12791 * config/s390/s390.md (DF, SF): New mode attributes.
12792 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
12793 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
12794 SFmode support for VRs.
12795 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
12797 (VFT, VF_HW): New mode iterators.
12798 (vw, sdx): New mode attributes.
12799 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
12800 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
12801 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
12802 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
12803 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
12804 also the new vector floating point modes. Renaming to ...
12806 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
12807 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
12808 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
12809 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
12810 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
12811 ("vec_unordered<mode>"): ... these.
12813 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
12814 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
12815 ("*vec_extendv2df"): New insn definitions.
12817 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12819 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
12820 ("mulditi3_2", "*muldi3_sign"): New patterns.
12821 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
12822 rename the pattern definition.
12824 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12826 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
12828 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
12830 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12832 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
12833 instruction if possible.
12834 * config/s390/vector.md (vec_halfnumelts): New mode
12836 ("*vec_vllezlf<mode>"): New pattern.
12838 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12840 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
12841 ("popcountv4si2", "popcountv2di2"): Rename to ...
12842 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
12843 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
12845 ("popcount<mode>2_vxe"): New pattern.
12847 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12849 * common/config/s390/s390-common.c (processor_flags_table): Add
12851 * config.gcc: Add arch12.
12852 * config/s390/driver-native.c (s390_host_detect_local_cpu):
12853 Default to arch12 for unknown CPU model numbers.
12854 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
12855 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
12856 PROCESSOR_max sanity check.
12857 * config/s390/s390-opts.h (enum processor_type): Add
12859 * config/s390/s390.c (processor_table): Add arch12.
12860 (s390_expand_builtin): Add check for B_VXE flag.
12861 (s390_issue_rate): Add PROCESSOR_ARCH12.
12862 (s390_get_sched_attrmask): Likewise.
12863 (s390_get_unit_mask): Likewise.
12864 (s390_sched_score): Enable z13 scheduling for arch12.
12865 (s390_sched_reorder): Likewise.
12866 (s390_sched_variable_issue): Likewise.
12867 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
12869 (s390_tune_attr): Use z13 scheduling also for arch12.
12870 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
12871 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
12872 (TARGET_VXE_P): New macros.
12873 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
12874 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
12875 * config/s390/s390.opt: Add arch12 as processor_type.
12877 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12879 * config/s390/s390.md
12880 ("fixuns_truncdddi2", "fixuns_trunctddi2")
12881 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
12882 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
12884 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
12885 Rename expanders to ...
12887 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
12888 ("fixuns_truncdddi2_emu"): ... these.
12890 ("fixuns_trunc<mode>si2_emu"): New expander.
12892 ("*fixuns_truncdfdi2_z13"): Rename to ...
12893 ("*fixuns_truncdfdi2_vx"): ... this.
12895 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12897 * config/s390/2964.md: Remove the single element vector compare
12898 instructions which are no longer used.
12899 * config/s390/s390.c (s390_select_ccmode): Remove handling of
12901 (s390_canonicalize_comparison): Remove handling of DFmode
12903 (s390_expand_vec_compare_scalar): Remove function.
12904 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
12905 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
12907 ("*cmp<mode>_ccs"): Add wfcdb instruction.
12909 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12911 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
12913 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
12914 will anyway by matched by mov<mode>_64dfp.
12916 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12918 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
12919 vlef/vstef. Add missing operand to vleif.
12921 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12923 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
12924 pair for all vector types with 64 bit elements.
12925 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
12926 * config/s390/vector.md (V_HW_64): ... here.
12927 (V_128_NOSINGLE): New mode iterator.
12928 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
12929 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
12930 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
12931 ("*vec_load_pairv2di"): Change to ...
12932 ("*vec_load_pair<mode>"): ... this one.
12934 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12936 * config/s390/constraints.md: Add comments.
12937 (jKK): Reject element sizes > 8 bytes.
12938 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
12940 * config/s390/s390.md: Add the s_operand checks formerly in
12941 s390_split_ok_p to various splitters where they are still
12943 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
12944 for 128 bit vectors. Plus two splitters.
12946 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12948 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
12951 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12954 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
12955 error if the boundary argument is not constant.
12957 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12959 PR rtl-optimization/80112
12960 * loop-doloop.c (doloop_condition_get): Don't check condition
12961 if cmp isn't SET with IF_THEN_ELSE src.
12963 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12965 PR tree-optimization/80158
12966 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
12967 replacing a candidate statement, also replace it for the
12968 candidate's alternate interpretation.
12969 (replace_rhs_if_not_dup): Likewise.
12970 (replace_one_candidate): Likewise.
12972 2017-03-24 Richard Biener <rguenther@suse.de>
12974 PR tree-optimization/80167
12975 * graphite-isl-ast-to-gimple.c
12976 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
12978 (translate_isl_ast_to_gimple::get_rename): Likewise.
12980 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
12982 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
12983 handling of certain combinations of target options, including the
12984 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
12985 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
12987 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12990 * config/arm/arm.md (*load_multiple): Add reload_completed to
12991 matching condition.
12993 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12994 Richard Biener <rguenth@suse.de>
12996 PR tree-optimization/79908
12997 PR tree-optimization/80136
12998 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
12999 been cast away, gimplify_and_add suffices.
13001 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
13003 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
13005 2017-03-23 Richard Biener <rguenther@suse.de>
13007 PR tree-optimization/80032
13008 * gimplify.c (gimple_push_cleanup): Forced unconditional
13009 cleanups still have to go to the conditional_cleanups
13012 2017-03-22 Jakub Jelinek <jakub@redhat.com>
13014 PR tree-optimization/80072
13015 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
13017 (next_operand_entry_id): Change type to unsigned int.
13018 (sort_by_operand_rank): Make sure to return the right return value
13019 even if unsigned fields are bigger than INT_MAX.
13020 (struct oecount): Change cnt and id type to unsigned int.
13021 (oecount_hasher::equal): Formatting fix.
13022 (oecount_cmp): Make sure to return the right return value
13023 even if unsigned fields are bigger than INT_MAX.
13024 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
13027 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
13028 TREE_READONLY on result if writing it more than once.
13031 * doc/invoke.texi (-fsanitize=thread): Document that with
13032 -fnon-call-exceptions atomics are not able to throw
13036 * tsan.c: Include tree-eh.h.
13037 (instrument_builtin_call): Call maybe_clean_eh_stmt or
13038 maybe_clean_or_replace_eh_stmt where needed.
13039 (instrument_memory_accesses): Add cfg_changed argument.
13040 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
13041 if it returned true.
13042 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
13044 PR rtl-optimization/63191
13045 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
13046 wrapper function, moved the whole old content into ...
13047 (ix86_delegitimize_address_1): ... this. New inline function.
13048 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
13049 true as last argument instead of ix86_delegitimize_address.
13051 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
13053 * config/aarch64/aarch64.c (generic_branch_cost): Copy
13054 cortexa57_branch_cost.
13056 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
13058 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
13060 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13063 * doc/md.texi (Constraints): Document wA constraint.
13064 * config/rs6000/constraints.md (wA): New.
13065 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
13066 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
13067 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
13068 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
13070 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
13073 * gimplify.c (is_oacc_declared): New function.
13074 (oacc_default_clause): Use it to set default flags for acc declared
13075 variables inside parallel regions.
13076 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
13077 declared variables.
13078 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
13079 declare attribute to any decl as necessary.
13081 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
13084 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
13085 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
13086 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
13087 (arm_arch_lpae): This.
13088 * config/arm/arm.c (arm_arch7ve): Rename into ...
13089 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
13090 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
13093 2017-03-22 Martin Liska <mliska@suse.cz>
13096 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
13097 error message instead of an ICE.
13099 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13101 * doc/extend.texi (6.11 Additional Floating Types): Revise.
13103 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
13105 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
13107 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
13110 2017-03-21 Martin Sebor <msebor@redhat.com>
13112 * doc/extend.texi: Use "cannot" instead of "can't."
13113 * doc/hostconfig.texi: Same.
13114 * doc/install.texi: Same.
13115 * doc/invoke.texi: Same.
13116 * doc/loop.texi: Same.
13117 * doc/md.texi: Same.
13118 * doc/objc.texi: Same.
13119 * doc/rtl.texi: Same.
13120 * doc/tm.texi: Same.
13121 * doc/tm.texi.in: Same.
13122 * doc/trouble.texi: Same.
13124 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
13127 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
13128 (collect_checksum_attributes): Set it.
13129 (die_checksum_ordered): Use it.
13131 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13133 PR tree-optimization/79908
13134 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
13135 change: For a VA_ARG whose LHS has been cast away, use
13136 force_gimple_operand to construct the side effects.
13138 2017-03-21 David Malcolm <dmalcolm@redhat.com>
13140 PR translation/80001
13141 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
13142 more amenable to translation.
13143 (oacc_loop_auto_partitions): Likewise.
13145 2017-03-21 Marek Polacek <polacek@redhat.com>
13146 Martin Sebor <msebor@redhat.com>
13148 PR tree-optimization/80109
13149 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
13150 on INTEGRAL_TYPE_P.
13152 2017-03-21 Jakub Jelinek <jakub@redhat.com>
13153 Segher Boessenkool <segher@kernel.crashing.org>
13156 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
13157 check reg_used_between_p between insn and one of succ or succ2
13158 depending on if succ is artificial insn not inserted into insn
13161 2017-03-21 Martin Liska <mliska@suse.cz>
13163 PR gcov-profile/80081
13164 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
13165 * doc/gcc.texi: Include gcov-dump stuff.
13166 * doc/gcov-dump.texi: New file.
13168 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
13170 PR rtl-optimization/79150
13171 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
13172 conditional jump, if the jump is the last insn of the loop.
13174 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13175 Richard Biener <rguenth@suse.de>
13177 PR tree-optimization/79908
13178 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
13179 been cast away, use force_gimple_operand to construct the side
13182 2017-03-21 Martin Liska <mliska@suse.cz>
13184 PR libfortran/79956
13185 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
13188 2017-03-21 Brad Spengler <spender@grsecurity.net>
13191 * plugin.c (htab_hash_plugin): New function.
13192 (add_new_plugin): Use it and adjust.
13193 (parse_plugin_arg_opt): Adjust.
13194 (init_one_plugin): Likewise.
13196 2017-03-21 Richard Biener <rguenther@suse.de>
13198 PR tree-optimization/80032
13199 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
13200 if set force the cleanup to happen unconditionally.
13201 (gimplify_target_expr): Push inserted clobbers with force_uncond
13202 to avoid them being removed by control-dependent DCE.
13204 2017-03-21 Richard Biener <rguenther@suse.de>
13206 PR tree-optimization/80122
13207 * tree-inline.c (copy_bb): Do not expans va-arg packs or
13208 va_arg_pack_len when the inlined call stmt requires pack
13210 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
13212 2017-03-21 Jakub Jelinek <jakub@redhat.com>
13215 * tsan.c (instrument_builtin_call): If the memory model argument
13216 is not a constant, assume it is valid.
13219 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
13222 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
13224 PR rtl-optimization/79910
13225 * combine.c (can_combine_p): Do not allow combining an I0 or I1
13226 if its dest is used by an insn before I2 (other than the combined
13227 insns themselves, which are properly handled already).
13229 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
13232 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
13234 * combine.c (record_used_regs): New static function.
13235 (try_combine): Handle situations where there is an additional
13236 instruction between I2 and I3 which needs to have a LOG_LINK
13240 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
13242 * combine.c (try_combine): Delete redundant i1 test. Call
13243 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
13245 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13248 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
13249 alternatives 13/14.
13251 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13253 PR tree-optimization/80054
13254 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
13255 the optimization if a PHI or any of its arguments is not dominated
13256 by the candidate's basis. Use gphi* rather than gimple* as
13258 (replace_profitable_candidates): Clean up a gimple* variable that
13259 should be a gphi* variable.
13261 2017-03-20 Martin Sebor <msebor@redhat.com>
13264 * doc/extend.texi (attribute constructor): Document present limitation.
13266 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
13269 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
13270 __POWER9_VECTOR__ #ifdef control, change template definition to
13271 use Power9-specific built-in function.
13272 (vec_any_eq): Likewise.
13273 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
13274 to control outcomes from this test.
13275 (vector_ae_<mode>p): For VEC_F modes, likewise.
13277 2017-03-20 Ian Lance Taylor <iant@google.com>
13279 * config/i386/i386.c (ix86_function_regparm): Save an extra
13280 register for -fsplit-stack with DECL_STATIC_CHAIN.
13282 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
13285 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
13286 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
13288 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
13290 * config/riscv/riscv.c (riscv_print_operand): Use "fence
13292 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
13295 2017-03-20 Marek Polacek <polacek@redhat.com>
13298 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
13300 2017-03-20 Richard Biener <rguenther@suse.de>
13302 PR tree-optimization/80113
13303 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
13304 allocate extra SSA name for PHI def.
13305 (add_close_phis_to_outer_loops): Likewise.
13306 (add_close_phis_to_merge_points): Likewise.
13307 (copy_loop_close_phi_args): Likewise.
13308 (copy_cond_phi_nodes): Likewise.
13310 2017-03-20 Martin Liska <mliska@suse.cz>
13312 PR middle-end/79753
13313 * tree-chkp.c (chkp_build_returned_bound): Do not build
13314 returned bounds for a LHS that's not a BOUNDED_P type.
13316 2017-03-20 Martin Liska <mliska@suse.cz>
13320 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
13321 COMPLEX_CST and VECTOR_CST.
13323 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13326 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
13327 target operand. A new splitter adds the clobber statement in case
13328 the target operand is dead anyway.
13330 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
13332 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
13333 to age-old versions of binutils and glibc.
13335 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
13337 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
13339 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
13341 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
13343 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
13345 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
13346 requirement for binutils 2.13.
13348 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
13350 * combine.c (try_combine): Delete redundant i1 test. Call
13351 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
13353 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
13355 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
13356 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
13358 <riscv64-*-elf>: Re-arrange section
13359 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
13360 <riscv32-*-linux>: Likewise.
13361 <riscv64-*-elf>: Likewise
13362 <riscv64-*-linux>: Likewise.
13364 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
13367 * aarch64.opt(verbose-cost-dump): Fix typo.
13369 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
13372 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
13373 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
13375 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
13377 * reload.c (find_reloads): When reloading a nonoffsettable address,
13378 use RELOAD_OTHER for it and its address reloads.
13380 PR rtl-optimization/79910
13381 * combine.c (record_used_regs): New static function.
13382 (try_combine): Handle situations where there is an additional
13383 instruction between I2 and I3 which needs to have a LOG_LINK
13386 2017-03-17 Jeff Law <law@redhat.com>
13388 PR tree-optimization/71437
13389 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
13390 conditional in the hash table first.
13391 (vrp_dom_walker::before_dom_children): Extract condition from
13392 ASSERT_EXPR. Record condition, its inverion and any implied
13393 conditions as well.
13395 2017-03-17 Marek Polacek <polacek@redhat.com>
13396 Markus Trippelsdorf <markus@trippelsdorf.de>
13398 PR tree-optimization/80079
13399 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
13402 2017-03-17 Richard Biener <rguenther@suse.de>
13404 PR middle-end/80075
13405 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
13406 Properly verify the LHS before the RHS possibly claims to be
13408 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
13411 2017-03-17 Martin Jambor <mjambor@suse.cz>
13413 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
13414 (List of -O2 options): Likewise.
13415 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
13418 2017-03-17 Tom de Vries <tom@codesourcery.com>
13420 * gcov-dump.c (print_usage): Print bug_report_url.
13422 2017-03-17 Richard Biener <rguenther@suse.de>
13424 PR middle-end/80050
13425 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
13426 (parser::peek): Likewise.
13428 2017-03-17 Richard Biener <rguenther@suse.de>
13430 PR tree-optimization/80048
13431 * sese.c (free_sese_info): Properly release rename_map and
13432 copied_bb_map elements.
13434 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
13436 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
13437 Add linked-list forward and backlinks. Insert on
13438 construction, remove on destruction.
13439 (class pass_store_merging): Add m_stores_head field.
13440 (pass_store_merging::terminate_and_process_all_chains):
13441 Iterate over m_stores_head list.
13442 (pass_store_merging::terminate_all_aliasing_chains):
13444 (pass_store_merging::execute): Check for debug stmts first.
13445 Push new chains onto the m_stores_head stack.
13447 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
13450 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
13451 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
13452 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
13454 2017-03-16 Jeff Law <law@redhat.com>
13456 PR tree-optimization/71437
13457 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
13458 member function. Implementation moved into after_dom_children
13459 member function and into the threader's thread_outgoing_edges
13461 (dom_opt_dom_walker::after_dom_children): Simplify by moving
13462 some code into new thread_outgoing_edges.
13463 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
13464 definition. Simplify marker handling (do it here). Assume we always
13465 have the available expression and the const/copies tables.
13466 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
13468 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
13469 * tree-vrp.c (equiv_stack): No longer file scoped.
13470 (vrp_dom_walker): New class.
13471 (vrp_dom_walker::before_dom_children): New member function.
13472 (vrp_dom_walker::after_dom_children): Likewise.
13473 (identify_jump_threads): Setup domwalker. Use it rather than
13474 walking edges in a random order by hand. Simplify setup/finalization.
13475 (finalize_jump_threads): Remove.
13476 (vrp_finalize): Do not call identify_jump_threads here.
13477 (execute_vrp): Do it here instead and call thread_through_all_blocks
13480 PR tree-optimization/71437
13481 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
13483 (simplify_stmt_for_jump_threading): Add basic_block argument. All
13485 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
13486 (dom_opt_dom_walker::thread_across_edge): Remove
13487 handle_dominating_asserts argument. All callers changed.
13488 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
13489 changes. Remove calls to lhs_of_dominating_assert. Other
13490 uses of handle_dominating_asserts turn into unconditional code
13491 (simplify_control_stmt_condition_1): Likewise.
13492 (simplify_control_stmt_condition): Likewise.
13493 (thread_through_normal_block, thread_across_edge): Likewise.
13494 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
13495 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
13496 object if it is not an SSA_NAME.
13497 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
13498 before calling into the VRP specific simplifiers.
13499 (identify_jump_threads): Remove handle_dominating_asserts
13502 2017-03-16 Jakub Jelinek <jakub@redhat.com>
13505 * tree-diagnostic.c (default_tree_printer): No longer static.
13506 * tree-diagnostic.h (default_tree_printer): New prototype.
13508 2017-03-16 Tamar Christina <tamar.christina@arm.com>
13510 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
13511 Change ins into fmov.
13513 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13515 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
13516 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
13517 Use h_con constraint for operand 1.
13518 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
13519 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
13521 2017-03-15 Jeff Law <law@redhat.com>
13523 PR tree-optimization/71437
13524 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
13525 (record_temporary_equivalences): Use it.
13527 PR tree-optimization/71437
13528 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
13529 tree-ssa-scopedtables.
13530 (lookup_avail_expr, build_and_record_new_cond): Likewise.
13531 (record_conditions, record_cond, vuse_eq): Likewise.
13532 (record_edge_info): Adjust to API tweak of record_conditions.
13533 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
13534 (record_temporary_equivalences, optimize_stmt): Likewise.
13535 (eliminate_redundant_computations): Likewise.
13536 (record_equivalences_from_stmt): Likewise.
13537 * tree-ssa-scopedtables.c: Include options.h and params.h.
13538 (vuse_eq): New function, moved from tree-ssa-dom.c
13539 (build_and_record_new_cond): Likewise.
13540 (record_conditions): Likewise. Accept vector of conditions rather
13541 than edge_equivalence structure for first argument.
13542 for the first argument.
13543 (avail_exprs_stack::lookup_avail_expr): New member function, moved
13544 from tree-ssa-dom.c.
13545 (avail_exprs_stack::record_cond): Likewise.
13546 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
13547 from tree-ssa-dom.c.
13548 (avail_exprs_stack): Add new member functions lookup_avail_expr
13550 (record_conditions): Declare.
13552 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
13555 * lra-constraints.c (process_alt_operands): Increase reject for
13556 reloading an input/output operand.
13558 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
13561 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
13562 insns to convert from signed/unsigned char/short to IEEE 128-bit
13564 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
13566 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
13569 * config/i386/i386.c (ix86_vector_duplicate_value): Create
13570 subreg of inner mode for values already in registers.
13572 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
13574 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
13575 iteration reg is used after the loop.
13577 2017-03-14 Martin Sebor <msebor@redhat.com>
13579 PR tree-optimization/79800
13580 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
13581 precision in negative-positive range.
13582 (format_floating): Call non-const overload with adjusted precision.
13584 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
13587 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
13590 2017-03-14 Martin Sebor <msebor@redhat.com>
13592 PR middle-end/80020
13593 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
13594 * builtins.def (aligned_alloc): Use it.
13597 * Makefile.in (GTFILES): Add calls.c.
13598 * calls.c: Include "gt-calls.h".
13600 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
13602 PR rtl-optimization/79728
13603 * regs.h (struct target_regs): New field
13604 x_contains_allocatable_regs_of_mode.
13605 (contains_allocatable_regs_of_mode): New macro.
13606 * reginfo.c (init_reg_sets_1): Initialize it, and change
13607 contains_reg_of_mode so it includes global regs as well.
13608 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
13609 rather than contains_regs_of_mode.
13611 2017-03-14 Martin Liska <mliska@suse.cz>
13613 * doc/invoke.texi: Document options that can't be combined with
13614 -fcheck-pointer-bounds.
13616 2017-03-14 Martin Liska <mliska@suse.cz>
13618 PR middle-end/79831
13619 * doc/invoke.texi (-Wchkp): Document the option.
13621 2017-03-14 Martin Liska <mliska@suse.cz>
13623 * Makefile.in: Install gcov-dump.
13625 2017-03-14 Martin Liska <mliska@suse.cz>
13627 * multiple_target.c (expand_target_clones): Bail out for
13628 an invalid attribute.
13630 2017-03-14 Richard Biener <rguenther@suse.de>
13632 * alias.c (struct alias_set_entry): Pack properly.
13633 * cfgloop.h (struct loop): Likewise.
13634 * cse.c (struct set): Likewise.
13635 * ipa-utils.c (struct searchc_env): Likewise.
13636 * loop-invariant.c (struct invariant): Likewise.
13637 * lra-remat.c (struct cand): Likewise.
13638 * recog.c (struct change_t): Likewise.
13639 * rtl.h (struct address_info): Likewise.
13640 * symbol-summary.h (function_summary): Likewise.
13641 * tree-loop-distribution.c (struct partition): Likewise.
13642 * tree-object-size.c (struct object_size_info): Likewise.
13643 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
13644 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
13645 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
13646 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
13647 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
13648 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
13649 (struct _stmt_vec_info): Likewise.
13651 2017-03-14 Martin Liska <mliska@suse.cz>
13654 * multiple_target.c (create_dispatcher_calls): Check that
13655 a target can create a function dispatcher.
13657 2017-03-14 Martin Liska <mliska@suse.cz>
13660 * multiple_target.c (expand_target_clones): Drop local.local
13661 flag for default implementation.
13663 2017-03-14 Richard Biener <rguenther@suse.de>
13665 PR tree-optimization/80030
13666 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
13668 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
13670 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
13671 gcc_fallthrough() instead of __attribute__((fallthrough));
13673 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13675 * doc/gcc.texi: Remove "up" link to (DIR).
13676 * doc/gccint.texi: Ditto.
13678 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13680 * doc/install.texi (Specific) <avr>: Remove reference to
13683 2017-03-13 Jeff Law <law@redhat.com>
13685 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
13686 attribute rather than comments.
13688 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
13689 match_scratch operand is highest.
13691 2017-03-13 Martin Liska <mliska@suse.cz>
13693 PR middle-end/78339
13694 * ipa-pure-const.c (warn_function_noreturn): If the declarations
13695 is a CHKP clone, use original declaration.
13697 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13699 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
13700 (arc_conditional_register_usage): Use a different allocation order
13701 when optimizing for size.
13702 * common/config/arc/arc-common.c (arc_option_optimization_table):
13703 Section anchors default on when optimizing for size.
13705 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13707 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
13709 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13711 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
13712 * config/arc/arc.md (cpu_facility): Add cd variant.
13713 (*movqi_insn): Add code density variant.
13714 (*movhi_insn): Likewise.
13715 (*movqi_insn): Likewise.
13716 (*addsi3_mixed): Likewise.
13717 (subsi3_insn): Likewise.
13719 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13721 * config/arc/arc.md (movsi_cond_exec): Update constraint.
13723 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13725 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
13726 expressions with MINUS and UNARY ops.
13728 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13731 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
13733 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
13734 between vec_select and vector argument.
13735 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
13736 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
13737 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
13738 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
13739 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
13740 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
13742 2017-03-13 Richard Biener <rguenther@suse.de>
13745 * params.def (vect-max-peeling-for-alignment): Fix typo.
13747 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13749 * doc/install.texi (Specific) <mips-*-*>: Remove description of
13750 issue that only occurred with binutils below 2.18.
13752 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13754 * doc/install.texi (Specific) <cris-axis-elf>: No longer
13755 refer to binutils 2.11/2.12 minimum.
13757 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13759 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
13760 ftp.kernel.org and simplify binutils requirement.
13762 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
13764 * doc/invoke.texi (Warning Options): Fix spelling of link-time
13766 (Optimize Options): Ditto. Also remove redundancy.
13768 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13770 PR translation/79848
13771 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
13773 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
13774 to G_ to avoid double translation.
13776 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13778 PR translation/79923
13779 * auto-profile.c (get_combined_location): Convert leading
13780 character of diagnostics to lower case and remove trailing period.
13781 (read_profile): Likewise for various diagnostics.
13782 * config/arm/arm.c (arm_option_override): Remove trailing period
13783 from various diagnostics.
13784 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
13785 (msp430_expand_delay_cycles): Likewise.
13787 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13790 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
13791 full command-line argument, rather than just "str".
13792 (aarch64_validate_march): Likewise.
13793 (aarch64_validate_mtune): Likewise.
13795 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
13797 PR rtl-optimization/78911
13798 * lra-assigns.c (must_not_spill_p): New function.
13799 (spill_for): Use it.
13801 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13803 PR tree-optimization/79981
13804 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
13805 ATOMIC_COMPARE_EXCHANGE ifn result.
13806 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
13807 IFN_ATOMIC_COMPARE_EXCHANGE.
13809 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13812 * opts.c (parse_sanitizer_options): Add missing question mark to
13813 "did you mean" message.
13815 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13817 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
13819 (VMULEUH_UNS): Likewise.
13820 (VMULOUB_UNS): Likewise.
13821 (VMULOUH_UNS): Likewise.
13822 * config/rs6000/rs6000.c (builtin_function_type): Remove
13823 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
13825 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13828 * read-rtl-function.c (function_reader::read_rtx_operand): Update
13829 x with result of extra_parsing_for_operand_code_0.
13830 (function_reader::extra_parsing_for_operand_code_0): Convert
13831 return type from void to rtx, returning x. When reading
13832 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
13833 larger size containing struct block_symbol.
13835 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
13837 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
13838 -mfloat128-hardware without -m64.
13840 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
13843 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
13844 entries to the case statement that marks unsigned arguments to
13845 overloaded functions.
13847 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
13849 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
13850 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
13852 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
13855 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
13856 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
13858 2017-03-10 Martin Liska <mliska@suse.cz>
13862 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
13863 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
13866 2017-03-10 Olivier Hainque <hainque@adacore.com>
13868 * tree-switch-conversion (array_value_type): Start by resetting
13869 candidate type to it's main variant.
13871 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13873 PR rtl-optimization/79909
13874 * combine.c (try_combine): Use simplify_replace_rtx on individual
13875 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
13876 of the whole CALL_INSN_FUNCTION_USAGE.
13878 PR tree-optimization/79972
13879 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
13880 get_range_info on SSA_NAMEs. Formatting fixes.
13882 2017-03-10 Richard Biener <rguenther@suse.de>
13883 Jakub Jelinek <jakub@redhat.com>
13885 PR tree-optimization/77975
13886 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
13887 edge to be constant.
13888 (get_val_for): For constant x return it. Formatting fix.
13889 (loop_niter_by_eval): Avoid pointless looping if the next iteration
13890 would use the same bases as the current one.
13892 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13894 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
13895 instead of vec_select for V1TImode.
13896 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
13898 (VSX_LE_128): Add V1TI to this mode iterator.
13899 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
13900 (*vsx_le_perm_store_<mode>): Likewise.
13901 (pre-reload splitter for VSX stores): Likewise.
13902 (post-reload splitter for VSX stores): Likewise.
13903 (*vsx_xxpermdi2_le_<mode>): Likewise.
13904 (*vsx_lxvd2x2_le_<mode>): Likewise.
13905 (*vsx_stxvd2x2_le_<mode>): Likewise.
13907 2017-03-09 Michael Eager <eager@eagercon.com>
13909 Correct failures with --enable-checking=yes,rtl.
13911 * config/microblaze/microblaze.c (microblaze_expand_shift):
13912 Replace GET_CODE test with CONST_INT_P and INTVAL test with
13913 test for const0_rtx.
13914 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
13915 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
13917 2017-03-09 Richard Biener <rguenther@suse.de>
13919 PR tree-optimization/79977
13920 * graphite-scop-detection.c (scop_detection::merge_sese):
13921 Handle the case of extra exits to blocks dominating the entry.
13923 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
13925 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13928 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
13930 PR rtl-optimization/79949
13931 * lra-constraints.c (process_alt_operands): Check memory when
13932 trying to predict a cycle. Print about the overall increase.
13934 2017-03-09 Richard Biener <rguenther@suse.de>
13936 PR middle-end/79971
13937 * gimple-expr.c (useless_type_conversion_p): Preserve
13938 TYPE_SATURATING for fixed-point types.
13940 2017-03-09 Richard Biener <rguenther@suse.de>
13943 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
13944 alignment of BLKmode params.
13946 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13949 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
13950 (VALL_NO_V2Q): Likewise.
13952 * config/aarch64/aarch64-simd.md
13953 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
13955 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
13956 VALL_NO_V2Q mode iterator.
13957 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
13959 2017-03-09 Martin Liska <mliska@suse.cz>
13961 PR tree-optimization/79631
13962 * tree-chkp-opt.c (chkp_is_constant_addr): Call
13963 tree_int_cst_sign_bit just for INTEGER constants.
13965 2017-03-09 Martin Liska <mliska@suse.cz>
13969 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
13972 2017-03-09 Marek Polacek <polacek@redhat.com>
13975 * tree.c (inchash::add_expr): Handle TREE_VEC.
13977 2017-03-09 Martin Liska <mliska@suse.cz>
13980 (chkp_narrow_size_and_offset): New function.
13981 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
13982 (void chkp_parse_bit_field_ref): New function.
13983 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
13984 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
13986 2017-03-09 Martin Liska <mliska@suse.cz>
13989 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
13990 (chkp_find_bounds_1): Remove gcc_unreachable.
13992 2017-03-09 Jakub Jelinek <jakub@redhat.com>
13995 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
13996 BUILT_IN_SYNC*, determine the access type from the size suffix and
13997 always build a MEM_REF with that type. Handle forgotten
13998 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
14001 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
14002 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
14003 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
14004 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
14005 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
14006 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
14007 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
14008 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
14009 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
14010 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
14011 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
14012 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
14013 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
14014 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
14015 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
14016 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
14017 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
14018 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
14019 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
14020 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
14021 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
14022 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
14023 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
14024 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
14025 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
14026 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
14027 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
14028 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
14029 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
14030 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
14031 definitions outside of __OPTIMIZE__ guarded section.
14034 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
14035 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
14036 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
14037 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
14040 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14042 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
14043 ("vfenez<mode>"): Add missing constraints.
14045 2017-03-08 Martin Sebor <msebor@redhat.com>
14048 * config/nds32/nds32.c (nds32_option_override):
14049 Fix misspelled diagnostic.
14051 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14054 * gimplify.c (gimplify_omp_for): Replace index var in outer
14055 taskloop statement with an artificial variable and add
14056 OMP_CLAUSE_PRIVATE clause for it.
14058 2017-03-08 Richard Biener <rguenther@suse.de>
14060 PR tree-optimization/79955
14061 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
14062 for accesses that are completely outside of the variable.
14064 2017-03-08 Andrew Haley <aph@redhat.com>
14066 PR tree-optimization/79943
14067 * tree-ssa-loop-split.c (compute_new_first_bound): When
14068 calculating the new upper bound, (END-BEG) should be added, not
14071 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14073 * config/avr/avr.md (setmemhi): Make sure match_dup
14074 operand number comes before match_scratch.
14076 2017-03-08 Richard Biener <rguenther@suse.de>
14078 PR tree-optimization/79920
14079 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
14080 with ncopies == 1 to ...
14081 (vect_transform_slp_perm_load): ... here. Properly compute
14082 all element loads by iterating VF times over the group. Do
14083 not handle ncopies (computed in a broken way) in
14084 vect_create_mask_and_perm.
14086 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14089 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
14090 is a uniform vector, use uniform_vector_p return value instead of
14091 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
14093 2017-03-07 Marek Polacek <polacek@redhat.com>
14095 PR middle-end/79809
14096 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
14097 (alloca_call_type): Likewise.
14099 2017-03-07 Martin Liska <mliska@suse.cz>
14101 * gcov.c (process_args): Put comment to correct location.
14103 2017-03-07 Martin Liska <mliska@suse.cz>
14105 PR middle-end/68270
14106 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
14107 Use array_at_struct_end_p instead of DECL_CHAIN (field).
14108 (chkp_narrow_bounds_for_field): Likewise.
14109 (chkp_parse_array_and_component_ref): Pass one more argument to
14112 2017-03-07 Richard Biener <rguenther@suse.de>
14114 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
14117 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
14119 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
14120 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
14122 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14125 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
14126 to end of description.
14127 (PARAM_MAX_STORES_TO_MERGE): Likewise.
14129 2017-03-07 Jakub Jelinek <jakub@redhat.com>
14131 PR rtl-optimization/79901
14132 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
14134 (*avx512f_<code><mode>3<mask_name>): ... this.
14135 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
14136 iterator instead of VI8_AVX2_AVX512BW.
14138 PR rtl-optimization/79901
14139 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
14140 min/max expander, expand it using expand_vec_cond_expr.
14143 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
14146 2017-03-06 Jakub Jelinek <jakub@redhat.com>
14149 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
14150 to void * for PCH reasons.
14151 * dwarf2out.c (output_loc_operands, output_die): Cast
14152 v.val_vec.array to unsigned char *.
14154 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
14157 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
14160 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
14162 PR rtl-optimization/79571
14163 * lra-constraints.c (process_alt_operands): Calculate static
14164 reject and subtract it from overall when only addresses will be
14167 2017-03-06 Julia Koval <julia.koval@intel.com>
14170 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
14171 incoming stack boundary to 128 for 64-bit targets.
14173 2017-03-06 Richard Biener <rguenther@suse.de>
14175 PR tree-optimization/79894
14176 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
14177 to NULL after folding it.
14179 2017-03-06 Richard Biener <rguenther@suse.de>
14181 PR tree-optimization/79824
14182 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
14183 check disabling peeling for gaps.
14185 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
14187 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
14188 attributes): Document gettimeofday.
14190 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
14192 * config/s390/s390.c (s390_option_override_internal): Set
14193 PARAM_MIN_VECT_LOOP_BOUND
14195 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
14197 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
14198 * config/s390/s390.md: Likewise.
14200 2017-03-06 Jakub Jelinek <jakub@redhat.com>
14203 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
14204 (<avx2_avx512>_perm<mode>): Rename to ...
14205 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
14207 (<avx512>_perm<mode>_mask): Rename to ...
14208 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
14210 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
14211 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
14212 instead of VI8F_256_512.
14213 (avx512f_perm<mode>): New define_expand.
14214 (avx512f_perm<mode>_mask): Likewise.
14215 (avx512f_perm<mode>_1<mask_name>): New define_insn.
14216 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
14218 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14220 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
14221 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
14223 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
14225 2017-03-06 Martin Liska <mliska@suse.cz>
14228 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
14229 when having a SSA NAME w/o VAR_DECL assigned to it.
14231 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14233 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
14234 msa_dpsub_<su>_d): Fix MODE for vec_select.
14236 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14238 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
14240 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
14242 2017-03-06 Richard Biener <rguenther@suse.de>
14244 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
14245 * plugin.c (register_plugin_info): Likewise.
14246 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
14248 2017-03-05 Jakub Jelinek <jakub@redhat.com>
14250 * config/i386/sse.md (sse_storehps, sse_storelps,
14251 avx_<castmode><avxsizesuffix>_<castmode>,
14252 avx512f_<castmode><avxsizesuffix>_<castmode>,
14253 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
14254 in condition that at least one operand is not a MEM.
14256 2017-03-03 Jakub Jelinek <jakub@redhat.com>
14258 PR middle-end/79805
14259 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
14260 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
14262 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
14263 gimple_call_nothrow_p flag based on whether original builtin can throw.
14264 If it can, emit following stmts on the fallthrough edge.
14265 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
14266 don't create new bb if inserting just debug stmts on the edge, try to
14267 insert them on the fallthru bb or just reset debug stmts.
14269 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
14272 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
14273 restore recog_data (including the operand rtxes inside it) around
14274 the call to get_insn_template.
14276 2017-03-03 Martin Sebor <msebor@redhat.com>
14278 PR tree-optimization/79699
14279 * context.c (context::~context): Free MPFR caches to avoid
14280 a memory leak on program exit.
14282 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14284 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
14285 Use wide_int::ulow () instead of .elt (0).
14287 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
14289 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
14290 (*pushxf): Limit oF constraint to 32bit targets and add oC
14291 constraint for 64bit targets.
14292 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
14293 (*pushdf): Change rmF constraint to rmC.
14295 2017-03-03 Martin Liska <mliska@suse.cz>
14297 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
14298 Remove unused variable.
14300 2017-03-03 Jakub Jelinek <jakub@redhat.com>
14303 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
14304 is a memory operand, increase num_memory.
14305 (ix86_expand_args_builtin): Likewise.
14307 2017-03-03 Jan Hubicka <jh@suse.cz>
14310 * ipa-devirt.c (maybe_record_node): Properly handle
14311 __cxa_pure_virtual visibility.
14313 2017-03-03 Martin Liska <mliska@suse.cz>
14315 PR tree-optimization/79803
14316 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
14318 (pass_loop_prefetch::execute): Disabled optimization if an
14319 assumption about L1 cache size is not met.
14321 2017-03-03 Martin Liska <mliska@suse.cz>
14323 PR rtl-optimization/79574
14324 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
14325 (hash_scan_set): Likewise.
14326 (dump_hash_table): Likewise.
14327 (hoist_code): Likewise.
14329 2017-03-03 Richard Biener <rguenther@suse.de>
14331 * fixed-value.c (fixed_from_string): Restore use of elt (1)
14332 in place of uhigh ().
14333 (fixed_convert_from_real): Likewise.
14335 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
14338 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
14340 2017-03-03 Richard Biener <rguenther@suse.de>
14342 PR middle-end/79818
14343 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
14344 TYPE_OVERFLOW_UNDEFINED check.
14346 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14348 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
14350 (vector_ae_<mode>_p): Likewise.
14351 (vector_nez_<mode>_p): Likewise.
14352 (vector_ne_v2di_p): Likewise.
14353 (vector_ae_v2di_p): Likewise.
14354 (vector_ne_<mode>_p): Likewise.
14355 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
14357 (vsx_tsqrt<mode>2_fe): Likewise.
14359 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
14362 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
14364 2017-03-02 Jakub Jelinek <jakub@redhat.com>
14366 PR rtl-optimization/79780
14367 * cprop.c (one_cprop_pass): When second and further conditional trap
14368 in a single basic block is turned into an unconditional trap, turn it
14369 into a deleted note to avoid RTL verification failures.
14371 2017-03-02 Richard Biener <rguenther@suse.de>
14373 * fold-const.c (const_binop): Use ulow () instead of elt (0).
14375 2017-03-02 Richard Biener <rguenther@suse.de>
14377 PR tree-optimization/79345
14379 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
14380 param and abort the walk, returning -1 if it is hit.
14381 (walk_aliased_vdefs): Take a limit param and pass it on.
14382 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
14383 defaulting to 0 and return a signed int.
14384 * tree-ssa-uninit.c (struct check_defs_data): New struct.
14385 (check_defs): New helper.
14386 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
14387 about uninitialized memory.
14388 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
14389 bogus uninitialized warning.
14390 (fixed_convert_from_real): Likewise.
14392 2017-03-02 Bin Cheng <bin.cheng@arm.com>
14394 PR tree-optimization/66768
14395 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
14396 iv_use if base object can't be determined.
14398 2017-03-02 Jakub Jelinek <jakub@redhat.com>
14400 PR tree-optimization/79345
14401 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
14402 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
14403 (get_pattern_stats): Initialize it.
14404 * genemit.c (gen_expand): Verify match_scratch numbers come after
14405 match_operand/match_dup numbers.
14406 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
14407 match_scratch numbers.
14408 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
14410 * config/s390/s390.md (trunctdsd2): Likewise.
14412 2017-03-02 Richard Biener <rguenther@suse.de>
14414 * wide-int.h (wide_int_storage::operator=): Implement in terms
14417 2017-03-02 Richard Biener <rguenther@suse.de>
14419 PR tree-optimization/79777
14420 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
14421 the to insert expression to sth existing.
14423 2017-03-01 Martin Sebor <msebor@redhat.com>
14425 PR middle-end/79692
14426 * gimple-ssa-sprintf.c
14427 (directive::known_width_and_precision): New function.
14428 (format_integer): Use it.
14429 (get_mpfr_format_length): Consider the full range of precision
14430 when computing %g output with the # flag. Set the likely byte
14431 count to 3 rather than 1 when precision is indeterminate.
14432 (format_floating): Correct the lower bound of precision.
14434 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14436 * doc/invoke.texi: Document default code model for 64-bit Linux.
14438 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14441 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
14442 udiv rather than div since input pattern is unsigned.
14444 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
14446 * config/i386/i386.c (print_reg): Warn for values of
14447 unsupported size in integer register.
14449 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
14452 * config/rs6000/predicates.md (current_file_function_operand): Do
14453 not allow self calls to be local if the function is replaceable.
14455 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
14458 * config/rs6000/altivec.h (vec_ctz and others): Change the
14459 preprocessor macro that controls conditional compilation from
14460 _ARCH_PWR9 to __POWER9_VECTOR__.
14461 (vec_all_ne): Change parameterization of __altivec_scalar_pred
14462 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
14463 control (instead of _ARCH_PWR9 control) so that template
14464 definition uses power9-specific function.
14465 (vec_any_eq): Likewise.
14466 (vec_all_ne): Change macro definition to use a power9-specific
14467 expansion under #ifdef __POWER9_VECTOR__ control (instead of
14468 _ARCH_PWR9 control).
14469 (vec_any_eq) Likewise.
14470 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
14471 expansion for CMPNEF to remove support for xvcmpnesp instruction.
14472 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
14473 support for xvcmpnedp instruction.
14474 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
14475 macro expansion so that Power9 implementation of vec_all_ne does
14476 not use the AltiVec predicate framework.
14477 (VCMPNEH_P): Likewise.
14478 (VCMPNEW_P): Likewise.
14479 (VCMPNED_P): Likewise.
14480 (VCMPNEFP_P): Likewise.
14481 (VCMPNEDP_P): Likewise.
14482 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
14483 implementation of vec_any_eq to not use AltiVec predicate
14485 (VCMPAEH_P): Likewise.
14486 (VCMPAEW_P): Likewise.
14487 (VCMPAED_P): Likewise.
14488 (VCMPAEFP_P): Likewise.
14489 (VCMPAEDP_P): Likewise.
14490 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
14491 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
14492 not use the AltiVec predicate framework.
14493 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
14494 of vec_any_eq to not use AltiVec predicate framework.
14495 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
14496 support for predefined __POWER9_VECTOR__ macro to indicate that
14497 Power9 instruction selection is enabled.
14498 (altivec_overloaded_builtins): Remove extraneous
14499 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
14500 function argument types RS6000_BTI_bool_V16QI and
14501 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
14502 entry for overloaded function argument types RS6000_BTI_bool_V4SI
14503 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
14504 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
14505 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
14506 Power9 for implementations of vec_cmpne. Change the signature for
14507 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
14508 (representing vec_all_ne) to remove the previously described first
14509 argument of type RS6000_BTI_INTSI, as this was an artifact of
14510 reliance on the AltiVec predicate framework, which is no longer
14511 used in the implementation of these functions. Add
14512 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
14513 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
14514 since, unlike the AltiVec predicate framework implementation, we
14515 do not share function descriptors between vec_alle and vec_anyeq.
14516 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
14517 set of modes that receive special treatment even when
14518 TARGET_P9_VECTOR is true. The special treatment emits code that
14519 does not depend on Power9 instructions.
14520 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
14521 define_expand to not rely on AltiVec predicate framework.
14522 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
14524 (vector_ne_v2di_p): Change this define_expand to not rely on
14525 AltiVec predicate framework.
14526 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
14528 (vector_ne_<mode>_p): Change this define_expand to not rely on
14529 AltiVec predicate framework.
14530 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
14532 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
14533 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
14534 define_insn pattern.
14535 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
14536 define_insn pattern because the xvcmpne<VSs>. instruction is not
14538 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
14539 instruction is not supported.
14541 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14543 * config/nvptx/nvptx.c: Include intl.h.
14545 2017-03-01 Martin Jambor <mjambor@suse.cz>
14548 * ipa-prop.h (ipa_bits): Removed field known.
14549 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
14550 to pointers. Adjusted their comments to warn about their sharing.
14551 (ipcp_transformation_summary): Change bits to a vector of pointers.
14552 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
14553 (ipa_get_ipa_bits_for_value): Declare.
14554 * tree-vrp.h (value_range): Mark as GTY((for_user)).
14555 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
14556 (ipa_bits_hash_table): Likewise.
14557 (ipa_vr_ggc_hash_traits): Likewise.
14558 (ipa_vr_hash_table): Likewise.
14559 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
14560 being pointers and vr_known being removed.
14561 (ipa_set_jf_unknown): Likewise.
14562 (ipa_get_ipa_bits_for_value): New function.
14563 (ipa_set_jfunc_bits): Likewise.
14564 (ipa_get_value_range): New overloaded functions.
14565 (ipa_set_jfunc_vr): Likewise.
14566 (ipa_compute_jump_functions_for_edge): Use the above functions to
14567 construct bits and vr parts of jump functions.
14568 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
14569 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14571 (ipcp_grow_transformations_if_necessary): Also allocate
14572 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14574 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
14575 them. Fix too long lines.
14576 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
14577 vr_known being removed.
14578 (ipa_read_jump_function): Use new setter functions to construct bits
14579 and vr parts of jump functions or set them to NULL.
14580 (write_ipcp_transformation_info): Adjust for bits being pointers.
14581 (read_ipcp_transformation_info): Likewise.
14582 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
14584 Include gt-ipa-prop.h.
14585 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
14587 (ipcp_store_bits_results): Likewise.
14588 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
14589 Do not write to existing jump functions but use a temporary instead.
14591 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14594 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
14595 attempt to use its first operand as BIT_FIELD_REF base.
14597 2017-03-01 Richard Biener <rguenther@suse.de>
14599 PR middle-end/79721
14600 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
14601 interpolating formula in wrapping arithmetic.
14602 (chrec_apply): Convert chrec_evaluate return value to wanted type.
14604 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14606 PR tree-optimization/79734
14607 * tree-vect-generic.c (expand_vector_condition): Optimize
14608 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
14609 Handle VEC_COND_EXPR where comparison has different inner width from
14610 type's inner width.
14612 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
14614 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
14615 markup, and similar issues. Remove @opindex entries for things
14616 that aren't options. Add missing -mmpy-option entries.
14618 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14620 PR tree-optimization/79737
14621 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
14622 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
14623 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
14624 instead of byte_size. Formatting fix.
14625 (shift_bytes_in_array_right): Formatting fix.
14627 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
14630 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
14631 condition on optimize for the leaf function test.
14633 2017-02-28 Martin Liska <mliska@suse.cz>
14636 * read-rtl-function.c (function_reader::handle_unknown_directive):
14637 Bail out when one uses -flto.
14639 2017-02-28 Martin Liska <mliska@suse.cz>
14641 * common.opt: Replace space with tabular for options of <number>
14643 * config/i386/i386.opt: Show <number> value for
14644 -mlarge-data-threshold.
14645 * opts.c (print_filtered_help): Do not display number in hexadecimal
14648 2017-02-28 Martin Liska <mliska@suse.cz>
14650 * common.opt: Fix --help=option -Q for options which are of
14653 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
14655 * config/i386/i386.c (print_reg): Error out for values
14656 of 8-bit size in invalid integer register.
14658 2017-02-28 Martin Sebor <msebor@redhat.com>
14660 PR tree-optimization/79691
14661 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
14663 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14666 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
14667 gcc_unreachable with output_operand_lossage.
14669 2017-02-28 Richard Biener <rguenther@suse.de>
14671 PR tree-optimization/79740
14672 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
14674 (visit_nary_op): Insert the nary into the hashtable if we
14675 pattern-matched sth.
14676 * tree-ssa-pre.c (eliminate_insert): Robustify.
14678 2017-02-28 Richard Biener <rguenther@suse.de>
14680 PR middle-end/79731
14681 * fold-const.c (decode_field_reference): Reject out-of-bound
14684 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14686 * config/i386/i386.c: Include intl.h.
14687 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
14688 instead of just cond ? "..." : "...".
14689 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
14690 * coverage.c (read_counts_file): Likewise.
14691 * omp-offload.c: Include intl.h.
14692 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
14693 of just cond ? "..." : "...".
14694 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
14695 of just cond ? "..." : "...".
14697 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
14700 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
14702 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
14703 'tune for' CPU name.
14704 * config/arm/arm-cpu-data.h: Regenerated.
14706 2017-02-28 Richard Biener <rguenther@suse.de>
14708 PR tree-optimization/79732
14709 * tree-inline.c (expand_call_inline): Do not shadow var.
14711 2017-02-28 Richard Biener <rguenther@suse.de>
14713 PR tree-optimization/79723
14714 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
14715 address-space properly.
14717 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
14719 * doc/optinfo.texi (Optimization groups): Fix option used for
14721 * doc/invoke.texi (-fopt-info): Document "omp".
14722 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
14723 (OPTGROUP_ALL): Add OPTGROUP_OMP.
14724 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
14725 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
14726 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
14728 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
14730 * dumpfile.c (optgroup_options): Instead of "openmp", associate
14731 OPTGROUP_OMP with "omp".
14733 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
14736 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
14737 for arithmetic shift of unsigned V2DI.
14739 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
14741 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
14742 arc/linux.h headers.
14743 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
14744 (LINK_SPEC): Likewise.
14745 (ARC_TLS_EXTRA_START_SPEC): Likewise.
14746 (EXTRA_SPECS): Likewise.
14747 (STARTFILE_SPEC): Likewise.
14748 (ENDFILE_SPEC): Likewise.
14749 (LIB_SPEC): Likewise.
14750 (TARGET_SDATA_DEFAULT): Likewise.
14751 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
14752 (MULTILIB_DEFAULTS): Likewise.
14753 (DWARF2_UNWIND_INFO): Likewise.
14754 * config/arc/big.h: New file.
14755 * config/arc/elf.h: Likewise.
14756 * config/arc/linux.h: Likewise.
14757 * config/arc/t-uClibc: Remove.
14759 2017-02-27 Bin Cheng <bin.cheng@arm.com>
14761 PR tree-optimization/77536
14762 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
14763 (tree_transform_and_unroll_loop): Use above function to compute the
14764 estimated niter of unrolled loop and use it when scaling profile.
14765 Also use count info rather than frequency if it's non-zero.
14766 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
14767 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
14768 (vect_transform_loop): Call above function.
14770 2017-02-27 Richard Biener <rguenther@suse.de>
14772 PR tree-optimization/45397
14773 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
14774 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
14775 (visit_nary_op): Add pattern matching for CSEing sign-changed
14776 or truncated operations with wider ones.
14778 2017-02-27 Richard Biener <rguenther@suse.de>
14780 PR tree-optimization/79690
14781 * tree-vect-stmts.c (vectorizable_store): Use vector type
14782 built from the DR with address-space.
14784 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
14786 * doc/invoke.texi (Optimize Options): Refine the description
14787 of asan-use-after-return.
14789 2017-02-25 Alan Modra <amodra@gmail.com>
14791 PR rtl-optimization/79584
14792 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
14793 base, not ad->base_term, the reg within base. Remove assertion
14794 that ad->base == ad->base_term. Replace gen_int_mode using
14795 bogus mode with const0_rtx.
14797 2017-02-25 Jakub Jelinek <jakub@redhat.com>
14799 PR middle-end/79396
14800 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
14801 FMA_EXPR like tcc_binary or tcc_unary.
14803 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
14806 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
14808 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
14809 (output_loc_operands): Handle DW_OP_call_ref and
14810 DW_OP_GNU_variable_value.
14811 (struct variable_value_struct): New type.
14812 (struct variable_value_hasher): Likewise.
14813 (variable_value_hash): New variable.
14814 (string_types): Remove.
14815 (copy_loc_descr): New function.
14816 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
14817 (prepend_loc_descr_to_each): New function.
14818 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
14819 instead of add_loc_descr_to_each if the first argument is single
14820 location list and the second has multiple.
14821 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
14822 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
14823 when looking for variable value which doesn't have other location info.
14824 (loc_list_from_tree): Formatting fix.
14825 (gen_array_type_die): Simplify DW_AT_string_length handling.
14826 (adjust_string_types): Remove.
14827 (gen_subprogram_die): Don't call adjust_string_types nor test/set
14828 string_types. Call resolve_variable_values.
14829 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
14830 (resolve_addr_in_expr): Likewise. Add A argument.
14831 (copy_deref_exprloc): Remove deref argument. Adjust for the
14832 original expression being DW_OP_GNU_variable_value with optionally
14833 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
14834 optionally after it.
14835 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
14836 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
14837 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
14838 (variable_value_hasher::hash, variable_value_hasher::equal): New
14840 (resolve_variable_value_in_expr, resolve_variable_value,
14841 resolve_variable_values, note_variable_value_in_expr,
14842 note_variable_value): New functions.
14843 (dwarf2out_early_finish): Call note_variable_value on all toplevel
14846 2017-02-24 Jakub Jelinek <jakub@redhat.com>
14849 * opts.h (handle_generated_option): Add GENERATED_P argument.
14850 * opts-common.c (handle_option): Adjust function comment.
14851 (handle_generated_option): Add GENERATED_P argument, pass it to
14853 (control_warning_option): Pass false to handle_generated_option
14855 * opts.c (maybe_default_option): Pass true to handle_generated_option
14857 * optc-gen.awk: Likewise.
14859 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14861 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
14862 a REG, look at the REG it is a SUBREG of.
14863 (splitter for cmpeqsi_t): Ditto.
14865 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14867 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
14868 the special USEs with the pattern of the insn, not the insn itself.
14870 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
14873 * doc/invoke.texi: Document -mload-store-pairs.
14875 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14876 Sandra Loosemore <sandra@codesourcery.com>
14878 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
14879 argument isn't a CONST_INT.
14880 (nios2_alternate_compare_const): Assert op is a CONST_INT.
14881 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
14882 (nios2_validate_compare): Bypass alternate compare logic if *op2
14883 is not a CONST_INT.
14884 (ldstwm_operation_p): Return false if first_base is not a REG or
14885 if first_offset is not a CONST_INT.
14887 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14889 * config/cris/cris.md: Use correct operand in a define_peephole2.
14891 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14893 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
14895 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14897 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
14898 this_insn if it is an INSN or JUMP_INSN.
14899 (force_offsettable): Look at base, not at addr.
14900 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
14901 on things that aren't necessarily CONST_INTs.
14903 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
14905 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
14906 -mfpmath=sse is the default also for x86-32 targets with SSE2
14907 instruction set when @option{-ffast-math} is enabled
14909 2017-02-24 Jeff Law <law@redhat.com>
14911 PR rtl-optimizatoin/79286
14912 * ira.c (update_equiv_regs): Drop may_trap_p exception to
14915 2017-02-24 Richard Biener <rguenther@suse.de>
14917 PR tree-optimization/79389
14918 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
14921 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
14923 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
14924 function comment to reflect reality.
14925 (loop_exits_before_overflow): Fix typo in function description.
14927 2017-02-24 Richard Biener <rguenther@suse.de>
14929 PR tree-optimization/79389
14930 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
14931 properly that a threading opportunity exists. Detect conditional
14932 copy/constant propagation opportunities.
14934 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
14936 * config/visium/visium.md (type): Add trap.
14937 (b): New mode attribute.
14938 (*btst): Rename into...
14939 (*btst<mode>): ...this and adjust.
14940 (*cbranchsi4_btst_insn): Rename into...
14941 (*cbranch<mode>4_btst_insn): ...this and adjust.
14942 (trap): New define_insn.
14944 2017-02-23 Jakub Jelinek <jakub@redhat.com>
14946 PR tree-optimization/79389
14947 * ifcvt.c (struct noce_if_info): Add rev_cond field.
14948 (noce_reversed_cond_code): New function.
14949 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
14950 reversed_comparison_code. Formatting fix.
14951 (noce_try_store_flag): Test rev_cond != NULL in addition to
14952 reversed_comparison_code.
14953 (noce_try_store_flag_constants): Likewise.
14954 (noce_try_store_flag_mask): Likewise.
14955 (noce_try_addcc): Use rev_cond if non-NULL instead of
14956 reversed_comparison_code.
14957 (noce_try_cmove_arith): Likewise. Formatting fixes.
14958 (noce_try_minmax, noce_try_abs): Clear rev_cond.
14959 (noce_find_if_block): Initialize rev_cond.
14960 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
14961 instead of false as last argument never attempt to reverse it
14964 2017-02-23 Bin Cheng <bin.cheng@arm.com>
14966 PR tree-optimization/79663
14967 * tree-predcom.c (combine_chains): Process refs in reverse order
14968 only for ZERO length chains, and add explaining comment.
14970 2017-02-23 Jeff Law <law@redhat.com>
14972 PR tree-optimization/79578
14973 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
14974 in call to operand_equal_p.
14976 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
14979 * config/i386/cpuid.h: Fix another undefined behavior.
14981 2017-02-23 Richard Biener <rguenther@suse.de>
14983 PR tree-optimization/79683
14984 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
14985 vector types for data-refs.
14987 2017-02-23 Martin Liska <mliska@suse.cz>
14989 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
14991 2017-02-23 Jakub Jelinek <jakub@redhat.com>
14993 PR middle-end/79665
14994 * internal-fn.c (get_range_pos_neg): Moved to ...
14995 * tree.c (get_range_pos_neg): ... here. No longer static.
14996 * tree.h (get_range_pos_neg): New prototype.
14997 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
14998 are known to be in between 0 and signed maximum inclusive, try to
14999 expand both unsigned and signed divmod and use the cheaper one from
15002 2017-02-22 Jeff Law <law@redhat.com>
15004 PR tree-optimization/79578
15005 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
15006 to compare base operands.
15008 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
15011 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
15012 gpc_reg_operand instead of fpr_reg_operand.
15014 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
15016 * config/mips/mips.c (mips_return_in_memory): Force FP
15017 vector types to be returned in memory for o32 ABI.
15019 2017-02-22 Jakub Jelinek <jakub@redhat.com>
15021 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
15022 instead of DW_TAG_member for static data member declarations and don't
15023 set no_linkage_name for static inline data members.
15024 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
15027 2017-02-22 Martin Liska <mliska@suse.cz>
15029 * doc/invoke.texi: Replace inequality signs with square brackets
15032 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15034 PR tree-optimization/68644
15035 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
15037 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
15040 * lra-constraints.c (simplify_operand_subreg): Handle
15041 WORD_REGISTER_OPERATIONS targets.
15043 2017-02-22 Jakub Jelinek <jakub@redhat.com>
15046 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
15047 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
15048 elimination by swapping fld*.
15050 2017-02-22 Richard Biener <rguenther@suse.de>
15052 PR tree-optimization/79673
15053 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
15054 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
15055 irrelevant address-space qualifiers and avoiding a
15056 ADDR_SPACE_CONVERT_EXPR from fold_convert.
15058 2017-02-22 Richard Biener <rguenther@suse.de>
15060 PR tree-optimization/79666
15061 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
15062 to not symbolically negate if that may introduce undefined
15065 2017-02-22 Martin Liska <mliska@suse.cz>
15068 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
15069 * data-streamer-out.c (streamer_write_gcov_count_stream):
15071 * value-prof.c (stream_out_histogram_value): Make assert more
15072 precise based on type of counter.
15074 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
15077 * config/i386/i386.md (standard_x87sse_constant_load splitter):
15078 Use nonimmediate_operand instead of memory_operand for operand 1.
15079 (float-extend standard_x87sse_constant_load splitter): Ditto.
15081 2017-02-21 Jeff Law <law@redhat.com>
15083 PR tree-optimization/79621
15084 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
15085 blocks with edges to themselves.
15087 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15090 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
15091 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
15092 Use gimple_call_builtin_p.
15095 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
15096 on temporarily removed DEBUG_INSNs.
15098 PR tree-optimization/79649
15099 * tree-loop-distribution.c (classify_partition): Give up on
15100 non-generic address space loads/stores.
15102 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
15104 * doc/loop.texi (Loop manipulation): Remove nonexistent
15105 tree_ssa_loop_version from the documentation.
15106 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
15108 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15111 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
15112 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
15113 * config/rs6000/rs6000.c: Include except.h.
15114 (rs6000_expand_split_stack_prologue): Call
15115 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
15117 2017-02-21 Martin Jambor <mjambor@suse.cz>
15120 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
15121 have been analyzed.
15123 2017-02-21 Martin Jambor <mjambor@suse.cz>
15125 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
15126 for backward compatibility only.
15127 * doc/invoke.texi (Option Summary): Remove all references to
15128 -fipa-cp-alignment.
15130 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
15134 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15136 * lra-constraints.c (curr_insn_transform): Handle
15137 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
15139 2017-02-21 Martin Liska <mliska@suse.cz>
15141 * config/i386/i386.opt: Replace -masm-dialect with -masm.
15143 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
15145 PR translation/79638
15146 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
15148 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
15151 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
15152 (arm_function_ok_for_sibcall): Return false for an indirect call by
15153 descriptor if all the argument registers are used.
15154 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
15155 alignment of the function.
15157 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15159 PR tree-optimization/61441
15160 * simplify-rtx.c (simplify_const_unary_operation): For
15161 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
15162 the sNaN unmodified.
15164 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
15166 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
15167 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
15168 instead of SYSTEM_HEADER_DIR.
15170 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
15171 Martin Liška <mliska@suse.cz>
15173 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
15174 Fix typos and grammar, use active voice, and clarify.
15176 2017-02-20 Marek Polacek <polacek@redhat.com>
15178 PR middle-end/79537
15179 * gimplify.c (gimplify_expr): Handle unused *&&L;.
15182 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
15184 2017-02-20 Jakub Jelinek <jakub@redhat.com>
15187 * config/i386/i386.c (ix86_expand_builtin): Handle
15188 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
15189 ix86_builtins_isa[fcode].isa as a requirement of those
15190 flags and any other flag in the bitmask.
15191 (ix86_init_mmx_sse_builtins): Use 0 instead of
15192 ~OPTION_MASK_ISA_64BIT as mask.
15193 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
15194 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
15195 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
15196 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
15198 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15201 * lra-constraints.c (split_reg): Check requested split mode
15202 is supported by the register.
15204 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15206 * lra-constraints.c (simplify_operand_subreg): Remove early
15209 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15212 * lra-constraints.c (curr_insn_transform): Tighten condition
15213 for converting SUBREG reloads from OP_OUT to OP_INOUT.
15215 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15218 * lra-constraints.c (curr_insn_transform): Handle
15219 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
15221 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
15224 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
15226 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
15228 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
15231 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
15234 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15236 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
15239 2017-02-18 Jakub Jelinek <jakub@redhat.com>
15241 * final.c (last_columnnum, override_columnnum): New variables.
15242 (final_start_function): Set last_columnnum, pass it to begin_prologue
15243 hook and pass 0 to dwarf2out_begin_prologue.
15244 (final_scan_insn): Update override_columnnum. Pass last_columnnum
15245 to source_line debug hook.
15246 (notice_source_line): Compute last_columnnum and for debug_column_info
15247 return true on column changes.
15248 * debug.h (struct gcc_debug_hooks): Add column argument to
15249 source_line and begin_prologue hooks.
15250 (debug_nothing_int_charstar_int_bool): Remove prototype.
15251 (debug_nothing_int_int_charstar,
15252 debug_nothing_int_int_charstar_int_bool): New prototypes.
15253 (dwarf2out_begin_prologue): Add column argument.
15254 * debug.c (do_nothing_debug_hooks): Adjust source_line and
15255 begin_prologue hooks.
15256 (debug_nothing_int_charstar_int_bool): Remove.
15257 (debug_nothing_int_int_charstar,
15258 debug_nothing_int_int_charstar_int_bool): New functions.
15259 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
15260 through to dwarf2out_source_line.
15261 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
15262 (dwarf2out_source_line): Add column argument, emit it if requested.
15263 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
15265 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
15266 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
15267 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
15268 through to dwarf2out_begin_prologue.
15269 (vmsdbgout_source_line): Add column argument, pass it through to
15270 dwarf2out_source_line.
15271 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
15272 dbxout_source_line caller.
15273 (dbxout_source_line): Add column argument.
15275 * common.opt (gno-column-info, gcolumn-info): New options.
15276 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
15277 (check_die): Also test for multiple DW_AT_decl_column attributes.
15278 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
15279 DW_AT_decl_column if requested.
15280 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
15282 (gen_variable_die): Likewise.
15283 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
15284 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
15287 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
15288 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
15289 (ix86_handle_option): Handle OPT_m3dnowa.
15290 * doc/invoke.texi (-m3dnowa): Document.
15291 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
15292 -m3dnowa instead of -m3dnow -march=athlon.
15295 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
15296 instead of gcc_assert for K, r and R code checks. Formatting fixes.
15298 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15301 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
15302 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
15303 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
15304 generator for vsx_xxpermdi_<mode>_be.
15305 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
15306 force big-endian semantics.
15307 (vsx_xxpermdi_<mode>_be): New define_expand with same
15308 implementation as previous version of vsx_xxpermdi_<mode>.
15310 2017-02-17 Jakub Jelinek <jakub@redhat.com>
15312 PR tree-optimization/79327
15313 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
15314 variable, its initialization and use.
15316 2017-02-17 Julia Koval <julia.koval@intel.com>
15318 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
15319 (OPTION_MASK_ISA_PKU_UNSET): New.
15320 (ix86_handle_option): Handle -mrdpid.
15321 * config/i386/cpuid.h (bit_RDPID): New.
15322 * config/i386/driver-i386.c (host_detect_local_cpu):
15323 Detect RDPID feature.
15324 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
15325 * config/i386/i386-c.c (ix86_target_macros_internal):
15327 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
15328 (ix86_valid_target_attribute_inner_p): Add "rdpid".
15329 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
15330 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
15331 * config/i386/i386.md (define_insn "rdpid"): New.
15332 * config/i386/i386.opt Add -mrdpid.
15333 * config/i386/immintrin.h (_rdpid_u32): New.
15335 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
15337 PR rtl-optimization/79541
15338 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
15339 instead of transforming it into USE.
15341 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
15343 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
15344 If HONOR_SNANS (SFmode) force the input to a register.
15345 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
15346 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
15347 an frsp or similar insn.
15349 2017-02-17 Martin Liska <mliska@suse.cz>
15351 PR rtl-optimization/79577
15352 * params.def (selsched-max-sched-times): Increase minimum to 1.
15354 2017-02-17 Martin Liska <mliska@suse.cz>
15356 PR rtl-optimization/79574
15357 * gcse.c (want_to_gcse_p): Prevent integer overflow.
15359 2017-02-17 Martin Liska <mliska@suse.cz>
15361 PR tree-optimization/79529
15362 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
15363 ssa_defined_default_def_p to handle cases which are implicitly
15365 * tree-ssa.c (ssa_defined_default_def_p): New function.
15366 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
15367 which are implicitly defined.
15368 * tree-ssa.h (ssa_defined_default_def_p): Declare.
15370 2017-02-17 Richard Biener <rguenther@suse.de>
15372 PR middle-end/79576
15373 * params.def (max-ssa-name-query-depth): Limit to 10.
15375 2017-02-17 Richard Biener <rguenther@suse.de>
15377 PR tree-optimization/79552
15378 * tree-ssa-structalias.c (visit_loadstore): Properly verify
15381 2017-02-17 Richard Biener <rguenther@suse.de>
15384 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
15386 2017-02-17 Marek Polacek <polacek@redhat.com>
15388 PR middle-end/79536
15389 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
15390 (fold_negate_expr): New wrapper.
15392 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
15394 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
15395 Correct terminology and de-emphasize pre-standard behavior.
15397 2017-02-16 Alan Modra <amodra@gmail.com>
15399 PR rtl-optimization/79286
15400 * ira.c (def_dominates_uses): New function.
15401 (update_equiv_regs): Don't create an equivalence for insns that
15402 may trap where the register def does not dominate the use.
15404 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
15406 PR rtl-optimization/78127
15407 * lra.c (lra): Call lra_eliminate before finish the loop after
15410 2017-02-16 Richard Biener <rguenther@suse.de>
15412 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
15414 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
15415 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
15416 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
15417 (isl_val_int_from_wi): New function.
15418 (extract_affine_gmp): Rename to ...
15419 (extract_affine_wi): ... this, take a widest_int.
15420 (extract_affine_int): Just wrap extract_affine_wi.
15421 (add_param_constraints): Use isl_val_int_from_wi.
15422 (add_loop_constraints): Likewise, and extract_affine_wi.
15424 2017-02-15 Jeff Law <law@redhat.com>
15426 PR middle-end/79521
15427 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
15428 ira_init_register_move_cost_if_necessary.
15430 2017-02-15 Martin Sebor <msebor@redhat.com>
15432 PR middle-end/32003
15433 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
15434 removed in a prior commit.
15436 2017-02-15 Bin Cheng <bin.cheng@arm.com>
15438 PR tree-optimization/79347
15439 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
15440 counters during peeling.
15442 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
15444 * Makefile.in (site.exp): Remove "set ISLVER".
15446 2017-02-15 Jakub Jelinek <jakub@redhat.com>
15449 * real.c (real_from_integer): Call real_convert even for decimal.
15451 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
15454 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
15456 2017-02-14 Andrew Pinski <apinski@cavium.com>
15458 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
15459 cores and change the partno/implementer to be correct.
15460 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
15461 the 'B" as the implementer.
15462 * config/aarch64/aarch64-tune.md: Regenerate.
15464 2017-02-14 Carl Love <cel@us.ibm.com>
15466 * config/rs6000/rs6000.c: Add case statement entry to make the
15467 xvcvuxdsp built-in argument unsigned.
15468 * config/rs6000/vsx.md: Fix the source and return operand types so they
15469 match the instruction definitions from the ISA document. Fix typo
15470 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
15473 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
15476 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
15477 member early_clobber_alts.
15478 * lra-lives.c (reg_early_clobber_p): New.
15479 (process_bb_lives): Use it.
15480 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
15481 (debug_operand_data): Initialize early_clobber_alts.
15482 (setup_operand_alternative): Set up early_clobber_alts.
15483 (collect_non_operand_hard_regs): Ditto. Pass early clobber
15484 alternatives to new_insn_reg.
15485 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
15487 (lra_update_insn_regno_info): Pass the new arg.
15489 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15491 PR middle-end/79505
15492 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
15493 (new_oacc_loop_raw): Don't clear already cleared fields.
15496 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
15497 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
15498 _mm512_prefetch_i64gather_ps): New inline functions and macros.
15500 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
15503 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
15505 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
15508 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
15509 the extra instruction to the right place to store 128-bit constant
15512 2017-02-14 Martin Sebor <msebor@redhat.com>
15514 PR middle-end/79448
15515 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
15516 warning for strings of unknown length.
15518 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
15520 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
15522 2017-02-14 Jeff Law <law@redhat.com>
15525 * ira-costs.c (scan_one_insn): Initialize register move costs
15526 for pseudos seen in USE/CLOBBER insns.
15528 PR tree-optimization/79095
15529 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
15530 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
15531 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
15532 if the operands are known to be not equal, then the resulting range
15534 (intersect_ranges): If the new range is ~[0,0] and the old range is
15535 wide, then prefer ~[0,0].
15536 * tree-vrp.c (overflow_comparison_p_1): New function.
15537 (overflow_comparison_p): New function.
15538 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
15539 if NAME is used in an overflow test.
15540 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
15541 overflow check that can be expressed as an equality test, then adjust
15542 ops to be that equality test.
15544 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15546 * config/s390/s390-builtin-types.def: Remove flags argument.
15547 * config/s390/s390.c (s390_init_builtins): Likewise.
15549 2017-02-14 Martin Liska <mliska@suse.cz>
15551 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
15552 vector. Fix trailing white spaces.
15554 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
15556 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
15559 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15561 PR rtl-optimization/68664
15562 * config/arm/arm.c (arm_sched_can_speculate_insn):
15563 New function. Declare prototype.
15564 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15566 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15568 PR rtl-optimization/68664
15569 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
15571 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15573 2017-02-14 Amit Pawar <amit.pawar@amd.com>
15575 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
15576 max skip bytes for function, loop and jump.
15578 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15580 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
15581 ABS_EXPR for gimple dump.
15583 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15586 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
15588 PR tree-optimization/79408
15589 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
15590 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
15591 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
15592 also if rhs1 is INTEGER_CST.
15594 2017-02-14 Richard Biener <rguenther@suse.de>
15596 PR middle-end/79432
15597 * tree-into-ssa.c (insert_phi_nodes): When the function can
15598 have abnormal edges rewrite SSA names with broken use-def
15599 dominance out of SSA and register them for PHI insertion.
15601 2017-02-13 Martin Sebor <msebor@redhat.com>
15603 PR middle-end/79496
15604 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
15605 clearing info.nowrite flag when snprintf size argument is a range.
15607 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15609 * cprop.c (cprop_jump): Add missing space in string literal.
15610 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
15611 (get_constraint_for_component_ref): Likewise.
15612 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
15613 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
15614 * lra-constraints.c (process_alt_operands): Likewise.
15615 * ipa-inline.c (inline_small_functions): Likewise.
15616 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
15617 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15618 * trans-mem.c (diagnose_tm_1_op): Likewise.
15619 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
15620 (grid_parallel_clauses_gridifiable): Likewise.
15622 * config/nvptx/mkoffload.c (process): Add space in between
15625 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
15626 "MOD4_SSE_REGS" and "ALL_REGS".
15628 * spellcheck.c (test_data): Add , in between "foo" and "food".
15630 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15633 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
15634 boundary crossing check and subsequent code generation agree.
15636 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15638 * config/aarch64/aarch64.c (has_memory_op): Delete.
15639 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
15642 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15644 PR rtl-optimization/79388
15645 PR rtl-optimization/79450
15646 * combine.c (distribute_notes): When removing TEM_INSN for which
15647 corresponding dest has last value recorded, invalidate that last
15650 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15652 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
15653 of explicit '@'. Add missing assembly comment marker on branch costs
15656 2017-02-13 Nathan Sidwell <nathan@acm.org>
15658 * gengtype-lex.l (<in_struct>): Add '/'.
15660 2017-02-13 Martin Liska <mliska@suse.cz>
15663 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
15665 2017-02-13 Richard Biener <rguenther@suse.de>
15667 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
15669 * configure: Re-generate.
15670 * config.in: Likewise.
15671 * graphite-dependences.c: Simplify as if
15672 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
15673 * graphite-isl-ast-to-gimple.c: Likewise.
15674 * graphite-optimize-isl.c: Likewise.
15675 * graphite-poly.c: Likewise.
15676 * graphite-sese-to-poly.c: Likewise.
15677 * graphite.h: Likewise.
15678 * toplev.c: Include isl/version.h and use isl_version () for
15679 printing the ISL version.
15680 * doc/install.texi: Update ISL requirement.
15682 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15684 * doc/standards.texi (Standards): Update reference to
15687 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15689 * doc/extend.texi (Named Address Spaces): sourceware.org now
15691 * doc/install.texi (Binaries): Ditto.
15694 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15696 * doc/cpp.texi: Replace "stringify"/"stringification" with C
15697 standard terminology "stringize"/"stringizing" throughout.
15698 * doc/cppinternals.texi: Likewise.
15700 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15702 * doc/extend.texi: Fix some spelling mistakes and typos.
15703 * doc/invoke.texi: Likewise.
15705 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15708 * params.def (inline-min-speedup) Change from 10 to 8.
15710 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15712 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
15715 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15718 * ipa-inline-analysis.c (get_minimal_bb): New function.
15719 (record_modified): Use it.
15720 (remap_edge_change_prob): Handle also ancestor functions.
15722 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
15724 * doc/contrib.texi (Contributors): Remove broken link into
15725 the Mauve CVS repository.
15727 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15729 PR middle-end/79454
15730 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
15731 result computation whenever lhs doesn't have vector mode, not
15732 just when it has BLKmode.
15734 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15736 * doc/makefile.texi (profiledbootstrap): Refer to the
15737 installation instructions only in textual form.
15739 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15742 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
15744 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15746 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
15747 (Specific): Update mingw-w64 reference.
15749 (Specific): Remove broken link to Renesas RX processor.
15751 2017-02-10 Richard Biener <rguenther@suse.de>
15753 * toplev.c (process_options): Do not mention obsolete graphite
15754 options when printing sorry message about missing graphite support.
15755 Mention -floop-nest-optimize.
15757 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
15759 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
15760 (vtst_p16): Likewise.
15761 (vtstq_p8): Likewise.
15762 (vtstq_p16): Likewise.
15764 (vtstq_p64): Likewise.
15765 * config/arm/arm_neon.h (vgetq_lane_p64): New.
15766 (vset_lane_p64): New.
15767 (vsetq_lane_p64): New.
15769 2017-02-10 Jakub Jelinek <jakub@redhat.com>
15771 PR tree-optimization/79411
15772 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
15773 stmt operands are SSA_NAMEs used in abnormal phis.
15774 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
15777 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15780 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
15783 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15785 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
15787 2017-02-09 Jakub Jelinek <jakub@redhat.com>
15789 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
15793 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
15794 not arbitrary TREE_CONSTANT.
15797 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
15798 "omp declare target link" attribute unless is_global_var.
15799 * omp-offload.c (find_link_var_op): Likewise.
15801 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
15802 Chung-Lin Tang <cltang@codesourcery.com>
15804 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
15806 (gimplify_adjust_omp_clauses): Don't delete TILE.
15807 (gimplify_omp_for): Deal with TILE.
15808 * internal-fn.c (expand_GOACC_TILE): New function.
15809 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
15811 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
15812 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
15814 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
15815 avoid DIV for outermost collapse var.
15816 (expand_oacc_for): Insert tile element loop as needed. Adjust.
15817 Remove out of date comments, fix whitespace.
15818 * omp-general.c (omp_extract_for_data): Deal with tiling.
15819 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
15820 adjust OLF_DIM_BASE value.
15821 (struct omp_for_data): Add tiling field.
15822 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
15823 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
15824 for auto loops. Remove default auto determining, moved to
15825 oacc_loop_fixed_partitions.
15826 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
15827 stmts, add e_mask field.
15828 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
15829 (oacc_thread_numbers): Use oacc_dim_call.
15830 (oacc_xform_tile): New.
15831 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
15832 (finish_oacc_loop): Adjust for ifns vector.
15833 (oacc_loop_discover_walk): Append loop abstraction sites to list,
15834 add case for GOACC_TILE fns.
15835 (oacc_loop_xform_loop): Delete.
15836 (oacc_loop_process): Iterate over call list directly, and add
15837 handling for GOACC_TILE fns.
15838 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
15840 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
15841 vector partitioning to outer loops. Assign 2 partitions to loops
15842 when available. Add TILE handling.
15843 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
15844 (execite_oacc_device_lower): Process GOACC_TILE fns,
15845 ignore unknown specs.
15846 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
15847 * tree.c (omp_clause_num_ops): Adjust TILE ops.
15848 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
15850 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
15852 * configure.ac (ACX_BUGURL): Update.
15853 * configure: Regenerate.
15855 2017-02-09 Richard Biener <rguenther@suse.de>
15857 PR tree-optimization/69823
15858 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
15859 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
15861 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15863 * config/arc/arc-c.def: Add __NPS400__ definition.
15864 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
15865 (TARGET_NPS400): Define.
15867 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15869 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
15871 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
15872 pointer, arch_info.
15873 (arc_cpu_types): Fill the arch_info field with a pointer into the
15874 arc_arch_types table.
15875 (arc_selected_cpu): Declare.
15876 * config/arc/arc.c (arc_selected_cpu): Make global.
15877 (arc_selected_arch): Delete.
15878 (arc_base_cpu): Delete.
15879 (arc_override_options): Remove references to deleted variables,
15880 update access to arch information.
15881 (ARC_OPT): Update access to arch information.
15882 (ARC_OPTX): Likewise.
15883 * config/arc/arc.h (arc_base_cpu): Remove declaration.
15884 (TARGET_ARC600): Update access to arch information.
15885 (TARGET_ARC601): Likewise.
15886 (TARGET_ARC700): Likewise.
15887 (TARGET_EM): Likewise.
15888 (TARGET_HS): Likewise.
15889 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
15892 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
15895 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
15896 condition/operands for integer GE/LE/GEU/LEU operations.
15898 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
15900 PR translation/79397
15901 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
15904 2017-02-08 Martin Jambor <mjambor@suse.cz>
15907 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
15908 whether allocation happened.
15909 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
15910 nothing was allocated.
15912 2017-02-08 Jakub Jelinek <jakub@redhat.com>
15914 PR tree-optimization/79408
15915 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
15916 constant, but SSA_NAME with a known integer range, use the minimum
15917 of that range instead of op1 to determine if modulo can be replaced
15918 with its first operand.
15920 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15922 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
15924 2017-02-08 Richard Biener <rguenther@suse.de>
15926 PR tree-optimization/71824
15927 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
15928 Check all loops contained in the merged region.
15930 2017-02-07 Andrew Pinski <apinski@cavium.com>
15932 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
15934 2017-02-07 Andrew Pinski <apinski@cavium.com>
15936 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
15937 (thunderxt88): Likewise.
15938 (thunderxt81): Disable LSE and change v8.1 to v8.
15939 (thunderxt83): Likewise.
15941 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15942 Richard Biener <rguenther@suse.de>
15944 PR middle-end/79399
15945 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
15946 type from int to size_t.
15947 * ira-costs.c (struct_costs_size): Change type from int to size_t.
15949 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15951 PR rtl-optimization/79386
15952 * cprop.c (bypass_conditional_jumps): Initialize
15953 bypass_last_basic_block already before splitting bbs after
15954 unconditional traps...
15955 (bypass_conditional_jumps): ... rather than here.
15958 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
15959 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
15960 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
15961 fix -masm=intel patterns.
15963 2017-02-07 Richard Biener <rguenther@suse.de>
15965 PR tree-optimization/79256
15966 PR middle-end/79278
15967 * builtins.c (get_object_alignment_2): Use min_align_of_type
15968 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
15969 and ADJUST_FIELD_ALIGN.
15971 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
15973 * doc/tm.texi: Regenerate.
15974 * stor-layout.c (layout_decl): Adjust.
15975 (update_alignment_for_field): Likewise.
15976 (place_field): Likewise.
15977 (min_align_of_type): Likewise.
15978 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
15979 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
15980 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
15981 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
15982 * config/frv/frv.c (frv_adjust_field_align): Likewise.
15983 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
15984 * config/i386/i386.c (x86_field_alignment): Likewise.
15985 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
15986 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
15987 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
15988 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
15989 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
15990 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
15994 2017-01-30 Richard Biener <rguenther@suse.de>
15996 PR tree-optimization/79256
15997 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
15998 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
16001 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
16003 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
16004 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
16005 builtins to SImode and emit a zero-extend, if necessary.
16007 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
16009 * docs/invoke.texi (RISC-V Options): Alphabetize.
16011 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
16013 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
16016 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
16018 * config/riscv/riscv.c: New file.
16019 * common/config/riscv/riscv-common.c: Likewise.
16020 * config.gcc: Likewise.
16021 * config/riscv/constraints.md: Likewise.
16022 * config/riscv/elf.h: Likewise.
16023 * config/riscv/generic.md: Likewise.
16024 * config/riscv/linux.h: Likewise.
16025 * config/riscv/multilib-generator: Likewise.
16026 * config/riscv/peephole.md: Likewise.
16027 * config/riscv/pic.md: Likewise.
16028 * config/riscv/predicates.md: Likewise.
16029 * config/riscv/riscv-builtins.c: Likewise.
16030 * config/riscv/riscv-c.c: Likewise.
16031 * config/riscv/riscv-ftypes.def: Likewise.
16032 * config/riscv/riscv-modes.def: Likewise.
16033 * config/riscv/riscv-opts.h: Likewise.
16034 * config/riscv/riscv-protos.h: Likewise.
16035 * config/riscv/riscv.h: Likewise.
16036 * config/riscv/riscv.md: Likewise.
16037 * config/riscv/riscv.opt: Likewise.
16038 * config/riscv/sync.md: Likewise.
16039 * config/riscv/t-elf-multilib: Likewise.
16040 * config/riscv/t-linux: Likewise.
16041 * config/riscv/t-linux-multilib: Likewise.
16042 * config/riscv/t-riscv: Likewise.
16043 * configure.ac: Likewise.
16044 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
16045 Waterman as RISC-V maintainers.
16046 * doc/install.texi: Add RISC-V entries.
16047 * doc/invoke.texi: Add RISC-V options section.
16048 * doc/md.texi: Add RISC-V constraints section.
16049 * configure: Regenerated.
16051 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
16054 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
16055 false values to be constant vectors with all 0 or all 1 bits set.
16056 (vcondu<mode><mode>): Likewise.
16057 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
16059 (fpmask_comparison_operator): Update comment.
16060 (vecint_comparison_operator): New predicate.
16061 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
16062 vector conditionals when the true and false values are constant
16063 vectors with all 0 bits or all 1 bits set.
16065 2017-02-06 Martin Sebor <msebor@redhat.com>
16067 PR tree-optimization/79376
16068 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
16070 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
16072 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
16073 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
16074 to simplify split condition.
16076 2017-02-06 Jakub Jelinek <jakub@redhat.com>
16078 * omp-expand.c (oxpand_omp_atomic_fetch_op,
16079 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
16082 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
16084 PR rtl-optimization/68664
16085 * target.def (can_speculate_insn): New hook.
16086 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
16087 * doc/tm.texi: Regenerate.
16088 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
16089 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
16090 (rs6000_sched_can_speculate_insn): New function.
16092 2017-02-06 Jakub Jelinek <jakub@redhat.com>
16094 PR tree-optimization/79284
16095 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
16096 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
16097 vectorizable_mask_load_store, vectorizable_operation,
16098 vect_is_simple_cond, get_same_sized_vectype): Use it instead
16099 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
16100 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
16101 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
16102 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
16103 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
16104 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
16105 is_gimple_assign (stmt). Replace another such test with
16106 is_gimple_assign (stmt).
16108 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
16111 * config/avr/avr.c (rtl-iter.h): Include it.
16112 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
16113 (avr_legitimate_combined_insn): ...and implementation.
16115 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
16117 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
16118 * config/s390/s390.c (s390_const_operand_ok)
16119 (s390_canonicalize_comparison, s390_extract_part)
16120 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
16121 (s390_contiguous_bitmask_p, s390_rtx_costs)
16122 (legitimize_pic_address): Likewise.
16123 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
16124 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
16125 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
16126 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
16127 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
16129 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
16131 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
16132 REGNO($0) == REGNO($1).
16134 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16136 * config/s390/linux.h(SIZE_TYPE): Add comment.
16138 2017-02-06 Julian Brown <julian@codesourcery.com>
16139 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16140 Virendra Pathak <virendra.pathak@broadcom.com>
16142 * config/aarch64/aarch64-cores.def: Change the scheduler
16144 * config/aarch64/aarch64.md: Include thunderx2t99.md.
16145 * config/aarch64/thunderx2t99.md: New file.
16147 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16149 * doc/standards.texi (Go Language): Update link to language
16152 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
16154 * tree-eh.c (lower_resx): Sanitize profile.
16155 (cleanup_empty_eh_move_lp): Likewise.
16157 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
16160 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
16162 * cfgloopmanip.h (loop_version): Update prototype.
16163 * modulo-sched.c (sms_schedule): Update call of loop_version.
16164 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
16165 * tree-parloops.c (gen_parallel_loop): Likewise.
16166 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
16167 * tree-ssa-loop-split.c (split_loop): Likewise.
16168 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
16169 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
16171 2017-02-05 Martin Liska <mliska@suse.cz>
16174 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
16176 (print_operand_address): Initialize a struct to zero.
16178 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16180 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
16181 garbage collector only in textual form.
16183 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16185 * doc/extend.texi (x86 specific memory model extensions for
16186 transactional memory): Simplify a phrase.
16188 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
16191 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
16192 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
16193 (atomic_storedi_1): Likewise.
16195 2017-02-04 Jakub Jelinek <jakub@redhat.com>
16197 PR tree-optimization/79338
16198 * tree-parloops.c (gather_scalar_reductions): Don't call
16199 vect_analyze_loop_form for loop->inner before destroying loop's
16202 2017-02-03 Martin Sebor <msebor@redhat.com>
16204 PR tree-optimization/79327
16205 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
16206 when precision has resulted in leading zeros.
16207 (format_integer): Adjust the likely counter to assume an unknown
16208 argument that may be zero is non-zero.
16210 2017-02-03 Jason Merrill <jason@redhat.com>
16213 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
16214 avoid copying non-taken branch.
16216 2017-02-03 Jakub Jelinek <jakub@redhat.com>
16218 PR tree-optimization/79340
16219 * tree-vect-loop.c (vectorizable_reduction): Release
16220 vec_defs elements after safe_splicing them into other vectors.
16223 PR tree-optimization/79327
16224 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
16225 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
16227 (format_integer): Use wide_int_to_tree instead of build_int_cst
16228 + to_?hwi. If argmin is NULL, just set argmin and argmax to
16229 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
16230 of shortest and longest sequence.
16232 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
16234 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
16235 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
16237 2017-02-03 Walter Lee <walt@tilera.com>
16240 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
16241 after initial stackframe link reg save.
16242 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
16244 2017-02-03 Jakub Jelinek <jakub@redhat.com>
16247 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
16248 wu for stxssp alternative.
16250 2017-02-03 Martin Sebor <msebor@redhat.com>
16252 PR tree-optimization/79352
16253 * gimple-fold.c (get_range_strlen): Add argument.
16254 (get_range_strlen): Change return type to bool.
16255 (get_maxval_strlen): Pass in a dummy argument.
16256 * gimple-fold.h (get_range_strlen): Change return type to bool.
16257 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
16258 * tree.h (array_at_struct_end_p): Add argument.
16259 * tree.c (array_at_struct_end_p): Handle it.
16261 2017-02-03 Martin Liska <mliska@suse.cz>
16264 * multiple_target.c (create_dispatcher_calls): Redirect edge
16265 from a caller of a dispatcher.
16266 (expand_target_clones): Make the clones local.
16267 (ipa_target_clone): Do both target clones and resolvers.
16268 (ipa_dispatcher_calls): Remove the pass.
16269 (pass_dispatcher_calls::gate): Likewise.
16270 (make_pass_dispatcher_calls): Likewise.
16271 * passes.def (pass_target_clone): Put as very first IPA early
16274 2017-02-03 Martin Liska <mliska@suse.cz>
16276 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
16277 in case of a function with ifunc attribute.
16279 2017-02-03 Martin Liska <mliska@suse.cz>
16281 * cgraph.c (cgraph_node::dump): Dump function version info.
16282 * symtab.c (symtab_node::dump_base): Add missing new line.
16284 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
16286 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
16287 (ifcombine_ifandif): Use it.
16289 2017-02-03 Martin Liska <mliska@suse.cz>
16291 * doc/invoke.texi: Document default value for
16292 use-after-scope-direct-emission-threshold.
16294 2017-02-03 Martin Liska <mliska@suse.cz>
16296 PR tree-optimization/79339
16297 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
16298 (format_floating): Likewise.
16300 2017-02-03 Martin Liska <mliska@suse.cz>
16303 * ipa-prop.c (ipa_node_params_t::insert): Remove current
16305 (ipa_node_params_t::remove): Likewise.
16306 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
16307 initialization from removed ipa_node_params_t::insert.
16308 (ipa_node_params::~ipa_node_params): Move from removed
16309 ipa_node_params_t::release.
16310 * symbol-summary.h (symbol_summary::m_released): New member.
16311 Do not release a summary twice. Do not allow to call finalizer
16312 for types of a summary that live in GGC memory.
16314 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16316 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
16319 2017-02-02 Martin Sebor <msebor@redhat.com>
16321 PR middle-end/79275
16322 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
16323 (format_string): Tighten up the range of output for non-constant
16324 strings and correct the expected range for wide non-constant strings.
16326 2017-02-02 Martin Sebor <msebor@redhat.com>
16328 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
16330 PR middle-end/32003
16331 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
16333 (-fdump-tree-@var): Add to index and document how to come up
16334 with pass-specific option and dump file names.
16335 (-fdump-passes): Clarify where to look for output.
16337 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
16339 PR middle-end/77445
16340 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
16341 statistics of the analyzed path; allow threading for speed when
16342 any of BBs along the path are optimized for speed.
16344 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
16346 PR middle-end/78468
16347 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
16348 settings of the virtual registers.
16351 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
16353 * explow.c (get_dynamic_stack_size): Take known alignment of stack
16354 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
16357 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16359 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
16360 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
16362 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16364 * config/s390/s390.md: Add missing comments with the expanded
16366 * config/s390/vector.md: Likewise.
16367 * config/s390/vx-builtins.md: Likewise.
16369 2017-02-02 Jakub Jelinek <jakub@redhat.com>
16372 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
16373 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
16374 conditions on a single line.
16376 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16378 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
16379 __S390_VX__ to __VX__.
16381 2017-02-01 Andrew Pinski <apinski@cavium.com>
16383 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
16384 stmt_info to record_stmt_cost.
16385 (vect_get_known_peeling_cost): Pass stmt_info if known to
16387 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
16388 cpu_vector_cost field into
16389 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
16390 field into vec_int_stmt_cost and vec_fp_stmt_cost.
16391 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
16392 splitting of scalar_stmt_cost and vec_stmt_cost.
16393 (thunderx_vector_cost): Likewise.
16394 (cortexa57_vector_cost): LIkewise.
16395 (exynosm1_vector_cost): Likewise.
16396 (xgene1_vector_cost): Likewise.
16397 (thunderx2t99_vector_cost): Improve after the splitting of the two
16399 (aarch64_builtin_vectorization_cost): Update for the splitting of
16400 scalar_stmt_cost and vec_stmt_cost.
16402 2017-02-01 Torvald Riegel <triegel@redhat.com>
16403 Richard Henderson <rth@redhat.com>
16405 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
16406 conditional on existance of a fast atomic load.
16407 * optabs-query.c (can_atomic_load_p): New function.
16408 * optabs-query.h (can_atomic_load_p): Declare it.
16409 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
16410 no fast atomic load is available for the particular size of access.
16411 (expand_atomic_compare_and_swap): Likewise.
16412 (expand_atomic_load): Likewise.
16413 (expand_atomic_store): Likewise.
16414 (expand_atomic_fetch_op): Likewise.
16415 * testsuite/lib/target-supports.exp
16416 (check_effective_target_sync_int_128): Remove x86 because it provides
16417 no fast atomic load.
16418 (check_effective_target_sync_int_128_runtime): Likewise.
16420 2017-02-01 Richard Biener <rguenther@suse.de>
16422 * graphite.c: Include tree-vectorizer.h for find_loop_location.
16423 (graphite_transform_loops): Provide opt-info for optimized nests.
16424 * tree-parloop.c (parallelize_loops): Provide opt-info for
16425 parallelized loops.
16427 2017-02-01 Richard Biener <rguenther@suse.de>
16429 PR middle-end/79315
16430 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
16431 was not set before.
16433 2017-02-01 Richard Biener <rguenther@suse.de>
16435 PR tree-optimization/71824
16436 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
16437 Verify the loops are valid in the merged SESE region.
16438 (scop_detection::can_represent_loop_1): Check analyzing the
16439 evolution of the number of iterations in the region succeeds.
16441 2017-01-31 Ian Lance Taylor <iant@golang.org>
16443 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
16444 REG_ARGS_SIZE note to 32-bit push insns and call insn.
16446 2017-01-31 David Malcolm <dmalcolm@redhat.com>
16448 PR preprocessor/79210
16449 * input.c (get_substring_ranges_for_loc): Replace line_width
16450 assertion with error-handling.
16452 2017-01-31 Richard Biener <rguenther@suse.de>
16454 PR tree-optimization/77318
16455 * graphite-sese-to-poly.c (extract_affine): Fix assert.
16456 (create_pw_aff_from_tree): Take loop parameter.
16457 (add_condition_to_pbb): Pass loop of the condition to
16458 create_pw_aff_from_tree.
16460 2017-01-31 Jakub Jelinek <jakub@redhat.com>
16462 * config/s390/s390.c (s390_asan_shadow_offset): New function.
16463 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
16465 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
16469 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
16471 (convert_int_to_float128): Likewise.
16472 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
16473 (convert_int_to_float128): Likewise.
16474 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
16475 (UNSPEC_IEEE128_CONVERT): Likewise.
16476 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
16477 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
16478 Use local variables for IBM extended format.
16479 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
16480 (fix_trunc<mode>si2_fprs): Likewise.
16481 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
16482 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
16483 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
16484 to know that we can now have integers of all sizes in vector
16486 (fix<uns>_<mode>di2_hw): Likewise.
16487 (float<uns>_<mode>si2_hw): Likewise.
16488 (fix_<mode>si2_hw): Likewise.
16489 (fixuns_<mode>si2_hw): Likewise.
16490 (float<uns>_<mode>di2_hw): Likewise.
16491 (float_<mode>di2_hw): Likewise.
16492 (float_<mode>si2_hw): Likewise.
16493 (floatuns_<mode>di2_hw): Likewise.
16494 (floatuns_<mode>si2_hw): Likewise.
16495 (xscvqp<su>wz_<mode>): Delete, no longer used.
16496 (xscvqp<su>dz_<mode>): Likewise.
16497 (xscv<su>dqp_<mode>): Likewise.
16498 (ieee128_mfvsrd_64bit): Likewise.
16499 (ieee128_mfvsrd_32bit): Likewise.
16500 (ieee128_mfvsrwz): Likewise.
16501 (ieee128_mtvsrw): Likewise.
16502 (ieee128_mtvsrd_64bit): Likewise.
16503 (ieee128_mtvsrd_32bit): Likewise.
16505 2017-01-31 Martin Liska <mliska@suse.cz>
16508 * ipa-prop.c (ipa_free_all_node_params): Call release method
16509 instead of ~sumbol_summary to not to trigger double times
16512 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
16514 PR tree-optimization/71691
16515 * bitmap.h (class auto_bitmap): New.
16516 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
16517 is_maybe_undefined instead of ssa_undefined_value_p.
16519 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16521 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
16522 __S390_ARCH_LEVEL__ to __ARCH__.
16524 2017-01-31 Jakub Jelinek <jakub@redhat.com>
16526 PR tree-optimization/79267
16527 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
16528 if should_remove_lhs_p is true.
16530 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
16533 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
16534 (add_alignment_attribute): New.
16535 (base_type_die): Add alignment attribute.
16536 (subrange_type_die): Likewise.
16537 (modified_type_die): Likewise.
16538 (gen_array_type_die): Likewise.
16539 (gen_descr_array_type_die: Likewise.
16540 (gen_enumeration_type_die): Likewise.
16541 (gen_subprogram_die): Likewise.
16542 (gen_variable_die): Likewise.
16543 (gen_field_die): Likewise.
16544 (gen_ptr_to_mbr_type_die): Likewise.
16545 (gen_struct_or_union_type_die): Likewise.
16546 (gen_subroutine_type_die): Likewise.
16547 (gen_typedef_die): Likewise.
16548 (base_type_cmp): Compare alignment attribute.
16550 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16553 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
16554 (setb_unsigned) New pattern for setb with CCUNS.
16555 * config/rs6000/rs6000.c (expand_block_compare): Use a different
16556 subfc./subfe sequence to avoid overflow problems. Generate a
16557 shorter sequence with cmpld/setb for power9.
16558 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
16559 for generating subfc. instruction.
16560 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
16561 now uses this instruction.
16563 2017-01-30 Ian Lance Taylor <iant@google.com>
16566 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
16567 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
16569 2017-01-30 Martin Sebor <msebor@redhat.com>
16571 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
16572 Move constant to the right of a relational operator.
16573 (get_mpfr_format_length, format_character, format_string): Ditto.
16574 (should_warn_p, maybe_warn): Same.
16576 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
16578 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
16581 * asan.c (get_translation_unit_decl): Remove function.
16582 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
16584 2017-01-30 Martin Liska <mliska@suse.cz>
16586 PR gcov-profile/79259
16587 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
16588 -fprofile-generate.
16590 2017-01-30 Martin Liska <mliska@suse.cz>
16593 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
16594 Initialize variables with NULL value.
16596 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
16599 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
16601 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
16603 2017-01-30 Richard Biener <rguenther@suse.de>
16605 PR tree-optimization/79276
16606 * tree-vrp.c (process_assert_insertions): Properly adjust common
16607 when removing a duplicate.
16609 * gcc.dg/torture/pr79276.c: New testcase.
16611 2017-01-30 Richard Biener <rguenther@suse.de>
16613 PR tree-optimization/79256
16614 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
16615 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
16617 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
16619 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
16622 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
16623 ("*r<noxa>sbg_<mode>_sll_bitmask")
16624 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
16625 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
16626 Use contiguous_bitmask_nowrap_operand.
16628 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16631 * config/rs6000/altivec.h (vec_xl): Revise #define.
16632 (vec_xst): Likewise.
16634 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
16636 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
16638 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
16640 PR rtl-optimization/79194
16641 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
16642 traps before call to bypass_conditional_jumps.
16644 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16646 PR tree-optimization/71374
16647 * lra-constraints.c (check_conflict_input_operands): New.
16648 (match_reload): Use it.
16650 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16653 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
16654 account to calculate conflict_set.
16656 2017-01-27 Bin Cheng <bin.cheng@arm.com>
16658 PR rtl-optimization/78559
16659 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
16660 other_insn in combine.
16662 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16664 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
16665 uint16_type_node for BT_UINT16.
16667 2017-01-27 David Malcolm <dmalcolm@redhat.com>
16669 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
16670 "RTL Tests" to menu.
16671 (GIMPLE Tests): New node.
16672 (RTL Tests): New node.
16674 2017-01-27 Richard Biener <rguenther@suse.de>
16676 PR tree-optimization/79245
16677 * tree-loop-distribution.c (distribute_loop): Apply cost
16678 modeling also to detected patterns.
16680 2017-01-27 Richard Biener <rguenther@suse.de>
16682 PR tree-optimization/71433
16683 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
16684 (compare_assert_loc): New function.
16685 (process_assert_insertions): Sort and optimize assert locations
16686 to remove duplicates and push down identical assertions on
16687 edges to their destination block.
16689 2017-01-27 Richard Biener <rguenther@suse.de>
16691 PR tree-optimization/79244
16692 * tree-vrp.c (remove_range_assertions): Forcefully propagate
16693 out SSA names even if abnormal.
16695 2017-01-27 Jakub Jelinek <jakub@redhat.com>
16697 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
16698 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
16699 instead of MPFR_RNDN.
16701 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
16704 * arm.c (arm_option_override): Don't call build_target_option_node
16705 until after doing all option overrides.
16706 (arm_valid_target_attribute_tree): Likewise.
16708 2017-01-27 Martin Liska <mliska@suse.cz>
16710 * doc/invoke.texi (-fprofile-arcs): Document profiling support
16711 for {cd}tors and C++ {cd}tors.
16713 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
16715 * config/s390/s390.md ("*setmem_long_and")
16716 ("*setmem_long_and_31z"): Use zero_extend instead of and.
16718 2017-01-26 Martin Sebor <msebor@redhat.com>
16720 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
16723 2017-01-26 Martin Sebor <msebor@redhat.com>
16725 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
16726 HAVE_DFmode before using XFmode or DFmode.
16727 (parse_directive): Avoid using the z length modifier to avoid
16728 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
16730 PR middle-end/78703
16731 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
16732 to accept adjustment as an array.
16733 (get_int_range): New function.
16734 (struct directive): Make width and prec arrays.
16735 (directive::set_width, directive::set_precision): Call get_int_range.
16736 (format_integer, format_floating): Handle width and precision ranges.
16737 (format_string, parse_directive): Same.
16739 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16742 * dwarf2out.c (generate_skeleton_bottom_up): For children with
16743 comdat_type_p set, just clone them, but keep the children in the
16747 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
16748 which have direct callers with -fvar-tracking-assignments enabled
16750 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
16751 inside of type units.
16753 2017-01-26 Martin Sebor <msebor@redhat.com>
16755 PR middle-end/78703
16756 * gimple-ssa-sprintf.c (struct result_range): Add likely and
16758 (struct format_result): Replace number_chars, number_chars_min,
16759 and number_chars_max with a single member of struct result_range.
16761 (format_result::operator+=): Adjust.
16762 (struct fmtresult): Remove bounded. Handle likely and unlikely
16764 (fmtresult::adjust_for_width_or_precision): New function.
16765 (fmtresult:type_max_digits): New function.
16766 (bytes_remaining): Handle likely and unlikely counters.
16767 (min_bytes_remaining): Remove.
16768 (format_percent): Simplify.
16769 (format_integer, format_floating): Set likely and unlikely counters.
16770 (get_string_length, format_character, format_string): Same.
16771 (format_plain, should_warn_p): New function.
16772 (maybe_warn): Call should_warn_p. Update diagnostic messages
16773 and handle those for all directives, including plain strings.
16774 (format_directive): Handle likely and unlikely counters.
16775 Remove unnecessary quoting from diagnostics. Add an informational
16777 (add_bytes): Remove.
16778 (pass_sprintf_length::compute_format_length): Simplify.
16779 (try_substitute_return_value): Handle likely and unlikely counters.
16781 2017-01-26 Carl Love <cel@us.ibm.com>
16783 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
16784 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
16786 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
16789 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
16790 endianess for subregs into account.
16791 * lra-constraints.c (lra_constraints): Do risky transformations
16792 always on the first iteration.
16793 * lra-lives.c (check_pseudos_live_through_calls): Add arg
16794 last_call_used_reg_set.
16795 (process_bb_lives): Define and use last_call_used_reg_set.
16796 * lra.c (lra): Always continue after lra_constraints on the first
16799 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
16801 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
16803 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
16805 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16807 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
16808 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
16809 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
16810 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
16811 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
16812 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
16813 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
16814 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
16815 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
16817 2017-01-26 Marek Polacek <polacek@redhat.com>
16820 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
16821 for the third operand.
16823 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16825 PR middle-end/79236
16826 * omp-low.c (struct omp_context): Add simt_stmt field.
16827 (scan_omp_for): Return omp_context *.
16828 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
16829 context to the _simt_ SIMD stmt.
16830 (lower_omp_for): For combined SIMD with sibling _simt_
16831 SIMD, make sure to use the same decls in _looptemp_
16832 clauses as in the sibling.
16834 2017-01-26 David Sherwood <david.sherwood@arm.com>
16836 PR middle-end/79212
16837 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
16840 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16843 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
16844 emit fld b; fld a; if possible.
16846 * brig-builtins.def: Update copyright years.
16847 * config/arm/arm_acle_builtins.def: Update copyright years.
16849 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16852 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
16853 constraint instead of o for the stxsd instruction.
16855 2017-01-25 Carl Love <cel@us.ibm.com>
16857 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
16858 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
16860 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
16862 * doc/invoke.texi (C++ Dialect Options): Fix typo.
16864 2017-01-25 Richard Biener <rguenther@suse.de>
16866 PR tree-optimization/69264
16867 * target.def (vector_alignment_reachable): Improve documentation.
16868 * doc/tm.texi: Regenerate.
16869 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
16871 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
16872 earlier changes with respect to TYPE_USER_ALIGN.
16873 (vector_alignment_reachable_p): Likewise. Improve dumping.
16875 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16878 * config/arm/arm.md (xordi3): Force constant operand into a register
16881 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16883 * doc/invoke.texi (-fstore-merging): Correct default optimization
16884 levels at which it is enabled.
16885 (-O): Move -fstore-merging from list to...
16888 2017-01-25 Richard Biener <rguenther@suse.de>
16891 * omp-expand.c: Include debug.h.
16892 (expand_omp_taskreg): Make sure to generate early debug before
16893 outlining anything from a function.
16894 (expand_omp_target): Likewise.
16895 (grid_expand_target_grid_body): Likewise.
16897 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
16900 * asan.c (get_translation_unit_decl): New function.
16901 (asan_add_global): Extract modules file name from globals
16902 TRANSLATION_UNIT_DECL name.
16904 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
16907 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
16908 for long calls with APCS frame and VFP.
16910 2017-01-24 David Malcolm <dmalcolm@redhat.com>
16912 * cfg.c (original_copy_tables_initialized_p): New function.
16913 * cfg.h (original_copy_tables_initialized_p): New decl.
16914 * cfgrtl.c (relink_block_chain): Guard the call to
16915 free_original_copy_tables with a call to
16916 original_copy_tables_initialized_p.
16917 * cgraph.h (symtab_node::native_rtl_p): New decl.
16918 * cgraphunit.c (symtab_node::native_rtl_p): New function.
16919 (symtab_node::needed_p): Don't assert for early assembly output
16920 for __RTL functions.
16921 (cgraph_node::finalize_function): Set "force_output" for __RTL
16923 (cgraph_node::analyze): Bail out early for __RTL functions.
16924 (analyze_functions): Update assertion to support __RTL functions.
16925 (cgraph_node::expand): Bail out early for __RTL functions.
16926 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
16928 * function.h (struct function): Update comment for field
16930 * gimple-expr.c: Include "tree-pass.h".
16931 (gimple_has_body_p): Return false for __RTL functions.
16932 * Makefile.in (OBJS): Add run-rtl-passes.o.
16933 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
16935 (gcc::pass_manager::get_clean_slate): New accessor.
16936 * passes.c: Include "insn-addr.h".
16937 (should_skip_pass_p): Add logging. Update logic for running
16938 "expand" to be compatible with both __GIMPLE and __RTL. Guard
16939 property-provider override so it is only done for gimple passes.
16941 (skip_pass): New function.
16942 (execute_one_pass): Call skip_pass when skipping passes.
16943 * read-md.c (md_reader::read_char): Support filtering
16944 the input to a subset of line numbers.
16945 (md_reader::md_reader): Initialize fields
16946 m_first_line and m_last_line.
16947 (md_reader::read_file_fragment): New function.
16948 * read-md.h (md_reader::read_file_fragment): New decl.
16949 (md_reader::m_first_line): New field.
16950 (md_reader::m_last_line): New field.
16951 * read-rtl-function.c (function_reader::create_function): Only
16952 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
16953 curr_properties. Set DECL_INITIAL to a dummy block.
16954 (read_rtl_function_body_from_file_range): New function.
16955 * read-rtl-function.h (read_rtl_function_body_from_file_range):
16957 * run-rtl-passes.c: New file.
16958 * run-rtl-passes.h: New file.
16960 2017-01-24 Jeff Law <law@redhat.com>
16962 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
16965 2017-01-24 Bin Cheng <bin.cheng@arm.com>
16967 PR tree-optimization/79159
16968 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
16969 (record_nonwrapping_iv): Improve boundary using above function if no
16970 value range information.
16972 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
16973 Martin Jambor <mjambor@suse.cz>
16975 * brig-builtins.def: New file.
16976 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
16977 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
16978 (DEF_HSAIL_SAT_BUILTIN): Likewise.
16979 (DEF_HSAIL_INTR_BUILTIN): Likewise.
16980 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
16981 * builtin-types.def (BT_INT8): New.
16982 (BT_INT16): Likewise.
16983 (BT_UINT8): Likewise.
16984 (BT_UINT16): Likewise.
16985 (BT_FN_ULONG): Likewise.
16986 (BT_FN_UINT_INT): Likewise.
16987 (BT_FN_UINT_ULONG): Likewise.
16988 (BT_FN_UINT_LONG): Likewise.
16989 (BT_FN_UINT_PTR): Likewise.
16990 (BT_FN_ULONG_PTR): Likewise.
16991 (BT_FN_INT8_FLOAT): Likewise.
16992 (BT_FN_INT16_FLOAT): Likewise.
16993 (BT_FN_UINT32_FLOAT): Likewise.
16994 (BT_FN_UINT16_FLOAT): Likewise.
16995 (BT_FN_UINT8_FLOAT): Likewise.
16996 (BT_FN_UINT64_FLOAT): Likewise.
16997 (BT_FN_UINT16_UINT32): Likewise.
16998 (BT_FN_UINT32_UINT16): Likewise.
16999 (BT_FN_UINT16_UINT16_UINT16): Likewise.
17000 (BT_FN_INT_PTR_INT): Likewise.
17001 (BT_FN_UINT_PTR_UINT): Likewise.
17002 (BT_FN_LONG_PTR_LONG): Likewise.
17003 (BT_FN_ULONG_PTR_ULONG): Likewise.
17004 (BT_FN_VOID_UINT64_UINT64): Likewise.
17005 (BT_FN_UINT8_UINT8_UINT8): Likewise.
17006 (BT_FN_INT8_INT8_INT8): Likewise.
17007 (BT_FN_INT16_INT16_INT16): Likewise.
17008 (BT_FN_INT_INT_INT): Likewise.
17009 (BT_FN_UINT_FLOAT_UINT): Likewise.
17010 (BT_FN_FLOAT_UINT_UINT): Likewise.
17011 (BT_FN_ULONG_UINT_UINT): Likewise.
17012 (BT_FN_ULONG_UINT_PTR): Likewise.
17013 (BT_FN_ULONG_ULONG_ULONG): Likewise.
17014 (BT_FN_UINT_UINT_UINT): Likewise.
17015 (BT_FN_VOID_UINT_PTR): Likewise.
17016 (BT_FN_UINT_UINT_PTR: Likewise.
17017 (BT_FN_UINT32_UINT64_PTR): Likewise.
17018 (BT_FN_INT_INT_UINT_UINT): Likewise.
17019 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
17020 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
17021 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
17022 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
17023 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
17024 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
17025 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
17026 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
17027 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
17028 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
17029 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
17030 * doc/frontends.texi: List BRIG FE.
17031 * doc/install.texi (Testing): Add BRIG tesring requirements.
17032 * doc/invoke.texi (Overall Options): Mention BRIG.
17033 * doc/standards.texi (Standards): Doucment BRIG HSA version.
17035 2017-01-24 Richard Biener <rguenther@suse.de>
17037 PR translation/79208
17038 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
17040 2017-01-24 Martin Jambor <mjambor@suse.cz>
17043 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
17044 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
17045 and known_contexts.
17047 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
17049 PR middle-end/79123
17050 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
17051 casts from signed to unsigned really don't have a range.
17053 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
17055 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
17056 GMP_RNDx for compatiblity.
17058 2017-01-24 Martin Liska <mliska@suse.cz>
17061 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
17062 that would prevent us to call alloca with -1 as argument.
17064 2017-01-24 Jakub Jelinek <jakub@redhat.com>
17066 * dwarf2out.c (output_compilation_unit_header, output_file_names):
17067 Avoid -Wformat-security warning.
17069 2017-01-23 Andrew Pinski <apinski@cavium.com>
17071 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
17074 2017-01-23 Martin Sebor <msebor@redhat.com>
17076 PR middle-end/78703
17077 * gimple-ssa-sprintf.c (warn_level): New global.
17078 (format_integer): Use it here and throughout the rest of the file.
17079 Use the same switch to compute sign as base.
17080 (maybe_warn): New function.
17081 (format_directive): Factor out warnings into maybe_warn.
17082 Add debugging output. Use warn_level.
17083 (add_bytes): Use warn_level.
17084 (pass_sprintf_length::compute_format_length): Add debugging output.
17085 (try_substitute_return_value): Same.
17086 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
17088 PR middle-end/78703
17089 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
17090 (struct fmtresult, format_integer, format_floating): Adjust.
17091 (fmtresult::fmtresult): Set max correctly in two argument ctor.
17092 (get_string_length, format_string,format_directive): Same.
17093 (pass_sprintf_length::compute_format_length): Same.
17094 (try_substitute_return_value): Simplify slightly.
17096 PR middle-end/78703
17097 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
17098 (fmtresult::operator+=): Outlined.
17099 (struct fmtresult): Add ctors.
17100 (struct conversion_spec): Rename...
17101 (struct directive): ...to this. Add and remove data members.
17102 (directive::set_width, directive::set_precision): New functions.
17103 (format_percent): Use fmtresult ctor.
17104 (get_width_and_precision): Remove.
17105 (format_integer): Make naming changes. Avoid computing width and
17107 (format_floating): Same. Adjust indentation.
17108 (format_character, format_none): New functions.
17109 (format_string): Moved character handling to format_character.
17110 (format_directive): Remove arguments, change return type.
17111 (parse_directive): New function.
17112 (pass_sprintf_length::compute_format_length): Move directive
17113 parsing to parse_directive.
17115 2017-01-23 Jakub Jelinek <jakub@redhat.com>
17117 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
17118 (assign_assembler_name_if_needed): ... this.
17119 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
17120 (assign_assembler_name_if_needed): ... this.
17121 (free_lang_data_in_cgraph): Adjust callers.
17122 * cgraphunit.c (cgraph_node::analyze): Likewise.
17123 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
17126 2017-01-23 Richard Biener <rguenther@suse.de>
17128 PR tree-optimization/79088
17129 PR tree-optimization/79188
17130 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
17131 resetting loop bounds after last path deletion. Reset loop
17132 bounds of the target loop, make code match the comments.
17133 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
17134 Make sure loops need no fixups.
17136 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
17138 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
17139 exponent support with double type for first argument.
17140 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
17141 type returned by __builtin_vec_extract_sig,
17142 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
17143 functions from "vector int" to "vector unsigned int" or from
17144 "vector long long int" to "vector unsigned long long int".
17145 Changed type returned by __builtin_vec_extract_exp,
17146 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
17147 functions from "vector int" to "vector unsigned int" or from
17148 "vector long long int" to "vector unsigned long long int".
17149 Changed return type of __builtin_vec_test_data_class,
17150 __builtin_vec_test_data_class_sp, and
17151 __builtin_vec_test_data_class_dp from "vector int" to
17152 "vector bool int" or from "vector long long int" to "vector bool
17153 long long int" and changed second argument type from "unsigned
17154 int" to "int". Added new overloaded function forms "vector float
17155 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
17156 "vector float __builtin_vec_insert_exp_sp (vector float, vector
17157 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
17158 double, vector unsigned long long int)" and "vector double
17159 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
17160 long int)". Changed return type of
17161 __builtin_scalar_test_data_class and
17162 __builtin_scalar_test_data_class_sp and
17163 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
17164 int" and changed second argument from "unsigned int" to "int".
17165 Changed type returned by __builtin_scalar_test_neg,
17166 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
17167 from "int" to "bool int". Added new overloaded function form
17168 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
17169 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
17170 exponent double-precision with floating point first argument.
17171 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
17172 documentation of scalar_test_data_class, scalar_test_neg,
17173 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
17174 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
17175 vec_test_data_class built-in functions to reflect refinements in
17176 their type signatures.
17178 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
17180 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
17182 (aarch64_elf_asm_destructor): Likewise.
17184 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
17186 PR rtl-optimization/78634
17187 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
17188 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
17189 * ifcvt.c (noce_try_cmove): Add missing cost check.
17191 PR rtl-optimization/71724
17192 * combine.c (if_then_else_cond): Look for situations where it is
17193 beneficial to undo the work of one of the recursive calls.
17195 2017-01-23 Bin Cheng <bin.cheng@arm.com>
17197 PR tree-optimization/70754
17198 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
17199 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
17200 combined stmt before it if not NULL.
17201 (combine_chains): Process refs reversely and compute dominance point
17204 2017-01-23 Martin Liska <mliska@suse.cz>
17206 PR tree-optimization/79196
17207 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
17208 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
17210 (strlen_optimize_stmt): Call the renamed function.
17212 2017-01-23 Michael Matz <matz@suse.de>
17214 PR tree-optimization/78384
17215 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
17217 2017-01-23 Richard Biener <rguenther@suse.de>
17219 PR tree-optimization/79186
17220 * tree-vrp.c (register_new_assert_for): Make sure we've seen
17221 both incoming edges before moving an assert.
17223 2017-01-23 Martin Jambor <mjambor@suse.cz>
17225 * ipa-prop.c (load_from_param_1): Removed.
17226 (load_from_unmodified_param): Bits from load_from_param_1 put back
17228 (load_from_param): Removed.
17229 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
17230 with stmt. Reverted back to use of load_from_unmodified_param.
17232 2017-01-23 Martin Jambor <mjambor@suse.cz>
17235 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
17236 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
17237 field a pointer to garbage collected vector, mark lattices and
17238 ipcp_orig_node with GTY((skip)).
17239 (ipa_get_param_count): Adjust to descriptors being a pointer.
17240 (ipa_get_param): Likewise.
17241 (ipa_get_type): Likewise.
17242 (ipa_get_param_move_cost): Likewise.
17243 (ipa_set_param_used): Likewise.
17244 (ipa_get_controlled_uses): Likewise.
17245 (ipa_set_controlled_uses): Likewise.
17246 (ipa_is_param_used): Likewise.
17247 (ipa_node_params_t): Move into garbage collector. New methods insert
17249 (ipa_node_params_sum): Annotate wth GTY(()).
17250 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
17252 (ipa_load_from_parm_agg): Adjust declaration.
17253 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
17254 * ipa-profile.c (ipa_profile): Likewise.
17255 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
17256 (ipa_populate_param_decls): Make descriptors parameter garbage
17258 (ipa_dump_param): Adjust to descriptors being a pointer.
17259 (ipa_alloc_node_params): Likewise.
17260 (ipa_initialize_node_params): Likewise.
17261 (load_from_param_1): Make descriptors parameter garbage collected.
17262 (load_from_unmodified_param): Likewise.
17263 (load_from_param): Likewise.
17264 (ipa_load_from_parm_agg): Likewise.
17265 (ipa_node_params::~ipa_node_params): Removed.
17266 (ipa_free_all_node_params): Remove call to delete operator.
17267 (ipa_node_params_t::insert): New.
17268 (ipa_node_params_t::remove): Likewise.
17269 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
17270 copy known_csts and known_contexts vectors.
17271 (ipa_read_node_info): Adjust to descriptors being a pointer.
17272 (ipcp_modif_dom_walker): Make m_descriptors field garbage
17274 (ipcp_transform_function): Make descriptors variable garbage
17277 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
17279 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
17280 * config/i386/avx512dqintrin.h: Ditto.
17281 * config/i386/avx512fintrin.h: Ditto.
17282 * config/i386/i386.c: Handle new builtins.
17283 * config/i386/i386-builtin.def: Add new builtins.
17284 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
17285 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
17287 2017-01-23 Jakub Jelinek <jakub@redhat.com>
17288 Martin Liska <mliska@suse.cz>
17290 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
17291 * asan.c (asan_expand_poison_ifn): Support stores and use
17292 appropriate ASAN report function.
17293 * internal-fn.c (expand_ASAN_POISON_USE): New function.
17294 * internal-fn.def (ASAN_POISON_USE): Declare.
17295 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
17296 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
17297 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
17298 ASAN_POISON calls w/o LHS.
17299 * tree-ssa.c (execute_update_addresses_taken): Create clobber
17300 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
17301 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
17302 * gimplify.c (asan_poison_variables): Add attribute
17303 use_after_scope_memory to variables that really needs to live
17305 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
17306 having the attribute.
17308 2017-01-23 Martin Liska <mliska@suse.cz>
17310 * asan.c (create_asan_shadow_var): New function.
17311 (asan_expand_poison_ifn): Likewise.
17312 * asan.h (asan_expand_poison_ifn): New declaration.
17313 * internal-fn.c (expand_ASAN_POISON): Likewise.
17314 * internal-fn.def (ASAN_POISON): New builtin.
17315 * sanopt.c (pass_sanopt::execute): Expand
17316 asan_expand_poison_ifn.
17317 * tree-inline.c (copy_decl_for_dup_finish): Make function
17319 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
17320 * tree-ssa.c (is_asan_mark_p): New function.
17321 (execute_update_addresses_taken): Rewrite local variables
17322 (identified just by use-after-scope as addressable) into SSA.
17324 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
17326 * doc/install.texi (Specific): opensource.apple.com uses https
17327 now. Remove trailing slash.
17329 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
17331 * README.Portability: Remove note on an Irix compatibility issue.
17333 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
17335 * gcov.c (INCLUDE_ALGORITHM): Define.
17336 (INCLUDE_VECTOR): Define.
17337 No longer include <vector> and <algorithm> directly.
17339 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
17341 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
17343 * doc/invoke.texi (Code Gen Options): Ditto.
17345 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
17348 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
17350 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
17352 rtl-optimization/79125
17353 * cprop.c (local_cprop_pass): Handle cases where we make an
17354 unconditional trap.
17356 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
17360 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
17361 read from, for big endian.
17363 2017-01-20 Jiong Wang <jiong.wang@arm.com>
17365 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
17366 register pauth builtins for LP64 only.
17368 2017-01-20 Marek Polacek <polacek@redhat.com>
17371 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
17374 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
17376 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
17378 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
17379 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
17380 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
17382 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17385 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
17386 in big-endian mode when they are not a single duplicated value.
17388 2017-01-20 Richard Biener <rguenther@suse.de>
17390 * BASE-VER: Bump to 7.0.1.
17392 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
17394 * omp-low.c (omplow_simd_context): New struct. Use it...
17395 (lower_rec_simd_input_clauses): ...here and...
17396 (lower_rec_input_clauses): ...here to hold common data. Adjust all
17397 references to idx, lane, max_vf, is_simt.
17399 2017-01-20 Graham Markall <graham.markall@embecosm.com>
17401 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
17404 2017-01-20 Martin Jambor <mjambor@suse.cz>
17406 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
17407 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
17409 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
17410 (GTFILES): Rename hsa.c to hsa-common.c.
17411 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
17412 * hsa-dump.c: Likewise.
17413 * hsa-gen.c: Likewise.
17414 * hsa-regalloc.c: Likewise.
17415 * ipa-hsa.c: Likewise.
17416 * omp-expand.c: Likewise.
17417 * omp-low.c: Likewise.
17418 * toplev.c: Likewise.
17420 2017-01-20 Marek Polacek <polacek@redhat.com>
17423 * doc/invoke.texi: Document -Wduplicated-branches.
17424 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
17425 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
17426 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
17427 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
17428 return 0 only when not OEP_LEXICOGRAPHIC.
17429 (fold_build_cleanup_point_expr): Use the expression
17430 location when building CLEANUP_POINT_EXPR.
17431 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
17432 * tree.c (add_expr): Handle error_mark_node.
17434 2017-01-20 Martin Liska <mliska@suse.cz>
17437 * tree-profile.c (init_ic_make_global_vars): Do not call
17439 (gimple_init_gcov_profiler): Likewise.
17441 2017-01-20 Martin Liska <mliska@suse.cz>
17444 * cgraph.h (maybe_create_reference): Remove argument and
17446 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
17448 * ipa-cp.c (create_specialized_node): Likewise.
17449 * symtab.c (symtab_node::maybe_create_reference): Handle
17450 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
17452 2017-01-20 Martin Liska <mliska@suse.cz>
17454 * read-rtl-function.c (function_reader::create_function): Use
17455 build_decl instread of build_decl_stat.
17457 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
17459 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
17460 * config/i386/avx512dqintrin.h: Ditto.
17461 * config/i386/avx512fintrin.h: Ditto.
17462 * config/i386/i386-builtin-types.def: Add new types.
17463 * config/i386/i386.c: Handle new types.
17464 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
17465 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
17466 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
17467 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
17468 (__builtin_ia32_kshiftridi): New.
17469 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
17471 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
17475 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
17476 define to rs6000_init_stack_protect_guard.
17477 (rs6000_init_stack_protect_guard): New function.
17479 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
17480 Yunqiang Su <yunqiang.su@imgtec.com>
17482 * config.gcc (supported_defaults): Add madd4.
17483 (with_madd4): Add validation.
17484 (all_defaults): Add madd4.
17485 * config/mips/mips.opt (mmadd4): New option.
17486 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17488 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17490 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
17491 (ISA_HAS_FUSED_MADD4): Likewise.
17492 * doc/invoke.texi (-mmadd4): Document the new option.
17493 * doc/install.texi (--with-madd4): Document the new option.
17495 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17497 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
17498 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
17499 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
17500 (aarch64_init_pauth_hint_builtins): New.
17501 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
17502 (aarch64_expand_builtin): Expand new builtins.
17504 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17506 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
17507 * combine-stack-adj.c (no_unhandled_cfa): Handle
17508 REG_CFA_TOGGLE_RA_MANGLE.
17509 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
17510 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
17511 info for return address signing.
17512 (aarch64_expand_epilogue): Likewise.
17514 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17516 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
17517 * config/aarch64/aarch64-protos.h
17518 (aarch64_return_address_signing_enabled): New declaration.
17519 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
17521 (aarch64_expand_prologue): Sign return address before it's pushed onto
17523 (aarch64_expand_epilogue): Authenticate return address fetched from
17525 (aarch64_override_options): Sanity check for ILP32 and ISA level.
17526 (aarch64_attributes): New function attributes for "sign-return-address".
17527 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
17528 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
17529 ("*do_return"): Generate combined instructions according to key index.
17530 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
17531 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
17533 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
17534 * config/aarch64/aarch64.opt (msign-return-address=): New.
17535 * doc/extend.texi (AArch64 Function Attributes): Documents
17536 "sign-return-address=".
17537 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
17539 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
17541 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
17542 overall option summary.
17544 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17546 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
17547 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
17548 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
17549 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
17551 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
17553 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
17554 -mpower9-minmax by default for -mcpu=power9.
17555 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
17556 128-bit floating point.
17558 2017-01-20 Alan Modra <amodra@gmail.com>
17560 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
17561 optimizing for size.
17563 2017-01-20 Alan Modra <amodra@gmail.com>
17566 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
17567 for strcmp and strncmp from corresponding builtin decl.
17569 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17571 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
17572 instead of i386/rtems-64.h.
17573 * config/i386/rtems-64.h: Remove.
17575 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17579 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
17581 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
17583 2017-01-19 Tamar Christina <tamar.christina@arm.com>
17585 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
17586 Change int to HOST_WIDE_INT.
17587 * config/aarch64/aarch64-protos.h
17588 (aarch64_simd_gen_const_vector_dup): Likewise.
17589 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
17591 2017-01-19 David Malcolm <dmalcolm@redhat.com>
17593 * langhooks-def.h (lhd_type_for_size): New decl.
17594 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
17595 * langhooks.c (lhd_type_for_size): New function, taken from
17598 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
17600 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
17601 define_bypass for CR latency.
17602 (power9-cracked-alu): Update bypass latency and remove power9-branch.
17603 (power9-alu2): Add define_bypass for CR latency.
17605 (power9-mul): Update insn latency.
17606 (power9-mul-compare): Update insn latency, bypass latency and remove
17609 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17611 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
17613 * config/aarch64/aarch64.md
17614 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
17615 aarch64_nopcrelative_literal_loads.
17616 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
17618 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
17620 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
17621 TARGET_LOONGSON_3A.
17622 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
17624 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
17627 * config.gcc (supported_defaults): Add lxc1-sxc1.
17628 (with_lxc1_sxc1): Add validation.
17629 (all_defaults): Add lxc1-sxc1.
17630 * config/mips/mips.opt (mlxc1-sxc1): New option.
17631 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17633 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17634 __mips_no_lxc1_sxc1.
17635 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
17636 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
17637 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
17639 2017-01-19 Richard Biener <rguenther@suse.de>
17641 PR tree-optimization/72488
17642 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
17643 sure to restore SSA info.
17644 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
17646 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
17648 PR rtl-optimization/79121
17649 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
17650 of the inner type when shifting an extended value.
17652 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
17655 * symtab.c (symtab_node::equal_address_to): Fix comparing of
17656 interposable aliases.
17658 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
17661 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
17662 Use the evmergelohi instruction.
17663 (mov_si<mode>_e500_subreg4_2_le): Likewise.
17664 (mov_sitf_e500_subreg8_2_be): Likewise.
17665 (mov_sitf_e500_subreg12_2_le): Likewise.
17666 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
17667 (mov_si<mode>_e500_subreg4_2_be): Likewise.
17668 (mov_sitf_e500_subreg8_2_le): Likewise.
17669 (mov_sitf_e500_subreg12_2_be): Likewise.
17671 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17673 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
17674 attribute from vecsimple to vecperm.
17675 (altivec_vbpermq2): Likewise.
17677 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17680 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
17682 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17683 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
17684 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
17685 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
17686 case where N arg is SIZE_MAX.
17687 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
17688 (cmpstrsi): Add pattern.
17690 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17692 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17693 __builtin_vec_revb builtins.
17694 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
17695 built-in functions to support generation of the ISA 3.0 XXBR<x>
17696 vector byte reverse instructions.
17697 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
17698 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
17699 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
17700 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
17701 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
17702 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
17703 (P9V_BUILTIN_VEC_REVB): Likewise.
17704 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
17705 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
17706 (p9_xxbrq_v16qi): Likewise.
17707 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
17708 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
17709 (p9_xxbrh_v8hi): Likewise.
17710 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
17711 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
17712 vec_revb built-in functions.
17714 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
17716 PR rtl-optimization/78952
17717 * config/i386/i386.md (any_extract): New code iterator.
17718 (*insvqi_2): Use any_extract for source operand.
17719 (*insvqi_3): Use any_shiftrt for source operand.
17721 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
17723 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
17725 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
17727 2017-01-18 Matthias Klose <doko@ubuntu.com>
17729 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
17731 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17733 * config/rs6000/altivec.h (vec_bperm): Change #define.
17734 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
17735 (altivec_vbpermq2): New define_insn.
17736 (altivec_vbpermd): Likewise.
17737 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
17738 function interface.
17739 (VBPERMD): Likewise.
17740 (VBPERM): New polymorphic function interface.
17741 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
17742 Add entries for P9V_BUILTIN_VEC_VBPERM.
17743 * doc/extend.texi: Add interfaces for vec_bperm.
17745 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17747 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
17748 first letter of error messages.
17749 (s390_resolve_overloaded_builtin): Likewise.
17750 * config/s390/s390.c (s390_expand_builtin): Likewise.
17751 (s390_invalid_arg_for_unprototyped_fn): Likewise.
17752 (s390_valid_target_attribute_inner_p): Likewise.
17753 * config/s390/s390.md ("tabort"): Likewise.
17755 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
17757 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
17758 (ISA_AVOID_DIV_HILO): New macro.
17759 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
17760 (ISA_HAS_DDIV): Likewise.
17762 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17764 * doc/invoke.texi (fabi-version): Correct number of occurrences.
17766 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17768 * doc/invoke.texi (fabi-version): Spelling fix.
17770 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17773 * doc/invoke.texi (fabi-version): Mention mangling fix for
17776 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17779 * doc/invoke.texi (fabi-version): Document discriminator mangling.
17781 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
17784 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
17785 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
17787 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
17789 (stack_protect_test): Ditto.
17790 * config/rs6000/rs6000.opt (mstack-protector-guard=,
17791 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
17793 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
17794 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
17795 -mstack-protector-guard-offset=.
17796 (RS/6000 and PowerPC Options): Ditto.
17798 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
17800 * config/i386/i386.h (MASK_CLASS_P): New define.
17801 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
17802 there are no registers from different register sets also when
17803 mask registers are used. Update function comment.
17804 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
17805 to (*k/*r) and (*k/*km) alternatives.
17807 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
17809 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
17810 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
17811 (EH_RETURN_HANDLER_RTX): New define.
17812 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
17813 Force frame pointer in EH return functions.
17814 (aarch64_expand_epilogue): Add barrier for eh_return.
17815 (aarch64_final_eh_return_addr): Remove.
17816 (aarch64_eh_return_handler_rtx): New function.
17817 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
17819 (aarch64_eh_return_handler_rtx): New prototype.
17821 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17823 * config/rs6000/altivec.h (vec_rlmi): New #define.
17824 (vec_vrlnm): Likewise.
17825 (vec_rlnm): Likewise.
17826 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
17827 (UNSPEC_VRLNM): Likewise.
17828 (VIlong): New mode iterator.
17829 (altivec_vrl<VI_char>mi): New define_insn.
17830 (altivec_vrl<VI_char>nm): Likewise.
17831 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
17833 (VRLDNM): Likewise.
17834 (RLNM): New polymorphic function entry.
17835 (VRLWMI): New monomorphic function entry.
17836 (VRLDMI): Likewise.
17837 (RLMI): New polymorphic function entry.
17838 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
17839 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
17840 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
17843 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17846 * dwarf2out.c (field_byte_offset): Restore the
17847 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
17848 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
17849 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
17852 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17855 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
17857 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17860 * dwarf2out.c (add_data_member_location_attribute): For constant
17861 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
17862 instead of DW_AT_data_member_location, DW_AT_bit_offset and
17863 DW_AT_byte_size attributes.
17865 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17867 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
17868 after forcing to constant memory when the code model is medium.
17870 2017-01-17 Julia Koval <julia.koval@intel.com>
17873 * config/i386/avx512fintrin.h
17874 (_mm512_i32gather_ps): Change __addr type to void const*.
17875 (_mm512_mask_i32gather_ps): Ditto.
17876 (_mm512_i32gather_pd): Ditto.
17877 (_mm512_mask_i32gather_pd): Ditto.
17878 (_mm512_i64gather_ps): Ditto.
17879 (_mm512_mask_i64gather_ps): Ditto.
17880 (_mm512_i64gather_pd): Ditto.
17881 (_mm512_mask_i64gather_pd): Ditto.
17882 (_mm512_i32gather_epi32): Ditto.
17883 (_mm512_mask_i32gather_epi32): Ditto.
17884 (_mm512_i32gather_epi64): Ditto.
17885 (_mm512_mask_i32gather_epi64): Ditto.
17886 (_mm512_i64gather_epi32): Ditto.
17887 (_mm512_mask_i64gather_epi32): Ditto.
17888 (_mm512_i64gather_epi64): Ditto.
17889 (_mm512_mask_i64gather_epi64): Ditto.
17890 (_mm512_i32scatter_ps): Change __addr type to void*.
17891 (_mm512_mask_i32scatter_ps): Ditto.
17892 (_mm512_i32scatter_pd): Ditto.
17893 (_mm512_mask_i32scatter_pd): Ditto.
17894 (_mm512_i64scatter_ps): Ditto.
17895 (_mm512_mask_i64scatter_ps): Ditto.
17896 (_mm512_i64scatter_pd): Ditto.
17897 (_mm512_mask_i64scatter_pd): Ditto.
17898 (_mm512_i32scatter_epi32): Ditto.
17899 (_mm512_mask_i32scatter_epi32): Ditto.
17900 (_mm512_i32scatter_epi64): Ditto.
17901 (_mm512_mask_i32scatter_epi64): Ditto.
17902 (_mm512_i64scatter_epi32): Ditto.
17903 (_mm512_mask_i64scatter_epi32): Ditto.
17904 (_mm512_i64scatter_epi64): Ditto.
17905 (_mm512_mask_i64scatter_epi64): Ditto.
17906 * config/i386/avx512pfintrin.h
17907 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
17908 (_mm512_mask_prefetch_i32gather_ps): Ditto.
17909 (_mm512_mask_prefetch_i64gather_pd): Ditto.
17910 (_mm512_mask_prefetch_i64gather_ps): Ditto.
17911 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
17912 (_mm512_prefetch_i32scatter_ps): Ditto.
17913 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
17914 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
17915 (_mm512_prefetch_i64scatter_pd): Ditto.
17916 (_mm512_prefetch_i64scatter_ps): Ditto.
17917 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
17918 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
17919 * config/i386/avx512vlintrin.h
17920 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
17921 (_mm_mmask_i32gather_ps): Ditto.
17922 (_mm256_mmask_i32gather_pd): Ditto.
17923 (_mm_mmask_i32gather_pd): Ditto.
17924 (_mm256_mmask_i64gather_ps): Ditto.
17925 (_mm_mmask_i64gather_ps): Ditto.
17926 (_mm256_mmask_i64gather_pd): Ditto.
17927 (_mm_mmask_i64gather_pd): Ditto.
17928 (_mm256_mmask_i32gather_epi32): Ditto.
17929 (_mm_mmask_i32gather_epi32): Ditto.
17930 (_mm256_mmask_i32gather_epi64): Ditto.
17931 (_mm_mmask_i32gather_epi64): Ditto.
17932 (_mm256_mmask_i64gather_epi32): Ditto.
17933 (_mm_mmask_i64gather_epi32): Ditto.
17934 (_mm256_mmask_i64gather_epi64): Ditto.
17935 (_mm_mmask_i64gather_epi64): Ditto.
17936 (_mm256_i32scatter_ps): Change __addr type to void*.
17937 (_mm256_mask_i32scatter_ps): Ditto.
17938 (_mm_i32scatter_ps): Ditto.
17939 (_mm_mask_i32scatter_ps): Ditto.
17940 (_mm256_i32scatter_pd): Ditto.
17941 (_mm256_mask_i32scatter_pd): Ditto.
17942 (_mm_i32scatter_pd): Ditto.
17943 (_mm_mask_i32scatter_pd): Ditto.
17944 (_mm256_i64scatter_ps): Ditto.
17945 (_mm256_mask_i64scatter_ps): Ditto.
17946 (_mm_i64scatter_ps): Ditto.
17947 (_mm_mask_i64scatter_ps): Ditto.
17948 (_mm256_i64scatter_pd): Ditto.
17949 (_mm256_mask_i64scatter_pd): Ditto.
17950 (_mm_i64scatter_pd): Ditto.
17951 (_mm_mask_i64scatter_pd): Ditto.
17952 (_mm256_i32scatter_epi32): Ditto.
17953 (_mm256_mask_i32scatter_epi32): Ditto.
17954 (_mm_i32scatter_epi32): Ditto.
17955 (_mm_mask_i32scatter_epi32): Ditto.
17956 (_mm256_i32scatter_epi64): Ditto.
17957 (_mm256_mask_i32scatter_epi64): Ditto.
17958 (_mm_i32scatter_epi64): Ditto.
17959 (_mm_mask_i32scatter_epi64): Ditto.
17960 (_mm256_i64scatter_epi32): Ditto.
17961 (_mm256_mask_i64scatter_epi32): Ditto.
17962 (_mm_i64scatter_epi32): Ditto.
17963 (_mm_mask_i64scatter_epi32): Ditto.
17964 (_mm256_i64scatter_epi64): Ditto.
17965 (_mm256_mask_i64scatter_epi64): Ditto.
17966 (_mm_i64scatter_epi64): Ditto.
17967 (_mm_mask_i64scatter_epi64): Ditto.
17968 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
17969 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
17970 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
17971 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
17972 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
17973 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
17974 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
17975 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
17976 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
17977 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
17978 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
17979 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
17980 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
17981 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
17982 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
17983 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
17984 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
17985 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
17986 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
17987 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
17988 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
17989 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
17990 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
17991 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
17992 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
17993 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
17994 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
17995 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
17996 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
17997 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
17998 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
17999 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
18000 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
18001 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
18002 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
18003 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
18004 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
18005 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
18006 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
18007 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
18008 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
18009 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
18010 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
18011 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
18012 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
18013 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
18014 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
18015 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
18016 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
18017 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
18018 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
18019 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
18020 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
18021 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
18022 definitions accordingly.
18024 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
18025 Kuan-Lin Chen <kuanlinchentw@gmail.com>
18028 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
18031 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
18034 * ira-conflicts.c (ira_build_conflicts): Update total conflict
18035 hard regs for inner regno.
18037 2017-01-17 Martin Liska <mliska@suse.cz>
18040 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
18041 assumption and add comment.
18043 2017-01-17 Nathan Sidwell <nathan@acm.org>
18045 * ipa-visibility.c (localize_node): New function, broken out of ...
18046 (function_and_variable_visibility): ... here. Call it.
18048 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
18050 PR middle-end/77445
18051 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
18052 correctly set frequency of oudgoing edge.
18053 (duplicate_thread_path): Fix profile updating.
18055 2017-01-17 Jakub Jelinek <jakub@redhat.com>
18058 * configure.ac: Add GCC_BASE_VER.
18059 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
18060 version from BASE-VER file.
18061 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
18062 (gcc.o): Depend on $(BASEVER).
18063 * common.opt (dumpfullversion): New option.
18064 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
18065 * doc/invoke.texi: Document -dumpfullversion.
18066 * doc/install.texi: Document --with-gcc-major-version-only.
18067 * configure: Regenerated.
18069 2017-01-17 Richard Biener <rguenther@suse.de>
18071 PR tree-optimization/71433
18072 * tree-vrp.c (register_new_assert_for): Merge same asserts
18073 on all incoming edges.
18074 (process_assert_insertions_for): Handle insertions at the
18077 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
18079 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
18080 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
18082 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
18085 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
18088 2017-01-17 Alan Modra <amodra@gmail.com>
18091 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
18092 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
18093 symbolic stack limit when pic.
18095 2017-01-16 Martin Sebor <msebor@redhat.com>
18097 PR tree-optimization/78608
18098 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
18100 2017-01-16 Jeff Law <law@redhat.com>
18103 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
18104 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
18105 for several include directories that may be relative to sysroot.
18106 * config/i386/x-mingw32 (gplus_includedir): Define.
18107 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
18108 (native_system_includedir): Likewise.
18109 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
18110 override if TARGET_SYSTEM_ROOT is defined.
18111 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
18113 PR tree-optimization/79090
18114 PR tree-optimization/33562
18115 PR tree-optimization/61912
18116 PR tree-optimization/77485
18117 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
18118 and computed trims into the dump file.
18120 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
18122 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
18124 2017-01-16 Jakub Jelinek <jakub@redhat.com>
18127 * gimplify.c (gimplify_init_constructor): If want_value and
18128 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
18132 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
18133 sequence. Formatting fixes.
18134 (doloop_optimize): Formatting fixes.
18137 * gcc.c (debug_level_greater_than_spec_func): New function.
18138 (static_spec_functions): Add debug-level-gt spec function.
18139 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
18141 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
18142 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
18143 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
18144 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
18145 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
18146 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
18148 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
18150 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
18151 QImode fixups to general and mask registers only.
18153 2017-01-16 Carl Love <cel@us.ibm.com>
18155 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
18156 for built-in functions
18157 vector signed char vec_nabs (vector signed char)
18158 vector signed short vec_nabs (vector signed short)
18159 vector signed int vec_nabs (vector signed int)
18160 vector signed long long vec_nabs (vector signed long long)
18161 vector float vec_nabs (vector float)
18162 vector double vec_nabs (vector double)
18163 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
18165 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
18166 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
18167 * doc/extend.texi: Update the documentation file for the new built-in
18170 2017-01-16 Martin Sebor <msebor@redhat.com>
18172 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
18175 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18177 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
18178 UNSPEC_VSX__XXSPLTD to require special splat handling.
18180 2017-01-16 David Malcolm <dmalcolm@redhat.com>
18183 * system.h: Poison strndup.
18185 2017-01-16 Alan Modra <amodra@gmail.com>
18188 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
18191 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
18193 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
18195 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
18197 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
18198 call recog here. Assert that INSN_CODE (insn) is non-negative.
18200 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
18203 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
18205 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
18206 in the currently scheduled RTL fragment.
18208 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
18210 PR rtl-optimization/78751
18211 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
18214 2017-01-14 Jeff Law <law@redhat.com>
18216 PR tree-optimization/79090
18217 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
18218 variable length stores.
18219 (compute_trims): Delete dead assignment to *trim_tail.
18220 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
18223 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
18225 PR rtl-optimization/78626
18226 PR rtl-optimization/78727
18227 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
18228 of a block, and split such blocks after everything else is finished.
18230 2017-01-14 Alan Modra <amodra@gmail.com>
18233 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
18234 target legitimate_combined_insn.
18235 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
18236 (rs6000_legitimate_combined_insn): New function.
18237 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
18239 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
18240 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
18241 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
18243 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
18245 * doc/frontends.texi (G++ and GCC): Remove references to Java.
18247 2017-01-13 Jeff Law <law@redhat.com>
18249 PR tree-optimization/33562
18250 PR tree-optimization/61912
18251 PR tree-optimization/77485
18252 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
18254 (delete_dead_assignment): Likewise.
18255 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
18256 statement to delete_dead_call and delete_dead_assignment.
18258 2017-01-13 David Malcolm <dmalcolm@redhat.com>
18261 * substring-locations.c (format_warning_va): Strengthen case 1 so
18262 that both endpoints of the substring must be within the format
18263 range for just the substring to be printed.
18265 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
18267 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
18268 * config/i386/i386.c (ix86_target_string): Add missing options
18269 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
18270 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
18271 flags_other and ix86_target_other to flags2_other. Display unknown
18273 (ix86_valid_target_attribute_inner_p): Add missing options and
18274 reorder options by implied ISAs, as in ix86_target_string.
18276 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
18278 * hash-table.h (hash_table::too_empty_p): New function.
18279 (hash_table::expand): Use it.
18280 (hash_table::traverse): Likewise.
18281 (hash_table::empty_slot): Use sizeof (value_type) instead of
18282 sizeof (PTR) to convert bytes to elements. Shrink the table
18283 if the current size is excessive for the current number of
18286 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
18288 * ira-costs.c (record_reg_classes): Break from the inner loop
18289 early once alt_fail is known to be true. Update outer loop
18290 handling accordingly.
18292 2017-01-13 Jeff Law <law@redhat.com>
18294 * tree-ssa-dse.c (decrement_count): New function.
18295 (increment_start_addr, maybe_trim_memstar_call): Likewise.
18296 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
18297 when we know the partially dead statement is a mem* function.
18299 PR tree-optimization/61912
18300 PR tree-optimization/77485
18301 * tree-ssa-dse.c: Include expr.h.
18302 (maybe_trim_constructor_store): New function.
18303 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
18305 PR tree-optimization/33562
18306 PR tree-optimization/61912
18307 PR tree-optimization/77485
18308 * doc/invoke.texi: Document new dse-max-object-size param.
18309 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
18310 * tree-ssa-dse.c: Include params.h.
18311 (dse_store_status): New enum.
18312 (initialize_ao_ref_for_dse): New, partially extracted from
18314 (valid_ao_ref_for_dse, normalize_ref): New.
18315 (setup_live_bytes_from_ref, compute_trims): Likewise.
18316 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
18317 (maybe_trim_partially_dead_store): Likewise.
18318 (maybe_trim_complex_store): Likewise.
18319 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
18320 Track what bytes live from the original store. Return tri-state
18321 for dead, partially dead or live.
18322 (dse_dom_walker): Add constructor, destructor and new private members.
18323 (delete_dead_call, delete_dead_assignment): New extracted from
18325 (dse_optimize_stmt): Make a member of dse_dom_walker.
18326 Use initialize_ao_ref_for_dse.
18328 PR tree-optimization/33562
18329 PR tree-optimization/61912
18330 PR tree-optimization/77485
18331 * sbitmap.h (bitmap_count_bits): Prototype.
18332 (bitmap_clear_range, bitmap_set_range): Likewise.
18333 * sbitmap.c (bitmap_clear_range): New function.
18334 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
18336 2017-01-13 Martin Liska <mliska@suse.cz>
18339 * function.c (set_cfun): Add new argument force.
18340 * function.h (set_cfun): Likewise.
18341 * ipa-inline-transform.c (inline_call): Use the function when
18342 strict alising from is dropped for function we inline to.
18344 2017-01-13 Richard Biener <rguenther@suse.de>
18346 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
18347 for dumping GIMPLE INTEGER_CSTs.
18349 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18351 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
18352 to 201112L since C++17.
18354 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
18357 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
18358 if -fsanitize=kernel-address is present.
18360 2017-01-13 Richard Biener <rguenther@suse.de>
18362 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
18363 as _Literal ( type ) number in case usual suffixes do not
18364 preserve all information.
18366 2017-01-13 Richard Biener <rguenther@suse.de>
18368 PR tree-optimization/77283
18369 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
18370 and ssa-iterators.h.
18371 (is_feasible_trace): Implement a cost model based on joiner
18374 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
18377 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
18378 char or short to __float128/_Float128 directly.
18380 2017-01-12 Martin Sebor <msebor@redhat.com>
18382 to -Wformat-overflow.
18383 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
18384 (min_bytes_remaining): Same.
18385 (get_string_length): Same.
18386 (format_string): Same.
18387 (format_directive): Same.
18389 (pass_sprintf_length::handle_gimple_call): Same.
18391 2017-01-12 Jakub Jelinek <jakub@redhat.com>
18393 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
18394 info.nowrite calls with no lhs that can't throw. Return bool
18395 whether gsi_remove has been called or not.
18396 (pass_sprintf_length::handle_gimple_call): Return bool whether
18397 try_substitute_return_value called gsi_remove. Formatting fix.
18398 (pass_sprintf_length::execute): Don't use gsi_remove if
18399 handle_gimple_call returned true.
18402 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
18403 be removed due to side-effects, don't remove following barrier nor
18404 turn the successor edge into fallthru edge.
18406 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18409 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
18410 element-reversing loads and stores as not swappable.
18412 2017-01-12 Nathan Sidwell <nathan@acm.org>
18413 Nicolai Stange <nicstange@gmail.com>
18415 * combine.c (try_combine): Don't ignore result of overlap checking
18416 loop. Combine overlap & asm check into single loop.
18418 2017-01-12 Richard Biener <rguenther@suse.de>
18420 * tree-pretty-print.c (dump_generic_node): Provide -gimple
18421 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
18423 2017-01-12 Richard Biener <rguenther@suse.de>
18425 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
18426 and TS_TARGET_OPTION directly derive from TS_BASE.
18427 * tree-core.h (tree_optimization_option): Derive from tree_base.
18428 (tree_target_option): Likewise.
18430 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
18432 * config/i386/i386.c (memory_address_length): Increase len
18433 only when rip_relative_addr_p returns false.
18435 2017-01-11 Julia Koval <julia.koval@intel.com>
18437 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
18438 (OPTION_MASK_ISA_SGX_SET): New.
18439 (ix86_handle_option): Handle OPT_msgx.
18440 * config.gcc: Added sgxintrin.h.
18441 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
18442 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
18443 * config/i386/i386.c (ix86_target_string): Add -msgx.
18445 (ix86_option_override_internal): Handle new options.
18446 (ix86_valid_target_attribute_inner_p): Add sgx.
18447 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
18448 * config/i386/i386.opt: Add msgx.
18449 * config/i386/sgxintrin.h: New file.
18450 * config/i386/x86intrin.h: Add sgxintrin.h.
18452 2017-01-11 Jakub Jelinek <jakub@redhat.com>
18455 * fold-const.c (maybe_nonzero_address): Return 1 for function
18457 (tree_single_nonzero_warnv_p): Don't handle function local objects
18461 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
18464 2017-01-11 David Malcolm <dmalcolm@redhat.com>
18467 * opts.c: Include "spellcheck.h"
18468 (struct string_fragment): New struct.
18469 (struct edit_distance_traits<const string_fragment &>): New
18471 (get_closest_sanitizer_option): New function.
18472 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
18474 2017-01-11 Jakub Jelinek <jakub@redhat.com>
18476 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
18478 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
18479 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
18480 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
18481 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
18482 for initial die_offset if dwarf_split_debug_info.
18483 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
18484 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
18486 (output_skeleton_debug_sections): Formatting fix. Use
18487 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
18488 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
18490 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
18492 * config/arm/cortex-a53.md: Add bypasses for
18493 cortex_a53_r2f_cvt.
18494 (cortex_a53_r2f): Only use for transfers.
18495 (cortex_a53_f2r): Likewise.
18496 (cortex_a53_r2f_cvt): Add reservation for conversions.
18497 (cortex_a53_f2r_cvt): Likewise.
18499 2017-01-11 Tamar Christina <tamar.christina@arm.com>
18501 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
18502 to all inlined functions, change static to extern.
18504 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
18507 * config/arm/arm.c (legitimize_pic_address): Handle reference to
18509 (arm_assemble_integer): Likewise.
18511 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
18513 * config.gcc: Use new awk script to check CPU, FPU and architecture
18514 parameters for --with-... options.
18515 * config/arm/parsecpu.awk: New file
18516 * config/arm/arm-cpus.in: New file.
18517 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
18519 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
18521 * config/arm/t-arm: Update dependency rules.
18522 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
18523 of processing .def files.
18524 * config/arm/genopt.sh: Deleted.
18525 * config/arm/gentune.sh: Deleted.
18526 * config/arm/arm-cores.def: Deleted.
18527 * config/arm/arm-arches.def: Deleted.
18528 * config/arm/arm-fpus.def: Deleted.
18529 * config/arm/arm-tune.md: Regenerated.
18530 * config/arm/arm-tables.opt: Regenerated.
18531 * config/arm/arm-cpu.h: New generated file.
18532 * config/arm/arm-cpu-data.h: New generated file.
18533 * config/arm/arm-cpu-cdata.h: New generated file.
18535 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
18538 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
18540 (input_varpool_node): Unpack dynamically_initialized bit.
18542 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
18544 PR rtl-optimization/79032
18545 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
18546 the alignment of the adjusted memory reference against that of MODE,
18547 instead of the alignment of the original memory reference.
18549 2017-01-11 Martin Jambor <mjambor@suse.cz>
18551 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
18553 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
18554 decorated functions.
18556 2017-01-11 Richard Biener <rguenther@suse.de>
18558 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
18559 set range/nonnull info for PHI results. Do not set it on
18560 stmts marked for removal.
18562 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
18564 * expr.c (store_field): In the bitfield case, fetch the return value
18565 from the registers before applying a single big-endian adjustment.
18566 Always do a final load for a BLKmode value not larger than a word.
18568 2017-01-10 David Malcolm <dmalcolm@redhat.com>
18571 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18572 that we correctly handle column numbers greater than
18573 LINE_MAP_MAX_COLUMN_NUMBER.
18575 2017-01-10 Martin Sebor <msebor@redhat.com>
18577 PR middle-end/78245
18578 * gimple-ssa-sprintf.c (get_destination_size): Call
18579 {init,fini}object_sizes.
18580 * tree-object-size.c (addr_object_size): Adjust.
18581 (pass_through_call): Adjust.
18582 (pass_object_sizes::execute): Adjust.
18583 * tree-object-size.h (fini_object_sizes): Declare.
18585 2017-01-10 Martin Sebor <msebor@redhat.com>
18587 PR tree-optimization/78775
18588 * builtins.c (get_size_range): Move...
18589 * calls.c: ...to here.
18590 (alloc_max_size): Accept zero argument.
18591 (operand_signed_p): Remove.
18592 (maybe_warn_alloc_args_overflow): Call get_size_range.
18593 * calls.h (get_size_range): Declare.
18595 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
18597 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
18598 from TI's devices.csv file as of September 2016.
18599 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
18601 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
18603 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
18604 * doc/invoke.texi: Likewise.
18605 * doc/md.texi: Likewise.
18606 * doc/objc.texi: Likewise.
18608 2017-01-10 Joshua Conner <joshconner@google.com>
18610 * config/arm/fuchsia-elf.h: New file.
18611 * config/fuchsia.h: New file.
18612 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
18613 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
18615 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
18617 2016-01-10 Richard Biener <rguenther@suse.de>
18619 PR tree-optimization/79034
18620 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18621 Propagate out degenerate PHIs in the joiner.
18623 2017-01-10 Martin Liska <mliska@suse.cz>
18625 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
18626 (sort_congruence_classes_by_decl_uid): Likewise.
18627 (sort_congruence_class_groups_by_decl_uid): Likewise.
18628 (sem_item_optimizer::merge_classes): Sort class, groups in these
18629 classes and members in the groups by DECL_UID of declarations.
18630 This would make merge operations stable.
18632 2017-01-10 Martin Liska <mliska@suse.cz>
18634 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
18635 usage of m_classes_vec.
18636 (sem_item_optimizer::~sem_item_optimizer): Likewise.
18637 (sem_item_optimizer::get_group_by_hash): Likewise.
18638 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
18639 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
18640 (sem_item_optimizer::verify_classes): Likewise.
18641 (sem_item_optimizer::process_cong_reduction): Likewise.
18642 (sem_item_optimizer::dump_cong_classes): Likewise.
18643 (sem_item_optimizer::merge_classes): Likewise.
18644 * ipa-icf.h (congruence_class_hash): Rename from
18645 congruence_class_group_hash. Remove declaration of m_classes_vec.
18647 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
18649 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
18650 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
18651 * config.gcc: Add avx512vpopcntdqintrin.h.
18652 * config/i386/avx512vpopcntdqintrin.h: New.
18653 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
18654 * config/i386/i386-builtin-types.def: Add new types.
18655 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
18656 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
18657 __builtin_ia32_vpopcountq_v8di_mask): New.
18658 * config/i386/i386-c.c (ix86_target_macros_internal): Define
18659 __AVX512VPOPCNTDQ__.
18660 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
18661 (PTA_AVX512VPOPCNTDQ): Define.
18662 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
18663 TARGET_AVX512VPOPCNTDQ_P): Define.
18664 * config/i386/i386.opt: Add mavx512vpopcntdq.
18665 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
18666 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
18668 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18670 PR middle-end/77484
18671 * predict.def (PRED_CALL): Set to 67.
18673 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
18675 * expr.c (store_field): In the bitfield case, if the value comes from
18676 a function call and is of an aggregate type returned in registers, do
18677 not modify the field mode; extract the value in all cases if the mode
18678 is BLKmode and the size is not larger than a word.
18680 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
18683 * config/i386/cpuid.h: Fix undefined behavior.
18685 2017-01-04 Jeff Law <law@redhat.com>
18687 PR tree-optimization/79007
18688 PR tree-optimization/67955
18689 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
18690 conservative for pt.null when flag_non_call_exceptions is on.
18692 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18694 PR translation/79019
18695 PR translation/79020
18696 * params.def (PARAM_INLINE_MIN_SPEEDUP,
18697 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
18698 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
18700 * config/avr/avr.opt (maccumulate-args): Likewise.
18701 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
18702 * common.opt (freport-bug): Likewise.
18703 * cif-code.def (CIF_FINAL_ERROR): Likewise.
18704 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
18705 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
18706 translatable string.
18707 * config/i386/i386.c (function_value_32): Likewise.
18708 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
18709 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
18711 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
18712 * common/config/msp430/msp430-common.c (msp430_handle_option):
18714 * symtab.c (symtab_node::verify_base): Likewise.
18715 * opts.c (set_debug_level): Likewise.
18716 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
18717 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
18718 missing whitespace to translatable strings.
18719 * config/avr/avr.md (bswapsi2): Fix typo in comment.
18720 * config/sh/superh.h: Likewise.
18721 * config/i386/xopintrin.h: Likewise.
18722 * config/i386/znver1.md: Likewise.
18723 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
18724 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
18725 * double-int.h (struct double_int): Likewise.
18726 * double-int.c (div_and_round_double): Likewise.
18727 * wide-int.cc: Likewise.
18728 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
18729 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
18730 * cfgcleanup.c (crossjumps_occured): Renamed to ...
18731 (crossjumps_occurred): ... this.
18732 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
18735 PR tree-optimization/78899
18736 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
18737 returning bool return struct loop *, NULL for failure and the new
18739 (versionable_outer_loop_p): Don't version outer loop if it has
18740 dont_vectorized bit set.
18741 (tree_if_conversion): When versioning outer loop, ensure
18742 tree_if_conversion is performed also on the inner loop of the
18743 non-vectorizable outer loop copy.
18744 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
18745 LOOP_VECTORIZED in inner loop of the scalar outer loop and
18746 prevent vectorization of it.
18747 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
18748 the outer loop vectorization of the non-scalar version is attempted
18749 before vectorization of the inner loop in scalar version. If
18750 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
18751 vectorization of its inner loop.
18752 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
18753 has 2 inner loops, rename also on edges from bb whose single pred
18754 is outer_loop->header. Fix typo in function comment.
18756 2017-01-09 Martin Sebor <msebor@redhat.com>
18759 * asan.c (asan_emit_stack_protection): Increase local buffer size
18760 to avoid snprintf truncation warning.
18762 2017-01-09 Andrew Pinski <apinski@cavium.com>
18764 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
18765 to reference thunderx2t99 for the tuning structure
18766 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
18768 (thunderx2t99_extra_costs): This.
18769 * config/aarch64/aarch64-tune.md: Regenerate.
18770 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
18771 (vulcan_addrcost_table): This.
18772 (vulcan_regmove_cost): Rename to ...
18773 (thunderx2t99_regmove_cost): This.
18774 (vulcan_vector_cost): Rename to ...
18775 (thunderx2t99_vector_cost): this.
18776 (vulcan_branch_cost): Rename to ...
18777 (thunderx2t99_branch_cost): This.
18778 (vulcan_tunings): Rename to ...
18779 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
18780 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
18782 2017-01-09 Martin Jambor <mjambor@suse.cz>
18786 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
18787 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
18788 (propagate_vr_accross_jump_function): Use the above function for all
18789 value range computations for pass-through jump functions and type
18790 converasion from explicit value range values.
18791 (ipcp_propagate_stage): Do not attempt to deduce types of formal
18792 parameters from TYPE_ARG_TYPES.
18793 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
18794 (ipa_write_node_info): Stream type of the actual argument.
18795 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
18797 2017-01-09 Martin Liska <mliska@suse.cz>
18800 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
18801 (lookup_compiler): Do not show error message with have_E.
18803 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18805 PR tree-optimization/78938
18806 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
18807 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
18808 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
18809 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
18812 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18814 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
18817 2017-01-09 Richard Biener <rguenther@suse.de>
18819 PR tree-optimization/78997
18820 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
18821 name condition properly.
18823 2017-01-09 Richard Biener <rguenther@suse.de>
18826 * dwarf2out.c (is_cxx): New overload with context.
18827 (is_naming_typedef_decl): Use it.
18829 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18831 * invoke.texi (Option Summary): Correct spacing in option lists
18832 and add line breaks to fix over-long lines.
18834 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18836 PR middle-end/17660
18838 * extend.texi (Common Variable Attributes): Add xref to GCC
18839 Internals manual to explain mode attribute keywords.
18841 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18844 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
18845 and Preprocessor Options.
18846 (Options for Linking): Document -pthread here....
18847 (RS/6000 and PowerPC Options): ...not here.
18848 (Solaris 2 Options): ...or here.
18849 * doc/cppopts.texi: Document -pthread.
18851 2017-01-08 Martin Sebor <msebor@redhat.com>
18853 PR middle-end/77708
18854 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
18855 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
18856 New member functions.
18857 (format_directive): Used them.
18859 (pass_sprintf_length::handle_gimple_call): Same.
18860 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
18861 to avoid truncation for any argument.
18862 (extract_affine_mul): Same.
18863 * tree.c (get_file_function_name): Same.
18865 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18867 PR middle-end/77484
18868 * predict.def (PRED_INDIR_CALL): Set to 86.
18870 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18872 PR preprocessor/54124
18873 * doc/cppopts.texi: Reformat -d subtable to list the full name
18874 of the options. Add cross-reference to the docs for the general
18875 compiler -d options.
18876 * doc/invoke.texi (Developer Options): Add cross-reference to the
18877 preprocessor-specific -d option documentation.
18879 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18881 PR preprocessor/13498
18882 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
18883 redudant material, and reflect new command-line options.
18884 (System Headers): Likewise.
18886 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18888 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
18889 -isystem, and -idirafter. Copy-edit.
18890 * doc/cppopts.texi: Copy-edit. Remove contradiction about
18891 default for -ftrack-macro-expansion. Delete obsolete and
18892 badly-formatted implementation details about -fdebug-cpp output.
18893 * doc/cppwarnopts.texi: Copy-edit.
18895 2017-01-07 David Malcolm <dmalcolm@redhat.com>
18898 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18899 that the transition from a max line width >= 1<<10 to narrower
18900 lines works correctly.
18902 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
18904 * doc/options.texi (PerFunction): New.
18905 * opt-functions.awk (switch_flags): Map both Optimization and
18906 PerFunction to CL_OPTIMIZATION.
18907 * opth-gen.awk: Test for PerFunction flag along with
18909 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
18910 it only when the latter is present. Skip those that don't in
18911 the hash function generator.
18912 * common.opt (fvar-tracking): Mark as PerFunction instead of
18914 (fvar-tracking-assignments): Likewise.
18915 (fvar-tracking-assignments-toggle): Likewise.
18916 (fvar-tracking-uninit): Likewise.
18918 2017-01-07 Jakub Jelinek <jakub@redhat.com>
18920 PR translation/79018
18921 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
18924 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
18927 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
18928 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
18929 TARGET_LONG_JUMP_TABLE_OFFSETS.
18930 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
18931 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
18932 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
18933 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
18934 * config/m68k/m68k.md (tablejump expander): Likewise.
18935 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
18936 TARGET_LONG_JUMP_TABLE_OFFSETS.
18937 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
18938 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
18940 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
18941 David Holsgrove <david.holsgrove@xilinx.com>
18943 * common/config/microblaze/microblaze-common.c
18944 (TARGET_EXCEPT_UNWIND_INFO): Remove.
18945 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
18947 * config/microblaze/microblaze.c (microblaze_must_save_register)
18948 (microblaze_expand_epilogue, microblaze_return_addr): Handle
18950 (microblaze_eh_return): New function.
18951 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
18952 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
18953 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
18954 * config/microblaze/microblaze.md (eh_return): New pattern.
18956 2017-01-06 Jakub Jelinek <jakub@redhat.com>
18958 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
18959 GCC_DIAGNOSTIC_STRINGIFY): Define.
18961 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
18963 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18965 * config/arm/arm.md (<mcrr>): New.
18967 * config/arm/arm.c (arm_arch5te): New.
18968 (arm_option_override): Set arm_arch5te.
18969 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
18971 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
18972 (arm_mcrr_qualifiers): ... this. New.
18973 (MRRC_QUALIFIERS): Define to...
18974 (arm_mrrc_qualifiers): ... this. New.
18975 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
18977 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
18978 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
18979 (MRRCI, mrrc, MRRC): New.
18980 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
18981 VUNSPEC_MRRC2): New.
18983 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18985 * config/arm/arm.md (<mcr>): New.
18987 * config/arm/arm.c (arm_coproc_builtin_available): Add
18988 support for mcr, mrc, mcr2 and mrc2.
18989 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
18990 (arm_mcr_qualifiers): ... this. New.
18991 (MRC_QUALIFIERS): Define to ...
18992 (arm_mrc_qualifiers): ... this. New.
18993 (MCR_QUALIFIERS): Define to ...
18994 (arm_mcr_qualifiers): ... this. New.
18995 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
18997 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
18998 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
18999 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
19000 VUNSPEC_MRC2): New.
19002 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19004 * config/arm/arm.md (*ldc): New.
19008 * config/arm/arm.c (arm_coproc_builtin_available): Add
19009 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
19010 (arm_coproc_ldc_stc_legitimate_address): New.
19011 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
19012 'qualifier_const_pointer'.
19013 (LDC_QUALIFIERS): Define to...
19014 (arm_ldc_qualifiers): ... this. New.
19015 (STC_QUALIFIERS): Define to...
19016 (arm_stc_qualifiers): ... this. New.
19017 * config/arm/arm-protos.h
19018 (arm_coproc_ldc_stc_legitimate_address): New.
19019 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
19020 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
19021 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
19022 stc2, stcl, stc2l): New.
19023 * config/arm/constraints.md (Uz): New.
19024 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
19025 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
19026 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
19027 VUNSPEC_STC2L): New.
19029 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19031 * config/arm/arm.md (<cdp>): New.
19032 * config/arm/arm.c (neon_const_bounds): Rename this ...
19033 (arm_const_bounds): ... this.
19034 (arm_coproc_builtin_available): New.
19035 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
19036 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
19037 (CDP_QUALIFIERS): Define to...
19038 (arm_cdp_qualifiers): ... this. New.
19040 (arm_expand_builtin_args): Add case for 6 arguments.
19041 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
19042 (arm_const_bounds): ... this.
19043 (arm_coproc_builtin_available): New.
19044 * config/arm/arm_acle.h (__arm_cdp): New.
19046 * config/arm/arm_acle_builtins.def (cdp): New.
19048 * config/arm/iterators.md (CDPI,CDP,cdp): New.
19049 * config/arm/neon.md: Rename all 'neon_const_bounds' to
19050 'arm_const_bounds'.
19051 * config/arm/types.md (coproc): New.
19052 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
19053 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
19054 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
19055 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
19057 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19059 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
19060 (UBINOP_QUALIFIERS): New.
19062 (acle_builtin_data): New. Change comment.
19063 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
19064 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
19065 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
19066 arm_acle_builtins.def.
19067 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
19068 (arm_init_acle_builtins): New.
19069 (CRC32_BUILTIN): Remove.
19070 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
19071 crc32cb, crc32ch and crc32cw.
19072 (arm_init_crc32_builtins): Remove.
19073 (arm_init_builtins): Use arm_init_acle_builtins rather
19074 than arm_init_crc32_builtins.
19075 (arm_expand_acle_builtin): New.
19076 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
19077 * config/arm/arm_acle_builtins.def: New.
19079 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19081 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
19082 (arm_builtin_datum): ... this.
19083 (arm_init_neon_builtin): Rename to ...
19084 (arm_init_builtin): ... this. Add a new parameters PREFIX
19085 and USE_SIG_IN_NAME.
19086 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
19087 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
19088 'arm_builtin_datum'.
19089 (arm_init_vfp_builtins): Likewise.
19090 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
19091 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
19092 (arm_expand_neon_args): Rename to ...
19093 (arm_expand_builtin_args): ... this. Rename builtin_arg
19094 enum values and differentiate between ARG_BUILTIN_MEMORY
19095 and ARG_BUILTIN_NEON_MEMORY.
19096 (arm_expand_neon_builtin_1): Rename to ...
19097 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
19098 values, arm_expand_builtin_args and add bool parameter NEON.
19099 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
19100 (arm_expand_vfp_builtin): Likewise.
19101 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
19103 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19105 PR middle-end/77484
19106 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
19107 * predict.c (tree_estimate_probability_bb): Reverse direction of
19108 polymorphic call predictor.
19110 2017-01-06 David Malcolm <dmalcolm@redhat.com>
19112 * passes.c (execute_one_pass): Split out pass-skipping logic into...
19113 (determine_pass_name_match): ...this new function and...
19114 (should_skip_pass_p): ...this new function.
19116 2017-01-06 Nathan Sidwell <nathan@acm.org>
19118 * ipa-visibility.c (function_and_variable_visibility): Reformat
19119 comments and long lines. Remove extrneous if.
19120 * symtab.c (symtab_node::make_decl_local): Fix code format.
19121 (symtab_node::set_section_for_node): Fix comment typo.
19123 2017-01-06 Martin Liska <mliska@suse.cz>
19126 * lra-constraints.c: Rename invariant to lra_invariant.
19127 * predict.c (set_even_probabilities): Initialize e to NULL.
19129 2017-01-05 Martin Sebor <msebor@redhat.com>
19131 PR tree-optimization/78910
19132 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
19133 (format_integer): Correct off-by-one error in the handling
19134 of precision with negative numbers in signed conversions..
19136 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
19138 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
19140 2017-01-05 Jakub Jelinek <jakub@redhat.com>
19142 PR tree-optimization/71016
19143 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
19144 factor_out_conditional_conversion. Formatting fix.
19145 (factor_out_conditional_conversion): Add cond_stmt argument.
19146 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
19147 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
19150 2017-01-05 David Malcolm <dmalcolm@redhat.com>
19152 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
19153 read-rtl-function.o, and selftest-rtl.o.
19154 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
19155 (selftest::aarch64_test_loading_full_dump): New function.
19156 (selftest::aarch64_run_selftests): New function.
19157 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
19158 selftest::aarch64_run_selftests.
19159 * config/i386/i386.c
19160 (selftest::ix86_test_loading_dump_fragment_1): New function.
19161 (selftest::ix86_test_loading_call_insn): New function.
19162 (selftest::ix86_test_loading_full_dump): New function.
19163 (selftest::ix86_test_loading_unspec): New function.
19164 (selftest::ix86_run_selftests): Call the new functions.
19165 * emit-rtl.c (maybe_set_max_label_num): New function.
19166 * emit-rtl.h (maybe_set_max_label_num): New decl.
19167 * function.c (instantiate_decls): Guard call to
19168 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
19169 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
19171 * gensupport.c (gen_reader::gen_reader): Pass "false"
19172 for new "compact" param of rtx_reader.
19173 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
19174 rather than an empty string for NULL strings.
19175 * read-md.c: Potentially include config.h rather than bconfig.h.
19176 Wrap include of errors.h with #ifdef GENERATOR_FILE.
19177 (have_error): New global, copied from errors.c.
19178 (md_reader::read_name): Rename to...
19179 (md_reader::read_name_1): ...this, adding "out_loc" param,
19180 and converting "missing name or number" to returning false, rather
19182 (md_reader::read_name): Reimplement in terms of read_name_1.
19183 (md_reader::read_name_or_nil): New function.
19184 (md_reader::read_string): Handle "(nil)" by returning NULL.
19185 (md_reader::md_reader): Add new param "compact".
19186 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
19187 (md_reader::read_file): New method.
19188 * read-md.h (md_reader::md_reader): Add new param "compact".
19189 (md_reader::read_file): New method.
19190 (md_reader::is_compact): New accessor.
19191 (md_reader::read_name): Convert return type from void to file_location.
19192 (md_reader::read_name_or_nil): New decl.
19193 (md_reader::read_name_1): New decl.
19194 (md_reader::m_compact): New field.
19195 (noop_reader::noop_reader): Pass "false" for new "compact" param
19197 (rtx_reader::rtx_reader): Add new "compact" param.
19198 (rtx_reader::read_rtx_operand): Make virtual and convert return
19199 type from void to rtx.
19200 (rtx_reader::read_until): New decl.
19201 (rtx_reader::handle_any_trailing_information): New virtual function.
19202 (rtx_reader::postprocess): New virtual function.
19203 (rtx_reader::finalize_string): New virtual function.
19204 (rtx_reader::m_in_call_function_usage): New field.
19205 (rtx_reader::m_reuse_rtx_by_id): New field.
19206 * read-rtl-function.c: New file.
19207 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
19208 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
19209 (selftest::verify_three_block_rtl_cfg): New decl.
19210 * read-rtl-function.h: New file.
19211 * read-rtl.c: Potentially include config.h rather than bconfig.h.
19212 For host, include function.h, memmodel.h, and emit-rtl.h.
19213 (one_time_initialization): New function.
19214 (struct compact_insn_name): New struct.
19215 (compact_insn_names): New array.
19216 (find_code): Handle insn codes in compact dumps.
19217 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
19218 (bind_subst_iter_and_attr): Likewise.
19219 (add_condition_to_string): Likewise.
19220 (add_condition_to_rtx): Likewise.
19221 (apply_attribute_uses): Likewise.
19222 (add_current_iterators): Likewise.
19223 (apply_iterators): Likewise.
19224 (initialize_iterators): Guard usage of apply_subst_iterator with
19225 #ifdef GENERATOR_FILE.
19226 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
19227 (md_reader::read_mapping): Likewise.
19228 (add_define_attr_for_define_subst): Likewise.
19229 (add_define_subst_attr): Likewise.
19230 (read_subst_mapping): Likewise.
19231 (check_code_iterator): Likewise.
19232 (rtx_reader::read_rtx): Likewise. Move one-time initialization
19234 (one_time_initialization): New function.
19235 (rtx_reader::read_until): New method.
19236 (read_flags): New function.
19237 (parse_reg_note_name): New function.
19238 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
19239 Handle reuse_rtx ids.
19240 Wrap iterator lookup within #ifdef GENERATOR_FILE.
19241 Add parsing support for RTL dumps, mirroring the special-cases in
19242 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
19243 values, and calling handle_any_trailing_information.
19244 (rtx_reader::read_rtx_operand): Convert return type from void
19245 to rtx, returning return_rtx. Handle case 'e'. Call
19246 finalize_string on XSTR and XTMPL fields.
19247 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
19248 "(nil)" values were omitted. Call the postprocess vfunc on the
19250 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
19251 class ctor. Initialize m_in_call_function_usage. Call
19252 one_time_initialization.
19253 * rtl-tests.c (selftest::test_uncond_jump): Call
19254 set_new_first_and_last_insn.
19255 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
19256 * selftest-rtl.c: New file.
19257 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
19258 (selftest::get_insn_by_uid): New decl.
19259 * selftest-run-tests.c (selftest::run_tests): Call
19260 read_rtl_function_c_tests.
19261 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
19262 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
19265 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
19267 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
19268 operands in a special way. Assert that pos+len <= mode precision.
19270 2017-01-05 Jakub Jelinek <jakub@redhat.com>
19272 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
19273 3 argument Alias with unlimited for the negative form.
19274 (fno-vect-cost-model): Removed.
19276 2017-01-05 Martin Liska <mliska@suse.cz>
19278 * hsa-gen.c (gen_hsa_divmod): New function.
19279 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
19281 2017-01-05 Martin Liska <mliska@suse.cz>
19284 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
19287 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19289 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
19290 small constant length operands.
19292 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19294 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
19295 between loop iterations.
19297 2017-01-05 Martin Liska <mliska@suse.cz>
19300 * gimplify.c (gimplify_decl_expr): Compare to
19301 asan_poisoned_variables instread of checking flags.
19302 (gimplify_target_expr): Likewise.
19303 (gimplify_expr): Likewise.
19304 (gimplify_function_tree): Conditionally initialize
19305 asan_poisoned_variables.
19307 2017-01-04 Jeff Law <law@redhat.com>
19309 PR tree-optimizatin/78812
19310 * rtl.h (contains_mem_rtx_p): Prototype.
19311 * ifcvt.c (containts_mem_rtx_p): Move from here to...
19312 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
19313 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
19314 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
19315 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
19317 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19319 * input.c (assert_char_at_range): Default-initialize actual_range.
19321 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19323 * df-scan.c (df_ref_create_structure): Make regno unsigned,
19324 to match the caller.
19326 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19328 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
19329 insns after final jump in test to emit dummy move.
19331 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19333 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
19334 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
19336 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19338 * multiple_target.c (create_dispatcher_calls): Init e_next.
19339 * tree-ssa-loop-split.c (split_loop): Init border.
19340 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
19343 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
19348 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
19349 (altivec_register_operand): Do not return true if the operand
19350 contains a SUBREG mixing SImode and SFmode.
19351 (vsx_register_operand): Likewise.
19352 (vsx_reg_sfsubreg_ok): New predicate.
19353 (vfloat_operand): Do not return true if the operand contains a
19354 SUBREG mixing SImode and SFmode.
19355 (vint_operand): Likewise.
19356 (vlogical_operand): Likewise.
19357 (gpc_reg_operand): Likewise.
19358 (int_reg_operand): Likewise.
19359 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
19360 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
19361 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
19363 (rs6000_emit_move_si_sf_subreg): New helper function.
19364 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
19365 fixup SUBREGs involving SImode and SFmode.
19366 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
19367 numbers for the new peephole2 optimization.
19368 (peephole2 for SFmode unions): New peephole2 to optimize cases in
19369 the GLIBC math library that do AND/IOR/XOR operations on single
19370 precision floating point.
19371 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
19372 target macros to say whether we need to avoid SUBREGs mixing
19374 (TARGET_ALLOW_SF_SUBREG): Likewise.
19375 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
19376 (UNSPEC_SI_FROM_SF): Likewise.
19377 (iorxor): Change spacing.
19378 (and_ior_xor): New iterator for AND, IOR, and XOR.
19379 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
19380 (movdi_from_sf_zero_ext): Likewise.
19381 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
19382 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
19383 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
19384 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
19385 (fms<mode>4): Likewise.
19386 (fnma<mode>4): Likewise.
19387 (fnms<mode>4): Likewise.
19388 (nfma<mode>4): Likewise.
19389 (nfms<mode>4): Likewise.
19391 2017-01-04 Marek Polacek <polacek@redhat.com>
19394 * doc/invoke.texi: Document -Wpointer-compare.
19396 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19398 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
19401 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
19402 descriptions for -gdwarf-5 and emit them as uleb128 instead of
19405 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
19408 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
19409 documentation of the powerpc_popcntb_ok attribute.
19410 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19411 code to issue warning messages if a requested CPU configuration is
19412 not supported by the binary (assembler and loader) toolchain.
19413 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
19414 made to define a built-in function that has been disabled.
19415 (paired_init_builtins): Add assertion to prevent ICE if attempt is
19416 made to define a built-in function that has been disabled.
19417 (altivec_init_builtins): Add comment explaining why definition
19418 of the DST built-in functions is not preceded by an assertion
19419 check. Add assertions to prevent ICE if attempts are made to
19420 define an altivec predicate or an abs* built-in function that has
19422 (htm_init_builtins): Add comment explaining why definition of the
19423 htm built-in functions is not preceded by an assertion check.
19425 2017-01-04 Jeff Law <law@redhat.com>
19427 PR tree-optimizatin/67955
19428 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
19429 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
19430 the points-to solution does not include pt_null. Use DECL_PT_UID
19433 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
19435 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
19436 Use gen_int_mode instead of gen_lopwart for const_int operands.
19438 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19440 PR tree-optimization/71563
19441 * match.pd: Simplify X << Y into X if Y is known to be 0 or
19442 out of range value - has low bits known to be zero.
19444 2017-01-04 Alan Modra <amodra@gmail.com>
19446 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
19447 * configure: Regenerate.
19448 * config.in: Regenerate.
19450 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19453 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
19454 a substring of the message, but strcmp with the whole message. Ifdef
19455 ENABLE_NLS, translate the message first using dgettext.
19457 2017-01-03 Jeff Law <law@redhat.com>
19459 PR tree-optimizatin/78856
19460 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
19461 (mark_threaded_blocks): Remove code to truncate thread paths that
19462 cross multiple loop headers. Instead invalidate the cached loop
19463 iteration information and handle case of a thread path walking
19464 into an irreducible region.
19466 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
19469 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
19470 assertions. Add support for doing the signbit if the IEEE 128-bit
19471 floating point value is in a GPR.
19472 * config/rs6000/rs6000.md (Fsignbit): Delete.
19473 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
19474 Update the length attribute if the value is in a GPR.
19475 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
19476 the sign or zero extension instruction, since the value is always 0/1.
19477 (signbit<mode>2_dm2): Delete using <Fsignbit>.
19480 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
19481 extracting SImode to a GPR register so that we can generate a
19482 store, limit the vector to be in a traditional Altivec register
19483 for the vextuwrx instruction.
19485 2017-01-03 Ian Lance Taylor <iant@google.com>
19487 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
19489 2017-01-03 Martin Sebor <msebor@redhat.com>
19491 PR tree-optimization/78696
19492 * gimple-ssa-sprintf.c (format_floating): Correct handling of
19493 precision. Use MPFR for %f for greater fidelity. Correct handling
19495 (pass_sprintf_length::compute_format_length): Set width and precision
19496 specified by asrerisk to void_node for vararg functions.
19497 (try_substitute_return_value): Adjust dump output.
19499 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
19501 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
19503 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
19505 * doc/invoke.texi (SPARC options): Document -mlra as the default.
19506 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
19507 -mlra/-mno-lra was passed to the compiler.
19509 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
19511 PR rtl-optimization/65618
19512 * emit-rtl.c (try_split): Move initialization of "before" and
19513 "after" to just before the call to emit_insn_after_setloc.
19515 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
19517 * doc/md.texi (Standard Names): Remove reference to Java frontend.
19519 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
19521 * dwarf2out.c (gen_enumeration_type_die): When
19522 -gno-strict-dwarf, add a DW_AT_encoding attribute.
19524 2017-01-03 Jakub Jelinek <jakub@redhat.com>
19526 PR tree-optimization/78965
19527 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
19528 Change first argument from const call_info & to call_info &. For %n
19529 set info.nowrite to false.
19531 PR middle-end/78901
19532 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
19533 possibly throwing calls.
19535 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
19536 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
19537 and fns handling, rather than in a separate case SSA_NAME.
19539 2017-01-02 Jeff Law <law@redhat.com>
19541 * config/darwin-driver.c (darwin_driver_init): Const-correctness
19542 fixes for first_period and second_period variables.
19544 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
19547 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
19548 (*insvqi_1): New insn pattern.
19549 (*insvqi_1_mem_rex64): Ditto.
19550 (*insvqi_2): Ditto.
19551 (*insvqi_3): Rename from *insvqi.
19553 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
19555 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
19557 * doc/cfg.texi (Edges): Remove reference to Java.
19558 (Maintaining the CFG): Ditto.
19560 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19562 PR middle-end/77674
19563 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
19564 transparent aliases.
19566 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19568 PR middle-end/77484
19569 * predict.def (PRED_CALL): Update hitrate.
19570 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
19571 * predict.c (tree_estimate_probability_bb): Split CALL predictor
19572 into direct/indirect/polymorphic variants.
19574 2017-01-01 Jakub Jelinek <jakub@redhat.com>
19576 Update copyright years.
19578 * gcc.c (process_command): Update copyright notice dates.
19579 * gcov-dump.c (print_version): Ditto.
19580 * gcov.c (print_version): Ditto.
19581 * gcov-tool.c (print_version): Ditto.
19582 * gengtype.c (create_file): Ditto.
19583 * doc/cpp.texi: Bump @copying's copyright year.
19584 * doc/cppinternals.texi: Ditto.
19585 * doc/gcc.texi: Ditto.
19586 * doc/gccint.texi: Ditto.
19587 * doc/gcov.texi: Ditto.
19588 * doc/install.texi: Ditto.
19589 * doc/invoke.texi: Ditto.
19591 Copyright (C) 2017 Free Software Foundation, Inc.
19593 Copying and distribution of this file, with or without modification,
19594 are permitted in any medium without royalty provided the copyright
19595 notice and this notice are preserved.