1 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3 * tree.c (make_int_cst_stat): Remove _stat from name.
4 * tree.h (make_int_cst_stat): Adjust prototype.
5 (make_int_cst): Remove macro.
7 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9 * tree.c (make_tre_binfo_stat): Remove _stat from name.
10 * tree.h (make_tree_binfo_stat): Adjust prototype.
11 (make_tree_binfo): Remove.
13 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15 * tree.c (copy_node_stat): Rename to copy_node.
16 (build_distinct_type_copy): Adjust.
17 * tree.h (copy_node_stat): Adjust prototype.
18 (copy_node): Remove macro.
20 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22 * tree.c (make_node_stat): rename to make_node.
23 (build_tree_list_stat): Adjust.
24 (build0_stat): Likewise.
25 (build2_stat): Likewise.
26 (build3_stat): Likewise.
27 (build4_stat): Likewise.
28 (build5_stat): Likewise.
29 (build_decl_stat): Likewise.
30 * tree.h (make_node_stat): Adjust prototype.
31 (make_node): remove macro.
33 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
35 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
36 (PPC_FEATURE2_SCV): Likewise.
37 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
39 2017-07-28 Tamar Christina <tamar.christina@arm.com>
41 * config/aarch64/aarch64.c
42 (aarch64_internal_mov_immediate): Add new special pattern.
43 * config/aarch64/aarch64.md (*movdi_aarch64):
44 Add reg/32bit const mov case.
46 2017-07-28 Tamar Christina <tamar.christina@arm.com>
47 Richard Sandiford <richard.sandiford@linaro.org>
49 * config/aarch64/aarch64.md (mov<mode>): Generalize.
50 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
51 Add integer and movi cases.
52 (movi-split-hf-df-sf split, fp16): New.
53 (enabled): Added TARGET_FP_F16INST.
54 * config/aarch64/iterators.md (GPF_HF): New.
55 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
57 2017-07-28 Tamar Christina <tamar.christina@arm.com>
59 * config/aarch64/aarch64.c
60 (aarch64_simd_container_mode): Add prototype.
61 (aarch64_expand_mov_immediate): Add HI support.
62 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
63 (aarch64_can_const_movi_rtx_p): New.
64 (aarch64_preferred_reload_class):
65 Remove restrictions of using FP registers for certain SIMD operations.
66 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
67 (aarch64_valid_floating_const): Add integer move validation.
68 (aarch64_simd_imm_scalar_p): Remove.
69 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
70 (aarch64_legitimate_constant_p): Expand list of supported cases.
71 * config/aarch64/aarch64-protos.h
72 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
73 (aarch64_reinterpret_float_as_int): New.
74 (aarch64_simd_imm_scalar_p): Remove.
75 * config/aarch64/constraints.md (Uvi): New.
76 (Dd): Split into Ds and new Dd.
77 * config/aarch64/aarch64.md (*movsi_aarch64):
79 (*movdi_aarch64): Add SIMD mov case.
81 2017-07-28 Bin Cheng <bin.cheng@arm.com>
83 * tree-predcom.c: (struct chain): Handle store-store chain in which
84 stores for elimination only store loop invariant values.
85 (execute_pred_commoning_chain): Ditto.
86 (prepare_initializers_chain_store_elim): Ditto.
87 (prepare_finalizers): Ditto.
88 (is_inv_store_elimination_chain): New function.
89 (initialize_root_vars_store_elim_1): New function.
91 2017-07-28 Bin Cheng <bin.cheng@arm.com>
93 * tree-predcom.c: Revise general description of the pass.
94 (enum chain_type): New enum type for store elimination.
95 (struct chain): New field supporting store elimination.
96 (struct component): Ditto.
97 (dump_chain): Dump store-stores chain.
98 (release_chain): Release resources.
99 (split_data_refs_to_components): Compute and create component
100 contains only stores for elimination.
101 (get_chain_last_ref_at): New function.
102 (make_invariant_chain): Initialization.
103 (make_rooted_chain): Specify chain type in parameter and record it.
104 (add_looparound_copies): Skip for store-stores chain.
105 (determine_roots_comp): Compute type of chain and pass it to
107 (initialize_root_vars_store_elim_2): New function.
108 (finalize_eliminated_stores): New function.
109 (remove_stmt): Handle store for elimination.
110 (execute_pred_commoning_chain): Execute predictive commoning on
112 (determine_unroll_factor): Skip unroll for store-stores chain.
113 (prepare_initializers_chain_store_elim): New function.
114 (prepare_initializers_chain): Hanlde store-store chain.
115 (prepare_finalizers_chain, prepare_finalizers): New function.
116 (tree_predictive_commoning_loop): Return integer value indicating
117 if loop is unrolled or lcssa form is corrupted.
118 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
120 2017-07-28 Bin Cheng <bin.cheng@arm.com>
122 * tree-predcom.c (initialize_root): Delete.
123 (execute_pred_commoning_chain): Initialize root vars and replace
124 reference of non-combined chain directly, rather than call above
127 2017-07-28 Bin Cheng <bin.cheng@arm.com>
129 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
130 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
132 2017-07-28 Bin Cheng <bin.cheng@arm.com>
134 * tree-predcom.c (struct chain): New field init_seq.
135 (release_chain): Release init_seq.
136 (prepare_initializers_chain): Record intialization stmts in above
138 (insert_init_seqs): New function.
139 (tree_predictive_commoning_loop): Call insert_init_seqs.
141 2017-07-28 Bin Cheng <bin.cheng@arm.com>
143 * tree-predcom.c (determine_roots_comp): Skip trivial components.
145 2017-07-28 Richard Biener <rguenther@suse.de>
147 * match.pd: Remove superfluous :c.
148 * genmatch.c (simplify::id): Add member.
149 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
151 (current_id): New global.
152 (dt_node::parent): Move from ...
153 (dt_operand::parent): ... here. Add for_id member.
154 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
155 (decision_tree::find_node): Relax order requirement when
156 merging DT_TRUE nodes to ones inbetween the current simplify
157 and the one we try to merge with. Add diagnostic whenever
158 we need to enforce pattern order by not merging.
159 (decision_tree::insert): Set current_id.
160 (decision_tree::print_node): Dump parent node and for_id.
161 (parser::last_id): Add member.
162 (parser::push_simplify): Assign unique id.
163 (parser::parser): Initialize last_id.
165 2017-07-28 Martin Liska <mliska@suse.cz>
168 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
169 gimple_build_debug_bind.
171 2017-07-28 Richard Biener <rguenther@suse.de>
173 PR tree-optimization/81502
174 * match.pd: Add pattern combining BIT_INSERT_EXPR with
176 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
178 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
179 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
180 for BIT_FIELD_REF args.
181 * fold-const.c (make_bit_field_ref): Likewise.
182 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
184 2017-07-28 Jakub Jelinek <jakub@redhat.com>
187 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
188 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
189 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
190 Or it into SANITIZER_UNDEFINED.
191 * ubsan.c: Include gimple-fold.h and varasm.h.
192 (ubsan_expand_ptr_ifn): New function.
193 (instrument_pointer_overflow): New function.
194 (maybe_instrument_pointer_overflow): New function.
195 (instrument_object_size): Formatting fix.
196 (pass_ubsan::execute): Call instrument_pointer_overflow
197 and maybe_instrument_pointer_overflow.
198 * internal-fn.c (expand_UBSAN_PTR): New function.
199 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
200 * sanitizer.def (__ubsan_handle_pointer_overflow,
201 __ubsan_handle_pointer_overflow_abort): New builtins.
202 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
203 * internal-fn.def (UBSAN_PTR): New internal function.
204 * opts.c (sanitizer_opts): Add pointer-overflow.
205 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
206 * fold-const.c (build_range_check): Compute pointer range check in
207 integral type if pointer arithmetics would be needed. Formatting
210 2017-07-28 Martin Liska <mliska@suse.cz>
213 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
214 parameters that are of a variable-length.
216 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
218 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
220 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
221 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
222 (CRT_CALL_STATIC_FUNCTION): Likewise.
223 (ASM_DEFAULT_SPEC): New define.
224 (ASM_SPEC32): Likewise.
225 (ASM_SPEC64): Likewise.
226 (ASM_SPEC_COMMON): Likewise.
227 (ASM_SPEC): Likewise.
228 (INVALID_64BIT): Likewise.
229 (LINK_OS_DEFAULT_SPEC): Likewise.
230 (LINK_OS_SPEC32): Likewise.
231 (LINK_OS_SPEC64): Likewise.
232 (POWERPC_LINUX): Likewise.
233 (PTRDIFF_TYPE): Likewise.
234 (RESTORE_FP_PREFIX): Likewise.
235 (RESTORE_FP_SUFFIX): Likewise.
236 (SAVE_FP_PREFIX): Likewise.
237 (SAVE_FP_SUFFIX): Likewise.
238 (SIZE_TYPE): Likewise.
239 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
240 (TARGET_64BIT): Likewise.
241 (TARGET_64BIT): Likewise.
242 (TARGET_AIX): Likewise.
243 (WCHAR_TYPE_SIZE): Likewise.
244 (WCHAR_TYPE): Undefine.
245 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
246 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
247 (CPP_OS_RTEMS_SPEC): Delete.
248 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
249 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
251 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
253 2017-07-28 Jakub Jelinek <jakub@redhat.com>
255 PR tree-optimization/81578
256 * tree-parloops.c (build_new_reduction): Bail out if
257 reduction_code isn't one of the standard OpenMP reductions.
258 Move the details printing after that decision.
260 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
262 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
263 related to reload_in_progress.
264 (splat_input_operand): Likewise.
265 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
267 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
269 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
270 (TARGET_INSTANTIATE_DECLS): Likewise.
271 (legitimate_indexed_address_p): Delete reload_in_progress code.
272 (rs6000_debug_legitimate_address_p): Likewise.
273 (rs6000_eliminate_indexed_memrefs): Likewise.
274 (rs6000_emit_le_vsx_store): Likewise.
275 (rs6000_emit_move_si_sf_subreg): Likewise.
276 (rs6000_emit_move): Likewise.
277 (register_to_reg_type): Likewise.
278 (rs6000_pre_atomic_barrier): Likewise.
279 (rs6000_machopic_legitimize_pic_address): Likewise.
280 (rs6000_allocate_stack_temp): Likewise.
281 (rs6000_address_for_fpconvert): Likewise.
282 (rs6000_address_for_altivec): Likewise.
283 (rs6000_secondary_memory_needed_rtx): Delete function.
284 (rs6000_check_sdmode): Likewise.
285 (rs6000_alloc_sdmode_stack_slot): Likewise.
286 (rs6000_instantiate_decls): Likewise.
287 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
288 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
289 Delete reload_in_progress.
290 (*vec_reload_and_plus_<mptrsize>): Likewise.
291 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
292 (vsx_div_v2di): Likewise.
293 (vsx_udiv_v2di): Likewise.
295 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
297 * config/rs6000/rs6000.opt (mlra): Replace with stub.
298 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
299 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
300 (rs6000_debug_reg_global): Delete print of LRA status.
301 (rs6000_option_override_internal): Delete dead LRA related code.
302 (rs6000_lra_p): Delete function.
303 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
305 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
307 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
308 * config/riscv/rtems.h: New file.
310 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
311 Sudakshina Das <sudi.das@arm.com>
313 * config/aarch64/aarch64.md
314 (define_split for and<mode>3nr_compare): Move
315 non aarch64_logical_operand to a register.
316 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
317 register immediate operand to a register.
318 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
320 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
323 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
325 2017-07-27 Richard Biener <rguenther@suse.de>
327 PR tree-optimization/81573
328 PR tree-optimization/81494
329 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
330 multi defuse cycle case.
332 2017-07-27 Richard Biener <rguenther@suse.de>
334 PR tree-optimization/81571
335 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
338 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
340 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
341 earlier and only if MASK_FPU is set. Adjust formatting.
343 2017-07-27 Martin Liska <mliska@suse.cz>
345 * opt-functions.awk: Add validation of value of Init.
346 * optc-gen.awk: Pass new argument.
348 2017-07-27 Martin Liska <mliska@suse.cz>
350 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
353 2017-07-27 Martin Liska <mliska@suse.cz>
355 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
356 BBs and edges seen by autoFDO.
358 2017-07-27 Richard Biener <rguenther@suse.de>
360 PR tree-optimization/81502
361 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
362 with incompatible but same sized type.
363 (execute_update_addresses_taken): Likewise.
365 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
367 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
368 flag_tree_loop_vectorize rather than flag_tree_vectorize.
370 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
373 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
374 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
375 Change s_operand to memory_operand.
377 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
379 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
380 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
381 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
382 Emit instructions rather than returning an expression. Handle TFmode
383 and KFmode by casting to TImode.
384 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
385 (rs6000_emit_le_vsx_store): Likewise.
386 * config/rs6000/vsx.md (VSX_TI): New iterator.
387 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
388 (*vsx_le_undo_permute_<mode>): Likewise.
389 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
390 emit the split sequence.
391 (*vsx_le_perm_store_<mode>): Likewise.
393 2017-07-27 Jakub Jelinek <jakub@redhat.com>
395 PR tree-optimization/81555
396 PR tree-optimization/81556
397 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
398 if true, force CHANGED for the recursive invocation.
399 (reassociate_bb): Remember original length of ops array, pass
400 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
402 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
403 attributes for noipa attribute. For naked attribute use
404 lookup_attribute first before lookup_attribute_spec.
405 * final.c (rest_of_handle_final): Disable IPA RA for functions with
407 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
408 for functions with noipa attribute.
409 (cgraph_externally_visible_p): Return true for functions with noipa
411 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
412 for functions with noipa attribute.
413 * doc/extend.texi: Document noipa function attribute.
414 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
415 also for functions with noipa attribute.
416 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
418 2017-07-26 Andrew Pinski <apinski@cavium.com>
420 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
421 vec_unalign_load_cost and vec_unalign_store_cost.
423 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
425 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
426 -mvsx-small-integer option.
427 (ISA_3_0_MASKS_IEEE): Likewise.
428 (OTHER_VSX_VECTOR_MASKS): Likewise.
429 (POWERPC_MASKS): Likewise.
430 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
431 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
432 code, only testing for DImode being allowed in non-VSX floating
434 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
435 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
437 (rs6000_option_override_internal): Delete -mvsx-small-integer.
438 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
439 TARGET_P8_VECTOR test.
440 (rs6000_secondary_reload_simple_move): Likewise.
441 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
442 since TARGET_P9_VECTOR was already tested.
443 (rs6000_opt_masks): Remove -mvsx-small-integer.
444 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
445 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
447 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
448 test for TARGET_VEXTRACTUB was used, and that uses
450 (p9 extract splitter): Likewise.
451 (vsx_extract_<mode>_di_p9): Likewise.
452 (vsx_extract_<mode>_store_p9): Likewise.
453 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
454 for TARGET_P9_VECTOR was used. Delete code that is now dead with
455 the elimination of TARGET_VSX_SMALL_INTEGER.
456 (vsx_extract_<mode>_p8): Likewise.
457 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
458 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
459 (vsx_set_<mode>_p9): Likewise.
460 (vsx_set_v4sf_p9): Likewise.
461 (vsx_set_v4sf_p9_zero): Likewise.
462 (vsx_insert_extract_v4sf_p9): Likewise.
463 (vsx_insert_extract_v4sf_p9_2): Likewise.
464 * config/rs6000/rs6000.md (sign extend splitter): Change
465 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
466 (floatsi<mode>2_lfiwax_mem): Likewise.
467 (floatunssi<mode>2_lfiwzx_mem): Likewise.
468 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
469 since a test for TARGET_P9_VECTOR was used.
470 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
471 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
472 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
473 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
474 TARGET_P8_VECTOR test.
475 (fix_trunc<mode>si2_stfiwx): Likewise.
476 (fix_trunc<mode>si2_internal): Likewise.
477 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
478 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
480 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
481 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
482 TARGET_P8_VECTOR test.
483 (fixuns_trunc<mode>si2_stfiwx): Likewise.
484 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
485 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
487 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
488 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
489 since a test for TARGET_P9_VECTOR was used.
490 (splitter for loading small constants): Likewise.
492 2017-07-26 Andrew Pinski <apinski@cavium.com>
494 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
497 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
500 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
501 (fp_valid_at): Likewise.
503 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
505 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
506 (qdf24xx_addrcost_table): Likewise.
507 (cortexa57_tunings): Update to use generic_branch_cost.
508 (cortexa72_tunings): Likewise.
509 (cortexa73_tunings): Likewise.
510 (qdf24xx_tunings): Likewise.
512 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
514 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
515 (thunderx2t99_branch_cost): Likewise.
516 (cortexa35_tunings): Update to use generic_branch_cost.
517 (cortexa53_tunings): Likewise.
518 (cortexa57_tunings): Likewise.
519 (cortexa72_tunings): Likewise.
520 (cortexa73_tunings): Likewise.
521 (thunderx2t99_tunings): Likewise.
523 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
525 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
526 (sparc_option_override): Honour MASK_FSMULD.
527 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
528 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
529 * config/sparc/sparc.opt (mfsmuld): New option.
530 * doc/invoke.texi (mfsmuld): Document option.
532 2017-07-26 Marek Polacek <polacek@redhat.com>
535 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
537 2017-07-26 Richard Biener <rguenther@suse.de>
539 * gimple-match-head.c (do_valueize): Return OP if valueize
541 (get_def): New helper to get at the def stmt of a SSA name
543 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
544 do_valueize to get at the def stmt.
545 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
547 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
550 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
552 2017-07-26 Martin Liska <mliska@suse.cz>
555 * function.c (expand_function_start): Make expansion of
556 nonlocal_goto_save_area after parm_birth_insn.
558 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
560 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
561 from all CPU target flags enable members.
563 2017-07-26 Richard Biener <rguenther@suse.de>
565 * genmatch.c (dt_simplify::gen): Make iterator vars const.
566 (decision_tree::gen): Make 'type' const.
567 (write_predicate): Likewise.
569 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
571 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
572 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
573 (rs6000_option_override_internal): Likewise.
574 (rs6000_expand_vector_set): Likewise.
575 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
576 (TARGET_UPPER_REGS_SF): Likewise.
577 (TARGET_UPPER_REGS_DI): Likewise.
578 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
579 (TARGET_DIRECT_MOVE_64BIT): Likewise.
580 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
581 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
582 (Splitters for DI constants in Altivec registers): Likewise.
583 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
584 (vsx_set_v4sf_p9): Likewise.
585 (vsx_set_v4sf_p9_zero): Likewise.
586 (vsx_insert_extract_v4sf_p9): Likewise.
587 (vsx_insert_extract_v4sf_p9_2): Likewise.
589 2017-07-25 Carl Love <cel@us.ibm.com>
591 * doc/extend.texi: Update the built-in documentation file for the
592 existing built-in functions
593 vector signed char vec_cnttz (vector signed char);
594 vector unsigned char vec_cnttz (vector unsigned char);
595 vector signed short vec_cnttz (vector signed short);
596 vector unsigned short vec_cnttz (vector unsigned short);
597 vector signed int vec_cnttz (vector signed int);
598 vector unsigned int vec_cnttz (vector unsigned int);
599 vector signed long long vec_cnttz (vector signed long long);
600 vector unsigned long long vec_cnttz (vector unsigned long long);
602 2017-07-25 Andrew Pinski <apinski@cavium.com>
604 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
605 accesses where the use is for the first operand of a BIT_INSERT.
607 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
610 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
611 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
613 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
615 * config/i386/gstabs.h: Delete.
616 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
618 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
620 * config/i386/i386.c (ix86_decompose_address): Do not check for
621 register RTX when looking at index_reg or base_reg.
622 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
624 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
626 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
627 to update EH info here.
629 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
631 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
633 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
635 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
637 2017-07-25 Torsten Duwe <duwe@suse.de>
639 * common.opt: Introduce -fpatchable-function-entry
640 command line option, and its variables function_entry_patch_area_size
641 and function_entry_patch_area_start.
642 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
643 including a two-value parser.
644 * target.def (print_patchable_function_entry): New target hook.
645 * targhooks.h (default_print_patchable_function_entry): New function.
646 * targhooks.c (default_print_patchable_function_entry): Likewise.
647 * toplev.c (process_options): Switch off IPA-RA if
648 patchable function entries are being generated.
649 * varasm.c (assemble_start_function): Look at the
650 patchable-function-entry command line switch and current
651 function attributes and maybe generate NOP instructions by
652 calling the print_patchable_function_entry hook.
653 * doc/extend.texi: Document patchable_function_entry attribute.
654 * doc/invoke.texi: Document -fpatchable_function_entry
656 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
658 * doc/tm.texi: Re-generate.
660 2017-07-25 Jakub Jelinek <jakub@redhat.com>
663 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
664 TARGET_AVX512DQ rather than TARGET_AVX512BW.
666 2017-07-25 Tamar Christina <tamar.christina@arm.com>
668 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
670 2017-07-25 Richard Biener <rguenther@suse.de>
672 PR tree-optimization/81455
673 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
674 not walk in cycles when looking for guards.
676 2017-07-25 Richard Biener <rguenther@suse.de>
678 PR tree-optimization/81529
679 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
680 when optimizing backedge uses.
682 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
684 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
686 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
687 to dl_section_ref. On AIX, append an expression to subtract
688 the size of the section length to dl_section_ref.
690 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
692 * configure.ac: If any of the config.* scripts fail, exit 1.
693 * configure: Regenerate.
695 2017-07-25 Richard Biener <rguenther@suse.de>
698 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
699 of immediate uses, be more verbose on errors.
701 2017-07-25 Richard Biener <rguenther@suse.de>
703 PR tree-optimization/81510
704 * tree-vect-loop.c (vect_is_simple_reduction): When the
705 reduction stmt is not inside the loop bail out.
707 2017-07-25 Richard Biener <rguenther@suse.de>
709 PR tree-optimization/81303
710 * tree-vect-loop-manip.c (vect_loop_versioning): Build
711 profitability check against LOOP_VINFO_NITERSM1.
713 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
715 * domwalk.c (cmp_bb_postorder): Simplify.
716 (sort_bbs_postorder): New function. Use it...
717 (dom_walker::walk): ...here to optimize common cases.
719 2017-07-25 Martin Liska <mliska@suse.cz>
722 * ipa-visibility.c (function_and_variable_visibility): Make the
723 redirection just on target that supports aliasing.
724 Fix GNU coding style.
726 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
729 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
731 * config/arm/rtems.h: Likewise.
732 * config/bfin/rtems.h: Likewise.
733 * config/i386/rtemself.h: Likewise.
734 * config/lm32/rtems.h: Likewise.
735 * config/m32c/rtems.h: Likewise.
736 * config/m68k/rtemself.h: Likewise.
737 * config/microblaze/rtems.h: Likewise.
738 * config/mips/rtems.h: Likewise.
739 * config/moxie/rtems.h: Likewise.
740 * config/nios2/rtems.h: Likewise.
741 * config/powerpcspe/rtems.h: Likewise.
742 * config/rs6000/rtems.h: Likewise.
743 * config/rtems.h: Likewise.
744 * config/sh/rtems.h: Likewise.
745 * config/sh/rtemself.h: Likewise.
746 * config/sparc/rtemself.h: Likewise.
748 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
751 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
752 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
753 * tree-ssa-structalias.c (alias_get_name): Same.
755 2017-07-25 Bin Cheng <bin.cheng@arm.com>
758 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
759 instructions if no du chain is found.
761 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
763 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
765 2017-07-25 Richard Biener <rguenther@suse.de>
768 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
771 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
773 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
775 (ISA_2_7_MASKS_SERVER): Likewise.
776 (ISA_3_0_MASKS_IEEE): Likewise.
777 (OTHER_P8_VECTOR_MASKS): Likewise.
778 (OTHER_VSX_VECTOR_MASKS): Likewise.
779 (POWERPC_MASKS): Likewise.
780 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
781 duplicate list of options.
782 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
783 explicit -mupper-regs options.
784 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
785 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
786 alias for -mupper-regs-df.
787 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
788 (rs6000_init_hard_regno_mode_ok): Likewise.
789 (rs6000_option_override_internal): Likewise.
790 (rs6000_opt_masks): Likewise.
791 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
792 options in terms of whether -mvsx or -mpower8-vector was used.
793 (TARGET_UPPER_REGS_DI): Likewise.
794 (TARGET_UPPER_REGS_SF): Likewise.
795 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
796 -mupper-regs-* options.
798 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
800 * passes.c (emergency_dump_function): Print some empty lines and a
801 header before the RTL dump.
803 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
805 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
807 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
810 * config/aarch64/aarch64.c (aarch64_classify_symbol):
811 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
813 2017-07-24 Carl Love <cel@us.ibm.com>
815 * config/rs6000/rs6000-c.c: Add support for built-in functions
816 vector float vec_extract_fp32_from_shorth (vector unsigned short);
817 vector float vec_extract_fp32_from_shortl (vector unsigned short);
818 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
819 vec_extract_fp_from_shortl): Add defines for the two builtins.
820 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
821 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
823 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
824 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
825 * doc/extend.texi: Update the built-in documentation file for the
826 new built-in function.
828 2017-07-24 Jakub Jelinek <jakub@redhat.com>
831 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
833 * doc/generic.texi: Likewise.
834 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
835 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
837 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
839 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
840 (aarch64_mls_elt_merge<mode>): Likewise.
842 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
844 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
847 2017-07-23 Michael Collison <michael.collison@arm.com>
849 * config/arm/arm.c (arm_option_override): Deprecate
850 use of -mstructure-size-boundary.
851 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
852 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
854 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
857 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
858 Reduce cost estimate for direct moves.
860 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
863 * config/i386/i386.c (ix86_option_override_internal): Disable
864 BMI, BMI2 and TBM instructions for -m16.
866 2017-07-21 Carl Love <cel@us.ibm.com>
868 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
869 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
870 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
871 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
872 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
873 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
874 VMULOSW): New enum "unspec" values.
875 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
876 altivec_vmulosw): New patterns.
877 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
878 VMULOSW): Add definitions.
880 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
882 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
884 * config/aarch64/aarch64-options-extensions.def (rdma); New.
885 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
886 (AARCH64_FL_V8_1): Renumber.
887 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
888 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
889 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
890 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
891 rdma to feature modifiers list.
893 2017-07-21 Yury Gribov <tetra2005@gmail.com>
896 * ipa-visibility (function_and_variable_visibility): Convert
897 recursive PLT call to direct call if appropriate.
899 2017-07-21 Andrew Pinski <apinski@cavium.com>
901 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
902 operand 1 to see if the types precision matches.
903 * fold-const.c (operand_equal_p): Likewise.
905 2017-07-21 Richard Biener <rguenther@suse.de>
907 PR tree-optimization/81303
908 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
909 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
910 (vect_peeling_hash_get_lowest_cost): Adjust.
911 (vect_enhance_data_refs_alignment): Likewise. Use
912 vect_get_peeling_costs_all_drs to compute the penalty for no
913 peeling to match up costs.
915 2017-07-21 Richard Biener <rguenther@suse.de>
917 PR tree-optimization/81500
918 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
919 we didn't identify a reduction path.
921 2017-07-21 Tom de Vries <tom@codesourcery.com>
922 Cesar Philippidis <cesar@codesourcery.com>
924 PR gcov-profile/81442
925 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
928 2017-07-21 Tom de Vries <tom@codesourcery.com>
931 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
933 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
934 nvptx_override_options_after_change.
936 2017-07-21 Ulrich Drepper <drepper@redhat.com>
938 * dwarf2out.c (output_file_names): Avoid double testing for
941 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
943 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
945 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
947 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
949 (try_crossjump_to_edge): Do not punt on partitioned functions.
951 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
953 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
954 Put all BBs reachable only via paths crossing cold region to cold
956 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
958 2016-07-21 Richard Biener <rguenther@suse.de>
960 PR tree-optimization/81303
961 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
962 into account prologue and epilogue iterations when raising
963 min_profitable_iters to sth at least covering one vector iteration.
965 2017-07-21 Tamar Christina <tamar.christina@arm.com>
967 * config/arm/arm.c (arm_test_cpu_arch_dat):
970 2017-07-20 Nathan Sidwell <nathan@acm.org>
973 * tree.h (TYPE_METHODS): Delete.
974 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
975 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
976 (dbxout_type_methods): Scan TYPE_FIELDS.
977 (dbxout_type): Don't check TYPE_METHODS here.
978 * function.c (use_register_for_decl): Always ignore register for
979 class types when not optimizing.
980 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
981 * tree.c (free_lang_data_in_type): Stitch out member functions and
982 templates from TYPE_FIELDS.
983 (build_distinct_type_copy, verify_type_variant,
984 verify_type): Member fns are on TYPE_FIELDS.
985 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
986 * tree-pretty-print.c (dump_generic_node): Likewise.
988 2017-07-20 Jakub Jelinek <jakub@redhat.com>
991 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
992 V2TImode and V4TImode.
993 (ix86_expand_vector_extract): Likewise.
994 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
995 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
996 (ssescalarmode): Handle V4TImode and V2TImode.
997 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
998 (*vec_extractv2ti, *vec_extractv4ti): New insns.
999 (VEXTRACTI128_MODE): New mode iterator.
1000 (splitter for *vec_extractv?ti first element): New.
1001 (VEC_INIT_MODE): New mode iterator.
1002 (vec_init<mode>): Consolidate 3 expanders into one using
1003 VEC_INIT_MODE mode iterator.
1005 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
1007 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
1008 non_spilled_static_chain_regno_p.
1010 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
1012 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
1014 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
1016 * bb-reorder.c (connect_traces): Allow copying of blocks within
1019 2017-07-20 Richard Biener <rguenther@suse.de>
1021 * gimple.h (gimple_phi_result): Add gphi * overload.
1022 (gimple_phi_result_ptr): Likewise.
1023 (gimple_phi_arg): Likewise. Adjust index assert to only
1024 allow actual argument accesses rather than all slots available
1026 (gimple_phi_arg_def): Add gphi * overload.
1027 * tree-phinodes.c (make_phi_node): Initialize only actual
1029 (resize_phi_node): Clear memory not covered by old node,
1030 do not initialize excess argument slots.
1031 (reserve_phi_args_for_new_edge): Initialize new argument slot
1034 2017-07-20 Bin Cheng <bin.cheng@arm.com>
1036 PR tree-optimization/81388
1038 2016-07-21 Bin Cheng <bin.cheng@arm.com>
1040 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
1041 by removing computation of may_be_zero.
1043 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1044 Tom de Vries <tom@codesourcery.com>
1047 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
1048 when gimple level profile disagrees with what RTL expander did.
1050 2017-07-20 Richard Biener <rguenther@suse.de>
1052 PR tree-optimization/61171
1053 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
1054 (vect_analyze_stmt): Add slp instance parameter.
1055 (vectorizable_reduction): Likewise.
1056 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
1057 (vect_is_simple_reduction): Deal with chains not detected
1058 as SLP reduction chain, specifically not properly associated
1059 chains containing a mix of plus/minus.
1060 (get_reduction_op): Remove.
1061 (get_initial_defs_for_reduction): Simplify, pass in whether
1062 this is a reduction chain, pass in the SLP node for the PHIs.
1063 (vect_create_epilog_for_reduction): Get the SLP instance as
1065 (vectorizable_reduction): Get the SLP instance as arg.
1066 During analysis remember the SLP node with the PHIs in the
1067 instance. Simplify getting at the vectorized reduction PHIs.
1068 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
1069 through SLP instance.
1070 (vect_slp_analyze_operations): Likewise.
1071 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
1072 (vect_transform_stmt): Likewise.
1074 2017-07-20 Tom de Vries <tom@codesourcery.com>
1076 PR tree-optimization/81489
1077 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
1078 read of phi arg location to before loop that modifies phi.
1080 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
1082 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
1085 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1088 * except.c (execute): Fix ordering issue.
1090 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
1092 PR rtl-optimization/81423
1093 * combine.c (make_compound_operation_int): Don't try to optimize
1094 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
1096 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
1098 PR rtl-optimization/81423
1099 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
1100 with a constant that is -1 in the truncated to mode.
1102 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1104 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
1105 (determine_unlikely_bbs): ... here.
1106 * predict.h (propagate_unlikely_bbs_forward): Declare.
1107 * cfgexpand.c (pass_expand::execute): Use it.
1108 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
1110 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
1111 propagate_unlikely_bbs_forward.
1113 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1116 * except.c (maybe_add_nop_after_section_switch): New function.
1119 2017-07-19 Tom de Vries <tom@codesourcery.com>
1121 * gimple.h (gimple_phi_set_arg): Make assert more strict.
1123 2017-07-19 Tom de Vries <tom@codesourcery.com>
1125 * gimple.h (gimple_phi_arg): Make assert more strict.
1127 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
1129 * config.gcc (powerpc*-*-*): Add mmintrin.h.
1130 * config/rs6000/mmintrin.h: New file.
1131 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
1133 2017-07-19 Jakub Jelinek <jakub@redhat.com>
1135 PR tree-optimization/81346
1136 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
1138 2017-07-19 Tom de Vries <tom@codesourcery.com>
1140 * config/nvptx/nvptx.md (VECIM): Add V2DI.
1142 2017-07-19 Tom de Vries <tom@codesourcery.com>
1144 * config/nvptx/nvptx-modes.def: Add V2DImode.
1145 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
1146 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
1147 (nvptx_output_mov_insn): Handle lack of mov.b128.
1148 (nvptx_print_operand): Handle 'H' and 'L' codes.
1149 (nvptx_vector_mode_supported): Allow V2DImode.
1150 (nvptx_preferred_simd_mode): New function.
1151 (nvptx_data_alignment): New function.
1152 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
1153 nvptx_preferred_simd_mode.
1154 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
1156 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
1158 2017-07-19 Tom de Vries <tom@codesourcery.com>
1160 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
1161 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
1162 (nvptx_vector_mode_supported): New function. Allow V2SImode.
1163 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
1164 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
1165 (mov<VECIM>_insn): New define_insn.
1166 (define_expand "mov<VECIM>): New define_expand.
1168 2017-07-19 Tom de Vries <tom@codesourcery.com>
1170 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
1172 2017-07-19 Jakub Jelinek <jakub@redhat.com>
1174 PR tree-optimization/81346
1175 * fold-const.h (fold_div_compare, range_check_type): Declare.
1176 * fold-const.c (range_check_type): New function.
1177 (build_range_check): Use range_check_type.
1178 (fold_div_compare): No longer static, rewritten into
1179 a match.pd helper function.
1180 (fold_comparison): Don't call fold_div_compare here.
1181 * match.pd (X / C1 op C2): New optimization using fold_div_compare
1184 2017-07-19 Nathan Sidwell <nathan@acm.org>
1186 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
1187 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
1188 * tree.c (find_decls_types_r, verify_type): Use
1189 TYPE_{MIN,MAX}_VALUE_RAW.
1190 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
1191 (hash_tree): Likewise.
1192 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
1194 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
1197 2017-07-18 Tom de Vries <tom@codesourcery.com>
1200 * omp-expand.c (expand_omp_for_static_chunk): Handle
1201 equal-argument loop exit phi.
1203 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
1206 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
1207 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
1208 operand 2 predicate.
1209 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
1210 operand 2 predicate.
1211 (ror,rol -> rorx splitters): Use const_int_operand as
1212 operand 2 predicate.
1214 2017-06-18 Richard Biener <rguenther@suse.de>
1216 PR tree-optimization/81410
1217 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
1218 the gap in the ! slp_perm SLP case after each group.
1220 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1223 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
1226 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1229 * predict.c (set_even_probabilities): Cleanup; do not affect
1230 probabilities that are already known.
1231 (combine_predictions_for_bb): Call even when count is set.
1233 2017-07-18 Nathan Sidwell <nathan@acm.org>
1235 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
1238 2017-07-18 Bin Cheng <bin.cheng@arm.com>
1241 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
1242 optimization for loop niter analysis.
1244 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
1247 * config/avr/avr.c (avr_optimize_casesi): Don't use
1248 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
1250 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1252 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
1253 body_cost_vec from _vect_peel_extended_info.
1254 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
1255 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
1258 2017-07-18 Bin Cheng <bin.cheng@arm.com>
1260 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
1262 2017-07-18 Richard Biener <rguenther@suse.de>
1264 PR tree-optimization/80620
1265 PR tree-optimization/81403
1266 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
1267 info when re-using a VN table entry.
1269 2017-07-18 Richard Biener <rguenther@suse.de>
1271 PR tree-optimization/81418
1272 * tree-vect-loop.c (vectorizable_reduction): Properly compute
1273 vectype_in. Verify that with lane-reducing reduction operations
1274 we have a single def-use cycle.
1276 2017-07-17 Carl Love <cel@us.ibm.com>
1278 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
1280 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1281 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
1282 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
1283 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
1284 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
1285 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
1286 VMULOSW): New enum "unspec" values.
1287 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
1288 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
1289 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
1290 altivec_vmulosw): New patterns.
1291 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
1292 VMULOSW): Add definitions.
1293 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
1295 * config/alpha/alpha.c: Include predict.h.
1297 2017-07-17 Yury Gribov <tetra2005@gmail.com>
1299 * tree-vrp.c (compare_assert_loc): Fix comparison function
1300 to return predictable results.
1302 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1304 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
1307 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
1308 * doc/invoke.texi (mexpand-adddi): Update text.
1310 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1312 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
1313 that also clobbers the CC register. The old expand code is moved
1315 (*arc_clzsi2): ... here.
1316 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
1317 the CC register. The old expand code is moved to ...
1318 (arc_ctzsi2): ... here.
1320 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1322 * config/arc/arc.opt (mindexed-loads): Use initial value
1323 TARGET_INDEXED_LOADS_DEFAULT.
1324 (mauto-modify-reg): Use initial value
1325 TARGET_AUTO_MODIFY_REG_DEFAULT.
1326 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
1327 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
1328 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
1329 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
1331 2017-07-17 Martin Liska <mliska@suse.cz>
1334 * opts.c (finish_options): Do not allow -fgnu-tm
1335 w/ -fsanitize={kernel-,}address. Say sorry.
1337 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1340 * tree-loop-distribution.c (classify_partition): Only assert on
1341 numer of iterations.
1342 (merge_dep_scc_partitions): Delete prameter. Update function call.
1343 (distribute_loop): Remove code handling loop with unknown niters.
1344 (pass_loop_distribution::execute): Skip loop with unknown niters.
1346 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1349 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
1350 function sort_partitions_by_post_order.
1352 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1354 PR tree-optimization/81374
1355 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
1356 the max index of basic blocks, rather than number of basic blocks.
1358 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1360 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
1362 (arc_legitimate_pic_operand_p): Likewise.
1363 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
1365 (arc_needs_pcl_p): Likewise.
1366 (arc_legitimate_pc_offset_p): Likewise.
1367 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
1368 function is also used in constrains.md.
1369 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
1370 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
1371 PLUS. Only return true/false in known cases, otherwise assert.
1372 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
1373 is already called in arc_legitimate_constant_p.
1374 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
1376 (LEGITIMATE_PIC_OPERAND_P): Use
1377 arc_raw_symbolic_reference_mentioned_p function.
1378 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
1383 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1384 Andrew Burgess <andrew.burgess@embecosm.com>
1386 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
1387 (arc_return_address_register): New function.
1388 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
1389 (arc_handle_fndecl_attribute): Add naked attribute.
1390 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
1391 (TARGET_WARN_FUNC_RETURN): Likewise.
1392 (arc_allocate_stack_slots_for_args): New function.
1393 (arc_warn_func_return): Likewise.
1394 (machine_function): Change type fn_type.
1395 (arc_compute_function_type): Consider new naked function type,
1396 change function return type.
1397 (arc_must_save_register): Adapt to handle new
1398 arc_compute_function_type's return type.
1399 (arc_expand_prologue): Likewise.
1400 (arc_expand_epilogue): Likewise.
1401 (arc_return_address_regs): Delete.
1402 (arc_return_address_register): New function.
1403 (arc_epilogue_uses): Use above function.
1404 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
1405 (arc_function_type): Change encoding, add naked type.
1406 (ARC_INTERRUPT_P): Change to handle the new encoding.
1407 (ARC_FAST_INTERRUPT_P): Likewise.
1408 (ARC_NORMAL_P): Define.
1409 (ARC_NAKED_P): Likewise.
1410 (arc_compute_function_type): Delete prototype.
1411 * config/arc/arc.md (in_ret_delay_slot): Use
1412 arc_return_address_register function.
1413 (simple_return): Likewise.
1414 (p_return_i): Likewise.
1416 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1418 PR tree-optimization/81428
1419 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
1420 can't be built for those types.
1422 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
1424 Remove stuff dead since r239246.
1426 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
1427 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
1428 (avr_inform_devices): Remove dead stuff.
1430 2017-07-17 Tamar Christina <tamar.christina@arm.com>
1432 * config/arm/arm_neon.h: Fix softp typo.
1434 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1436 PR tree-optimization/81365
1437 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
1438 aggregate moves onto bb predecessor edges, make sure there are no
1439 loads that could alias the lhs in between the start of bb and the
1442 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
1445 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
1446 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
1447 [LSHIFTRT, outer_code = TRUNCATE]: Same.
1449 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1451 PR tree-optimization/81396
1452 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
1453 (init_symbolic_number): Initialize it to 1.
1454 (perform_symbolic_merge): Add n_ops from both operands into the new
1456 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
1457 without base_addr as useless if they need more than one operation.
1458 (bswap_replace): Handle !bswap case for NULL base_addr.
1460 2017-07-17 Tom de Vries <tom@codesourcery.com>
1463 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
1466 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
1468 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
1469 conditional builtin define __FIX_LEON3FT_B2BST.
1471 2017-07-17 Daniel Cederman <cederman@gaisler.com>
1473 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
1474 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
1477 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1479 PR rtl-optimization/81424
1480 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
1481 to remove potential trapping from operands if -fnon-call-exceptions.
1483 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1485 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
1486 profile_proability for scalling.
1487 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
1489 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1491 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
1493 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1495 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
1496 fixpoint arithmetics.
1498 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1500 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
1501 fixpoint arithmetics.
1503 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1505 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
1506 fixpoint arithmetics.
1508 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1510 * profile-count.h (profile_probability::from_reg_br_prob_note,
1511 profile_probability::to_reg_br_prob_note): New functions.
1512 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
1513 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
1514 * predict.c (probability_reliable_p): Update.
1515 (edge_probability_reliable_p): Update.
1516 (br_prob_note_reliable_p): Update.
1517 (invert_br_probabilities): Update.
1518 (add_reg_br_prob_note): New function.
1519 (combine_predictions_for_insn): Update.
1520 * asan.c (asan_clear_shadow): Update.
1521 * cfgbuild.c (compute_outgoing_frequencies): Update.
1522 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
1523 (update_br_prob_note): Update.
1524 (rtl_verify_edges): Update.
1525 (purge_dead_edges): Update.
1526 (fixup_reorder_chain): Update.
1527 * emit-rtl.c (try_split): Update.
1528 * ifcvt.c (cond_exec_process_insns): Update.
1529 (cond_exec_process_if_block): Update.
1530 (dead_or_predicable): Update.
1531 * internal-fn.c (expand_addsub_overflow): Update.
1532 (expand_neg_overflow): Update.
1533 (expand_mul_overflow): Update.
1534 * loop-doloop.c (doloop_modify): Update.
1535 * loop-unroll.c (compare_and_jump_seq): Update.
1536 * optabs.c (emit_cmp_and_jump_insn_1): Update.
1537 * predict.h: Update.
1538 * reorg.c (mostly_true_jump): Update.
1540 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
1541 * config/alpha/alpha.c (emit_unlikely_jump): Update.
1542 * config/arc/arc.c: (emit_unlikely_jump): Update.
1543 * config/arm/arm.c: (emit_unlikely_jump): Update.
1544 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
1545 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
1546 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
1547 (ix86_print_operand): Update.
1548 (ix86_split_fp_branch): Update.
1549 (predict_jump): Update.
1550 * config/ia64/ia64.c (ia64_print_operand): Update.
1551 * config/mmix/mmix.c (mmix_print_operand): Update.
1552 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
1553 (rs6000_expand_split_stack_prologue): Update.
1554 * config/rs6000/rs6000.c: Update.
1555 * config/s390/s390.c (s390_expand_vec_strlen): Update.
1556 (s390_expand_vec_movstr): Update.
1557 (s390_expand_cs_tdsi): Update.
1558 (s390_expand_split_stack_prologue): Update.
1559 * config/sh/sh.c (sh_print_operand): Update.
1560 (expand_cbranchsi4): Update.
1561 (expand_cbranchdi4): Update.
1562 * config/sparc/sparc.c (output_v9branch): Update.
1563 * config/spu/spu.c (get_branch_target): Update.
1564 (ea_load_store_inline): Update.
1565 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
1566 * config/tilepro/tilepro.c: Update.
1568 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1570 * gimplify.c (mostly_copy_tree_r): Revert latest change.
1571 (gimplify_save_expr): Likewise.
1573 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1575 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
1577 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1579 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
1581 * timevar.def (TV_IPA_FNSUMMARY): Define.
1583 2017-07-16 Daniel Cederman <cederman@gaisler.com>
1585 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
1586 to back store errata sensitive sequence from being generated.
1587 (sqrtdf2_fix): Likewise.
1589 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1591 * tree-ssa-threadupdate.c (compute_path_counts,
1592 update_joiner_offpath_counts): Use profile_probability.
1594 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
1597 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1599 * config/arm/arm-c.c (arm_cpu_builtins): Define
1600 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1602 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
1604 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1605 array entries to represent __ieee128 versions of the
1606 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
1607 scalar_extract_sig, and scalar_insert_exp built-in functions.
1608 (altivec_resolve_overloaded_builtin): Add special case handling
1609 for the __builtin_scalar_insert_exp function, as represented by
1610 the P9V_BUILTIN_VEC_VSIEDP constant.
1611 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
1612 exponent support for __ieee128 argument.
1613 (VSESQP): Add scalar extract signature support for __ieee128
1615 (VSTDCNQP): Add scalar test negative support for __ieee128
1617 (VSIEQP): Add scalar insert exponent support for __int128 argument
1618 with __ieee128 result.
1619 (VSIEQPF): Add scalar insert exponent support for __ieee128
1620 argument with __ieee128 result.
1621 (VSTDCQP): Add scalar test data class support for __ieee128
1623 (VSTDCNQP): Add overload support for scalar test negative with
1625 (VSTDCQP): Add overload support for scalar test data class
1627 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
1629 (UNSPEC_VSX_SIEXPQP): New constant.
1630 (xsxexpqp): New insn for VSX scalar extract exponent quad
1632 (xsxsigqp): New insn for VSX scalar extract significand quad
1634 (xsiexpqpf): New insn for VSX scalar insert exponent quad
1635 precision with floating point argument.
1636 (xststdcqp): New expand for VSX scalar test data class quad
1638 (xststdcnegqp): New expand for VSX scalar test negative quad
1640 (xststdcqp): New insn to match expansions for VSX scalar test data
1641 class quad precision and VSX scalar test negative quad precision.
1642 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
1643 special case operand checking to enforce that second operand of
1644 VSX scalar test data class with quad precision argument is a 7-bit
1646 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
1647 prototypes and descriptions of __ieee128 versions of
1648 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
1649 scalar_test_data_class, and scalar_test_neg built-in functions.
1651 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1653 PR tree-optimization/81162
1654 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
1655 replace a negate with an add.
1657 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
1659 * doc/invoke.texi (arm/-mcpu): Document +crypto.
1661 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1663 * config/arm/arm-c.c (arm_cpu_builtins): Define
1664 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1666 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1668 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
1669 (armv8-r): Set ARM Cortex-R52 as default CPU.
1670 * config/arm/arm-tables.opt: Regenerate.
1671 * config/arm/arm-tune.md: Regenerate.
1672 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
1674 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
1675 extension for -mcpu=cortex-r52.
1677 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1679 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
1680 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
1681 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
1682 (fp-armv8): Define it as FP_ARMv8 only.
1683 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
1684 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
1686 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
1687 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
1688 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
1689 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
1690 than TARGET_FPU_ARMV8.
1691 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
1692 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
1693 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
1695 * config/arm/neon.md (neon_vrint): Likewise.
1696 (neon_vcvt): Likewise.
1697 (neon_<fmaxmin_op><mode>): Likewise.
1698 (<fmaxmin><mode>3): Likewise.
1699 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
1700 * config/arm/predicates.md (arm_cond_move_operator): Check against
1701 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
1703 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
1705 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
1708 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1710 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
1711 loop in comment with memset.
1713 2017-07-14 Martin Liska <mliska@suse.cz>
1715 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
1716 * dwarf2out.c (is_java): Remove the function.
1717 (output_pubname): Remove usage of the function.
1718 (lower_bound_default): Remove usage of DW_LANG_Java.
1719 (gen_compile_unit_die): Likewise.
1720 * gcc.c: Remove compiler defaults for .java and .zip files.
1721 * gimple-expr.c (remove_suffix): Change as there's no longer
1722 extension than 4-letter one.
1723 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
1724 (gimplify_save_expr): Likewise.
1725 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
1726 as it's possible even for other languages than Java.
1727 * langhooks.h (struct lang_hooks): Remove Java from a comment.
1728 * lto-opts.c (lto_write_options): Remove reference to Java.
1729 * opts.c (strip_off_ending): Update file extension handling.
1730 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
1731 * tree-eh.c (lower_resx): Likewise.
1732 * tree.c (free_lang_data_in_type): Remove dead code.
1733 (find_decls_types_r): Likewise.
1734 (build_common_builtin_nodes): Remove Java from a comment.
1735 (verify_type): Remove dead code.
1736 * varasm.c (assemble_external): Remove Java from a comment.
1738 2017-07-14 Martin Liska <mliska@suse.cz>
1740 * opts.c (finish_options): Add quotes.
1741 (common_handle_option): Likewise.
1743 2017-07-14 Martin Liska <mliska@suse.cz>
1745 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
1746 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
1748 * dwarf2out.c (lower_bound_default): Do not handle
1750 (gen_compile_unit_die): Likewise.
1751 * gcc.c: Remove default extension binding for GNU Pascal.
1752 * stmt.c: Remove Pascal language from a comment.
1753 * xcoffout.c: Likewise.
1755 2017-07-13 David Malcolm <dmalcolm@redhat.com>
1758 * diagnostic-show-locus.c (fixit_cmp): New function.
1759 (layout::layout): Sort m_fixit_hints.
1760 (column_range::column_range): Assert that the values are valid.
1761 (struct char_span): New struct.
1762 (correction::overwrite): New method.
1763 (struct source_line): New struct.
1764 (line_corrections::add_hint): Add assertions. Reimplement memcpy
1765 calls in terms of classes source_line and char_span, and
1766 correction::overwrite.
1767 (selftest::test_overlapped_fixit_printing_2): New function.
1768 (selftest::diagnostic_show_locus_c_tests): Call it.
1770 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
1772 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
1773 early if there is no lhs.
1775 2017-07-13 Martin Liska <mliska@suse.cz>
1777 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
1778 (gen_reference_type_die): Likewise.
1779 * stor-layout.c: Remove Pascal-related comment.
1781 2017-07-13 Martin Liska <mliska@suse.cz>
1783 * opts.c (finish_options): Add quotes to error messages.
1784 (parse_sanitizer_options): Likewise.
1786 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
1788 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
1790 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
1792 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
1794 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
1796 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
1798 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
1800 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1803 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
1804 provides the hardware capability bits, define the macro
1805 __BUILTIN_CPU_SUPPORTS__.
1806 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
1807 if GLIBC does not provide the hardware capability bits. Add a
1808 gcc_unreachable call if the built-in cpu function is neither
1809 __builtin_cpu_is nor __builtin_cpu_supports.
1810 (rs6000_get_function_versions_dispatcher): Change the warning
1811 that an old GLIBC is used which does not export the capability
1812 bits to be an error.
1813 * doc/extend.texi (target_clones attribute): Document the
1814 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
1815 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
1816 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
1817 the macros defined by GCC if the newer GLIBC is available.
1819 2017-07-12 Jeff Law <law@redhat.com>
1821 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
1822 remaining includes slightly.
1823 * config/riscv/riscv-builtins.c: Include profile-count.h.
1825 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1828 * config/avr/avr.c (avr_set_current_function): In diagnostic
1829 messages: Quote keywords and (parts of) identifiers.
1830 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
1833 2017-07-12 Carl Love <cel@us.ibm.com>
1835 * config/rs6000/rs6000-c.c: Add support for built-in functions
1836 vector bool char vec_revb (vector bool char);
1837 vector bool short vec_revb (vector short char);
1838 vector bool int vec_revb (vector bool int);
1839 vector bool long long vec_revb (vector bool long long);
1840 * doc/extend.texi: Update the built-in documentation file for the
1841 new built-in functions.
1843 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1845 * config/s390/s390.md: Remove movcc splitter.
1847 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1849 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
1850 load/store on condition.
1852 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1855 * config/avr/avr.c (avr_encode_section_info)
1856 [progmem && !TREE_READONLY]: Error if progmem object needs
1859 2017-07-11 Michael Collison <michael.collison@arm.com>
1861 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
1864 2017-07-11 Carl Love <cel@us.ibm.com>
1866 * config/rs6000/rs6000-c.c: Add support for builtins
1867 vector unsigned int vec_parity_lsbb (vector signed int);
1868 vector unsigned int vec_parity_lsbb (vector unsigned int);
1869 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
1870 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
1871 vector unsigned long long vec_parity_lsbb (vector signed long long);
1872 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
1873 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
1874 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
1875 * doc/extend.texi: Update the built-in documentation file for the
1876 new built-in functions.
1878 2017-07-11 David Malcolm <dmalcolm@redhat.com>
1880 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
1881 (layout::m_primary_loc): New field.
1882 (layout::layout): Initialize new field. Move location filtering
1883 logic from here to...
1884 (layout::maybe_add_location_range): ...this new method. Add
1885 support for filtering to just the lines already specified by other
1887 (layout::will_show_line_p): New method.
1888 (gcc_rich_location::add_location_if_nearby): New method.
1889 (selftest::test_add_location_if_nearby): New test function.
1890 (selftest::diagnostic_show_locus_c_tests): Call it.
1891 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
1894 2017-07-11 Tom de Vries <tom@codesourcery.com>
1896 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
1897 (bb_first_real_insn): New function.
1898 (nvptx_single): Add extra initialization of broadcasted condition
1901 2017-07-11 Nathan Sidwell <nathan@acm.org>
1903 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
1905 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
1907 * doc/extend.texi (AVR Function Attributes): Remove weblink to
1908 Binutils doc as TEXI will mess them up.
1909 * doc/invoke.texi (AVR Options): Same here.
1911 2017-07-11 Daniel Cederman <cederman@gaisler.com>
1913 * config/sparc/sparc.opt (mfix-ut700): New option.
1914 (mfix-gr712rc): Likewise.
1915 (sparc_fix_b2bst): New variable.
1916 * doc/invoke.texi (SPARC options): Document them.
1917 (ARM options): Fix warnings.
1918 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
1919 instructions to prevent sequences that can trigger the store-store
1920 errata for certain LEON3FT processors.
1921 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
1922 (sparc_option_override): Set sparc_fix_b2bst appropriately.
1923 * config/sparc/sparc.md (fix_b2bst): New attribute.
1924 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
1926 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
1929 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
1931 (*rsqrtsf2_sse): Ditto.
1933 (div<mode>3): Macroize insn from divdf3 and divsf3
1934 using MODEF mode iterator.
1936 2017-07-10 Martin Sebor <msebor@redhat.com>
1938 PR tree-optimization/80397
1939 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
1940 instead of testing for equality to INTEGER_TYPE.
1942 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
1944 * config.gcc: Remove uclibc from arc target spec.
1946 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
1948 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
1950 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1953 * lto-wrapper.c (remove_option): New function.
1954 (merge_and_complain): Merge PIC/PIE options more realistically.
1956 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
1958 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
1962 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
1963 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
1964 * config.in: Regenerate.
1965 * configure: Regenerate.
1966 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
1967 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
1968 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
1969 (TARGET_GASISR_PROLOGUES): ...target mask.
1970 * common/config/avr/avr-common.c
1971 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
1972 Set -mgas-isr-prologues.
1973 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
1974 INSERT_PASS_BEFORE for it.
1975 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
1976 * config/avr/avr.c (avr_option_override)
1977 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
1978 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
1979 (avr_attribute_table) <no_gccisr>: Add new function attribute.
1980 (avr_set_current_function) <is_no_gccisr>: Init machine field.
1981 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
1983 (make_avr_pass_pre_proep): New function.
1984 (emit_push_sfr) <treg>: Add argument to function and use it
1986 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
1987 and set machine->gasisr.yes.
1988 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
1989 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
1990 __gcc_isr.n_pushed to .L__stack_usage.
1991 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
1992 (avr_asm_final_postscan_insn): ...this new static function.
1993 * config/avr/avr.h (machine_function)
1994 <is_no_gccisr, use_L__stack_usage>: New fields.
1995 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
1996 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
1997 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
1998 (gasisr, *gasisr): New expander and insn.
1999 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
2000 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
2001 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
2003 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
2005 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
2008 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
2010 Move jump-tables out of .text again.
2013 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
2014 (ASM_OUTPUT_ADDR_VEC): New function.
2015 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
2016 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
2017 INSN_ADDRESSes as asm comment.
2018 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
2019 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
2020 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
2021 * config/avr/avr.md (*tablejump): Adjust comment.
2022 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
2023 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
2025 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
2026 (avr_output_addr_vec): New proto.
2027 (avr_log_t) <insn_addresses>: New field.
2029 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
2032 * config/i386/i386.c (ix86_function_arg_advance): Set
2033 outgoing_args_on_stack to true if there are outgoing arguments
2035 (ix86_function_arg): Likewise.
2036 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
2037 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
2038 * config/i386/i386.h (machine_function): Add
2039 outgoing_args_on_stack.
2041 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
2043 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
2045 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
2047 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
2049 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
2050 (REAL_H): Remove $(MACHMODE_H).
2051 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
2053 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
2054 $(MACHMODE_H) and double-int.h.
2055 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
2057 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
2060 2017-07-07 Andrew Pinski <apinski@cavium.com>
2062 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
2063 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
2065 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
2067 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
2068 Add warning if GCC was not configured to link against a GLIBC that
2069 exports the hardware capability bits.
2070 (make_resolver_func): Make resolver function private and not a
2071 COMDAT function. Create the name with clone_function_name instead
2072 of make_unique_name.
2075 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
2076 correct operand in doing the split.
2078 2017-07-07 Carl Love <cel@us.ibm.com>
2080 * config/rs6000/rs6000-c: Add support for built-in function
2081 vector unsigned short vec_pack_to_short_fp32 (vector float,
2083 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
2084 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
2085 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
2086 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
2087 (convert_4f32_8i16): Add define_expand.
2088 * doc/extend.texi: Update the built-in documentation file for the
2089 new built-in function.
2091 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2093 * config/sparc/m8.md: New file.
2094 * config/sparc/sparc.md: Include m8.md.
2096 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2098 * config/sparc/sparc.opt: New option -mvis4b.
2099 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
2100 (sparc_option_override): Handle VIS4B.
2101 (enum sparc_builtins): Define
2102 SPARC_BUILTIN_DICTUNPACK{8,16,32},
2103 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
2104 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
2105 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
2106 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
2107 (check_constant_argument): New function.
2108 (sparc_vis_init_builtins): Define builtins
2109 __builtin_vis_dictunpack{8,16,32},
2110 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
2111 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
2112 __builtin_vis_fpcmpde{8,16,32}shl and
2113 __builtin_vis_fpcmpur{8,16,32}shl.
2114 (sparc_expand_builtin): Check that the constant operands to
2115 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
2116 constant and in range.
2117 * config/sparc/sparc-c.c (sparc_target_macros): Handle
2119 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
2120 (SPARC_IMM5_P): Likewise.
2121 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
2122 (enabled): Handle vis4b.
2123 (UNSPEC_DICTUNPACK): New unspec.
2124 (UNSPEC_FPCMPSHL): Likewise.
2125 (UNSPEC_FPUCMPSHL): Likewise.
2126 (UNSPEC_FPCMPDESHL): Likewise.
2127 (UNSPEC_FPCMPURSHL): Likewise.
2128 (cpu_feature): New CPU feature `vis4b'.
2129 (dictunpack{8,16,32}): New insns.
2130 (FPCSMODE): New mode iterator.
2131 (fpcscond): New code iterator.
2132 (fpcsucond): Likewise.
2133 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
2134 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
2135 (fpcmpde{8,16,32}{si,di}shl): Likewise.
2136 (fpcmpur{8,16,32}{si,di}shl): Likewise.
2137 * config/sparc/constraints.md: Define constraints `q' for unsigned
2138 2-bit integer constants and `t' for unsigned 5-bit integer
2140 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
2142 (imm5_operand_dictunpack16): Likewise.
2143 (imm5_operand_dictunpack32): Likewise.
2144 (imm2_operand): Likewise.
2145 * doc/invoke.texi (SPARC Options): Document -mvis4b.
2146 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
2147 ditunpack* and fpcmp*shl builtins.
2149 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2151 * config.gcc: Handle m8 in --with-{cpu,tune} options.
2152 * config.in: Add HAVE_AS_SPARC6 define.
2153 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
2155 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
2157 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
2158 (CPP_CPU_SPEC): Handle m8.
2159 (ASM_CPU_SPEC): Likewise.
2160 * config/sparc/sparc-opts.h (enum processor_type): Add
2162 * config/sparc/sparc.c (m8_costs): New struct.
2163 (sparc_option_override): Handle TARGET_CPU_m8.
2164 (sparc32_initialize_trampoline): Likewise.
2165 (sparc64_initialize_trampoline): Likewise.
2166 (sparc_issue_rate): Likewise.
2167 (sparc_register_move_cost): Likewise.
2168 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
2169 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
2170 (ASM_CPU64_DEFAULT_SPEC): Likewise.
2171 (CPP_CPU_SPEC): Handle M8.
2172 (ASM_CPU_SPEC): Likewise.
2173 (AS_M8_FLAG): Define.
2174 * config/sparc/sparc.md: Add m8 to the cpu attribute.
2175 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
2176 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
2178 * configure: Regenerate.
2179 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
2182 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2184 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
2186 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
2187 ("*movdi_insn_sp32"): Do not set v3pipe.
2188 ("*movsi_insn"): Likewise.
2189 ("*movdi_insn_sp64"): Likewise.
2190 ("*movsf_insn"): Likewise.
2191 ("*movdf_insn_sp32"): Likewise.
2192 ("*movdf_insn_sp64"): Likewise.
2193 ("*zero_extendsidi2_insn_sp64"): Likewise.
2194 ("*sign_extendsidi2_insn"): Likewise.
2195 ("*mov<VM32:mode>_insn"): Likewise.
2196 ("*mov<VM64:mode>_insn_sp64"): Likewise.
2197 ("*mov<VM64:mode>_insn_sp32"): Likewise.
2198 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
2199 ("<vlop:code><VL:mode>3"): Likewise.
2200 ("*not_<vlop:code><VL:mode>3"): Likewise.
2201 ("*nand<VL:mode>_vis"): Likewise.
2202 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
2203 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
2204 ("one_cmpl<VL:mode>2"): Likewise.
2205 ("faligndata<VM64:mode>_vis"): Likewise.
2206 ("alignaddrsi_vis"): Likewise.
2207 ("alignaddrdi_vis"): Likweise.
2208 ("alignaddrlsi_vis"): Likewise.
2209 ("alignaddrldi_vis"): Likewise.
2210 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
2211 ("bmaskdi_vis"): Likewise.
2212 ("bmasksi_vis"): Likewise.
2213 ("bshuffle<VM64:mode>_vis"): Likewise.
2214 ("cmask8<P:mode>_vis"): Likewise.
2215 ("cmask16<P:mode>_vis"): Likewise.
2216 ("cmask32<P:mode>_vis"): Likewise.
2217 ("pdistn<P:mode>_vis"): Likewise.
2218 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
2220 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2222 * config/sparc/sparc.md ("subtype"): New insn attribute.
2223 ("*wrgsr_sp64"): Set insn subtype.
2224 ("*rdgsr_sp64"): Likewise.
2225 ("alignaddrsi_vis"): Likewise.
2226 ("alignaddrdi_vis"): Likewise.
2227 ("alignaddrlsi_vis"): Likewise.
2228 ("alignaddrldi_vis"): Likewise.
2229 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
2230 ("fexpand_vis"): Likewise.
2231 ("fpmerge_vis"): Likewise.
2232 ("faligndata<VM64:mode>_vis"): Likewise.
2233 ("bshuffle<VM64:mode>_vis"): Likewise.
2234 ("cmask8<P:mode>_vis"): Likewise.
2235 ("cmask16<P:mode>_vis"): Likewise.
2236 ("cmask32<P:mode>_vis"): Likewise.
2237 ("fchksm16_vis"): Likewise.
2238 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
2239 ("fmean16_vis"): Likewise.
2240 ("fp<plusminus_insn>64_vis"): Likewise.
2241 ("<plusminus_insn>v8qi3"): Likewise.
2242 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
2243 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
2244 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
2245 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
2246 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
2247 ("*movqi_insn"): Likewise.
2248 ("*movhi_insn"): Likewise.
2249 ("*movsi_insn"): Likewise.
2250 ("movsi_pic_gotdata_op"): Likewise.
2251 ("*movdi_insn_sp32"): Likewise.
2252 ("*movdi_insn_sp64"): Likewise.
2253 ("movdi_pic_gotdata_op"): Likewise.
2254 ("*movsf_insn"): Likewise.
2255 ("*movdf_insn_sp32"): Likewise.
2256 ("*movdf_insn_sp64"): Likewise.
2257 ("*zero_extendhisi2_insn"): Likewise.
2258 ("*zero_extendqihi2_insn"): Likewise.
2259 ("*zero_extendqisi2_insn"): Likewise.
2260 ("*zero_extendqidi2_insn"): Likewise.
2261 ("*zero_extendhidi2_insn"): Likewise.
2262 ("*zero_extendsidi2_insn_sp64"): Likewise.
2263 ("ldfsr"): Likewise.
2264 ("prefetch_64"): Likewise.
2265 ("prefetch_32"): Likewise.
2266 ("tie_ld32"): Likewise.
2267 ("tie_ld64"): Likewise.
2268 ("*tldo_ldub_sp32"): Likewise.
2269 ("*tldo_ldub1_sp32"): Likewise.
2270 ("*tldo_ldub2_sp32"): Likewise.
2271 ("*tldo_ldub_sp64"): Likewise.
2272 ("*tldo_ldub1_sp64"): Likewise.
2273 ("*tldo_ldub2_sp64"): Likewise.
2274 ("*tldo_ldub3_sp64"): Likewise.
2275 ("*tldo_lduh_sp32"): Likewise.
2276 ("*tldo_lduh1_sp32"): Likewise.
2277 ("*tldo_lduh_sp64"): Likewise.
2278 ("*tldo_lduh1_sp64"): Likewise.
2279 ("*tldo_lduh2_sp64"): Likewise.
2280 ("*tldo_lduw_sp32"): Likewise.
2281 ("*tldo_lduw_sp64"): Likewise.
2282 ("*tldo_lduw1_sp64"): Likewise.
2283 ("*tldo_ldx_sp64"): Likewise.
2284 ("*mov<VM32:mode>_insn"): Likewise.
2285 ("*mov<VM64:mode>_insn_sp64"): Likewise.
2286 ("*mov<VM64:mode>_insn_sp32"): Likewise.
2288 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2290 * config/sparc/sparc.md ("type"): New insn type viscmp.
2291 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
2293 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
2294 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
2295 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
2296 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
2298 ("n7_vis_logical_11cycle"): Likewise.
2299 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
2300 * config/sparc/niagara2.md ("niag3_vis": Likewise.
2301 * config/sparc/niagara.md ("niag_vis"): Likewise.
2302 * config/sparc/ultra3.md ("us3_fga"): Likewise.
2303 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
2305 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2307 * config/sparc/sparc.md: New instruction type `bmask'.
2308 (bmaskdi_vis): Use the `bmask' type.
2309 (bmasksi_vis): Likewise.
2310 * config/sparc/ultra3.md (us3_array): Likewise.
2311 * config/sparc/niagara7.md (n7_array): Likewise.
2312 * config/sparc/niagara4.md (n4_array): Likewise.
2313 * config/sparc/niagara2.md (niag2_vis): Likewise.
2314 (niag3_vis): Likewise.
2315 * config/sparc/niagara.md (niag_vis): Likewise.
2317 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2319 * ipa-comdats.c: Remove optimize check from gate.
2320 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
2321 for functions not optimized.
2322 (ipa_fn_summary_read): Skip optimize check.
2323 (ipa_fn_summary_write): Likewise.
2324 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
2326 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
2328 (can_inline_edge_p): Check flag_pcc_struct_return for match.
2329 (check_callers): Give up on caller which is not optimized.
2330 (inline_small_functions): Likewise.
2331 (ipa_inline): Do not give up when not optimizing.
2332 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
2333 away unoptimizes cdtors.
2334 (whole_program_function_and_variable_visibility): Do
2335 ipa_discover_readonly_nonaddressable_vars in LTO mode.
2336 * ipa.c (process_references): Do not check optimize.
2337 (symbol_table::remove_unreachable_nodes): Update optimize check.
2338 (set_writeonly_bit): Update optimize check.
2339 (pass_ipa_cdtor_merge::gate): Do not check optimize.
2340 (pass_ipa_single_use::gate): Remove.
2342 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2344 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
2345 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
2346 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
2347 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
2348 permute_load, permute_store, adjust_extract, adjust_splat,
2349 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
2350 replace_swap_with_copy, dump_swap_insn_table,
2351 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
2352 recombine_lvx_pattern, recombine_stvx_pattern,
2353 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
2354 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
2355 to file rs6000-p8swap.c.
2356 * config/rs6000/rs6000-p8swap.c: New file.
2357 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
2358 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
2359 and rs6000*-*-* targets.
2361 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2363 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
2365 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2367 * lto-wrapper.c (merge_and_complain): Do not merge
2368 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
2369 fsigned_zeros, ftrapping_math, fwrapv.
2370 (append_compiler_options): Do not track these options.
2371 (append_linker_options): Likewie
2373 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2375 * cgraphunit.c (cgraph_node::finalize_function): When
2376 !flag_toplevel_reorde set no_reorder flag.
2377 (varpool_node::finalize_decl): Likewise.
2378 (symbol_table::compile): Drop no toplevel reorder path.
2380 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2382 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
2383 edges; zero probability is not better than uninitialized.
2385 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
2387 * asan.h (asan_sanitize_allocas_p): Declare.
2388 * asan.c (asan_sanitize_allocas_p): New function.
2389 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
2390 (handle_builtin_alloca): Likewise.
2391 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
2392 if !asan_sanitize_allocas_p.
2393 * params.def (asan-instrument-allocas): Add new option.
2394 * params.h (ASAN_PROTECT_ALLOCAS): Define.
2395 * opts.c (common_handle_option): Disable allocas sanitization for
2398 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
2400 * asan.c: Include gimple-fold.h.
2401 (get_last_alloca_addr): New function.
2402 (handle_builtin_stackrestore): Likewise.
2403 (handle_builtin_alloca): Likewise.
2404 (asan_emit_allocas_unpoison): Likewise.
2405 (get_mem_refs_of_builtin_call): Add new parameter, remove const
2406 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
2407 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
2408 (instrument_builtin_call): Pass gimple iterator to
2409 get_mem_refs_of_builtin_call.
2410 (last_alloca_addr): New global.
2411 * asan.h (asan_emit_allocas_unpoison): Declare.
2412 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
2413 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
2414 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
2415 if function calls alloca.
2416 * gimple-fold.c (replace_call_with_value): Remove static keyword.
2417 * gimple-fold.h (replace_call_with_value): Declare.
2418 * internal-fn.c: Include asan.h.
2419 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
2420 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
2422 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2424 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
2425 (C_SELFTEST_FLAGS): New.
2426 (CPP_SELFTEST_FLAGS): New.
2427 (SELFTEST_DEPS): New, from deps of s-selftest.
2428 (C_SELFTEST_DEPS): New, from deps of s-selftest.
2429 (CPP_SELFTEST_DEPS): New.
2430 (selftest): Add dependency on s-selftest-c++.
2431 (s-selftest): Rename to...
2432 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
2433 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
2434 than SELFTEST_FLAGS.
2435 (selftest-gdb): Rename to...
2436 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
2438 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
2439 (selftest-valgrind): Rename to...
2440 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
2442 (selftest-valgrind): Reintroduce as an alias for
2443 selftest-c-valgrind.
2444 (s-selftest-c++): New.
2445 (selftest-c++-gdb): New.
2446 (selftest-c++-valgrind): New.
2448 2017-07-06 Olivier Hainque <hainque@adacore.com>
2450 * gcc.c (process_command): When deciding if undefined variables
2451 should be ignored when processing specs, accept "gcc -v" as well.
2453 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2455 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
2456 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
2458 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
2460 * config/arm/arm-cpus.in (armv8-r): Add new entry.
2461 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
2462 * config/arm/arm-tables.opt: Regenerate.
2463 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
2465 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
2467 2017-07-06 Carl Love <cel@us.ibm.com>
2469 * ChangeLog: Clean up from mid air collision
2471 2017-07-06 Carl Love <cel@us.ibm.com>
2473 * config/rs6000/rs6000-c.c: Add support for built-in functions
2474 vector signed int vec_subc (vector signed int, vector signed int);
2475 vector signed __int128 vec_subc (vector signed __int128,
2476 vector signed __int128);
2477 vector unsigned __int128 vec_subc (vector unsigned __int128,
2478 vector unsigned __int128);
2479 vector signed int vec_sube (vector signed int, vector signed int,
2481 vector unsigned int vec_sube (vector unsigned int,
2482 vector unsigned int,
2483 vector unsigned int);
2484 vector signed __int128 vec_sube (vector signed __int128,
2485 vector signed __int128,
2486 vector signed__int128);
2487 vector unsigned __int128 vec_sube (vector unsigned __int128,
2488 vector unsigned __int128,
2489 vector unsigned __int128);
2490 vector signed int vec_subec (vector signed int, vector signed int,
2492 vector unsigned int vec_subec (vector unsigned int,
2493 vector unsigned int,
2494 vector unsigned int);
2495 vector signed __int128 vec_subec (vector signed __int128,
2496 vector signed __int128,
2497 vector signed__int128);
2498 vector unsigned __int128 vec_subec (vector unsigned __int128,
2499 vector unsigned __int128,
2500 vector unsigned __int128);
2501 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
2502 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
2503 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
2504 BU_ALTIVEC_OVERLOAD_X definitions.
2505 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
2506 * doc/extend.texi: Update the built-in documentation file for the new
2509 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2512 * diagnostic-show-locus.c (layout::layout): Use start and finish
2513 spelling location for the start and finish of each range.
2514 * genmatch.c (linemap_client_expand_location_to_spelling_point):
2515 Add unused aspect param.
2516 * input.c (expand_location_1): Add "aspect" param, and use it
2517 to access the correct part of the location.
2518 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
2520 (expand_location_to_spelling_point): Likewise.
2521 (linemap_client_expand_location_to_spelling_point): Add "aspect"
2522 param, and pass it to expand_location_1.
2524 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
2526 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
2527 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
2528 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
2529 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
2530 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
2531 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
2532 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
2533 _mm_maskz_getmant_ss): New intrinsics.
2534 (__builtin_ia32_getexpss128_mask): Changed to ...
2535 __builtin_ia32_getexpss128_round ... this.
2536 (__builtin_ia32_getexpsd128_mask): Changed to ...
2537 __builtin_ia32_getexpsd128_round ... this.
2538 * config/i386/i386-builtin-types.def
2539 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
2540 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
2541 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
2542 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
2543 __builtin_ia32_getmantss_mask_round): New builtins.
2544 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
2545 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
2546 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
2547 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
2548 * config/i386/sse.md
2549 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
2550 avx512f_sgetexp<mode><mask_scalar_name>
2551 <round_saeonly_scalar_name> ... this.
2552 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
2553 %0, %1, %2<round_saeonly_op3>}): Changed to ...
2554 vgetexp<ssescalarmodesuffix>
2555 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2556 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
2557 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
2558 avx512f_vgetmant<mode><mask_scalar_name>
2559 <round_saeonly_scalar_name> ... this.
2560 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
2561 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
2562 vgetmant<ssescalarmodesuffix>
2563 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
2564 %0<mask_scalar_operand4>, %1, %2
2565 <round_saeonly_scalar_mask_op4>, %3} ... this.
2566 * config/i386/subst.md (mask_scalar_operand4,
2567 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
2568 round_saeonly_scalar_nimm_predicate): New subst attributes.
2570 2017-07-06 Julia Koval <julia.koval@intel.com>
2572 * config/i386/i386.c (ix86_erase_embedded_rounding):
2573 Remove code for old rounding pattern.
2575 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
2577 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
2579 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
2581 * doc/sourcebuild.texi (Test Directives, Variants of
2582 dg-require-support): Add documentation for dg-require-stack-check.
2584 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
2586 * config/i386/subst.md (mask_scalar, round_scalar,
2587 round_saeonly_scalar): New meta-templates.
2588 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
2589 round_scalar_mask_operand3, round_scalar_mask_op3,
2590 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
2591 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
2592 round_saeonly_scalar_constraint,
2593 round_saeonly_scalar_prefix): New subst attribute.
2594 * config/i386/sse.md
2595 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
2596 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
2597 <round_scalar_name> ... this.
2598 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
2599 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
2600 <round_scalar_name> ... this.
2601 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
2602 <sse>_vm<code><mode>3<mask_scalar_name>
2603 <round_saeonly_scalar_name> ... this.
2604 (v<plusminus_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<plusminus_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<multdiv_mnemonic><ssescalarmodesuffix>
2611 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2612 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2613 v<multdiv_mnemonic><ssescalarmodesuffix>
2614 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2615 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2616 (v<maxmin_float><ssescalarmodesuffix>
2617 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
2618 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
2619 v<maxmin_float><ssescalarmodesuffix>
2620 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2621 %0<mask_scalar_operand3>, %1, %<iptr>2
2622 <round_saeonly_scalar_mask_op3>} ... this.
2624 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
2626 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
2627 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
2629 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
2630 Alan Hayward <alan.hayward@arm.com>
2631 David Sherwood <david.sherwood@arm.com>
2633 * combine.c (simplify_if_then_else): Remove "enum" before
2635 * compare-elim.c (can_eliminate_compare): Likewise.
2636 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
2638 (aarch64_lookup_simd_builtin_type): Likewise.
2639 (aarch64_simd_builtin_type): Likewise.
2640 (aarch64_init_simd_builtin_types): Likewise.
2641 (aarch64_simd_expand_args): Likewise.
2642 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
2644 (aarch64_reverse_mask): Likewise.
2645 (aarch64_simd_emit_reg_reg_move): Likewise.
2646 (aarch64_gen_adjusted_ldpstp): Likewise.
2647 (aarch64_ccmp_mode_to_code): Likewise.
2648 (aarch64_operands_ok_for_ldpstp): Likewise.
2649 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2650 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
2652 (aarch64_min_divisions_for_recip_mul): Likewise.
2653 (aarch64_reassociation_width): Likewise.
2654 (aarch64_get_condition_code_1): Likewise.
2655 (aarch64_simd_emit_reg_reg_move): Likewise.
2656 (aarch64_simd_attr_length_rglist): Likewise.
2657 (aarch64_reverse_mask): Likewise.
2658 (aarch64_operands_ok_for_ldpstp): Likewise.
2659 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2660 (aarch64_gen_adjusted_ldpstp): Likewise.
2661 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
2663 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
2664 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
2665 (arm_lookup_simd_builtin_type): Likewise.
2666 (arm_simd_builtin_type): Likewise.
2667 (arm_init_simd_builtin_types): Likewise.
2668 (arm_expand_builtin_args): Likewise.
2669 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
2670 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
2671 (ft32_setup_incoming_varargs): Likewise.
2672 (ft32_function_arg): Likewise.
2673 (ft32_function_arg_advance): Likewise.
2674 (ft32_pass_by_reference): Likewise.
2675 (ft32_arg_partial_bytes): Likewise.
2676 (ft32_valid_pointer_mode): Likewise.
2677 (ft32_addr_space_pointer_mode): Likewise.
2678 (ft32_addr_space_legitimate_address_p): Likewise.
2679 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
2681 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
2682 (ix86_emit_outlined_ms2sysv_restore): Likewise.
2683 (iamcu_alignment): Likewise.
2684 (canonicalize_vector_int_perm): Likewise.
2685 (ix86_noce_conversion_profitable_p): Likewise.
2686 (ix86_mpx_bound_mode): Likewise.
2687 (ix86_operands_ok_for_move_multiple): Likewise.
2688 * config/microblaze/microblaze-protos.h
2689 (microblaze_expand_conditional_branch_reg): Likewise.
2690 * config/microblaze/microblaze.c
2691 (microblaze_expand_conditional_branch_reg): Likewise.
2692 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
2694 (rs6000_reassociation_width): Likewise.
2695 (rs6000_invalid_binary_op): Likewise.
2696 (fusion_p9_p): Likewise.
2697 (emit_fusion_p9_load): Likewise.
2698 (emit_fusion_p9_store): Likewise.
2699 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
2701 (riscv_hard_regno_mode_ok_p): Likewise.
2702 (riscv_address_insns): Likewise.
2703 (riscv_split_symbol): Likewise.
2704 (riscv_legitimize_move): Likewise.
2705 (riscv_function_value): Likewise.
2706 (riscv_hard_regno_nregs): Likewise.
2707 (riscv_expand_builtin): Likewise.
2708 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
2709 (riscv_build_integer): Likewise.
2710 (riscv_split_integer): Likewise.
2711 (riscv_legitimate_constant_p): Likewise.
2712 (riscv_cannot_force_const_mem): Likewise.
2713 (riscv_regno_mode_ok_for_base_p): Likewise.
2714 (riscv_valid_base_register_p): Likewise.
2715 (riscv_valid_offset_p): Likewise.
2716 (riscv_valid_lo_sum_p): Likewise.
2717 (riscv_classify_address): Likewise.
2718 (riscv_legitimate_address_p): Likewise.
2719 (riscv_address_insns): Likewise.
2720 (riscv_load_store_insns): Likewise.
2721 (riscv_force_binary): Likewise.
2722 (riscv_split_symbol): Likewise.
2723 (riscv_force_address): Likewise.
2724 (riscv_legitimize_address): Likewise.
2725 (riscv_move_integer): Likewise.
2726 (riscv_legitimize_const_move): Likewise.
2727 (riscv_legitimize_move): Likewise.
2728 (riscv_address_cost): Likewise.
2729 (riscv_subword): Likewise.
2730 (riscv_output_move): Likewise.
2731 (riscv_canonicalize_int_order_test): Likewise.
2732 (riscv_emit_int_order_test): Likewise.
2733 (riscv_function_arg_boundary): Likewise.
2734 (riscv_pass_mode_in_fpr_p): Likewise.
2735 (riscv_pass_fpr_single): Likewise.
2736 (riscv_pass_fpr_pair): Likewise.
2737 (riscv_get_arg_info): Likewise.
2738 (riscv_function_arg): Likewise.
2739 (riscv_function_arg_advance): Likewise.
2740 (riscv_arg_partial_bytes): Likewise.
2741 (riscv_function_value): Likewise.
2742 (riscv_pass_by_reference): Likewise.
2743 (riscv_setup_incoming_varargs): Likewise.
2744 (riscv_print_operand): Likewise.
2745 (riscv_elf_select_rtx_section): Likewise.
2746 (riscv_save_restore_reg): Likewise.
2747 (riscv_for_each_saved_reg): Likewise.
2748 (riscv_register_move_cost): Likewise.
2749 (riscv_hard_regno_mode_ok_p): Likewise.
2750 (riscv_hard_regno_nregs): Likewise.
2751 (riscv_class_max_nregs): Likewise.
2752 (riscv_memory_move_cost): Likewise.
2753 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
2754 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
2755 (rl78_addr_space_address_mode): Likewise.
2756 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2758 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2759 (rs6000_reassociation_width): Likewise.
2760 (rs6000_invalid_binary_op): Likewise.
2761 (fusion_p9_p): Likewise.
2762 (emit_fusion_p9_load): Likewise.
2763 (emit_fusion_p9_store): Likewise.
2764 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
2765 (ok_for_simple_move_operands): Likewise.
2766 (ok_for_simple_move_strict_operands): Likewise.
2767 (ok_for_simple_arith_logic_operands): Likewise.
2768 (visium_legitimize_reload_address): Likewise.
2769 (visium_select_cc_mode): Likewise.
2770 (output_cbranch): Likewise.
2771 (visium_split_double_move): Likewise.
2772 (visium_expand_copysign): Likewise.
2773 (visium_expand_int_cstore): Likewise.
2774 (visium_expand_fp_cstore): Likewise.
2775 * config/visium/visium.c (visium_pass_by_reference): Likewise.
2776 (visium_function_arg): Likewise.
2777 (visium_function_arg_advance): Likewise.
2778 (visium_libcall_value): Likewise.
2779 (visium_setup_incoming_varargs): Likewise.
2780 (visium_legitimate_constant_p): Likewise.
2781 (visium_legitimate_address_p): Likewise.
2782 (visium_legitimize_address): Likewise.
2783 (visium_secondary_reload): Likewise.
2784 (visium_register_move_cost): Likewise.
2785 (visium_memory_move_cost): Likewise.
2786 (prepare_move_operands): Likewise.
2787 (ok_for_simple_move_operands): Likewise.
2788 (ok_for_simple_move_strict_operands): Likewise.
2789 (ok_for_simple_arith_logic_operands): Likewise.
2790 (visium_function_value_1): Likewise.
2791 (rtx_ok_for_offset_p): Likewise.
2792 (visium_legitimize_reload_address): Likewise.
2793 (visium_split_double_move): Likewise.
2794 (visium_expand_copysign): Likewise.
2795 (visium_expand_int_cstore): Likewise.
2796 (visium_expand_fp_cstore): Likewise.
2797 (visium_split_cstore): Likewise.
2798 (visium_select_cc_mode): Likewise.
2799 (visium_split_cbranch): Likewise.
2800 (output_cbranch): Likewise.
2801 (visium_print_operand_address): Likewise.
2802 * expmed.c (flip_storage_order): Likewise.
2803 * expmed.h (emit_cstore): Likewise.
2804 (flip_storage_order): Likewise.
2805 * genrecog.c (validate_pattern): Likewise.
2806 * hsa-gen.c (gen_hsa_addr): Likewise.
2807 * internal-fn.c (expand_arith_overflow): Likewise.
2808 * ira-color.c (allocno_copy_cost_saving): Likewise.
2809 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2810 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
2811 (process_invariant_for_inheritance): Likewise.
2812 * lra-eliminations.c (move_plus_up): Likewise.
2813 * omp-low.c (lower_oacc_reductions): Likewise.
2814 * simplify-rtx.c (simplify_subreg): Likewise.
2815 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
2816 (TARGET_CHKP_BOUND_MODE): Likewise..
2817 * targhooks.c (default_chkp_bound_mode): Likewise.
2818 (default_setup_incoming_vararg_bounds): Likewise.
2819 * targhooks.h (default_chkp_bound_mode): Likewise.
2820 (default_setup_incoming_vararg_bounds): Likewise.
2821 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
2822 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
2823 (have_whole_vector_shift): Likewise.
2824 * tree-vect-stmts.c (vectorizable_load): Likewise.
2825 * doc/tm.texi: Regenerate.
2827 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2829 Graceful degrade if Binutils PR21472 is not available.
2832 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
2833 .rodata in flash test fails.
2834 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
2835 * confgure: Regenerate.
2836 * config.in: Regenerate.
2837 * config/avr/avr.c (avr_asm_named_section)
2838 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
2839 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
2840 (avr_asm_init_sections): Same.
2842 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2844 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
2845 (fma<VH:mode>4_intrinsic): Likewise.
2846 (*fmsub<VCVTF:mode>4): Likewise.
2847 (*fmsub<VH:mode>4_intrinsic): Likewise.
2849 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2852 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
2853 Don't depend on "optimize > 0".
2854 (out_movhi_r_mr, out_movqi_mr_r): Same.
2855 (out_movhi_mr_r, out_movqi_r_mr): Same.
2856 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
2857 io_address_operand on "optimize > 0".
2859 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2861 * tree-loop-distribution.c: Add general explanantion on the pass.
2862 (generate_loops_for_partition): Mark distributed loop.
2863 (pg_add_dependence_edges): New parameter. Handle alias data
2864 dependence specially and record it in the parameter if asked.
2865 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
2866 (init_partition_graph_vertices, add_partition_graph_edge): New.
2867 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
2868 (free_partition_graph_vdata, build_partition_graph): New.
2869 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
2870 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
2871 (data_ref_segment_size, latch_dominated_by_data_ref): New.
2872 (compute_alias_check_pairs, version_loop_by_alias_check): New.
2873 (version_for_distribution_p, finalize_partitions): New.
2874 (distribute_loop): Handle alias data dependence specially. Factor
2875 out loop fusion code as functions and call these functions.
2877 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2879 * tree-loop-distribution.c (classify_partition): New parameter and
2880 better handle reduction statement.
2881 (rdg_build_partitions): Revise comment.
2882 (distribute_loop): Compute statements in all partitions and pass it
2883 to classify_partition.
2885 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2887 * tree-loop-distribution.c (enum partition_type): New.
2888 (struct partition): New field type.
2889 (partition_merge_into): Add parameter. Update partition type.
2890 (data_dep_in_cycle_p, update_type_for_merge): New functions.
2891 (build_rdg_partition_for_vertex): Compute partition type.
2892 (rdg_build_partitions): Dump partition type.
2893 (distribute_loop): Update calls to partition_merge_into.
2895 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2897 * tree-loop-distribution.c (struct ddr_hasher): New.
2898 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
2900 (classify_partition): Call get_data_dependence.
2901 (pg_add_dependence_edges): Ditto.
2902 (distribute_loop): Release data dependence hash table.
2904 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2906 * tree-loop-distribution.c (ref_base_address): Delete.
2907 (similar_memory_accesses): Rename ...
2908 (share_memory_accesses): ... to this. Check if partitions access
2909 the same memory reference.
2910 (distribute_loop): Call share_memory_accesses.
2912 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2914 * tree-loop-distribution.c (struct partition): New field recording
2916 (partition_alloc, partition_free): Init and release data refs.
2917 (partition_merge_into): Merge data refs.
2918 (build_rdg_partition_for_vertex): Collect data refs for partition.
2919 (pg_add_dependence_edges): Change parameters from vector to bitmap.
2921 (distribute_loop): Remve data refs from vertice data of partition
2924 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2926 * tree-loop-distribution.c (params.h): Include header file.
2927 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
2928 (datarefs_vec): New global var.
2929 (create_rdg_vertices): Use datarefs_vec directly.
2930 (free_rdg): Don't free data references.
2931 (build_rdg): Update use. Don't free data references.
2932 (distribute_loop): Compute global variable for data references.
2933 Bail out if there are too many data references.
2935 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2937 * tree-loop-distribution.c (loop_nest): New global var.
2938 (build_rdg): Use loop directly, rather than loop nest.
2939 (pg_add_dependence_edges): Remove loop nest parameter. Use global
2941 (distribute_loop): Compute global variable loop nest. Update use.
2943 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2945 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
2946 (partition_merge_into): New parameter. Dump reason for fusion.
2947 (distribute_loop): Update use of partition_merge_into.
2949 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2951 * tree-loop-distribution.c (bb_top_order_index): New.
2952 (bb_top_order_index_size, bb_top_order_cmp): New.
2953 (stmts_from_loop): Use topological order.
2954 (pass_loop_distribution::execute): Compute and release topological
2955 order for basic blocks.
2957 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2959 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
2962 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2964 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
2965 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
2966 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
2967 * internal-fn.def (LOOP_DIST_ALIAS): New.
2968 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
2969 (fold_loop_internal_call): ... this.
2970 (vect_loop_dist_alias_call): New function.
2971 (set_uid_loop_bbs): Call fold_loop_internal_call.
2972 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
2975 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
2978 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
2979 Require dead FLAGS_REG at the beginning of a peephole.
2981 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
2984 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
2985 arguments in the call to __builtin_ia32_sbb_u32.
2986 (_subborrow_u64): Swap _X and _Y arguments in the call to
2987 __builtin_ia32_sbb_u64.
2989 2017-07-04 Jakub Jelinek <jakub@redhat.com>
2992 * tree-vrp.c (compare_assert_loc): Turn into a function template
2993 with stable template parameter. Only test if a->e is NULL,
2994 !a->e == !b->e has been verified already. Use e == NULL or
2995 e != NULL instead of e or ! e tests. If stable is true, don't use
2996 iterative_hash_expr, on the other side allow a or b or both NULL
2997 and sort the NULLs last.
2998 (process_assert_insertions): Sort using compare_assert_loc<false>
2999 instead of compare_assert_loc, later sort using
3000 compare_assert_loc<true> before calling process_assert_insertions_for
3001 in a loop. Use break instead of continue once seen NULL pointer.
3003 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
3005 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
3006 Cortex-R7 and Cortex-R8 processors.
3008 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
3010 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
3011 uninitialized while src is not.
3013 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
3015 * common/config/arm/arm-common.c: Adjust include path for
3017 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
3018 (arm-cpu.h): Create in build directory. Adjust dependency rules.
3019 (arm-cpu-data.h): Likewise.
3020 (arm-cpu-cdata.h): Likewise.
3021 * config/arm/arm-cpu.h: Delete.
3022 * config/arm/arm-cpu-cdata.h: Delete.
3023 * config/arm/arm-cpu-data.h: Delete.
3025 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
3027 * config/arm/arm-cpus.in (cortex-a55): New.
3028 (cortex-a75): Likewise.
3029 (cortex-a75.cortex-a55): Likewise.
3030 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
3032 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
3033 * config/arm/arm-cpu-cdata.h: Regenerate.
3034 * config/arm/arm-cpu-data.h: Regenerate.
3035 * config/arm/arm-cpu.h: Regenerate.
3036 * config/arm/arm-tables.opt: Regenerate.
3037 * config/arm/arm-tune.md: Regenerate.
3039 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
3041 * haifa-sched.c (sched_create_recovery_edges): Update profile.
3043 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
3045 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
3048 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
3050 PR tree-optimization/81292
3051 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
3052 full_string_p, also call adjust_related_strinfos if the adjustment
3053 is simple, otherwise invalidate related strinfos.
3055 2017-07-04 Martin Liska <mliska@suse.cz>
3058 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
3059 newly created variable as DECL_IGNORED_P.
3061 2017-07-04 Martin Liska <mliska@suse.cz>
3064 * ipa-inline.c (inline_small_functions):
3065 Use xstrdup_for_dump.
3067 2017-07-04 Tom de Vries <tom@codesourcery.com>
3069 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
3071 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
3074 * config/darwin.c (darwin_function_switched_text_sections):
3077 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
3079 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
3081 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
3083 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
3085 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3087 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
3088 min_profitable_iters, and th as inclusive lower bounds.
3089 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
3090 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
3091 for min_profitable_iters and min_profitable_estimate.
3092 (vect_transform_loop): Treat th as an inclusive lower bound.
3093 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
3095 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
3098 * config/darwin.c (darwin_function_switched_text_sections):
3099 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
3100 in two pieces, and suppress the use of buf.
3102 2017-07-03 Nathan Sidwell <nathan@acm.org>
3104 * hash-table.h (hash_table_mod1): Fix indentation.
3106 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3109 * predict.c (force_edge_cold): Be more careful about propagation
3111 * profile-count.h (profile_probability::guessed,
3112 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
3114 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
3116 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
3118 * doc/invoke.texi (rcpc architecture extension): Document it.
3120 2017-07-03 Richard Biener <rguenther@suse.de>
3122 PR tree-optimization/60510
3123 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
3124 the scalar reduction PHI and use it.
3125 (vectorizable_reduction): Properly guard the single_defuse_cycle
3126 path for non-SLP reduction chains where we cannot use it.
3127 Rework reduc_def/index and vector type deduction. Rework
3128 vector operand gathering during reduction op code-gen.
3129 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
3130 chains dissolve the chain and leave it to non-SLP reduction
3133 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3135 * tree-data-ref.h (dr_alignment): Declare.
3136 * tree-data-ref.c (dr_alignment): New function.
3137 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
3138 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
3140 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
3142 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3144 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
3145 and base_misalignment fields.
3146 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
3147 * tree-data-ref.c: Include builtins.h.
3148 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
3149 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
3150 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
3151 * tree-vect-data-refs.c: Include tree-cfg.h.
3152 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
3153 fields instead of calculating an alignment here.
3154 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
3155 innermost_loop_behavior fields.
3157 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3159 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
3161 (DR_STEP_ALIGNMENT): New macro.
3162 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
3163 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
3164 (create_data_ref): Print it.
3165 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
3166 to tell whether the step preserves vector (mis)alignment.
3167 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
3168 Move the check for an integer step and generalise to all INTEGER_CST.
3169 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
3170 Print the outer step alignment.
3172 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3174 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
3175 with offset_alignment.
3176 (DR_ALIGNED_TO): Delete.
3177 (DR_OFFSET_ALIGNMENT): New macro.
3178 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
3179 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
3180 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
3181 (create_data_ref): Likewise.
3182 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
3183 (vect_analyze_data_refs): Likewise.
3184 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
3185 creating dummy innermost behavior.
3187 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3189 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
3190 with a "innermost_loop_behavior *" and refeence tree.
3191 * tree-data-ref.c (dr_analyze_innermost): Likewise.
3192 (create_data_ref): Update call accordingly.
3193 * tree-predcom.c (find_looparound_phi): Likewise.
3195 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3197 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
3198 fields with dr_wrt_vec_loop.
3199 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
3200 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
3201 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
3202 (vect_dr_behavior): New function.
3203 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
3204 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
3205 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
3206 track whether the step preserves the misalignment.
3207 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
3208 Use vect_dr_behavior.
3209 (vect_setup_realignment): Update call accordingly.
3210 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
3211 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
3212 call to vect_create_addr_base_for_vector_ref.
3213 (vect_create_cond_for_align_checks): Likewise.
3214 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
3215 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
3216 (vect_recog_mask_conversion_pattern): Likewise.
3217 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
3218 (new_stmt_vec_info): Remove redundant zeroing.
3220 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
3222 * common/config/arm/arm-common.c (arm_be8_option): New function.
3223 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
3224 (ISA_ARMv6): Add isa_bit_be8.
3225 * config/arm/arm.h (arm_be8_option): Add prototype.
3226 (BE8_SPEC_FUNCTION): New define.
3227 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
3228 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
3229 (mlittle-endian): Similarly.
3230 (mbe8, mbe32): New options.
3231 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
3232 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
3234 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3236 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
3238 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3240 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
3241 (cleanup_tree_cfg_bb): Use it.
3242 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
3244 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
3246 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3249 * loop-doloop.c (add_test): Update profile.
3251 2017-07-03 Martin Liska <mliska@suse.cz>
3254 * sanopt.c (rewrite_usage_of_param): New function.
3255 (sanitize_rewrite_addressable_params): Likewise.
3256 (pass_sanopt::execute): Call rewrite_usage_of_param.
3258 2017-07-03 Richard Biener <rguenther@suse.de>
3260 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
3261 back to using VIEW_CONVERT_EXPR.
3263 2017-07-03 Martin Liska <mliska@suse.cz>
3266 * doc/extend.texi: Document when a resolver function is
3267 generated for target_clones.
3269 2017-07-03 Martin Liska <mliska@suse.cz>
3271 * asan.c (asan_emit_stack_protection): Unpoison just red zones
3272 and shadow memory of auto variables which are subject of
3273 use-after-scope sanitization.
3274 (asan_expand_mark_ifn): Add do set only when is_poison.
3276 2016-07-03 Richard Biener <rguenther@suse.de>
3278 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
3280 (vect_force_simple_reduction): Record reduction def -> phi mapping.
3281 (vectorizable_reduction): Perform reduction PHI creation when
3282 visiting a reduction PHI and adjust and simplify code generation
3283 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
3284 (vect_transform_loop): Visit reduction PHIs.
3285 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
3286 defs into the SLP tree.
3287 (vect_build_slp_tree): Reduction defs terminate the recursion.
3288 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
3290 (vect_get_vec_defs_for_stmt_copy): Export.
3291 (vect_get_vec_defs): Likewise.
3292 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
3294 (vect_get_vec_defs_for_stmt_copy): Declare.
3295 (vect_get_vec_defs): Likewise.
3297 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3299 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
3300 parameter with a "loop" parameter and use it instead of the
3301 loop containing DR_STMT. Don't check simple_iv when doing
3302 BB analysis. Describe the two analysis modes in the comment.
3304 2017-07-03 Tom de Vries <tom@codesourcery.com>
3306 PR tree-optimization/69468
3307 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
3308 (find_same_succ_bb): Handle ignore_edge_flags.
3310 2017-07-03 Tom de Vries <tom@codesourcery.com>
3312 PR tree-optimization/81192
3313 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
3315 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
3317 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
3319 2017-07-03 Tom de Vries <tom@codesourcery.com>
3321 PR tree-optimization/81192
3322 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
3323 BB_SAME_SUCC (bb) == NULL.
3325 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3327 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
3330 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3332 * dumpfile.c: Include profile-count.h
3333 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
3335 (insert_cond_bb): Update profile.
3336 * tree-cfg.h (insert_cond_bb): Update prototype.
3337 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
3338 * tree-dump.c: Do not include tree-cfg.
3340 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3342 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
3344 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3346 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
3349 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3351 * tree-complex.c (expand_complex_div_wide): update profile.
3353 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3354 Alan Hayward <alan.hayward@arm.com>
3355 David Sherwood <david.sherwood@arm.com>
3357 * Makefile.in (MACHMODE_H): Remove insn-modes.h
3358 (CORETYPES_H): New define.
3359 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
3360 (insn-modes-inline.h, s-modes-inline-h): New rules.
3361 (generated_files): Add insn-modes-inline.h.
3362 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
3363 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
3364 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
3365 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
3366 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
3367 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
3368 (build/gencodes.o, build/genconditions.o): Likewise.
3369 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
3370 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
3371 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
3372 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
3373 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
3374 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
3375 * coretypes.h: Include everything up to real.h for generators.
3376 Include insn-modes.h first. Include wide-int-print.h after
3377 wide-int.h. Include insn-modes-inline.h and then machmode.h.
3378 * machmode.h: Don't include insn-modes.h here.
3379 * function-tests.c: Remove includes of signop.h, machmode.h,
3380 double-int.h and wide-int.h.
3382 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
3384 * optc-save-gen.awk: Likewise.
3385 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
3386 * godump.c: Remove include of wide-int-print.h.
3387 * pretty-print.h: Likewise.
3388 * wide-int-print.cc: Likewise.
3389 * wide-int.cc: Likewise.
3390 * hash-map-tests.c: Remove include of signop.h.
3391 * hash-set-tests.c: Likewise.
3392 * rtl-tests.c: Likewise.
3393 * mkconfig.sh: Remove include of machmode.h.
3394 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
3396 (emit_insn_modes_inline_h): ...this new function. Emit the code
3397 into an insn-modes-inline.h header file, adding appropriate
3398 include guards and end comments.
3399 (emit_insn_modes_c_header): Remove include of machmode.h.
3400 (emit_min_insn_modes_c_header): Include coretypes.h rather than
3402 (main): Handle -i flag and call emit_insn_modes_inline_h when
3405 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3407 * tree-ssa-strlen.c (strinfo): Rename the length field to
3408 nonzero_chars. Add a full_string_p field.
3409 (compare_nonzero_chars, zero_length_string_p): New functions.
3410 (get_addr_stridx): Add an offset_out parameter.
3411 Use compare_nonzero_chars.
3412 (get_stridx): Update accordingly. Use compare_nonzero_chars.
3413 (new_strinfo): Update after above changes to strinfo.
3414 (set_endptr_and_length): Set full_string_p.
3415 (get_string_length): Update after above changes to strinfo.
3416 (unshare_strinfo): Update call to new_strinfo.
3417 (maybe_invalidate): Likewise.
3418 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
3419 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
3420 as a uhwi instead of an shwi. Update after above changes to
3421 strinfo and new_strinfo.
3422 (zero_length_string): Assert that chainsi contains full strings.
3423 Use zero_length_string_p. Update call to new_strinfo.
3424 (adjust_related_strinfos): Update after above changes to strinfo.
3425 Copy full_string_p from origsi.
3426 (adjust_last_stmt): Use zero_length_string_p.
3427 (handle_builtin_strlen): Update after above changes to strinfo and
3428 new_strinfo. Install the lhs as the string length if the previous
3429 entry didn't describe a full string.
3430 (handle_builtin_strchr): Update after above changes to strinfo
3432 (handle_builtin_strcpy): Likewise.
3433 (handle_builtin_strcat): Likewise.
3434 (handle_builtin_malloc): Likewise.
3435 (handle_pointer_plus): Likewise.
3436 (handle_builtin_memcpy): Likewise. Track nonzero characters
3437 that aren't necessarily followed by a nul terminator.
3438 (handle_char_store): Likewise.
3440 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3442 PR tree-optimization/80769
3443 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
3444 for malloc and calloc. Document the new invariant that all related
3445 strinfos have delayed lengths or none do.
3446 (verify_related_strinfos): Move earlier in file.
3447 (set_endptr_and_length): New function, split out from...
3448 (get_string_length): ...here. Also set the lengths of related
3450 (zero_length_string): Assert that chainsi has known (rather than
3452 (adjust_related_strinfos): Likewise.
3454 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3456 PR tree-optimization/81136
3457 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
3458 assert that two references with the same misalignment have the same
3459 compile-time misalignment if those compile-time misalignments
3462 2017-07-01 Andi Kleen <ak@linux.intel.com>
3464 * print-tree.c (print_node): Print all attributes.
3466 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3468 * cfg.c (scale_bbs_frequencies): New function.
3469 * cfg.h (scale_bbs_frequencies): Declare it.
3470 * cfgloopanal.c (single_likely_exit): Cleanup.
3471 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
3473 (scale_loop_profile): Likewise.
3474 (loop_version): Likewise.
3475 (create_empty_loop_on_edge): Update.
3476 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
3477 scale_loop_frequencies, scale_loop_profile, loopify,
3478 loop_version): Update prototypes.
3479 * modulo-sched.c (sms_schedule): Update.
3480 * predict.c (unlikely_executed_edge_p): Also check probability.
3481 (probably_never_executed_edge_p): Fix typo.
3482 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3483 * tree-parloops.c (gen_parallel_loop): Update.
3484 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
3485 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3486 * tree-ssa-loop-split.c (split_loop): Update.
3487 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3488 * tree-vect-loop-manip.c (vect_do_peeling): Update.
3489 (vect_loop_versioning): Update.
3490 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3492 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3494 * trans-mem.c (split_bb_make_tm_edge): Update profile.
3496 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3498 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
3499 to keep profile consistent.
3501 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3503 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
3504 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
3505 * profile-count.h (max_safe_multiplier): Make unsigned.
3506 (profile_count::guessed_zero): New.
3508 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3510 * bb-reorder.c (fix_up_crossing_landing_pad,
3511 fix_crossing_conditional_branches): Use make_single_succ_edge
3512 to keep profile consistent.
3514 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3516 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
3519 2017-07-01 Jakub Jelinek <jakub@redhat.com>
3522 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
3523 the right scopes, make sure cond_jump isn't preserved between multiple
3524 iterations. Search for fallthru edge whenever there are 3+ edges and
3525 use find_fallthru_edge for it.
3527 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3529 Patch by Alexander Monakov <amonakov@ispras.ru>
3530 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
3531 probabilities consistently.
3533 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3535 * pa.c (pa_expand_compare_and_swap_loop): Update call of
3536 emit_cmp_and_jump_insns.
3538 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3541 * tree-inline.c (expand_call_inline): Combine profile statuses.
3543 2017-06-30 Andrew Pinski <apinski@cavium.com>
3545 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
3546 fold_stmt returned true.
3548 2017-06-30 Nathan Sidwell <nathan@acm.org>
3550 * ggc.h (empty_string): Delete.
3551 * cfgexpand.c (expand_asm_stmt): Use plain "".
3552 * optabs.c (expand_asm_memory_barrier): Likewise.
3553 * stringpool.c (empty_string): Delete.
3554 (digit_vector, digit_string): Delete.
3555 (ggc_alloc_string): Use plain "", don't optimize single digit
3556 strings. Use ggc_alloc_atomic.
3558 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
3560 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
3561 comparison set and one other set, use the cost of the non-comparison
3564 2017-06-30 Nathan Sidwell <nathan@acm.org>
3566 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
3569 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
3571 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
3572 loops. Remove now unneeded calls to gimple_switch_set_label() that
3573 just set removed labels to NULL_TREE.
3575 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
3577 * tree-ssanames.c (set_range_info_raw): Abstract from ...
3578 (set_range_info): ...here. Only call set_range_info_raw if domain
3580 (set_nonzero_bits): Call set_range_info_raw.
3581 * tree-ssanames.h (set_range_info_raw): New.
3583 2017-06-30 Jakub Jelinek <jakub@redhat.com>
3586 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
3587 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
3588 of nonimmediate_operand and <store_mask_constraint> instead of m
3589 for the input operand. For V8FI iterator, always split if input
3590 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
3591 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
3592 <store_mask_predicate> instead of register_operand and
3593 <store_mask_constraint> instead of v for the input operand. Make
3594 sure both operands aren't MEMs for if not <mask_applied>.
3596 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
3598 * lto-wrapper.c (copy_file) Close both file descriptors before
3601 2017-06-30 Martin Liska <mliska@suse.cz>
3604 * multiple_target.c (create_dispatcher_calls): Make ifunc
3605 also for function that don't have calls or are not referenced.
3607 2017-06-30 Richard Biener <rguenther@suse.de>
3609 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
3610 analyze the first scalar stmt. Move vector type computation
3611 for the BB case here from ...
3612 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
3613 live operation processing in the SLP case properly.
3615 2017-06-30 Richard Biener <rguenther@suse.de>
3617 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
3619 2017-06-30 Martin Liska <mliska@suse.cz>
3622 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
3623 before BUILT_IN_UNWIND_RESUME when ASAN is used.
3625 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
3627 * doc/invoke.texi (AArch64): Add missing options and remove redundant
3630 2017-06-30 Richard Biener <rguenther@suse.de>
3632 PR tree-optimization/81249
3633 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
3634 condition reduction result to original scalar type.
3636 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3638 * profile-count.h (enum profile_quality): Fix typos and whitespace
3641 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3643 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
3644 type for branch probabilities.
3646 2017-06-29 Julian Brown <julian@codesourcery.com>
3647 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3649 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
3650 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
3651 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
3652 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
3654 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3656 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
3657 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
3658 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
3659 CC usage from generic code to here.
3660 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
3661 CC usage into the target macros.
3663 2017-06-29 Maya Rashish <coypu@sdf.org>
3665 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
3668 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3670 * arm/arm-builtins.c: Include profile-count.h
3671 * except.c (sjlj_emit_function_enter): Use
3672 profile_probability::unlikely.
3674 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3676 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
3677 and tocrel_offset be pointer args rather than implicitly using
3679 (legitimate_constant_pool_address_p, rs6000_emit_move,
3680 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
3681 tocrel_offset and use in toc_relative_expr_p call.
3682 (print_operand, print_operand_address): Use static tocrel_base_oac
3683 and tocrel_offset_oac.
3684 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
3687 2017-06-29 Maya Rashish <coypu@sdf.org>
3689 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
3691 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
3693 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
3694 objects, take into account only the alignment of 'op0' and 'mode1' if
3697 2017-06-29 Steve Ellcey <sellcey@cavium.com>
3699 * ccmp.c (ccmp_tree_comparison_p): New function.
3700 (ccmp_candidate_p): Update to use above function.
3701 (get_compare_parts): New function.
3702 (expand_ccmp_next): Update to use new functions.
3703 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
3705 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
3706 take mode as argument.
3707 * ccmp.h (expand_ccmp_expr): Add mode as argument.
3708 * expr.c (expand_expr_real_1): Pass mode as argument.
3710 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
3712 * combine.c (combine_instructions): Print insns to dump_file, together
3715 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3717 * asan.c (asan_emit_stack_protection): Update.
3718 (create_cond_insert_point): Update.
3719 * auto-profile.c (afdo_propagate_circuit): Update.
3720 * basic-block.h (struct edge_def): Turn probability to
3721 profile_probability.
3722 (EDGE_FREQUENCY): Update.
3723 * bb-reorder.c (find_traces_1_round): Update.
3724 (better_edge_p): Update.
3725 (sanitize_hot_paths): Update.
3726 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
3727 (make_single_succ_edge): Update.
3728 (check_bb_profile): Update.
3729 (dump_edge_info): Update.
3730 (update_bb_profile_for_threading): Update.
3731 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
3732 probabilitycount to 0.
3733 * cfgbuild.c (compute_outgoing_frequencies): Update.
3734 * cfgcleanup.c (try_forward_edges): Update.
3735 (outgoing_edges_match): Update.
3736 (try_crossjump_to_edge): Update.
3737 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
3738 (expand_gimple_tailcall): Update.
3739 (construct_init_block): Use make_single_succ_edge.
3740 (construct_exit_block): Use make_single_succ_edge.
3741 * cfghooks.c (verify_flow_info): Update.
3742 (redirect_edge_succ_nodup): Update.
3743 (split_edge): Update.
3744 (account_profile_record): Update.
3745 * cfgloopanal.c (single_likely_exit): Update.
3746 * cfgloopmanip.c (scale_loop_profile): Update.
3747 (set_zero_probability): Remove.
3748 (duplicate_loop_to_header_edge): Update.
3749 * cfgloopmanip.h (loop_version): Update prototype.
3750 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3751 (force_nonfallthru_and_redirect): Update.
3752 (update_br_prob_note): Update.
3753 (rtl_verify_edges): Update.
3754 (purge_dead_edges): Update.
3755 (rtl_lv_add_condition_to_bb): Update.
3756 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
3757 * cgraphunit.c (init_lowered_empty_function): Update.
3758 (cgraph_node::expand_thunk): Update.
3759 * cilk-common.c: Include profile-count.h
3760 * dojump.c (inv): Remove.
3761 (jumpifnot): Update.
3762 (jumpifnot_1): Update.
3763 (do_jump_1): Update.
3765 (do_jump_by_parts_greater_rtx): Update.
3766 (do_compare_rtx_and_jump): Update.
3767 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
3768 do_jump_1. do_compare_rtx_and_jump): Update prototype.
3769 * dwarf2cfi.c: Include profile-count.h
3770 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
3771 (sjlj_emit_dispatch_table): Likewise.
3772 * explow.c: Include profile-count.h
3773 * expmed.c (emit_store_flag_force): Update.
3774 (do_cmp_and_jump): Update.
3775 * expr.c (compare_by_pieces_d::generate): Update.
3776 (compare_by_pieces_d::finish_mode): Update.
3777 (emit_block_move_via_loop): Update.
3778 (store_expr_with_bounds): Update.
3779 (store_constructor): Update.
3780 (expand_expr_real_2): Update.
3781 (expand_expr_real_1): Update.
3782 * expr.h (try_casesi, try_tablejump): Update prototypes.
3783 * gimple-pretty-print.c (dump_probability): Update.
3784 (dump_profile): New.
3785 (dump_gimple_label): Update.
3786 (dump_gimple_bb_header): Update.
3787 * graph.c (draw_cfg_node_succ_edges): Update.
3788 * hsa-gen.c (convert_switch_statements): Update.
3789 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
3790 (find_if_case_1): Update.
3791 (find_if_case_2): Update.
3792 * internal-fn.c (expand_arith_overflow_result_store): Update.
3793 (expand_addsub_overflow): Update.
3794 (expand_neg_overflow): Update.
3795 (expand_mul_overflow): Update.
3796 (expand_vector_ubsan_overflow): Update.
3797 * ipa-cp.c (good_cloning_opportunity_p): Update.
3798 * ipa-split.c (split_function): Use make_single_succ_edge.
3799 * ipa-utils.c (ipa_merge_profiles): Update.
3800 * loop-doloop.c (add_test): Update.
3801 (doloop_modify): Update.
3802 * loop-unroll.c (compare_and_jump_seq): Update.
3803 (unroll_loop_runtime_iterations): Update.
3804 * lra-constraints.c (lra_inheritance): Update.
3805 * lto-streamer-in.c (input_cfg): Update.
3806 * lto-streamer-out.c (output_cfg): Update.
3807 * mcf.c (adjust_cfg_counts): Update.
3808 * modulo-sched.c (sms_schedule): Update.
3809 * omp-expand.c (expand_omp_for_init_counts): Update.
3810 (extract_omp_for_update_vars): Update.
3811 (expand_omp_ordered_sink): Update.
3812 (expand_omp_for_ordered_loops): Update.
3813 (expand_omp_for_generic): Update.
3814 (expand_omp_for_static_nochunk): Update.
3815 (expand_omp_for_static_chunk): Update.
3816 (expand_cilk_for): Update.
3817 (expand_omp_simd): Update.
3818 (expand_omp_taskloop_for_outer): Update.
3819 (expand_omp_taskloop_for_inner): Update.
3820 * omp-simd-clone.c (simd_clone_adjust): Update.
3821 * optabs.c (expand_doubleword_shift): Update.
3822 (expand_abs): Update.
3823 (emit_cmp_and_jump_insn_1): Update.
3824 (expand_compare_and_swap_loop): Update.
3825 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
3826 * predict.c (predictable_edge_p): Update.
3827 (edge_probability_reliable_p): Update.
3828 (set_even_probabilities): Update.
3829 (combine_predictions_for_insn): Update.
3830 (combine_predictions_for_bb): Update.
3831 (propagate_freq): Update.
3832 (estimate_bb_frequencies): Update.
3833 (force_edge_cold): Update.
3834 * profile-count.c (profile_count::dump): Add missing space into dump.
3835 (profile_count::debug): Add newline.
3836 (profile_count::differs_from_p): Explicitly convert to unsigned.
3837 (profile_count::stream_in): Update.
3838 (profile_probability::dump): New member function.
3839 (profile_probability::debug): New member function.
3840 (profile_probability::differs_from_p): New member function.
3841 (profile_probability::differs_lot_from_p): New member function.
3842 (profile_probability::stream_in): New member function.
3843 (profile_probability::stream_out): New member function.
3844 * profile-count.h (profile_count_quality): Rename to ...
3845 (profile_quality): ... this one.
3846 (profile_probability): New.
3847 (profile_count): Update.
3848 * profile.c (compute_branch_probabilities): Update.
3849 * recog.c (peep2_attempt): Update.
3850 * sched-ebb.c (schedule_ebbs): Update.
3851 * sched-rgn.c (find_single_block_region): Update.
3852 (compute_dom_prob_ps): Update.
3853 (schedule_region): Update.
3854 * sel-sched-ir.c (compute_succs_info): Update.
3855 * stmt.c (struct case_node): Update.
3856 (do_jump_if_equal): Update.
3857 (get_outgoing_edge_probs): Update.
3858 (conditional_probability): Update.
3859 (emit_case_dispatch_table): Update.
3860 (expand_case): Update.
3861 (expand_sjlj_dispatch_table): Update.
3862 (emit_case_nodes): Update.
3863 * targhooks.c: Update.
3864 * tracer.c (better_p): Update.
3865 (find_best_successor): Update.
3866 * trans-mem.c (expand_transaction): Update.
3867 * tree-call-cdce.c: Update.
3868 * tree-cfg.c (gimple_split_edge): Upate.
3869 (move_sese_region_to_fn): Upate.
3870 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
3871 * tree-eh.c (lower_resx): Upate.
3872 (cleanup_empty_eh_move_lp): Upate.
3873 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3874 * tree-inline.c (copy_edges_for_bb): Update.
3875 (copy_cfg_body): Update.
3876 * tree-parloops.c (gen_parallel_loop): Update.
3877 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
3878 (gimple_gen_time_profiler): Update.
3879 * tree-ssa-dce.c (remove_dead_stmt): Update.
3880 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3881 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3882 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3883 (unloop_loops): Update.
3884 (try_peel_loop): Update.
3885 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3886 * tree-ssa-loop-split.c (connect_loops): Update.
3887 (split_loop): Update.
3888 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3889 (hoist_guard): Update.
3890 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3891 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3892 (value_replacement): Update.
3893 * tree-ssa-reassoc.c (branch_fixup): Update.
3894 * tree-ssa-tail-merge.c (replace_block_by): Update.
3895 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3896 (create_edge_and_update_destination_phis): Update.
3897 (compute_path_counts): Update.
3898 (recompute_probabilities): Update.
3899 (update_joiner_offpath_counts): Update.
3900 (freqs_to_counts_path): Update.
3901 (duplicate_thread_path): Update.
3902 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3903 (struct switch_conv_info): Update.
3904 (gen_inbound_check): Update.
3905 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3906 (vect_do_peeling): Update.
3907 (vect_loop_versioning): Update.
3908 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3909 (optimize_mask_stores): Update.
3910 * ubsan.c (ubsan_expand_null_ifn): Update.
3911 * value-prof.c (gimple_divmod_fixed_value): Update.
3912 (gimple_divmod_fixed_value_transform): Update.
3913 (gimple_mod_pow2): Update.
3914 (gimple_mod_pow2_value_transform): Update.
3915 (gimple_mod_subtract): Update.
3916 (gimple_mod_subtract_transform): Update.
3917 (gimple_ic): Update.
3918 (gimple_stringop_fixed_value): Update.
3919 (gimple_stringops_transform): Update.
3920 * value-prof.h: Update.
3922 2017-06-29 Carl Love <cel@us.ibm.com>
3924 * config/rs6000/rs6000-c.c: Add support for built-in functions
3925 vector signed int vec_signed (vector float);
3926 vector signed long long vec_signed (vector double);
3927 vector signed int vec_signed2 (vector double, vector double);
3928 vector signed int vec_signede (vector double);
3929 vector signed int vec_signedo (vector double);
3930 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
3931 instruction generator.
3932 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3933 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
3934 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
3936 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
3937 vunsignede_v2df): Add define_expands.
3938 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
3939 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
3940 VEC_UNSIGNEDO): Add definitions.
3941 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3942 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
3943 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
3944 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
3945 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
3946 * config/rs6000/altivec.h (vec_signed, vec_signed2,
3947 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
3948 vec_unsignede, vec_unsignedo): Add builtin defines.
3949 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
3951 * doc/extend.texi: Update the built-in documentation file for the
3952 new built-in functions.
3954 2017-06-29 Richard Biener <rguenther@suse.de>
3956 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
3957 reduction chains to LOOP_VINFO_REDUCTIONS.
3958 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
3959 SLP reductions after processing reduction chains.
3961 2017-06-29 Nathan Sidwell <nathan@acm.org>
3963 * builtins.c (fold_builtin_FUNCTION): Use
3964 lang_hooks.decl_printable_name.
3966 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
3969 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
3970 with only one label.
3971 * stmt.c (expand_case): Assert NCASES is greater than one.
3973 2017-06-29 Richard Biener <rguenther@suse.de>
3975 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
3977 (group_case_labels): Likewise.
3978 (find_taken_edge): Push sanity checking on val to workers...
3979 (find_taken_edge_cond_expr): ... here
3980 (find_taken_edge_switch_expr): ... and here, handle cases
3981 with just a default label.
3982 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
3983 (group_case_labels): Likewise.
3984 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
3985 group_case_labels does anything cleanup the CFG again.
3987 2017-06-29 Bin Cheng <bin.cheng@arm.com>
3989 PR tree-optimization/81196
3990 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
3991 exit condition comparing two IVs.
3993 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
3995 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
3996 profile to the dummy entry at the end of the list of architectures.
3997 * config/arm/arm-cpu-cdata.h: Regenerated.
3999 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4000 Michael Collison <michael.collison@arm.com>
4003 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
4005 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
4006 (*aarch64_reg_<mode>3_minus_mask): New pattern.
4007 (*aarch64_<optab>_reg_di3_mask2): New pattern.
4008 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
4009 of shift when the shift amount is masked with constant equal to
4010 the size of the mode.
4011 * config/aarch64/predicates.md (subreg_lowpart_operator): New
4014 2017-06-29 Martin Liska <mliska@suse.cz>
4016 * config/i386/i386.opt: Change range from [1,5] to [0,5].
4018 2017-06-29 Yury Gribov <tetra2005@gmail.com>
4021 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
4024 (edge_growth_cache_entry::edge_growth_cache_entry): New
4026 (reset_edge_growth_cache): Update to use constructor.
4028 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4030 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
4031 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
4032 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
4034 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
4036 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
4037 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
4039 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
4041 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
4042 (*-linux-uclibc*): Add t-uclibc tmake_file.
4043 * config/t-musl: New.
4044 * config/t-uclibc: New.
4046 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
4048 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
4050 (gen_comm_data): Emit architectural setting of arch_prof.
4051 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
4053 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
4054 (armv8-m.base, armv8-m.main): Likewise.
4055 * arm-protos.h (arm_build_target): Add profile field.
4056 (arch_option): Likewise.
4057 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
4059 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
4060 arm_active_target.profile.
4062 2017-06-28 Richard Biener <rguenther@suse.de>
4065 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
4066 TYPE_OVERFLOW_WRAPS.
4067 * match.pd (negate_expr_p): Likewise.
4068 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
4069 fold_build2, not fold_binary.
4071 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
4073 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
4074 Convert memory address to Pmode.
4075 (aarch64_print_operand): Assert MEM operands are always Pmode.
4077 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
4080 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
4081 Remove redundant if.
4082 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
4083 * config/arm/aarch-common-protos.h
4084 (aarch_forward_to_shift_is_not_shifted_re): Remove.
4085 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
4087 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
4090 * multiple_target.c (create_dispatcher_calls): Set the default
4091 clone to be static, not public.
4093 2017-06-28 Richard Biener <rguenther@suse.de>
4095 * tree-vect-loop.c (vectorizable_reduction): Move special
4096 cond reduction IV var creation ...
4097 (vect_create_epilog_for_reduction): ... here. Remove induction_index
4098 parameter. Use STMT_VINFO_VECTYPE.
4099 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
4102 2017-06-28 Martin Liska <mliska@suse.cz>
4105 * ipa-visibility.c (non_local_p): Handle visibility.
4107 2017-06-28 Martin Liska <mliska@suse.cz>
4110 * common.opt: Add IntegerRange to various options.
4111 * opt-functions.awk (integer_range_info): New function.
4112 * optc-gen.awk: Add integer_range_info to cl_options struct.
4113 * opts-common.c (decode_cmdline_option): Handle
4114 CL_ERR_INT_RANGE_ARG.
4115 (cmdline_handle_error): Likewise.
4116 * opts.c (print_filtered_help): Show valid interval in
4117 when --help is provided.
4118 * opts.h (struct cl_option): Add range_min and range_max fields.
4119 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
4121 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
4123 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
4124 (x * C EQ/NE y * C): New transformation.
4126 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
4128 * genmultilib (combination_space): Accept '+' in option names.
4130 2017-06-28 Martin Liska <mliska@suse.cz>
4133 * asan.c (instrument_derefs): Bail out inner references
4134 that are hard register variables.
4136 2017-06-28 Jakub Jelinek <jakub@redhat.com>
4139 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
4140 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
4142 2017-06-28 Richard Biener <rguenther@suse.de>
4144 * tree-vectorizer.h (vect_get_vec_defs): Remove.
4145 (vect_get_slp_defs): Adjust.
4146 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
4148 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
4150 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4151 get_initial_defs_for_reduction instead of vect_get_vec_defs.
4152 (vectorizable_reduction): Adjust.
4153 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
4155 (vect_get_slp_defs): Likewise.
4156 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
4157 (vectorizable_bswap): Adjust.
4158 (vectorizable_call): Likewise.
4159 (vectorizable_conversion): Likewise.
4160 (vectorizable_assignment): Likewise.
4161 (vectorizable_shift): Likewise.
4162 (vectorizable_operation): Likewise.
4163 (vectorizable_store): Likewise.
4164 (vectorizable_condition): Likewise.
4165 (vectorizable_comparison): Likewise.
4167 2017-06-28 Michael Collison <michael.collison@arm.com>
4170 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
4172 (arm_gen_movmemqi): Removed unused variable 'i'.
4173 Convert 'for' loop into 'while' loop.
4174 (arm_expand_prologue): Remove last unnecessary set of insn.
4175 (thumb_pop): Remove unused variable 'pushed_words'.
4176 (thumb_exit): Remove last unnecessary set of regs_to_pop.
4178 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4180 * config/s390/predicates.md: Use s390_rel_address_ok_p.
4181 * config/s390/s390-protos.h: Add prototype of
4182 s390_rel_address_ok_p.
4183 * config/s390/s390.c (s390_got_symbol): New function.
4184 (s390_rel_address_ok_p): New function.
4185 (legitimize_pic_address): Use s390_rel_address_ok_p.
4186 (s390_load_got): Use s390_got_symbol.
4187 (s390_option_override): Issue error if
4188 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
4189 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
4191 * config/s390/s390.opt: New option mpic-data-is-text-relative.
4193 2017-06-27 Andrew Pinski <apinski@cavium.com>
4195 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
4196 (X * copysign (1.0, X)): New pattern.
4197 (X * copysign (1.0, -X)): New pattern.
4198 (copysign (-1.0, CST)): New pattern.
4200 2017-06-27 Joseph Myers <joseph@codesourcery.com>
4202 * genmultilib (combination_space): Remove variable.
4203 Validate reuse rules against regular expression for any sequence
4204 of multilib options in any order.
4206 2017-06-27 Michael Collison <michael.collison@arm.com>
4208 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
4209 call aarch64_split_simd_combine.
4210 * (aarch64_combine_internal<mode>): Delete pattern.
4211 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
4212 Allow register and subreg operands.
4214 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4216 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
4217 specific need, just fallback on defaults.
4218 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
4220 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4221 Olivier Hainque <hainque@adacore.com>
4223 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
4225 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
4226 targets. Pick a default if no particular attempt applied.
4227 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
4230 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4232 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
4233 (x86_64-wrs-vxworks7): Likewise.
4235 2017-06-27 Marek Polacek <polacek@redhat.com>
4238 * ubsan.c (instrument_null): Check get_base_address's result for null.
4240 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
4242 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
4244 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
4246 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
4247 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
4248 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
4249 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
4251 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
4252 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
4253 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
4254 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
4255 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
4256 BUILT_IN_FEUPDATEENV): New builtins.
4257 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
4258 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
4259 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
4260 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
4262 (builtin_structptr_types): Adjust size.
4263 * tree.c (builtin_structptr_types): Add four entries.
4265 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4266 Olivier Hainque <hainque@adacore.com>
4268 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
4269 (TLS_SYM): New local macro, forcing reference to __tls__ on
4270 link command lines for VxWorks 7 RTPs, triggering initialization
4272 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
4273 OS features TLS support, true for RTPs on VxWorks 7.
4274 * config/vxworks.c (vxworks_override_options): Setup emutls
4277 2017-06-27 Jakub Jelinek <jakub@redhat.com>
4279 * predict.c (test_prediction_value_range): Use -1U instead of -1
4280 to avoid narrowing conversion warning.
4281 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
4282 to avoid narrowing conversion warning.
4283 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
4285 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
4287 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4289 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
4290 64bit configurations.
4291 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
4292 (SIZE_TYPE): Likewise.
4293 * config/vxworks.c (vxworks_emutls_var_fields): Use
4294 long_unsigned_type_node instead of unsigned_type_node as the offset
4295 field type, which is "pointer" mode in emutls.c.
4297 2017-06-27 Jakub Jelinek <jakub@redhat.com>
4300 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
4303 * gimple-fold.c (replace_call_with_call_and_fold): Handle
4304 gimple_vuse copying separately from gimple_vdef copying.
4306 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4308 * value-prof.c (free_hist): Remove call to memset and the enclosing if
4311 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
4312 Olivier Hainque <hainque@adacore.com>
4314 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
4315 for all vxworks7 targets.
4316 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
4317 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
4318 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
4319 variations for VX6/VX7 and 32/64bits later on in ...
4320 (VXWORKS_LIB_SPEC): Leverage new macros.
4321 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
4322 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
4324 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
4326 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
4327 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
4329 2017-06-26 Carl Love <cel@us.ibm.com>
4331 * config/rs6000/rs6000-c.c: Add support for built-in functions
4332 vector bool char vec_reve (vector bool char);
4333 vector signed char vec_reve (vector signed char);
4334 vector unsigned char vec_reve (vector unsigned char);
4335 vector bool int vec_reve (vector bool int);
4336 vector signed int vec_reve (vector signed int);
4337 vector unsigned int vec_reve (vector unsigned int);
4338 vector bool long long vec_reve (vector bool long long);
4339 vector signed long long vec_reve (vector signed long long);
4340 vector unsigned long long vec_reve (vector unsigned long long);
4341 vector bool short vec_reve (vector bool short);
4342 vector signed short vec_reve (vector signed short);
4343 vector double vec_reve (vector double);
4344 vector float vec_reve (vector float);
4345 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
4346 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
4347 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
4348 (altivec_vreve): New pattern.
4349 * config/rs6000/altivec.h (vec_reve): New define.
4350 * doc/extend.texi (vec_rev): Update the built-in documentation file
4351 for the new built-in functions.
4353 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4355 PR tree-optimization/71815
4356 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
4358 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
4360 (slsr_process_phi): Likewise.
4361 (replace_uncond_cands_and_profitable_phis): Don't replace a
4362 multiply candidate with a stride of 1 (copy or cast).
4363 (phi_incr_cost): Call uses_consumed_by_stmt rather than
4365 (lowest_cost_path): Likewise.
4366 (total_savings): Likewise.
4368 2017-06-26 Richard Biener <rguenther@suse.de>
4371 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
4372 Use def_builtin_pure for all gather builtins.
4374 2017-06-26 Richard Biener <rguenther@suse.de>
4376 PR tree-optimization/81203
4377 * tree-tailcall.c (find_tail_calls): Do not move stmts into
4380 2017-06-26 Marek Polacek <polacek@redhat.com>
4383 * doc/invoke.texi: Document -Wmultistatement-macros.
4385 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
4387 * doc/sourcebuild.texi (ARM-specific attributes): Document new
4388 arm_neon_ok_no_float_abi effective target.
4390 2017-06-26 Richard Biener <rguenther@suse.de>
4392 PR tree-optimization/80928
4393 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
4394 (copy_bbs): Set BB_DUPLICATED flag early.
4395 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
4397 (execute_on_shrinking_pred): Likewise.
4398 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
4399 BB_DUPLICATED blocks.
4400 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
4401 iterate over all PHIs considering removal of *gsi.
4403 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
4405 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
4408 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4410 * config/rs6000/rs6000-string.c: (expand_block_clear,
4411 do_load_for_compare, select_block_compare_mode,
4412 compute_current_alignment, expand_block_compare,
4413 expand_strncmp_align_check, expand_strn_compare,
4414 expand_block_move, rs6000_output_load_multiple)
4415 Move functions related to string/block move/compare
4417 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
4418 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
4419 for this function which is now used in two files.
4420 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
4421 * config.gcc: Add rs6000-string.o to extra_objs for
4422 targets powerpc*-*-* and rs6000*-*-*.
4424 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4427 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
4428 32-bit, since indexed is not valid for DImode.
4429 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
4430 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
4431 (define_peephole2 for Altivec d-form load): Add 32-bit support.
4432 (define_peephole2 for Altivec d-form store): Likewise.
4435 * multiple_target.c (create_dispatcher_calls): Only create the
4436 dispatcher call if the function is the default clone of a
4439 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
4442 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
4443 a call, force the call to not be a tail call.
4445 2017-06-23 Jeff Law <law@redhat.com>
4447 * doc/contrib.texi: Add entry for Steven Pemberton's work on
4450 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
4452 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
4453 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
4454 handling for early expansion of vector shifts (sl,sr,sra,rl).
4455 (builtin_function_type): Add vector shift right instructions
4456 to the unsigned argument list.
4458 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
4460 rtl-optimizatoin/79286
4461 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
4462 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
4463 trap. PIC register plus a const unspec without offset can never trap.
4465 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
4467 * tree.h (builtin_structptr_type): New type.
4468 (builtin_structptr_types): Declare new array.
4469 * tree.c (builtin_structptr_types): New array.
4470 (free_lang_data, build_common_tree_nodes): Use it.
4472 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
4475 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
4478 2017-06-22 Matt Turner <mattst88@gmail.com>
4480 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
4481 Lake models to skylake case. Assume skylake for unknown
4482 models with clflushopt.
4484 2017-06-22 Jeff Law <law@redhat.com>
4486 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
4487 frame sizes that do not satisfy aarch64_uimm12_shift.
4489 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
4491 * profile-count.h (apply_probability,
4492 apply_scale, probability_in): Fix checks for zero.
4494 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4496 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
4497 * doc/cppdiropts.texi (-I @var{dir}): Document it.
4499 2016-06-22 Richard Biener <rguenther@suse.de>
4501 * tree-vect-loop.c (vect_model_reduction_cost): Handle
4502 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
4503 REDUC_MAX_EXPR support.
4504 (vectorizable_reduction): Likewise.
4505 (vect_create_epilog_for_reduction): Likewise.
4507 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
4509 * match.pd (A / (1 << B) -> A >> B): New.
4510 * generic-match-head.c: Include optabs-tree.h.
4511 * gimple-match-head.c: Likewise.
4512 * optabs-tree.h (target_supports_op_p): New.
4513 * optabs-tree.c (target_supports_op_p): New.
4515 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4517 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
4518 $gcc_cv_ld --help output.
4519 (gcc_cv_ld_demangle): Likewise.
4520 (gcc_cv_ld_eh_frame_hdr): Likewise.
4521 (gcc_cv_ld_pie): Likewise.
4522 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
4523 (gcc_cv_ld_buildid): Likewise.
4524 (gcc_cv_ld_sysroot): Likewise.
4525 (ld_bndplt_support): Likewise.
4526 (ld_pushpopstate_support): Likewise.
4527 * configure: Regenerate.
4528 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
4530 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4533 * config/i386/sse.md (round<mode>2): Renumber match_dup and
4534 operands indexes to avoid gap between operands and match_dups.
4536 2017-06-21 Andrew Pinski <apinski@cavium.com>
4538 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
4539 Increment Arith_shift and Arith_shift_reg by 1.
4540 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
4542 * config/aarch64/aarch64.c (thunderx_tunings): Enable
4543 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
4544 (aarch64_strip_extend): Add new argument and test for it.
4545 (aarch64_cheap_mult_shift_p): New function.
4546 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
4547 add a cost if it is true.
4548 Update calls to aarch64_strip_extend.
4549 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
4551 2017-06-21 Andrew Pinski <apinski@cavium.com>
4553 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
4555 (thunderxt88): Likewise.
4556 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
4557 (thunderx_prefetch_tune): New variable.
4558 (thunderx2t99_prefetch_tune): Update for the correct values.
4559 (thunderxt88_tunings): New variable.
4560 (thunderx_tunings): Use thunderx_prefetch_tune instead of
4561 generic_prefetch_tune.
4562 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
4564 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4566 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
4567 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
4568 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
4569 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
4570 (aarch64_atomic_cas<mode>, GPI): Likewise.
4572 2017-06-21 Martin Liska <mliska@suse.cz>
4574 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
4575 statements on cold and hot labels.
4576 * predict.c (tree_estimate_probability_bb): Remove the
4577 prediction from this place.
4579 2017-06-21 Martin Liska <mliska@suse.cz>
4581 PR tree-optimization/79489
4582 * gimplify.c (maybe_add_early_return_predict_stmt): New
4584 (gimplify_return_expr): Call the function.
4585 * predict.c (tree_estimate_probability_bb): Remove handling
4587 * predict.def: Update comment about early return predictor.
4588 * gimple-predict.h (is_gimple_predict): New function.
4589 * predict.def: Change default value of early return to 66.
4590 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
4592 * passes.def: Put pass_strip_predict_hints to the beginning of
4595 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
4597 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
4598 FUNCTION_DECL declarations.
4599 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
4601 (dwaf2out_decl): Likewise.
4602 * godump.c (go_early_global_decl): Skip call to the real debug hook
4603 for FUNCTION_DECL declarations.
4604 * passes.c (rest_of_decl_compilation): Skip call to the
4605 early_global_decl debug hook for FUNCTION_DECL declarations, unless
4606 -fdump-go-spec is passed.
4608 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4610 * config/i386/i386.c (struct builtin_isa): New field pure_p.
4611 Reorder for compactness.
4612 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
4613 (def_builtin_pure, def_builtin_pure2): New functions.
4614 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
4616 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4618 * match.pd (nop_convert): New predicate.
4619 ((A +- CST1) +- CST2): Allow some NOP conversions.
4621 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4624 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
4625 with ctors/dtors if GOVD_SHARED is set.
4627 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4629 * config/aarch64/aarch64.md (movti_aarch64):
4630 Emit mov rather than orr.
4631 (movtf_aarch64): Likewise.
4632 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
4633 Emit mov rather than orr.
4635 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4637 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
4638 Swap alternatives, make integer dup more expensive.
4640 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4642 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
4643 Return true for non-tls symbols.
4645 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
4647 * config/aarch64/aarch64-cores.def (cortex-a55): New.
4648 (cortex-a75): Likewise.
4649 (cortex-a75.cortex-a55): Likewise.
4650 * config/aarch64/aarch64-tune.md: Regenerate.
4651 * doc/invoke.texi (-mtune): Document new values for -mtune.
4653 2017-06-21 Tom de Vries <tom@codesourcery.com>
4655 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
4657 (Effective-Target Keywords, Other attributes): Suggest using
4658 dg-add-options stack_size feature to get stack limit in stack_size
4659 effective target documentation.
4661 2017-06-21 Julian Brown <julian@codesourcery.com>
4662 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4664 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
4665 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
4666 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
4668 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
4669 attribute type list for neon_multiply.
4670 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
4671 attribute type list for neon_multiply.
4672 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
4673 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
4674 attribute type list for neon_multiply.
4675 * config/arm/types.md (crypto_pmull): Add.
4676 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
4677 attribute type list.
4679 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
4681 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
4684 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4686 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
4687 to make sure not to dereference a NULL cost_classes_ptr pointer.
4689 2017-06-20 Carl Love <cel@us.ibm.com>
4691 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4692 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4693 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4694 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4695 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4696 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4697 VMULOSW): New enum "unspec" values.
4698 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
4699 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
4700 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4701 altivec_vmulosw): New patterns.
4702 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4703 VMULOSW): Add definitions.
4705 2017-06-20 Julia Koval <julia.koval@intel.com>
4707 * config/i386/i386.c: Fix rounding expand for new pattern.
4708 * config/i386/subst.md: Fix pattern (parallel -> unspec).
4710 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4712 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
4713 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
4715 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4717 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
4720 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4722 * config/aarch64/aarch64-cores.def: Rearrange to sort by
4723 architecture, then by implementer ID.
4724 * config/aarch64/aarch64-tune.md: Regenerate.
4726 2017-06-20 Richard Biener <rguenther@suse.de>
4729 * fold-const.c (split_tree): Fold to type before negating.
4731 2017-06-20 David Malcolm <dmalcolm@redhat.com>
4733 * diagnostic-show-locus.c
4734 (selftest::test_fixit_deletion_affecting_newline): New function.
4735 (selftest::diagnostic_show_locus_c_tests): Call it.
4737 2017-06-20 Andreas Schwab <schwab@suse.de>
4740 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
4743 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
4745 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
4746 __ARM_FEATURE_COPROC according to support.
4748 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4750 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
4751 Rewritten to avoid overflow for > 32-bit pointers.
4754 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
4755 by removing enum keyword.
4756 (ubsan_type_descriptor): Likewise. Formatting fix.
4759 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
4760 splitter): Require TARGET_SSE2 in the condition.
4762 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
4765 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
4766 for doing vector set of SFmode on ISA 3.0.
4767 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
4768 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
4770 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
4771 SFmode value into a V4SF variable that was extracted from another
4772 V4SF variable without converting the element to double precision
4773 and back to single precision vector format.
4774 (vsx_insert_extract_v4sf_p9_2): Likewise.
4776 2017-06-19 Jakub Jelinek <jakub@redhat.com>
4778 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
4779 in UWHI to avoid undefined overflow.
4782 * ubsan.h (enum ubsan_encode_value_phase): New.
4783 (ubsan_encode_value): Change second argument to
4784 enum ubsan_encode_value_phase with default value of
4785 UBSAN_ENCODE_VALUE_GENERIC.
4786 * ubsan.c (ubsan_encode_value): Change second argument to
4787 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
4788 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
4789 create_tmp_var_raw instead of create_tmp_var and use a
4791 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
4792 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
4793 ubsan_encode_value callers.
4796 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
4797 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
4798 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
4800 2017-06-19 Richard Biener <rguenther@suse.de>
4803 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
4804 estimates if we changed anything.
4806 2017-06-19 Richard Biener <rguenther@suse.de>
4808 PR tree-optimization/80887
4809 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
4810 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
4811 simplified lookups, then reset mprts_hook.
4812 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
4814 (try_to_simplify): Likewise.
4816 2017-06-19 Martin Liska <mliska@suse.cz>
4819 * gimplify.c (gimplify_switch_expr):
4820 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
4822 2017-06-19 Martin Liska <mliska@suse.cz>
4824 * doc/install.texi: Document that PGO runs in 4 stages.
4826 2017-06-19 Martin Liska <mliska@suse.cz>
4829 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
4830 to dispatcher function name.
4831 * multiple_target.c (replace_function_decl): New function.
4832 (create_dispatcher_calls): Redirect both edges and references.
4834 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
4836 * profile-count.c (profile_count::dump): Dump quality.
4837 (profile_count::differs_from_p): Update for unsigned val.
4838 * profile-count.h (profile_count_quality): New enum.
4839 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
4841 2017-06-19 Richard Biener <rguenther@suse.de>
4843 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
4844 struct function as arg.
4845 (estimate_numbers_of_iterations): Export overload with loop arg.
4846 (free_numbers_of_iterations_estimates_loop): Use an overload of
4847 free_numbers_of_iterations_estimates instead.
4848 * tree-cfg.c (remove_bb): Adjust.
4849 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
4850 * tree-parloops.c (gen_parallel_loop): Likewise.
4851 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
4853 (tree_unroll_loops_completely): Likewise.
4854 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
4855 Use an overload instead and export.
4856 (estimated_loop_iterations): Adjust.
4857 (max_loop_iterations): Likewise.
4858 (likely_max_loop_iterations): Likewise.
4859 (estimate_numbers_of_iterations): Take struct function as arg
4861 (loop_exits_before_overflow): Adjust.
4862 (free_numbers_of_iterations_estimates_loop): Use an overload.
4863 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
4864 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
4866 2017-06-19 Richard Biener <rguenther@suse.de>
4869 * ipa-prop.c (find_constructor_constant_at_offset): Handle
4870 RANGE_EXPR conservatively.
4872 2017-06-16 Carl Love <cel@us.ibm.com>
4874 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4875 definitions for vec_float, vec_float2, vec_floato,
4876 vec_floate built-ins.
4877 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
4878 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
4880 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
4881 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
4882 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
4883 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
4884 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
4885 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
4886 vec_floato): Add builtin defines.
4887 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
4888 Update the built-in documentation file for the new built-in
4891 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4893 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
4894 (mthumb): Mark as the negative of -marm.
4896 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4898 * doc/invoke.texi (ARM Options, -mcpu): Document supported
4900 (ARM Options, -mtune): Document that this accepts the same
4901 extension options as -mcpu.
4902 (ARM Options, -mfpu): Document addition of -mfpu=auto.
4904 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4906 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
4907 permitted extensions.
4909 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4911 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
4912 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
4913 (armv8-m.main): Add option +nodsp.
4914 * config/arm/arm-cpu-cdata.h: Regenerated.
4916 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4918 * config/arm/t-fuchsia: New file.
4919 * config.gcc (arm*-*-fuchsia*): Use it.
4921 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4923 * config/arm/t-symbian: Rewrite for new option infrastructure.
4925 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4927 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
4928 (MULTILIB_REQUIRED): Likewise.
4930 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4932 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
4933 (MULTILIB_RESUE): Likewise.
4934 (MULTILIB_MATCHES): Likewise.
4935 (MULTLIB_REQUIRED): Likewise.
4937 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4939 * config/arm/t-rtems: Rewrite for new option framework.
4941 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4943 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
4944 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
4945 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
4946 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
4947 * config/arm/t-multilib: ... here.
4948 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
4949 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
4950 armv7-a and armv8*-a when A-profile libraries have not been built.
4951 * config/arm/t-rmprofile: Rewrite.
4953 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4955 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
4956 with a backslash. Remove the backslash after substituting unescaped
4958 * doc/fragments.texi (MULTILIB_REUSE): Document it.
4960 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4962 * config.gcc: (arm*-*-*): When building a-profile libraries, force
4963 the driver to pass through the default setting of -mfloat-abi.
4964 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
4966 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
4967 (all_feat_combs): New rule.
4968 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
4970 * config/arm/t-aprofile: Rewrite.
4972 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4974 * config/arm/arm.h (FPUTYPE_AUTO): Define.
4975 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
4976 fpu is not specified by the user/command-line.
4977 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
4978 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
4979 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
4980 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
4981 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
4982 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
4984 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4986 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
4987 * config/arm/t-arm-elf: Rewritten.
4989 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4991 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
4992 have some floating-point instructions.
4993 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
4994 (TARGET_MAYBE_HARD_FLOAT): New macro.
4995 * config/arm/arm-builtins.c (arm_init_builtins): Use
4996 TARGET_MAYBE_HARD_FLOAT.
4997 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
4999 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5001 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
5002 (configargs.h): Include it.
5003 (arm_print_hint_for_fpu_option): New function.
5004 (arm_parse_fpu_option): New function.
5005 (candidate_extension): New class.
5006 (arm_canon_for_multilib): New function.
5007 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
5008 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
5009 (ARCH_CANONICAL_SPECS): New macro.
5010 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
5012 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5014 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
5015 are set after handling multilib fragments. Set target_cpu_default2
5018 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5020 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
5022 (arm*-*-*): Set target_cpu_default2 to a quoted string.
5023 * config/arm/parsecpu.awk (check_cpu): Validate any extension
5025 (check_arch): Likewise.
5026 * config/arm/arm.c (arm_configure_build_target): Handle
5027 TARGET_CPU_DEFAULT being a string constant. Scan any feature
5028 options in the default.
5030 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5032 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
5033 when an option is an alias of another.
5034 * config/arm/parsecpu.awk (optalias): New parser token.
5035 (gen_comm_data): Mark non-alias options as such. Emit entries
5036 for extension aliases.
5037 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
5038 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
5039 (armv6kz, armv6zk, armv6t2): Likewise.
5040 (armv7): Make vfpv3-d16 an alias.
5041 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
5043 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
5044 Sort in canonical order.
5045 (armv8-a): Sort in canonical order.
5046 (armv8.1-a, armv8.2-a): Likewise.
5047 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
5049 (cortex-a9): Sort in canonical order.
5050 * config/arm/arm.c (selftests.h): Include it.
5051 (arm_test_cpu_arch_data): New function.
5052 (arm_run_self_tests): New function.
5053 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
5054 (targetm): Move declaration to the end of the file.
5055 * arm-cpu-cdata.h: Regenerated.
5057 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5059 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
5060 call to target_mode_check describing the type of option passed.
5061 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
5062 (arm_target_thumb_only): Use arm_parse_arch_option_name or
5063 arm_parse_cpu_option_name to match parameters against list of
5065 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
5066 arm_arch_core_flags data structure.
5067 * config/arm/arm-cpu_cdata.h: Regenerated.
5069 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5071 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
5073 (arm_print_hint_for_cpu_option): Likewise.
5074 (arm_print_hint_for_arch_option): Likewise.
5075 (arm_parse_cpu_option_name): Likewise.
5076 (arm_parse_arch_option_name): Likewise.
5077 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
5078 of entries in the all_fpus list.
5079 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
5080 (arm_parse_cpu_option_name): Declare.
5081 (arm_parse_arch_option_name): Declare.
5082 (arm_parse_option_features): Declare.
5083 (arm_intialize_isa): Declare.
5084 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
5086 (gen_comm_data): ... here. Make definitions non-static.
5087 * config/arm/arm-cpu-data.h: Regenerated.
5088 * config/arm/arm-cpu-cdata.h: Regenerated.
5090 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5092 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
5093 (cpu_arch_extension): New structure.
5094 (cpu_arch_option, arch_option, cpu_option): New structures.
5095 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
5097 (gen_data): Generate new format data tables.
5098 * config/arm/arm.c (cpu_tune): New structure.
5099 (cpu_option, processors): Delete.
5100 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
5101 (arm_print_hint_for_cpu_option): ... this and ...
5102 (arm_print_hint_for_arch_option): ... this.
5103 (arm_parse_arch_cpu_name): Delete. Replace with ...
5104 (arm_parse_cpu_option_name): ... this and ...
5105 (arm_parse_arch_option_name): ... this.
5106 (arm_unrecognized_feature): Change type of target parameter to
5108 (arm_parse_arch_cpu_features): Delete. Replace with ...
5109 (arm_parse_option_features): ... this.
5110 (arm_configure_build_target): Rework to use new configuration data
5112 (arm_print_tune_info): Rework for new configuration data tables.
5113 * config/arm/arm-cpu-data.h: Regenerated.
5114 * config/arm/arm-cpu.h: Regenerated.
5116 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5118 * Makefile.in (OBJS): Move sbitmap.o from here ...
5119 (OBJS-libcommon): ... to here.
5121 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5123 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
5124 (ISA_ALL_CRYPTO): New macro.
5125 (ISA_ALL_SIMD): New macro
5126 (ISA_ALL_FP): New macro.
5127 * config/arm/arm.c (fpu_bitlist): Update initializer.
5128 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
5130 (arm9e): Add fpu. Add option for nofp
5131 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
5132 (arm926ej-s, arm1026ej-s): Likewise.
5133 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
5134 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
5135 neon-fp16, neon-vfpv4, nofp and nosimd.
5136 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
5137 (cortex-a8): Add fpu. Add option for nofp.
5138 (cortex-a9): Add fpu. Add options for nosimd and nofp.
5139 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
5140 (cortex-r4f): Add fpu.
5141 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
5142 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
5144 (cortex-r8): Likewise.
5145 (cortex-m4): Add fpu. Add option for nofp.
5146 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
5147 (cortex-a17.cortex-a7): Likewise.
5148 (cortex-a32): Add fpu. Add options for crypto and nofp.
5149 (cortex-a35, cortex-a53): Likewise.
5150 (cortex-a57): Add fpu. Add option for crypto.
5151 (cortex-a72, cortex-a73): Likewise.
5152 (exynos-m1): Likewise.
5153 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
5154 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
5155 (cortex-m33): Add fpu. Add option for nofp.
5156 * config/arm/arm-cpu-cdata.h: Regenerated
5157 * config/arm/arm-cpu-data.h: Regenerated.
5159 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5161 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
5162 (armv5te, armv5tej): Likewise.
5163 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
5164 (armv7): Add options fp and vfpv3-d16.
5165 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
5166 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
5168 (armv7ve): Likewise.
5169 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
5170 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
5171 (armv8-a): Add nocrypto option.
5172 (armv8.1-a, armv8.2-a): Likewise.
5173 (armv8-m.main): add options fp, fp.dp and nofp.
5175 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5177 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
5179 (armv8-a+crc): Delete.
5180 (armv8.1-a): Add options simd, crypto and nofp.
5181 (armv8.2-a): Add options fp16, simd, crypto and nofp.
5182 (armv8.2-a+fp16): Delete.
5183 (armv8-m.main): Add option dsp.
5184 (armv8-m.main+dsp): Delete.
5185 (cortex-a8): Add fpu. Add nofp option.
5186 (cortex-a9): Add fpu. Add nofp and nosimd options.
5187 * config/arm/parsecpu.awk (gen_data): Generate option tables and
5188 link to main cpu and architecture data structures.
5189 (gen_comm_data): Only put isa attributes from the main architecture
5191 (option): New statement for architecture and CPU entries.
5192 * arm.c (struct cpu_option): New structure.
5193 (struct processors): Add entry for options.
5194 (arm_unrecognized_feature): New function.
5195 (arm_parse_arch_cpu_name): Ignore any characters after the first
5197 (arm_parse_arch_cpu_feature): New function.
5198 (arm_configure_build_target): Separate out any CPU and architecture
5199 features and parse separately. Don't error out if -mfpu=auto is
5200 used with only an architecture string.
5201 (arm_print_asm_arch_directives): New function.
5202 (arm_file_start): Call it.
5203 * config/arm/arm-cpu-cdata.h: Regenerated.
5204 * config/arm/arm-cpu-data.h: Likewise.
5205 * config/arm/arm-tables.opt: Likewise.
5207 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5209 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
5210 assembler when it is not -mfpu=auto.
5212 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5214 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
5215 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
5216 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
5217 (ASM_CPU_SPEC): Rewrite.
5218 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
5219 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
5220 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
5221 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
5222 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
5223 copied string is NUL-terminated. Also strip any characters prefixed
5225 (arm_rewrite_selected_arch): New function.
5226 (arm_rewrite_march): New function.
5228 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5230 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
5231 (x_arm_cpu_string, x_arm_tune_string): Likewise.
5232 (march, mcpu, mtune): Convert to string-based options.
5233 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
5234 (arm_parse_arch_cpu_name): New function.
5235 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
5236 identify selected architecture or CPU.
5237 (arm_option_save): New function.
5238 (TARGET_OPTION_SAVE): Redefine.
5239 (arm_option_restore): Restore string options.
5240 (arm_option_print): Print string options.
5242 2017-06-16 Martin Sebor <msebor@redhat.com>
5244 PR tree-optimization/80933
5245 PR tree-optimization/80934
5246 * builtins.c (fold_builtin_3): Do not handle bcmp here.
5247 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
5248 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
5249 (gimple_fold_builtin): Call them.
5251 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5253 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
5254 as unlikely; update profile.
5256 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5258 * predict.c (force_edge_cold): Handle declaring edges impossible
5261 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5263 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
5265 (try_unroll_loop_completely): Fix reporting.
5267 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5269 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
5271 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
5274 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
5275 if given a non-constant argument for an intrinsic which requires a
5278 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5280 * profile.c (compare_freqs): New function.
5281 (branch_prob): Sort edge list.
5282 (find_spanning_tree): Assume that the list is priority sorted.
5284 2017-06-16 Richard Biener <rguenther@suse.de>
5286 PR tree-optimization/81090
5287 * passes.def (pass_record_bounds): Remove.
5288 * tree-pass.h (make_pass_record_bounds): Likewise.
5289 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
5290 make_pass_record_bounds): Likewise.
5291 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
5292 not free niter estimates at the beginning but at the end.
5293 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
5295 2017-06-16 Richard Biener <rguenther@suse.de>
5297 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
5298 initializer to workaround ICE in host GCC 4.8.
5300 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5302 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
5304 (clone_inlined_nodes): Update.
5306 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5308 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
5309 prefetch settings, and enable prefetching by default at -O3.
5311 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5313 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5314 Set flag_prefetch_loop_arrays according to tuning data.
5316 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5318 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
5320 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
5321 [Unrelated to main purpose of the patch] Place the pointer field last
5322 to enable type checking errors when tune structure are wrongly merged.
5323 * config/aarch64/aarch64.c (generic_prefetch_tune,)
5324 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
5325 (thunderx2t99_prefetch_tune): New tune constants.
5326 (tune_params *_tunings): Update all tunings (no functional change).
5327 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
5328 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
5329 from tunings structures.
5331 2017-06-16 Jakub Jelinek <jakub@redhat.com>
5334 * ubsan.c (instrument_null): Add T argument, use it instead
5335 of computing it based on IS_LHS.
5336 (instrument_object_size): Likewise.
5337 (pass_ubsan::execute): Adjust instrument_null and
5338 instrument_object_size callers to pass gimple_get_lhs or
5339 gimple_assign_rhs1 result to it. Use instrument_null instead of
5340 calling get_base_address and instrument_mem_ref. Handle
5341 aggregate call arguments for object-size sanitization.
5343 2017-06-16 Yury Gribov <tetra2005@gmail.com>
5345 PR tree-optimization/81089
5346 * tree-vrp.c (is_masked_range_test): Validate operands of
5349 2017-06-15 Martin Sebor <msebor@redhat.com>
5352 * dumpfile.c (dump_register): Avoid calling memset to initialize
5353 a class with a default ctor.
5354 * gcc.c (struct compiler): Remove const qualification.
5355 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
5356 * hash-table.h: Ditto.
5357 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
5359 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
5360 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
5362 * params.h (struct param_info): Make struct members non-const.
5363 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
5364 with default initialization.
5365 * vec.h (vec_copy_construct, vec_default_construct): New helper
5367 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
5368 with vec_copy_construct.
5369 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
5370 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
5371 * doc/invoke.texi (-Wclass-memaccess): Document.
5373 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5375 * emit-rtl.h (is_leaf): Update comment about local
5378 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5381 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
5382 for a variable to have a section before checking if the section has a
5384 Set section to.persistent if persistent attribute is set.
5385 Warn if .persistent attribute is used on an automatic variable.
5387 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
5389 PR rtl-optimization/80474
5390 * reorg.c (update_block): Do not ignore instructions in a delay slot.
5392 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
5394 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
5397 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
5399 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
5400 (casesi): Emit bounds checking as RTL.
5401 (casesi_internal_mips16_<mode>): Remove bounds checking.
5403 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
5405 * config/xtensa/xtensa.c (xtensa_option_override): Append
5406 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
5407 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
5408 xtensa_doloop_hooks): Define unconditionally.
5409 (xtensa_reorg_loops): Only call reorg_loops in the presence of
5411 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
5412 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
5413 for it in xtensa_option_override.
5414 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
5415 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
5417 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
5419 * doc/cppopts.texi: Document '-' special value to -MF.
5421 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
5423 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
5424 (cortex_a53_fconst): Likewise.
5425 (cortex_a53_fpmul): Likewise.
5426 (cortex_a53_f_load_64): Likewise.
5427 (cortex_a53_f_load_many): Likewise.
5428 (cortex_a53_advsimd_alu): Likewise.
5429 (cortex_a53_advsimd_alu_q): Likewise.
5430 (cortex_a53_advsimd_mul): Likewise.
5431 (cortex_a53_advsimd_mul_q): Likewise.
5432 (fpmac bypass): Add new bypass for fpmac-fpmac case.
5433 Add missing fmul, r2f_cvt and fconst cases.
5435 2017-06-14 Richard Biener <rguenther@suse.de>
5438 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
5440 (fold_binary_loc): When associating do not treat pre-existing
5441 TREE_OVERFLOW on literal constants as a reason to allow
5442 TREE_OVERFLOW on associated literal constants.
5444 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5446 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
5447 (MASK_FEATURES): New macro.
5448 * config/sparc/sparc.c (sparc_option_override): Remove the special
5449 handling of -mfpu and generalize it to all MASK_FEATURES switches.
5451 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5453 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
5454 a division of 0 if non-call exceptions are enabled.
5456 2017-06-14 Andrew Pinski <apinski@cavium.com>
5457 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5460 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
5461 Improve vector initialization code gen for only variable case.
5463 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5465 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
5467 2017-06-14 Richard Biener <rguenther@suse.de>
5469 PR tree-optimization/81083
5470 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
5473 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5475 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
5476 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
5477 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
5478 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
5479 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
5480 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
5482 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5484 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
5485 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
5487 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5489 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
5491 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5493 * config/rs6000/t-rtems: Don't handle SPE.
5495 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5497 * config/rs6000/t-linux: Don't handle SPE.
5499 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5501 * config/rs6000/eabispe.h: Delete file.
5503 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5505 * config/rs6000/t-spe: Delete file.
5507 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5509 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
5510 (rs6000_legitimate_offset_address_p): Return false for anything in
5511 V2SImode or V2SFmode.
5513 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5515 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
5516 except V2SF and V2SI. Rearrange the vector modes, and add comments.
5517 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
5519 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
5520 (rs6000_legitimate_offset_address_p): Ditto.
5521 (rs6000_emit_move): Ditto.
5522 (rs6000_init_builtins): Remove V4HI_type_node.
5524 2017-06-13 Martin Liska <mliska@suse.cz>
5527 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
5528 (gate_asan): Likewise.
5529 * asan.h (asan_no_sanitize_address_p): Remove the function.
5530 (sanitize_flags_p): New function.
5531 * builtins.def: Fix coding style.
5532 * common.opt: Use renamed enum value.
5533 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
5534 * doc/extend.texi: Document no_sanitize attribute.
5535 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
5536 to SANITIZE_UNDEFINED_NONDEFAULT.
5537 * gcc.c (sanitize_spec_function): Use the renamed enum value.
5538 * gimple-fold.c (optimize_atomic_compare_exchange_p):
5539 Use sanitize_flags_p.
5540 * gimplify.c (gimplify_function_tree): Likewise.
5541 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
5542 * opts.c (parse_no_sanitize_attribute): New function.
5543 (common_handle_option): Use renamed enum value.
5544 * opts.h (parse_no_sanitize_attribute): Declare.
5545 * tree.c (sanitize_flags_p): New function.
5546 * tree.h: Declared here.
5547 * tsan.c: Use sanitize_flags_p.
5548 * ubsan.c (ubsan_expand_null_ifn): Likewise.
5549 (instrument_mem_ref): Likewise.
5550 (instrument_bool_enum_load): Likewise.
5551 (do_ubsan_in_current_function): Remove the function.
5552 (pass_ubsan::execute): Use sanitize_flags_p.
5553 * ubsan.h: Remove do_ubsan_in_current_function
5554 * tree-cfg.c (print_no_sanitize_attr_value): New function.
5555 (dump_function_to_file): Use it here.
5557 2017-06-13 Martin Jambor <mjambor@suse.cz>
5559 PR tree-optimization/80803
5560 PR tree-optimization/81063
5561 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
5562 (propagate_subaccesses_across_link): Enqueue subtree whenever
5563 necessary instead of relying on the caller.
5565 2017-06-13 Martin Jambor <mjambor@suse.cz>
5567 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
5568 that have a first_link.
5569 (sort_and_splice_var_accesses): Do not check first_link before
5571 (subtree_mark_written_and_enqueue): Likewise.
5572 (propagate_all_subaccesses): Likewise and do not stop at first
5573 parent with a first_link.
5575 2017-06-13 Martin Jambor <mjambor@suse.cz>
5577 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
5580 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5582 * match.pd: New pattern.
5584 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5586 * tree-vrp.c (is_masked_range_test): New function.
5587 (register_edge_assert_for): Determine ranges for
5590 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5592 PR tree-optimization/67328
5593 * fold-const.c (maskable_range_p): New function.
5594 (build_range_check): Generate bittests if possible.
5596 2017-06-13 Martin Liska <mliska@suse.cz>
5598 * gimple-pretty-print.c (dump_probability): Add new argument.
5599 (dump_edge_probability): Dump both probability and count.
5600 (dump_gimple_label): Likewise.
5601 (dump_gimple_bb_header): Likewise.
5603 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
5606 * config/avr/avr-devices.c: Fix indentation.
5607 * config/avr/gen-avr-mmcu-specs.c: Dito.
5609 2017-06-13 Richard Biener <rguenther@suse.de>
5611 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
5612 instead get vector type from stmt_info.
5613 (vectorizable_reduction): Adjust. Remove dead code.
5615 2017-06-13 Richard Biener <rguenther@suse.de>
5618 * fold-const.c (extract_muldiv_1): Remove bogus distribution
5619 case of C * (x * C2 + C3).
5620 (fold_addr_of_array_ref_difference): Properly fold index difference.
5622 2017-06-12 David S. Miller <davem@davemloft.net>
5625 * config/sparc/sparc.md (return expander): Emit frame blockage if
5626 function uses alloca.
5628 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
5630 * combine.c (make_field_assignment): Check len rather than the mode
5631 precision when calling force_to_mode.
5633 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
5635 Support multilibs and devices that see flash in RAM address range.
5638 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
5639 (avr_mcu_t) <flash_pm_offset>: New field.
5640 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
5641 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
5642 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
5643 (AVR_TINY_PM_OFFSET): Remove macro.
5644 * config/avr/avr.opt (-mshort-calls): New option.
5645 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
5646 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
5647 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
5648 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
5649 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
5650 instead of avr_arch->have_jmp_call.
5651 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
5652 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
5653 avr_arch->flash_pm_offset to define.
5654 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
5655 new field flash_pm_offset. Add entry for avrxmega3.
5656 (avr_texinfo): Add entry for avrxmega3.
5657 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
5658 attiny212, attiny214,
5659 attiny412, attiny414, attiny416, attiny417,
5660 attiny814, attiny816, attiny817,
5661 attiny1614, attiny1616, attiny1617,
5662 attiny3214, attiny3216, attiny3217.
5663 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
5664 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
5665 (avr_print_operand_address) [AVR_TINY]: Same.
5666 (avr_asm_init_sections) <readonly_data_section>: Only patch
5667 callback if avr_arch->flash_pm_offset = 0.
5668 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
5669 for rodata if avr_arch->flash_pm_offset != 0.
5670 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
5671 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
5672 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
5673 (m_options): Append opt_rcall.
5674 (m_dirnames): Append dir_rcall.
5675 * config/avr/t-multilib: Regenerate.
5677 * configure.ac [target=avr]: Check whether avrxmega3 default
5678 linker description file works as needed.
5679 * configure: Regenerate.
5680 * doc/avr-mmcu.texi: Regenerate.
5681 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
5682 <__AVR_ARCH__>: Document avrxmega3 and 103.
5683 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
5684 <__AVR_SHORT_CALLS__>: Document it.
5685 <__AVR_PM_BASE_ADDRESS__>: Document it.
5686 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
5687 (AVR Variable Attributes) <progmem>: Document this is
5688 not needed for avrxmega3.
5689 (AVR Named Address Spaces) <__flash>: Dito.
5691 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
5693 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
5695 2017-06-12 Doug Rupp <rupp@adacore.com>
5697 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
5698 Append vxworks-stdint.h to the tm_file list.
5699 * config/vxworks-stdint.h: New file.
5701 2017-06-12 Martin Liska <mliska@suse.cz>
5703 PR tree-optimization/81041
5704 * tree-profile.c (gimple_gen_ic_func_profiler):
5705 Create an extra BB in profile-generate
5706 (gimple_gen_time_profiler): Likewise.
5708 2017-06-12 Jakub Jelinek <jakub@redhat.com>
5710 PR tree-optimization/81003
5711 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
5712 (update_range_test): Use it instead of force_gimple_operand_gsi.
5714 2017-06-12 Richard Biener <rguenther@suse.de>
5716 PR tree-optimization/81053
5717 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
5718 with backedge value not defined in loop. Simplify def stmt
5721 2017-06-11 Tom de Vries <tom@codesourcery.com>
5724 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
5726 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
5727 nvptx_cannot_force_const_mem.
5729 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5731 * opts.c (finish_options): Move test for flag_split_stack after
5732 it has been initialized.
5734 2017-06-11 Jason Merrill <jason@redhat.com>
5736 * tree.h (id_equal): New.
5737 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
5738 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
5739 instead of strcmp of IDENTIFIER_POINTER.
5741 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5743 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
5744 (mark_all_inlined_calls_cdtor): Fix formating.
5745 (inline_transform): Rescale profile before inlining.
5747 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5749 * cgraph.h (cgraph_edge::clone): Update prototype.
5750 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
5751 (cgraph_node::create_clone): Update.
5752 (cgraph_node::create_version_clone): Update.
5753 * tree-inline.c (copy_bb): Update.
5754 (expand_call_inline): Update.
5756 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5758 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
5759 factored out from ...
5760 (rs6000_emit_prologue): ... here.
5762 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5764 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
5765 factored out from ...
5766 (rs6000_emit_prologue): ... here.
5768 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5770 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
5772 (handle_missing_profiles): Fix computation of tp_first_run.
5773 (counts_to_freqs): Do not touch freqs when count is 0.
5775 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5777 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
5780 2017-06-10 Tom de Vries <tom@codesourcery.com>
5782 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
5783 attributes): Document signal effective target.
5785 2017-06-10 Tom de Vries <tom@codesourcery.com>
5787 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
5788 Document effective target stack_size.
5790 2017-06-09 David Malcolm <dmalcolm@redhat.com>
5792 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
5793 to the edit_context if they can be auto-applied.
5795 2017-06-9 Ian Lance Taylor <iant@golang.org>
5797 * opts.c (finish_options): If -fsplit-stack, disable implicit
5798 -forder-blocks-and-partition.
5799 * doc/invoke.texi (Optimize Options): Document that when using
5800 -fsplit-stack -forder-blocks-and-partition is not implicitly
5803 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5805 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
5806 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
5807 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
5808 * builtins.def (abort, trap, unreachable): Declare cold.
5809 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
5810 * tree-core.h (ECF_COLD): New.
5811 * tree.c (set_call_expr_flags): Handle ECF_COLD.
5812 (build_common_builtin_nodes): Mark unreachable and abort as cold.
5814 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5816 * predict.c (unlikely_executed_stmt_p): Cleanup.
5818 2017-06-09 Richard Biener <rguenther@suse.de>
5820 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
5821 model if the ref is always written to.
5823 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5825 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
5827 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5829 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
5832 2017-06-09 Tom de Vries <tom@codesourcery.com>
5835 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
5836 "target cannot support label values" when encountering LABEL_REF.
5838 2017-06-09 Martin Liska <mliska@suse.cz>
5840 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
5841 (gimple_gen_ic_func_profiler): Emit direct comparison
5842 of __gcov_indirect_call_callee with NULL.
5843 (gimple_gen_time_profiler): Change probability from
5844 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
5846 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5848 * profile.c (edge_gcov_counts): Turn to pointer.
5849 (compute_branch_probabilities, compute_branch_probabilities): Update.
5850 (branch_prob): Do not clear edge_gcov_count.
5851 * profile.h (edge_gcov_counts): Turn to pointer.
5852 (edge_gcov_count): Update.
5854 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5856 * gimple.h (gimple_check_failed): Mark cold.
5858 2017-06-09 Richard Biener <rguenther@suse.de>
5860 PR tree-optimization/66623
5861 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
5862 refactor check_reduction into two parts, properly computing
5863 whether we have to check reduction validity for outer loop
5866 2017-06-09 Richard Biener <rguenther@suse.de>
5868 PR tree-optimization/79483
5869 * graphite-scop-detection.c (order): New global.
5870 (get_order): Compute bb to order mapping that satisfies code
5871 generation constraints.
5872 (cmp_pbbs): New helper.
5873 (build_scops): Start domwalk at entry block, sort generated
5876 2017-06-09 Richard Biener <rguenther@suse.de>
5879 * ipa-polymorphic-call.c
5880 (ipa_polymorphic_call_context::restrict_to_inner_class):
5881 Skip FIELD_DECLs with error_mark_node type.
5882 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
5885 2017-06-09 Martin Liska <mliska@suse.cz>
5887 * predict.c (struct branch_predictor): New struct.
5888 (test_prediction_value_range): New test.
5889 (predict_c_tests): New function.
5890 * selftest-run-tests.c (selftest::run_tests): Run the function.
5891 * selftest.h: Declare new tests.
5893 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
5896 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
5897 gen_add3_insn did not fail.
5898 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
5899 r0, construct that number in a temporary reg and add that reg to r0.
5900 If asked to put the result in r0 as well, fail.
5902 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
5904 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5905 for early expansion of vec_eqv.
5907 2017-06-08 Jakub Jelinek <jakub@redhat.com>
5910 * ubsan.c (instrument_null): Avoid pointless code temporary.
5911 (pass_ubsan::execute): Instrument aggregate arguments of calls.
5913 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
5917 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
5920 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
5923 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5925 * predict.c (unlikely_executed_edge_p): Move ahead.
5926 (probably_never_executed_edge_p): Use it.
5928 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5931 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
5932 gimple_call_builtin_p call.
5934 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5936 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
5937 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
5938 rtl_check_failed_type2, rtl_check_failed_code1,
5939 rtl_check_failed_code2, rtl_check_failed_code_mode,
5940 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
5941 rtvec_check_failed_bounds, rtl_check_failed_flag,
5942 _fatal_insn_not_found, _fatal_insn): Likewise.
5943 * tree.h (tree_contains_struct_check_failed,
5944 tree_check_failed, tree_not_check_failed,
5945 tree_class_check_failed, tree_range_check_failed,
5946 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
5947 tree_vec_elt_check_failed, phi_node_elt_check_failed,
5948 tree_operand_check_failed, omp_clause_check_failed,
5949 omp_clause_operand_check_failed, omp_clause_range_check_failed):
5952 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5954 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
5955 flag_branch_probabilities.
5956 * ipa-inline.c (edge_badness): Likewise.
5957 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
5958 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
5959 * predict.c (maybe_hot_frequency_p): Likewise.
5960 (probably_never_executed): Likewise.
5961 * sched-ebb.c (schedule_ebbs): Likewise.
5962 * sched-rgn.c (find_single_block_region): Likewise.
5963 * tracer.c (tail_duplicate): Likewise.
5965 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5967 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
5968 longer requires x_flag_profile_use.
5970 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5972 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
5973 instead of flag_reorder_blocks_and_partition.
5974 * dbxout.c (dbxout_function_end): Likewise.
5975 * dwarf2out.c (gen_subprogram_die): Likewise.
5976 * haifa-sched.c (sched_create_recovery_edges): Likewise.
5977 * hw-doloop.c (reorg_loops): Likewise.
5978 * varasm.c (assemble_start_function,
5979 assemble_end_function): Likewise.
5980 (decide_function_section): Do not check for
5981 flag_reorder_blocks_and_partition.
5983 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5985 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
5987 (chkp_get_hard_register_fake_addr_expr): Ditto.
5988 (chkp_build_addr_expr): Add check for hard reg case.
5989 (chkp_parse_array_and_component_ref): Ditto.
5990 (chkp_find_bounds_1): Ditto.
5991 (chkp_process_stmt): Don't generate bounds store for
5994 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5996 * predict.c (maybe_hot_bb_p): Do not check profile status.
5997 (maybe_hot_edge_p): Likewise.
5998 (probably_never_executed): Check for zero counts even if profile
6000 (unlikely_executed_edge_p): New function.
6001 (unlikely_executed_stmt_p): New function.
6002 (unlikely_executed_bb_p): New function.
6003 (set_even_probabilities): Use unlikely predicates.
6004 (combine_predictions_for_bb): Likewise.
6005 (predict_paths_for_bb): Likewise.
6006 (predict_paths_leading_to_edge): Likewise.
6007 (determine_unlikely_bbs): New function.
6008 (estimate_bb_frequencies): Use it.
6009 (compute_function_frequency): Use zero counts even if profile is
6011 * profile-count.h: Fix typo.
6013 2017-08-08 Julia Koval <julia.koval@intel.com>
6015 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
6016 _mm512_mask_cvtsepi16_storeu_epi8,
6017 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
6018 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
6019 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
6020 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
6021 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
6022 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
6023 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
6024 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
6025 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
6026 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
6027 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
6028 __builtin_ia32_pmovuswb256mem_mask,
6029 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
6030 __builtin_ia32_pmovwb512mem_mask): New builtins.
6032 2017-08-08 Julia Koval <julia.koval@intel.com>
6034 PR target/73350,80862
6035 * config/i386/subst.md (round): Fix round pattern.
6036 * config/i386/i386.c (ix86_erase_embedded_rounding):
6037 Fix erasing rounding for the fixed pattern.
6039 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6041 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
6043 2017-06-08 Martin Liska <mliska@suse.cz>
6045 PR gcov-profile/80911
6046 * gcov.c (block_info::block_info): New constructor.
6048 2017-06-07 Carl Love <cel@us.ibm.com>
6050 * config/rs6000/rs6000-c: The return type of the following
6051 built-in functions was implemented as int not long long. Fix sign
6052 of return value for the unsigned version of vec_mulo and vec_mule.
6053 vector unsigned long long vec_bperm (vector unsigned long long,
6054 vector unsigned char)
6055 vector signed long long vec_mule (vector signed int,
6057 vector unsigned long long vec_mule (vector unsigned int,
6058 vector unsigned int)
6059 vector signed long long vec_mulo (vector signed int,
6061 vector unsigned long long vec_mulo (vector unsigned int,
6062 vector unsigned int)
6063 * doc/extend.texi: Fix the documentation for the built-in
6066 2017-06-07 Carl Love <cel@us.ibm.com>
6069 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
6072 2017-06-07 Carl Love <cel@us.ibm.com>
6074 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
6075 support, Generate doublehv for signed int/float for BE case only.
6077 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
6079 * doc/invoke.texi (mcx16): Rewrite.
6081 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6083 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
6084 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
6085 *mov<mode>_softfloat, and an anonymous splitter): Use
6086 nonimmediate_operand instead of rs6000_nonimmediate_operand.
6088 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6090 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
6092 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
6093 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
6094 (rs6000_debug_reg_global): Adjust.
6095 (rs6000_init_hard_regno_mode_ok): Adjust.
6096 (rs6000_dbx_register_number): Adjust.
6097 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
6098 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
6099 Remove SPE_ACC and SPEFSCR.
6100 (REG_ALLOC_ORDER): Ditto.
6101 (FRAME_POINTER_REGNUM): Change to 111.
6102 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
6103 (REG_CLASS_NAMES): Ditto.
6104 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
6105 (REGISTER_NAMES): Ditto.
6106 (ADDITIONAL_REG_NAMES): Ditto.
6107 (rs6000_reg_names): Ditto.
6108 * config/rs6000/rs6000.md: Renumber some register number
6111 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6113 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
6115 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
6116 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
6118 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
6119 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
6120 Delete the SPE high registers.
6121 (REG_ALLOC_ORDER): Ditto.
6122 (enum reg_class): Remove SPE_HIGH_REGS.
6123 (REG_CLASS_NAMES): Ditto.
6124 (REG_CLASS_CONTENTS): Delete the SPE high registers.
6125 (REGISTER_NAMES): Ditto.
6126 (rs6000_reg_names): Ditto.
6127 * doc/tm.texi.in: Remove SPE as example.
6128 * doc/tm.texi: Regenerate.
6130 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6132 * config/rs6000/8540.md (ppc8540_brinc): Delete.
6133 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
6134 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
6135 * config/rs6000/rs6000.md (type): Remove "brinc".
6137 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6139 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
6140 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
6141 * config/rs6000/linuxspe.h: Delete file.
6142 * config/rs6000/rs6000.md: Don't include spe.md.
6143 * config/rs6000/spe.h: Delete file.
6144 * config/rs6000/spe.md: Delete file.
6145 * config/rs6000/t-rs6000: Remove spe.md.
6147 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6149 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
6150 (reg_or_none500mem_operand): Delete.
6151 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
6152 instead of reg_or_none500mem_operand.
6154 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6156 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
6157 handling of SPE flags.
6158 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
6160 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6162 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
6164 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
6165 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
6166 paired_divv2sf3): Similar.
6167 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
6168 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
6169 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
6171 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
6172 Rename the paired_* instruction patterns.
6173 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
6175 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
6176 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
6177 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
6178 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
6179 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
6181 (struct machine_function): Delete field spe_insn_chain_scanned_p.
6182 (spe_func_has_64bit_regs_p): Delete.
6183 (spe_expand_predicate_builtin): Delete.
6184 (spe_expand_evsel_builtin): Delete.
6185 (TARGET_DWARF_REGISTER_SPAN): Do not define.
6186 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
6187 (invalid_e500_subreg): Delete.
6188 (rs6000_legitimize_address): Always force_reg op2 as well, for
6189 paired single memory accesses.
6190 (rs6000_member_type_forces_blk): Delete.
6191 (rs6000_spe_function_arg): Delete.
6192 (rs6000_expand_unop_builtin): Delete SPE handling.
6193 (rs6000_expand_binop_builtin): Ditto.
6194 (spe_expand_stv_builtin): Delete.
6195 (bdesc_2arg_spe): Delete.
6196 (spe_expand_builtin): Delete.
6197 (spe_expand_predicate_builtin): Delete.
6198 (spe_expand_evsel_builtin): Delete.
6199 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
6200 (spe_init_builtins): Delete.
6201 (spe_func_has_64bit_regs_p): Delete.
6202 (savres_routine_name): Delete "info" parameter. Adjust callers.
6203 (rs6000_emit_stack_reset): Ditto.
6204 (rs6000_dwarf_register_span): Delete.
6205 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
6206 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
6207 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
6209 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
6211 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
6212 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
6213 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
6214 mulv2sf3, divv2sf3): Delete expanders.
6216 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6218 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
6220 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6222 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
6223 * config/rs6000/rs6000.c: Ditto.
6225 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6227 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
6228 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
6229 comparison_operator.
6231 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6233 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
6234 * config/rs6000/rs6000.opt: Ditto.
6235 * config/rs6000/t-rtems: Ditto.
6237 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6239 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
6240 TARGET_E500_SINGLE by 0, simplify.
6241 * config/rs6000/rs6000.c: Ditto.
6242 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
6243 (spe_build_register_parallel): Delete.
6244 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
6245 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
6246 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
6247 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
6248 (E500_CONVERT): Delete.
6249 * config/rs6000/spe.md: Remove many patterns and all define_constants.
6251 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6253 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
6254 * config/rs6000/dfp.md: Ditto.
6255 (negdd2, *negdd2_fpr): Merge.
6256 (absdd2, *absdd2_fpr): Merge.
6257 (negtd2, *negtd2_fpr): Merge.
6258 (abstd2, *abstd2_fpr): Merge.
6259 * config/rs6000/e500.h: Delete file.
6260 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
6261 TARGET_FPRS by 1 and simplify.
6262 * config/rs6000/rs6000-c.c: Ditto.
6263 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
6265 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
6267 * config/rs6000/rs6000.md: Ditto.
6268 (floatdidf2, *floatdidf2_fpr): Merge.
6269 (move_from_CR_gt_bit): Delete.
6270 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
6271 (E500_CR_IOR_COMPARE): Delete.
6272 (All patterns that require !TARGET_FPRS): Delete.
6273 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
6275 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6277 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
6279 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6281 * graphds.c (add_edge): Intitialize edge's attached data.
6282 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
6283 pointer parameter. Call pointed function on each edge during
6284 graph traversing. Skip traversing the edge when the function
6286 (graphds_dfs, graphds_scc): Ditto.
6287 (for_each_edge): New parameter. Pass the new parameter to callback
6289 * graphds.h (skip_edge_callback): New function pointer type.
6290 (graphds_dfs, graphds_scc): New function pointer parameter.
6291 (graphds_edge_callback, for_each_edge): New parameter.
6293 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6295 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
6296 out code checking if runtime alias check is possible to below ...
6297 Call the new function.
6298 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
6299 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
6301 2017-06-07 Marek Polacek <polacek@redhat.com>
6304 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
6305 TYPE_OVERFLOW_WRAPS check.
6307 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6309 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
6310 if versioning is required.
6311 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
6312 peeling with the check for versioning.
6314 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6316 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
6317 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
6318 Set true to new parameter if new ssa variable is defined.
6319 (vect_gen_vector_loop_niters): Refactor. Set range information
6320 for the new vector loop bound variable.
6321 (vect_do_peeling): Ditto.
6323 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6325 * tree-affine.c (ssa.h): Include header file.
6326 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
6327 has wrapping overflow behavior.
6329 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6331 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
6333 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6335 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
6336 (tree_to_aff_combination): ... here.
6338 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6340 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
6341 reg_pressure model function.
6342 (ivopts_global_cost_for_size): Delete.
6343 (determine_set_costs, iv_ca_recount_cost): Call new model function
6344 ivopts_estimate_reg_pressure.
6346 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6348 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
6349 expensive than udiv. Remove floating point cases from mod.
6351 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6353 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
6356 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6358 * config/aarch64/aarch64.md
6359 (copysignsf3): Fix mask generation.
6361 2017-06-07 Jakub Jelinek <jakub@redhat.com>
6363 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
6365 (class dump_manager): Add register_dumps method.
6366 * dumpfile.c: Include langhooks.h.
6367 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
6368 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
6369 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
6370 (dump_manager::dump_register): Start with 512 entries instead of 32.
6371 (dump_manager::register_dumps): New method.
6372 * toplev.c (general_init): Instead of invoking register_dumps
6373 langhook, invoke register_dumps method on the dump manager.
6374 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
6377 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
6379 * doc/md.texi: Clarify the restrictions on a define_insn condition.
6380 Say that # requires an associated define_split to exist, and that
6381 the define_split must be suitable for use after register allocation.
6383 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6385 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
6386 (compute_outgoing_frequencies): Also initialize zero counts.
6387 (find_many_sub_basic_blocks): Do not produce uninitialized profile
6388 around loops; preserve more of profile when nothing changes.
6390 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
6392 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
6394 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
6395 * config/arm/arm-cpu-cdata.h: Regenerate.
6396 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
6397 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
6398 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
6399 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
6400 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
6402 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
6404 * config/arm/t-rmprofile: Likewise.
6405 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
6407 2017-06-06 David S. Miller <davem@davemloft.net>
6410 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
6411 blockage if function uses alloca.
6413 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
6415 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
6416 New "uid" fields to hold pretty-print IDs of group and ref.
6417 Memory references are now identified as <group_id>:<ref_id>
6418 instead of using [random] addresses.
6419 (dump_mem_details): Simplify, no functional change.
6420 (dump_mem_ref): Simplify and make output more concise.
6421 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
6422 (find_or_create_group): Initialize group uid.
6423 (record_ref): Initialize ref uid. Improve debug output.
6424 (prune_group_by_reuse, should_issue_prefetch_p,)
6425 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
6426 (mark_nontemporal_store, determine_loop_nest_reuse):
6427 Improve debug output.
6429 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
6431 * dbgcnt.def (prefetch): New debug counter.
6432 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
6433 (schedule_prefetches): Stop issueing prefetches if debug counter
6436 2017-06-06 Tom de Vries <tom@codesourcery.com>
6438 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
6439 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
6441 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6443 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
6444 Use aarch64_reg_or_zero predicate for operand 4.
6445 (aarch64_compare_and_swap<mode> define_insn_and_split):
6446 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
6447 (aarch64_store_exclusive<mode>): Likewise for operand 2.
6449 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
6451 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
6452 (arm_compute_save_core_reg_mask): This.
6453 (thumb1_compute_save_reg_mask): Rename into ...
6454 (thumb1_compute_save_core_reg_mask): This.
6455 (arm_compute_save_reg0_reg12_mask): Adapt comment.
6456 (arm_compute_frame_layout): Likewise.
6458 2017-06-06 Richard Biener <rguenther@suse.de>
6460 PR tree-optimization/80974
6461 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
6462 keep or clear leaders SSA info.
6464 2017-06-06 Tom de Vries <tom@codesourcery.com>
6466 * config/nvptx/nvptx.c (split_mode_p): New function.
6467 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
6469 2017-06-06 Tom de Vries <tom@codesourcery.com>
6471 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
6473 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6476 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
6479 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6481 * shrink-wrap.c (handle_simple_exit): Update profile.
6482 (try_shrink_wrapping): Upate profile.
6484 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6486 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
6487 (tree_guess_outgoing_edge_probabilities): New.
6488 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
6489 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
6491 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6493 * ipa-split.c (split_function): Initialize return bb profile.
6495 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6497 * profile.c (compute_branch_probabilities): Also initialize
6500 2017-06-06 Richard Biener <rguenther@suse.de>
6502 PR tree-optimization/80928
6503 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
6504 (vect_analyze_loop_operations): Properly guard analysis for
6506 (vect_transform_loop): Likewise.
6507 (vect_analyze_loop_2): Also reset SLP type on PHIs.
6508 (vect_model_induction_cost): Do not cost for pure SLP.
6509 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
6510 of induction in inner loop vectorization.
6511 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
6512 (vect_get_and_check_slp_defs): Handle vect_induction_def.
6513 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
6515 (vect_analyze_slp_cost_1): Cost induction.
6516 (vect_detect_hybrid_slp_stmts): Handle PHIs.
6517 (vect_get_slp_vect_defs): Likewise.
6518 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
6519 (vect_transform_stmt): Handle SLP reductions.
6520 * tree-vectorizer.h (vectorizable_induction): Adjust.
6522 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6524 * config/rs6000/rs6000.c (make_resolver_func): Update
6525 init_lowered_empty_function call.
6527 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
6529 * doc/invoke.texi: Document the -fprofile-abs-path option.
6530 * common.opt (fprofile-abs-path): New option.
6531 * gcov-io.h (gcov_write_filename): Declare.
6532 * gcov-io.c (gcov_write_filename): New function.
6533 * coverage.c (coverage_begin_function): Use gcov_write_filename.
6534 * profile.c (output_location): Likewise.
6536 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6538 * shring-wrap.c: Revert accidental commit.
6540 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
6542 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
6544 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6546 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
6548 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
6549 profile in callgraph edge.
6550 * profile-count.h (apply_probability): If THIS is 0, then result is 0
6551 (apply_scale): Likewise.
6552 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
6553 Also scale profile when inlining function with zero profile.
6554 (initialize_cfun): Update exit block profile even when it is zero.
6555 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
6556 when profile is read.
6558 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6560 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
6561 (CLONE_*): New constants to define the processors we can generate
6562 code for with the target_clone attribute.
6563 (rs6000_clone_map): New array to identify which clone processors
6564 the current program is running on.
6565 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
6566 target_clone attribute.
6567 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
6568 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
6569 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
6570 (cpu_expand_builtin): Add support for target_clone attribute.
6571 (rs6000_valid_attribute_p): Allow "default" attribute.
6572 (get_decl_name): New debug function to simplify printing the
6573 current function name in debugging statements.
6574 (rs6000_clone_priority): New functions to support the target_clone
6575 attribute, and be able to generate code to switch between ISA 2.05
6576 through ISA 3.0 (power6 through power9).
6577 (rs6000_compare_version_priority): Likewise.
6578 (rs6000_get_function_versions_dispatcher): Likewise.
6579 (make_resolver_func): Likewise.
6580 (add_condition_to_bb): Likewise.
6581 (dispatch_function_versions): Likewise.
6582 (rs6000_generate_version_dispatcher_body): Likewise.
6583 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
6584 (fusion_gpr_load_p): Fix a spacing issue.
6585 * doc/extend.texi (Common Function Attributes): Document that the
6586 PowerPC supports the target_clone attribute.
6588 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
6590 * config/arm/arm.h: explain F symbol found in description of ARM
6591 register allocation in its legend.
6593 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6595 * config/mips/frame-header-opt.c: Include profile-count.h.
6596 * config/riscv/riscv.c: Include profile-count.h
6598 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6600 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
6602 (sm_set_flag_if_changed): Add bbs field.
6603 (execute_sm_if_changed_flag_set): Pass BBS.
6604 (execute_sm): Update.
6606 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6608 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6611 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6613 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
6614 (peephole2): New peephole2 to emit the above.
6615 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
6617 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6619 * config/aarch64/aarch64.c (define_peephole2 above
6620 *sub_<shift>_<mode>): New peephole.
6622 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
6624 * config/i386/i386.c (make_resolver_func): Update.
6625 * Makefile.in: Add profile-count.h and profile-count.o
6626 * auto-profile.c (afdo_indirect_call): Update to new API.
6627 (afdo_set_bb_count): Update.
6628 (afdo_propagate_edge): Update.
6629 (afdo_propagate_circuit): Update.
6630 (afdo_calculate_branch_prob): Update.
6631 (afdo_annotate_cfg): Update.
6632 * basic-block.h: Include profile-count.h
6633 (struct edge_def): Turn count to profile_count.
6634 (struct basic_block_def): Likewie.
6635 (REG_BR_PROB_BASE): Move to profile-count.h
6636 (RDIV): Move to profile-count.h
6637 * bb-reorder.c (max_entry_count): Turn to profile_count.
6638 (find_traces): Update.
6639 (rotate_loop):Update.
6640 (connect_traces):Update.
6641 (sanitize_hot_paths):Update.
6642 * bt-load.c (migrate_btr_defs): Update.
6643 * cfg.c (RDIV): Remove.
6644 (init_flow): Use alloc_block.
6645 (alloc_block): Uninitialize count.
6646 (unchecked_make_edge): Uninitialize count.
6647 (check_bb_profile): Update.
6648 (dump_edge_info): Update.
6649 (dump_bb_info): Update.
6650 (update_bb_profile_for_threading): Update.
6651 (scale_bbs_frequencies_int): Update.
6652 (scale_bbs_frequencies_gcov_type): Update.
6653 (scale_bbs_frequencies_profile_count): New.
6654 * cfg.h (update_bb_profile_for_threading): Update.
6655 (scale_bbs_frequencies_profile_count): Declare.
6656 * cfgbuild.c (compute_outgoing_frequencies): Update.
6657 (find_many_sub_basic_blocks): Update.
6658 * cfgcleanup.c (try_forward_edges): Update.
6659 (try_crossjump_to_edge): Update.
6660 * cfgexpand.c (expand_gimple_tailcall): Update.
6661 (construct_exit_block): Update.
6662 * cfghooks.c (verify_flow_info): Update.
6663 (dump_bb_for_graph): Update.
6664 (split_edge): Update.
6665 (make_forwarder_block): Update.
6666 (duplicate_block): Update.
6667 (account_profile_record): Update.
6668 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
6669 (get_estimated_loop_iterations): Update.
6670 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
6671 (single_likely_exit): Update.
6672 * cfgloopmanip.c (scale_loop_profile): Update.
6674 (set_zero_probability): Update.
6675 (lv_adjust_loop_entry_edge): Update.
6676 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
6677 (purge_dead_edges): Update.
6678 (rtl_account_profile_record): Update.
6679 * cgraph.c (cgraph_node::create): Uninitialize count.
6680 (symbol_table::create_edge): Uninitialize count.
6681 (cgraph_update_edges_for_call_stmt_node): Update.
6682 (cgraph_edge::dump_edge_flags): Update.
6683 (cgraph_node::dump): Update.
6684 (cgraph_edge::maybe_hot_p): Update.
6685 * cgraph.h: Include profile-count.h
6686 (create_clone), create_edge, create_indirect_edge): Update.
6687 (cgraph_node): Turn count to profile_count.
6688 (cgraph_edge0: Likewise.
6689 (make_speculative, clone): Update.
6690 (create_edge): Update.
6691 (init_lowered_empty_function): Update.
6692 * cgraphclones.c (cgraph_edge::clone): Update.
6693 (duplicate_thunk_for_node): Update.
6694 (cgraph_node::create_clone): Update.
6695 * cgraphunit.c (cgraph_node::analyze): Update.
6696 (cgraph_node::expand_thunk): Update.
6697 * final.c (dump_basic_block_info): Update.
6698 * gimple-streamer-in.c (input_bb): Update.
6699 * gimple-streamer-out.c (output_bb): Update.
6700 * graphite.c (print_global_statistics): Update.
6701 (print_graphite_scop_statistics): Update.
6702 * hsa-brig.c: Include basic-block.h.
6703 * hsa-dump.c: Include basic-block.h.
6704 * hsa-gen.c (T sum_slice): Update.
6705 (convert_switch_statements):Update.
6706 * hsa-regalloc.c: Include basic-block.h.
6707 * ipa-chkp.c (chkp_produce_thunks): Update.
6708 * ipa-cp.c (struct caller_statistics): Update.
6709 (init_caller_stats): Update.
6710 (gather_caller_stats): Update.
6711 (ipcp_cloning_candidate_p): Update.
6712 (good_cloning_opportunity_p): Update.
6713 (get_info_about_necessary_edges): Update.
6714 (dump_profile_updates): Update.
6715 (update_profiling_info): Update.
6716 (update_specialized_profile): Update.
6717 (perhaps_add_new_callers): Update.
6718 (decide_about_value): Update.
6719 (ipa_cp_c_finalize): Update.
6720 * ipa-devirt.c (struct odr_type_warn_count): Update.
6721 (struct decl_warn_count): Update.
6722 (struct final_warning_record): Update.
6723 (possible_polymorphic_call_targets): Update.
6724 (ipa_devirt): Update.
6725 * ipa-fnsummary.c (redirect_to_unreachable): Update.
6726 * ipa-icf.c (sem_function::merge): Update.
6727 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
6728 * ipa-inline.c (compute_uninlined_call_time): Update.
6729 (compute_inlined_call_time): Update.
6730 (want_inline_small_function_p): Update.
6731 (want_inline_self_recursive_call_p): Update.
6732 (edge_badness): Update.
6733 (lookup_recursive_calls): Update.
6734 (recursive_inlining): Update.
6735 (inline_small_functions): Update.
6736 (dump_overall_stats): Update.
6737 (dump_inline_stats): Update.
6738 * ipa-profile.c (ipa_profile_generate_summary): Update.
6739 (ipa_propagate_frequency): Update.
6740 (ipa_profile): Update.
6741 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
6742 * ipa-utils.c (ipa_merge_profiles): Update.
6743 * loop-doloop.c (doloop_modify): Update.
6744 * loop-unroll.c (report_unroll): Update.
6745 (unroll_loop_runtime_iterations): Update.
6746 * lto-cgraph.c (lto_output_edge): Update.
6747 (lto_output_node): Update.
6748 (input_node): Update.
6749 (input_edge): Update.
6750 (merge_profile_summaries): Update.
6751 * lto-streamer-in.c (input_cfg): Update.
6752 * lto-streamer-out.c (output_cfg): Update.
6753 * mcf.c (create_fixup_graph): Update.
6754 (adjust_cfg_counts): Update.
6755 (sum_edge_counts): Update.
6756 * modulo-sched.c (sms_schedule): Update.
6757 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
6758 * predict.c (maybe_hot_count_p): Update.
6759 (probably_never_executed): Update.
6760 (dump_prediction): Update.
6761 (combine_predictions_for_bb): Update.
6762 (propagate_freq): Update.
6763 (handle_missing_profiles): Update.
6764 (counts_to_freqs): Update.
6765 (rebuild_frequencies): Update.
6766 (force_edge_cold): Update.
6767 * predict.h: Include profile-count.h
6768 (maybe_hot_count_p, counts_to_freqs): UPdate.
6769 * print-rtl-function.c: Do not include cfg.h
6770 * print-rtl.c: Include basic-block.h
6771 * profile-count.c: New file.
6772 * profile-count.h: New file.
6773 * profile.c (is_edge_inconsistent): Update.
6774 (correct_negative_edge_counts): Update.
6775 (is_inconsistent): Update.
6776 (set_bb_counts): Update.
6777 (read_profile_edge_counts): Update.
6778 (compute_frequency_overlap): Update.
6779 (compute_branch_probabilities): Update; Initialize and deinitialize
6781 (branch_prob): Update.
6782 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
6783 (edge_gcov_count): New.
6784 (bb_gcov_count): New.
6785 * shrink-wrap.c (try_shrink_wrapping): Update.
6786 * tracer.c (better_p): Update.
6787 * trans-mem.c (expand_transaction): Update.
6788 (ipa_tm_insert_irr_call): Update.
6789 (ipa_tm_insert_gettmclone_call): Update.
6790 * tree-call-cdce.c: Update.
6791 * tree-cfg.c (gimple_duplicate_sese_region): Update.
6792 (gimple_duplicate_sese_tail): Update.
6793 (gimple_account_profile_record): Update.
6794 (execute_fixup_cfg): Update.
6795 * tree-inline.c (copy_bb): Update.
6796 (copy_edges_for_bb): Update.
6797 (initialize_cfun): Update.
6798 (freqs_to_counts): Update.
6799 (copy_cfg_body): Update.
6800 (expand_call_inline): Update.
6801 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
6802 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
6803 (try_unroll_loop_completely): Update.
6804 (try_peel_loop): Update.
6805 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6806 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
6807 * tree-ssa-loop-split.c (connect_loops): Update.
6808 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
6809 * tree-ssa-reassoc.c (branch_fixup): Update.
6810 * tree-ssa-tail-merge.c (replace_block_by): Update.
6811 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
6812 (compute_path_counts): Update.
6813 (update_profile): Update.
6814 (recompute_probabilities): Update.
6815 (update_joiner_offpath_counts): Update.
6816 (estimated_freqs_path): Update.
6817 (freqs_to_counts_path): Update.
6818 (clear_counts_path): Update.
6819 (ssa_fix_duplicate_block_edges): Update.
6820 (duplicate_thread_path): Update.
6821 * tree-switch-conversion.c (case_bit_test_cmp): Update.
6822 (struct switch_conv_info): Update.
6823 * tree-tailcall.c (decrease_profile): Update.
6824 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
6825 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6826 * value-prof.c (check_counter): Update.
6827 (gimple_divmod_fixed_value): Update.
6828 (gimple_mod_pow2): Update.
6829 (gimple_mod_subtract): Update.
6830 (gimple_ic_transform): Update.
6831 (gimple_stringop_fixed_value): Update.
6832 * value-prof.h (gimple_ic): Update.
6834 2017-06-02 Carl Love <cel@us.ibm.com>
6836 * config/rs6000/rs6000-c: Add support for built-in functions
6837 vector double vec_doublee (vector signed int);
6838 vector double vec_doublee (vector unsigned int);
6839 vector double vec_doublee (vector float);
6840 vector double vec_doubleh (vector signed int);
6841 vector double vec_doubleh (vector unsigned int);
6842 vector double vec_doubleh (vector float);
6843 vector double vec_doublel (vector signed int);
6844 vector double vec_doublel (vector unsigned int);
6845 vector double vec_doublel (vector float);
6846 vector double vec_doubleo (vector signed int);
6847 vector double vec_doubleo (vector unsigned int);
6848 vector double vec_doubleo (vector float);.
6849 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
6850 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
6852 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
6853 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
6854 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
6856 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
6857 vec_doublel, vec_doubleh.
6858 * doc/extend.texi: Update the built-in documentation file for the
6859 new built-in functions.
6861 2017-06-02 David Malcolm <dmalcolm@redhat.com>
6864 * ipa-inline-analysis.c (free_growth_caches): Set
6865 edge_removal_hook_holder to NULL after removing it.
6867 2017-06-02 Sudakshina Das <sudi.das@arm.com>
6869 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
6870 comparision with zero.
6872 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6873 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6874 for early expansion of vec_min and vec_max builtins.
6875 (builtin_function_type): Add min/max unsigned variants to those
6876 identified as having unsigned arguments.
6878 2017-06-02 Olivier Hainque <hainque@adacore.com>
6880 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
6882 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6884 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
6885 Use VALL_F16 iterator rather than VALL.
6887 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6889 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
6890 Emit CBNZ inside loop when doing a strong exchange and comparing
6891 against zero. Generate the CC flags after the loop.
6893 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
6895 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
6896 (dl_section_ref): New.
6897 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
6898 On AIX, append an expression to subtract the size of the
6899 section length to dl_section_ref.
6901 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6903 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6904 for early expansion of vector absolute builtins.
6906 2017-06-02 Richard Biener <rguenther@suse.de>
6908 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
6909 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
6911 2017-06-02 Richard Biener <rguenther@suse.de>
6913 PR tree-optimization/80948
6914 * tree-tailcall.c (find_tail_calls): Track stmts to move in
6917 2017-06-02 Richard Biener <rguenther@suse.de>
6919 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
6921 * tree-vect-stmts.c (process_use): Do not mark backedge defs
6922 for inductions as relevant.
6924 2017-06-02 Richard Biener <rguenther@suse.de>
6926 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
6927 (vectorizable_induction): ... this. Remove dead code.
6929 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
6931 * builtins. (expand_builtin_alloca): Remove second parameter and
6932 infer its value from the first parameter instead.
6933 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
6935 2017-06-02 Jakub Jelinek <jakub@redhat.com>
6937 PR rtl-optimization/80903
6938 * loop-doloop.c (add_test): Unshare sequence.
6940 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
6942 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
6944 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
6946 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
6948 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
6949 xlogue_layout::get_instance, logue_layout::xlogue_layout,
6950 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
6951 (xlogue_layout::get_stub_rtx): Make static.
6952 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
6953 (xlogue_layout::compute_stub_managed_regs): Rename to...
6954 (xlogue_layout::count_stub_managed_regs): ...this.
6955 (xlogue_layout::is_stub_managed_reg): New function.
6956 (xlogue_layout::m_stub_names): Rename to...
6957 (xlogue_layout::s_stub_names): ...this, make static.
6958 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
6959 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
6960 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
6961 xlogue_layout::s_stub_names): Instantiate statics.
6962 (stub_managed_regs): Remove.
6963 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
6964 (disable_call_ms2sysv_xlogues): Rename to...
6965 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
6966 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
6968 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
6969 change after reload_completed.
6970 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
6972 (ix86_expand_prologue): Likewise.
6973 (ix86_expand_epilogue): Likewise.
6974 (ix86_expand_split_stack_prologue): Likewise.
6975 (ix86_compute_frame_layout): Remove frame parameter ...
6976 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
6977 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
6979 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
6980 (ix86_frame): Move from here ...
6981 * config/i386/i386.h (ix86_frame): ... to here.
6982 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
6983 complete ix86_frame data structure instead. Remove some_ld_name.
6985 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
6987 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
6988 symbols that hold a DECL_VALUE_EXPR.
6990 2017-06-01 Martin Jambor <mjambor@suse.cz>
6992 PR tree-optimization/80898
6993 * tree-sra.c (process_subtree_disqualification): Removed.
6994 (disqualify_candidate): Do not acll
6995 process_subtree_disqualification.
6996 (subtree_mark_written_and_enqueue): New function.
6997 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
6998 RHS has been disqualified and re-queue LHS if necessary. Apart
6999 from that, ignore disqualified RHS.
7001 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7003 * config/s390/s390.c (s390_emit_epilogue): Disable early return
7004 address fetch for z10 or later.
7006 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7008 * config/arc/arc.md (tst_movb): Add guard when splitting.
7010 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7012 * config/arc/arc.c (arc_can_eliminate): Test against
7013 arc_frame_pointer_needed.
7015 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7017 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
7018 to prevent store reordering.
7019 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
7020 (type): Add block type.
7021 (stack_tie): Define special instruction to be used in
7024 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7026 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
7027 constraint. It is not valid for the pattern.
7028 (noncommutative_binary_comparison): Likewise.
7030 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7032 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
7035 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7037 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
7038 be used by the reg-alloc.
7040 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7042 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
7043 reg-alloc when having mul64 or mul32x16 instructions.
7044 (mulsidi3): Likewise.
7045 (umulsidi3): Likewise.
7046 (mulsi32x16): New pattern.
7047 (mulsi64): Likewise.
7048 (mulsidi64): Likewise.
7049 (umulsidi64): Likewise.
7050 (MUL32x16_REG): Define.
7051 (mul64_600): Use MUL32x16_REG.
7052 (mac64_600): Likewise.
7053 (umul64_600): Likewise.
7054 (umac64_600): Likewise.
7056 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7058 * config/arc/arc.md (mulsi3_700): Make it commutative.
7060 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
7062 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
7064 (*sign_extendsidi2_insn): Likewise for movstosw.
7066 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
7068 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
7069 the type of the input discriminant value. Convert the
7070 discriminant value of signedness vary.
7072 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
7074 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
7075 Add to -Wall section.
7077 2017-06-01 Richard Biener <rguenther@suse.de>
7080 * fold-const.c (fold_plusminus_mult_expr): If the factored
7081 factor may be zero use a wrapping type for the inner operation.
7082 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
7083 and handle moved defs.
7084 (process_assignment): Properly guard the unary op case. Return a
7085 tri-state indicating that moving the stmt before the call may allow
7086 to continue. Pass through to_move.
7087 (find_tail_calls): Handle moving unrelated defs before
7090 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
7093 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
7094 splitter result in the canonical way.
7096 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
7098 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
7099 also for 32bit target. Update insn attributes.
7100 (zero-extendsidi2 splitter): Allow all registers for operand 1.
7102 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
7104 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
7105 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
7106 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
7107 (_mm_maskz_min_ss): New intrinsics.
7109 2017-05-31 Martin Liska <mliska@suse.cz>
7111 * tree-vect-loop.c (vect_create_epilog_for_reduction):
7112 Change comment style to one we normally use.
7113 (vectorizable_reduction): Likewise.
7114 (vectorizable_induction): Likewise.
7115 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7116 (vectorizable_call): Likewise.
7117 (vectorizable_simd_clone_call): Likewise.
7118 (vectorizable_conversion): Likewise.
7119 (vectorizable_assignment): Likewise.
7120 (vectorizable_shift): Likewise.
7121 (vectorizable_operation): Likewise.
7122 (vectorizable_store): Likewise.
7123 (vectorizable_load): Likewise.
7124 * tree-vectorizer.h: Likewise.
7126 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
7128 * passes.c (emergency_dump_function): New.
7129 * tree-pass.h (emergency_dump_function): Declare.
7130 * plugin.c (plugins_internal_error_function): Remove.
7131 * plugin.h (plugins_internal_error_function): Remove declaration.
7132 * toplev.c (internal_error_function): New static function. Use it...
7133 (general_init): ...here.
7135 2017-05-31 Graham Markall <graham.markall@embecosm.com>
7137 * config/arc/arc.c (arc_print_operand): Handle constant operands.
7138 (arc_rtx_costs): Add costs for new patterns.
7139 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
7140 * config/arc/predicates.md: Add _1_2_3_operand predicate.
7142 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
7144 * tree-ssa-strlen.c (get_next_strinfo): New function.
7145 (get_stridx_plus_constant): Use it.
7146 (zero_length_string): Likewise.
7147 (adjust_related_strinfos): Likewise.
7148 (adjust_last_stmt): Likewise.
7150 2017-05-31 Richard Biener <rguenther@suse.de>
7153 * config/i386/i386.c (ix86_expand_builtin): Remove assert
7154 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
7156 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
7158 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
7159 loop_vinfo argument and use of dependence distance vectors.
7160 Check instead whether the two references differ only in their
7161 initial value and assume that they have the same alignment if the
7162 difference is a multiple of the vector alignment.
7163 (vect_analyze_data_refs_alignment): Update call accordingly.
7165 2017-05-31 Martin Liska <mliska@suse.cz>
7168 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
7170 2017-05-31 Bin Cheng <bin.cheng@arm.com>
7172 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
7173 (create_intersect_range_checks): Move from ...
7174 * tree-data-ref.c (create_intersect_range_checks_index)
7175 (create_intersect_range_checks): ... to here.
7176 (create_runtime_alias_checks): New function factored from ...
7177 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
7178 here. Call above function.
7179 * tree-data-ref.h (create_runtime_alias_checks): New function.
7181 2017-05-31 Bin Cheng <bin.cheng@arm.com>
7183 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
7184 segment length for dr_b and compute it in wide_int.
7186 2017-05-31 Richard Biener <rguenther@suse.de>
7188 PR tree-optimization/80906
7189 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
7190 and pass through iv_map.
7191 (copy_bb_and_scalar_dependences): Adjust.
7192 (translate_pending_phi_nodes): Likewise.
7193 (copy_loop_close_phi_args): Handle code-generating IVs instead
7196 2017-05-30 David Malcolm <dmalcolm@redhat.com>
7198 * diagnostic-color.c (color_dict): Add "type-diff".
7199 (parse_gcc_colors): Update comment.
7200 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
7201 -fdiagnostics-show-template-tree and -fno-elide-type.
7202 (GCC_COLORS): Add type-diff to example.
7204 (-fdiagnostics-show-template-tree): New.
7205 (-fno-elide-type): New.
7206 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
7207 the pp_format_decoder callback. Call any m_format_postprocessor's
7209 (pretty_printer::pretty_printer): Initialize
7210 m_format_postprocessor.
7211 (pretty_printer::~pretty_printer): Delete any
7212 m_format_postprocessor.
7213 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
7214 (class format_postprocessor): New class.
7215 (struct pretty_printer::format_decoder): Document the new parameters.
7216 (struct pretty_printer::m_format_postprocessor): New field.
7217 * tree-diagnostic.c (default_tree_printer): Update for new
7218 bool and const char ** params.
7219 * tree-diagnostic.h (default_tree_printer): Likewise.
7221 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
7223 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
7224 (lwa_operand): Delete rs6000_gen_cell_microcode test.
7225 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
7226 rs6000_gen_cell_microcode code.
7227 (rs6000_final_prescan_insn): Delete.
7228 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
7229 "warn-cell-microcode" entries.
7230 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
7231 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
7232 throughout. Change cc_reg_not_micro_cr0_operand to
7233 cc_reg_not_cr0_operand throughout.
7234 (*extendhi<mode>2_noload): Delete.
7235 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
7236 (mwarn-cell-microcode): Delete.
7237 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
7238 -mgen-cell-microcode and -mwarn-cell-microcode.
7240 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
7243 * config/i386/constraints.md (Yd): New constraint.
7245 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
7246 and (?Yd, r) alternatives. Update insn attributes.
7247 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
7248 and (?*Yd, r) alternatives. Update insn attributes.
7249 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
7251 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
7253 * gimplify.c (gimplify_modify_expr): Don't create a
7254 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
7257 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
7259 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
7261 2017-05-30 Richard Biener <rguenther@suse.de>
7263 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
7264 and reduc_def fields.
7265 (STMT_VINFO_REDUC_TYPE): New define.
7266 (STMT_VINFO_REDUC_DEF): Likewise.
7267 (vect_force_simple_reduction): Adjust prototype.
7268 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
7269 (vect_is_simple_reduction): Remove check_reduction argument.
7270 (vect_force_simple_reduction): Adjust and set
7271 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
7272 (vectorizable_reduction): Do not re-do reduction analysis
7273 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
7274 * tree-parloops.c (gather_scalar_reductions): Adjust.
7276 2017-05-30 Richard Biener <rguenther@suse.de>
7279 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
7282 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
7284 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
7285 Introduce unknown_misalignment parameter and remove vf.
7286 (vect_peeling_hash_get_lowest_cost):
7287 Pass unknown_misalignment parameter.
7288 (vect_enhance_data_refs_alignment):
7289 Fix unsupportable data ref treatment.
7291 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7293 * tree-vect-data-refs.c (vect_get_data_access_cost):
7294 Workaround for SLP handling.
7295 (vect_enhance_data_refs_alignment):
7296 Compute costs for doing no peeling at all, compare to the best
7297 peeling costs so far and avoid peeling if cheaper.
7299 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7301 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
7302 Return peeling info and set costs to zero for unlimited cost
7304 (vect_enhance_data_refs_alignment): Also inspect all datarefs
7305 with unknown misalignment. Compute and costs for unknown
7306 misalignment, compare them to the costs for known misalignment
7307 and choose the cheapest for peeling.
7309 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7311 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
7312 (vect_get_peeling_costs_all_drs): Create function.
7313 (vect_peeling_hash_get_lowest_cost):
7314 Use vect_get_peeling_costs_all_drs.
7315 (vect_peeling_supportable): Create function.
7316 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
7318 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7320 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
7321 DR_HAS_NEGATIVE_STEP.
7322 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
7323 (vect_enhance_data_refs_alignment): Use.
7324 (vect_duplicate_ssa_name_ptr_info): Use.
7325 * tree-vectorizer.h (dr_misalignment): Use.
7326 (known_alignment_for_access_p): Use.
7328 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7331 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
7333 (has_section_name): New function.
7335 2017-05-30 Martin Liska <mliska@suse.cz>
7338 * auto-profile.c (get_function_decl_from_block): Fix
7341 2017-05-30 Richard Biener <rguenther@suse.de>
7344 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
7346 2017-05-30 Martin Liska <mliska@suse.cz>
7348 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
7349 * dumpfile.h (struct dump_file_info): Remove ctors.
7351 2017-05-30 Martin Liska <mliska@suse.cz>
7353 * predict.def: Fix GNU coding style.
7355 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
7357 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
7358 Mark 'to' argument with ATTRIBUTE_UNUSED.
7360 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
7362 * config/xtensa/xtensa.c (xtensa_emit_call): Use
7363 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
7364 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
7367 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
7369 * doc/install.texi (Options specification): Restore entry of
7370 --enable-sjlj-exceptions.
7372 2017-05-27 Michael Eager <eager@eagercon.com>
7375 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
7377 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
7379 * config/microblaze/microblaze.h
7380 (FIXED_REGISTERS): Update in macro.
7381 (CALL_USED_REGISTERS): Update in macro.
7383 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
7385 * doc/install.texi: Add links to macOS binary distributions.
7387 2017-05-27 Jakub Jelinek <jakub@redhat.com>
7391 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7393 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
7395 2017-05-26 Martin Liska <mliska@suse.cz>
7397 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
7399 2017-05-26 Martin Liska <mliska@suse.cz>
7401 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
7402 always leading ';; '.
7403 (dump_bb_info): Likewise.
7404 (brief_dump_cfg): Likewise.
7405 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
7406 * dumpfile.c: Remove usage of TDF_VERBOSE.
7407 * dumpfile.h (enum dump_kind): Likewise.
7408 (dump_gimple_bb_header): Do not use TDF_COMMENT.
7409 * print-tree.c (debug_verbose): Remove.
7410 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
7411 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
7412 * tree-diagnostic.c (default_tree_printer): Replace
7413 TDF_DIAGNOSTIC with TDF_SLIM.
7415 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7417 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
7418 in parameter loop, rather than loop_vinfo.
7419 (create_intersect_range_checks): Ditto.
7420 (vect_create_cond_for_alias_checks): Update call to above functions.
7422 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7424 PR tree-optimization/80815
7425 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
7426 for merging runtime alias checks. Handle negative DR_STEPs.
7428 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7430 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
7432 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
7433 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
7434 out code pruning runtime alias checks.
7435 * tree-data-ref.c (prune_runtime_alias_test_list): New function
7436 factored out from above.
7437 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
7439 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
7441 (prune_runtime_alias_test_list): New decalaration.
7443 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7445 * tree-vect-data-refs.c (compare_tree): Rename and move ...
7446 * tree-data-ref.c (data_ref_compare_tree): ... to here.
7447 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
7448 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
7449 (operator==, comp_dr_with_seg_len_pair): Ditto.
7450 (vect_prune_runtime_alias_test_list): Ditto.
7452 2017-05-26 Martin Liska <mliska@suse.cz>
7455 * params.def: Bound partial-inlining-entry-probability param.
7457 2017-05-26 Marek Polacek <polacek@redhat.com>
7460 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
7463 2017-05-26 Richard Biener <rguenther@suse.de>
7465 PR tree-optimization/80842
7466 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
7469 2017-05-26 Richard Biener <rguenther@suse.de>
7471 PR tree-optimization/80844
7472 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
7474 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
7476 * doc/md.texi (Machine Constraints): Update x86 family
7477 machine constraints section to match 'config/i386/constraints.md'.
7479 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
7481 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
7483 2017-05-25 Nathan Sidwell <nathan@acm.org>
7485 * doc/invoke.texi (--enable-languages): Update documentation.
7487 2017-05-25 Martin Liska <mliska@suse.cz>
7489 * dumpfile.c: Add TDF_FOLDING.
7490 * dumpfile.h (enum dump_kind): Likewise.
7491 * genmatch.c (dt_simplify::gen_1): Use it.
7493 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7495 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
7497 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7499 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
7500 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
7502 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7504 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
7505 * match.pd (X == C): Rewrite it here.
7506 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
7507 with_certain_nonzero_bits2): New predicates.
7508 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
7510 2017-05-24 Nathan Sidwell <nathan@acm.org>
7512 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
7515 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
7518 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
7520 * config/powerpcspe: New port. Files are copied from the rs6000
7521 port, with "rs6000" in filenames replaced by "powerpcspe".
7523 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
7525 PR rtl-optimization/80754
7526 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
7528 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
7530 * config/sparc/sparc.md (length): Return the correct value for -mflat
7531 sibcalls to match output_sibcall.
7533 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
7537 * config/rs6000/rs6000.c (struct machine_function): Add new field
7539 (rs6000_get_separate_components): Init that field, use it.
7540 (rs6000_components_for_bb): Use the field.
7542 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
7544 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
7546 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
7549 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
7551 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7554 * config/s390/s390.c (s390_check_qrst_address): Check incoming
7555 address against address_operand predicate.
7556 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
7558 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
7560 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
7561 parameters passed indirectly.
7563 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
7565 * config/i386/i386.md (*movdi_internal): Remove SSE4
7566 alternative 18 (?r, *v). Update insn attributes.
7567 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
7568 Update insn attributes.
7569 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
7570 Update insn attributes.
7571 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
7572 alternative 1 (r, v). Remove isa attribute.
7573 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
7574 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
7575 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
7577 2017-05-23 Tom de Vries <tom@codesourcery.com>
7579 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
7582 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7584 * cgraphunit.c (symbol_table::process_new_functions): Update.
7585 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
7586 (inline_generate_summary): Rename to ...
7587 (ipa_fn_summary_generate): ... this one.
7588 (inline_read_summary): Rename to ...
7589 (ipa_fn_summary_read): ... this one.
7590 (inline_write_summary): Rename to ...
7591 (ipa_fn_summary_write): ... this one.
7592 (inline_free_summary): Rename to ...
7593 (ipa_free_fn_summary): ... this one.
7594 (pass_data_local_fn_summary, pass_local_fn_summary,
7595 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
7596 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
7597 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
7598 make_pass_ipa_fn_summary): New.
7599 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
7600 inline_write_summary, inline_free_summary): Remove.
7601 (ipa_free_fn_summary) : New.
7602 * ipa-inline.c (ipa_inline): Update.
7603 (pass_ipa_inline): Do not generate summaries.
7604 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
7606 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
7607 and add pass_ipa_fn_summary.
7608 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
7610 (make_pass_inline_parameters): Remove.
7612 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
7614 * omp-low.c (struct omp_context): Remove "default_kind" member.
7617 * omp-offload.c (execute_oacc_device_lower): Remove the
7618 parallelism dimensions function attributes for unparallelized
7619 OpenACC kernels constructs.
7621 2017-05-23 Martin Liska <mliska@suse.cz>
7623 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
7625 (cgraph_edge::make_speculative): Likewise.
7626 (cgraph_edge::resolve_speculation): Likewise.
7627 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
7628 (cgraph_node::dump): Likewise.
7629 * cgraph.h: Likewise.
7630 * cgraphunit.c (analyze_functions): Likewise.
7631 (symbol_table::compile): Likewise.
7632 * ipa-cp.c (print_all_lattices): Likewise.
7633 (determine_versionability): Likewise.
7634 (initialize_node_lattices): Likewise.
7635 (ipcp_verify_propagated_values): Likewise.
7636 (estimate_local_effects): Likewise.
7637 (update_profiling_info): Likewise.
7638 (create_specialized_node): Likewise.
7639 (perhaps_add_new_callers): Likewise.
7640 (decide_about_value): Likewise.
7641 (decide_whether_version_node): Likewise.
7642 (identify_dead_nodes): Likewise.
7643 (ipcp_store_bits_results): Likewise.
7644 * ipa-devirt.c (dump_targets): Likewise.
7645 (ipa_devirt): Likewise.
7646 * ipa-icf.c (sem_item::dump): Likewise.
7647 (sem_function::equals): Likewise.
7648 (sem_variable::equals): Likewise.
7649 (sem_item_optimizer::read_section): Likewise.
7650 (sem_item_optimizer::execute): Likewise.
7651 (congruence_class::dump): Likewise.
7652 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
7653 (dump_inline_summary): Likewise.
7654 (estimate_node_size_and_time): Likewise.
7655 (inline_analyze_function): Likewise.
7656 * ipa-inline-transform.c (inline_call): Likewise.
7657 * ipa-inline.c (report_inline_failed_reason): Likewise.
7658 (want_early_inline_function_p): Likewise.
7659 (edge_badness): Likewise.
7660 (update_edge_key): Likewise.
7661 (inline_small_functions): Likewise.
7662 * ipa-profile.c (ipa_profile): Likewise.
7663 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
7664 (ipa_make_edge_direct_to_target): Likewise.
7665 (remove_described_reference): Likewise.
7666 (ipa_impossible_devirt_target): Likewise.
7667 (propagate_controlled_uses): Likewise.
7668 (ipa_print_node_params): Likewise.
7669 (ipcp_transform_function): Likewise.
7670 * ipa-pure-const.c (pure_const_read_summary): Likewise.
7671 (propagate_pure_const): Likewise.
7672 * ipa-reference.c (generate_summary): Likewise.
7673 (read_write_all_from_decl): Likewise.
7674 (propagate): Likewise.
7675 (ipa_reference_read_optimization_summary): Likewise.
7676 * ipa-utils.c (ipa_merge_profiles): Likewise.
7677 * ipa.c (walk_polymorphic_call_targets): Likewise.
7678 (symbol_table::remove_unreachable_nodes): Likewise.
7679 (ipa_single_use): Likewise.
7680 * passes.c (execute_todo): Likewise.
7681 * predict.c (drop_profile): Likewise.
7682 * symtab.c (symtab_node::get_dump_name): New function.
7683 (symtab_node::dump_name): Likewise.
7684 (symtab_node::dump_asm_name): Likewise.
7685 (symtab_node::dump_references): Likewise.
7686 (symtab_node::dump_referring): Likewise.
7687 (symtab_node::dump_base): Likewise.
7688 (symtab_node::debug_symtab): Likewise.
7689 * tree-sra.c (convert_callers_for_node): Likewise.
7690 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
7691 * value-prof.c (init_node_map): Likewise.
7693 2017-05-23 Martin Liska <mliska@suse.cz>
7695 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
7696 and symtab_node::debug_symtab to symbol_table::debug.
7697 * cgraphunit.c (analyze_functions): Use the renamed function.
7698 (symbol_table::compile): Likewise.
7699 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
7700 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
7701 * passes.c (execute_todo): Likewise.
7702 * symtab.c (symbol_table::dump): New function.
7703 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
7705 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7707 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
7708 that nonconst implies exec.
7710 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7712 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7713 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7714 (inline_edge_summary_vec): Turn into ...
7715 (ipa_call_summaries): ... this one.
7716 (redirect_to_unreachable, edge_set_predicate,
7717 evaluate_properties_for_edge, inline_summary_alloc,
7718 reset_ipa_call_summary, reset_inline_summary,
7719 inline_summary_t::duplicate): Update.
7720 (inline_edge_duplication_hook): Turn to ...
7721 (ipa_call_summary_t::duplicate): ... this one.
7722 (inline_edge_removal_hook): Turn to ...
7723 (ipa_call_summary_t::remove): ... this one.
7724 (dump_inline_edge_summary): Turn to ...
7725 (dump_ipa_call_summary): ... this one.
7726 (estimate_function_body_sizes): Update.
7727 (inline_update_callee_summaries): Update.
7728 (remap_edge_change_prob): Update.
7729 (remap_edge_summaries): Update.
7730 (inline_merge_summary): Update.
7731 (do_estimate_edge_time): Update.
7732 (inline_generate_summary): Update.
7733 (inline_read_section): Update.
7734 (inline_read_summary): Update.
7735 (inline_free_summary): Update.
7736 * ipa-inline.c (can_inline_edge_p): Update.
7737 (compute_inlined_call_time): Update.
7738 (want_inline_small_function_p): Update.
7739 (edge_badness): Update.
7740 (early_inliner): Update.
7741 * ipa-inline.h (inline_edge_summary): Turn to ...
7742 (ipa_call_summary): ... this one.
7743 (ipa_call_summary_t): New class.
7744 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
7745 (ipa_call_summaries): New.
7746 (inline_edge_summary): Remove.
7747 (estimate_edge_growth): Update.
7748 * ipa-profile.c (ipa_propagate_frequency_1): Update.
7749 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7750 * ipa-split.c (execute_split_functions): Update.
7751 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
7753 2017-05-23 Tom de Vries <tom@codesourcery.com>
7755 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7756 attributes): Document rdrand effective target.
7758 2017-05-23 Tom de Vries <tom@codesourcery.com>
7760 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7761 attributes): Sort alphabetically.
7763 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
7765 * config/avr/genmultilib.awk: Use gsub instead of gensub.
7767 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
7770 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
7771 V2DF/V2DI splat into two separate patterns, one that handles
7772 registers, and the other that only handles memory. Drop support
7773 for splatting from a GPR on ISA 2.07 and then splitting the
7774 splat into direct move and splat.
7775 (vsx_splat_<mode>_reg): Likewise.
7776 (vsx_splat_<mode>_mem): Likewise.
7778 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
7780 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
7782 2017-05-22 Jakub Jelinek <jakub@redhat.com>
7785 * omp-low.c (finish_taskreg_remap): New function.
7786 (finish_taskreg_scan): If unit size of ctx->record_type
7787 is non-constant, unshare the size expression and replace
7788 decls in it with possible outer var refs.
7791 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
7792 GOVD_SHARED rather than GOVD_PRIVATE with it.
7793 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
7794 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
7797 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
7798 as last argument to build_outer_var_ref for pointer bases of array
7801 2017-05-19 Martin Sebor <msebor@redhat.com>
7803 * print-tree.c (print_node): Print DECL_READ_P flag.
7805 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7807 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
7808 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
7809 * cgraph.c: Likewise.
7810 * cgraphunit.c: Likewise.
7811 * gengtype.c: Likewise.
7812 * ipa-cp.c: Likewise.
7813 * ipa-devirt.c: Likewise.
7814 * ipa-icf.c: Likewise.
7815 * ipa-predicate.c: Likewise.
7816 * ipa-profile.c: Likewise.
7817 * ipa-prop.c: Likewise.
7818 * ipa-split.c: Likewise.
7820 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
7821 edge_predicate_pool, dump_inline_hints,
7822 inline_summary::account_size_time, redirect_to_unreachable,
7823 edge_set_predicate, set_hint_predicate,
7824 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
7825 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
7826 inline_summary_t::remove, remap_hint_predicate_after_duplication,
7827 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
7828 ipa_call_summary_t::remove, initialize_growth_caches,
7829 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
7830 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
7831 mark_modified, unmodified_parm_1, unmodified_parm,
7832 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
7833 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
7834 compute_bb_predicates, will_be_nonconstant_expr_predicate,
7835 will_be_nonconstant_predicate, record_modified_bb_info,
7836 get_minimal_bb, record_modified, param_change_prob,
7837 phi_result_unknown_predicate, predicate_for_phi_result,
7838 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
7839 estimate_function_body_sizes, compute_inline_parameters,
7840 compute_inline_parameters_for_curren, pass_data_inline_parameters,
7841 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
7842 inline_update_callee_summaries, remap_edge_change_prob,
7843 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
7844 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
7845 inline_analyze_function, inline_summary_t::insert,
7846 inline_generate_summary, read_ipa_call_summary, inline_read_section,
7847 inline_read_summary, write_ipa_call_summary, inline_write_summary,
7848 inline_free_summary): Move to ipa-fnsummary.h
7849 (predicate_t): Remove.
7850 * ipa-fnsummary.c: New file.
7851 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
7852 (enum inline_hints_vals, inline_hints, agg_position_info,
7853 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
7854 inline_summaries, ipa_call_summary, ipa_call_summary_t,
7855 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
7856 dump_inline_summary, dump_inline_hints, inline_generate_summary,
7857 inline_read_summary, inline_write_summary, inline_free_summary,
7858 inline_analyze_function, initialize_inline_failed,
7859 inline_merge_summary, inline_update_overall_summary,
7860 compute_inline_parameters): Move to ipa-fnsummary.h
7861 * ipa-fnsummary.h: New file.
7862 * ipa-inline-transform.h: Include ipa-inline.h.
7863 * ipa-inline.c: LIkewise.
7865 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7867 * ipa-inline.c (edge_badness): Use inlined_time instead of
7868 inline_summaries->get.
7870 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7872 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
7874 2017-05-22 Nathan Sidwell <nathan@acm.org>
7876 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
7877 (fdump-lang): Document 'raw' option.
7878 * dumpfile.h (TDI_tu): Delete.
7879 * dumpfile.c (dump_files): Remove translation-unit.
7880 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
7882 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
7884 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
7885 command option from $(AWK) call.
7886 * config/avr/genmultilib.awk: Simplify and rewrite so that it
7887 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
7888 [FORMAT]: Remove handling of variable.
7889 * config/avr/t-multilib: Regenerate.
7891 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7893 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
7895 (dump_inline_summary): Do not print self_time.
7896 (estimate_function_body_sizes): Do not set self_time.
7897 (compute_inline_parameters): Likewise.
7898 (inline_read_section, inline_write_summary): Do not stream self_time.
7899 * ipa-inline.h (inline_summary): Drop self_time.
7901 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7903 * ipa-inline-analysis.c (account_size_time): Rename to ...
7904 (inline_summary::account_size_time): ... this one.
7905 (reset_ipa_call_summary): Turn to ...
7906 (ipa_call_summary::reset): ... this one.
7907 (reset_inline_summary): Turn to ...
7908 (inline_summary::reset): ... this one.
7909 (inline_summary_t::remove): Update.
7910 (inline_summary_t::duplicate): Update.
7911 (ipa_call_summary_t::remove): Update.
7912 (dump_inline_summary): Update.
7913 (estimate_function_body_sizes): Update.
7914 (compute_inline_parameters): Update.
7915 (estimate_node_size_and_time): Update.
7916 (inline_merge_summary): Update.
7917 (inline_update_overall_summary): Update.
7918 (inline_read_section): Update.
7919 (inline_write_summary): Update.
7920 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
7921 add account_size_time and reset member functions.
7922 (ipa_call_summary): Add reset function.
7923 * ipa-predicate.h (predicate::operator &): Constify.
7925 2017-05-22 Richard Biener <rguenther@suse.de>
7927 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
7929 2017-05-19 Jason Merrill <jason@redhat.com>
7931 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
7933 2017-05-19 Marek Polacek <polacek@redhat.com>
7936 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
7937 TYPE_OVERFLOW_WRAPS checks.
7939 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
7941 * tree-core.h (enum omp_clause_default_kind): Add
7942 "OMP_CLAUSE_DEFAULT_PRESENT".
7943 * tree-pretty-print.c (dump_omp_clause): Handle it.
7944 * gimplify.c (enum gimplify_omp_var_data): Add
7945 "GOVD_MAP_FORCE_PRESENT".
7946 (gimplify_adjust_omp_clauses_1): Map it to
7947 "GOMP_MAP_FORCE_PRESENT".
7948 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
7950 * gimplify.c (oacc_default_clause): Clarify.
7952 2017-05-19 Nathan Sidwell <nathan@acm.org>
7954 LANG_HOOK_REGISTER_DUMPS
7955 * toplev.c (general_init): Call register dump lang hook.
7956 * doc/invoke.texi: Document -fdump-lang option family.
7957 * dumpfile.c (dump_files): Remove class dump here.
7958 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
7959 * dumpfile.h (tree_dump_index): Remove TDI_class.
7960 * langhooks-def.h (lhd_register_dumps): Declare.
7961 (LANG_HOOKS_REGISTER_DUMPS): Define.
7962 (LANG_HOOKS_INITIALIZER): Add it.
7963 * langhooks.c (lhd_register_dumps): Define.
7964 * langhooks.h (struct lang_hooks): Add register_dumps.
7966 2017-05-19 Nathan Sidwell <nathan@acm.org>
7968 * context.h (context::set_passes): New.
7969 * context.c (context::context): Do not create pass manager.
7970 * toplev.c (general_init): Create pass manager here.
7972 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
7974 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
7975 use this splitter if two add or or instructions would also work for
7976 the constant we want to generate.
7978 2017-05-19 Richard Biener <rguenther@suse.de>
7981 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
7982 predicate evaluation.
7984 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
7986 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
7988 * ipa-inline.c (want_inline_small_function_p): Do not cast to
7991 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
7993 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7994 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7995 (inline_edge_summary_vec): Turn into ...
7996 (ipa_call_summaries): ... this one.
7997 (redirect_to_unreachable, edge_set_predicate,
7998 evaluate_properties_for_edge, inline_summary_alloc,
7999 reset_ipa_call_summary, reset_inline_summary,
8000 inline_summary_t::duplicate): Update.
8001 (inline_edge_duplication_hook): Turn to ...
8002 (ipa_call_summary_t::duplicate): ... this one.
8003 (inline_edge_removal_hook): Turn to ...
8004 (ipa_call_summary_t::remove): ... this one.
8005 (dump_inline_edge_summary): Turn to ...
8006 (dump_ipa_call_summary): ... this one.
8007 (estimate_function_body_sizes): Update.
8008 (inline_update_callee_summaries): Update.
8009 (remap_edge_change_prob): Update.
8010 (remap_edge_summaries): Update.
8011 (inline_merge_summary): Update.
8012 (do_estimate_edge_time): Update.
8013 (inline_generate_summary): Update.
8014 (inline_read_section): Update.
8015 (inline_read_summary): Update.
8016 (inline_free_summary): Update.
8017 * ipa-inline.c (can_inline_edge_p): Update.
8018 (compute_inlined_call_time): Update.
8019 (want_inline_small_function_p): Update.
8020 (edge_badness): Update.
8021 (early_inliner): Update.
8022 * ipa-inline.h (inline_edge_summary): Turn to ...
8023 (ipa_call_summary): ... this one.
8024 (ipa_call_summary_t): New class.
8025 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
8026 (ipa_call_summaries): New.
8027 (inline_edge_summary): Remove.
8028 (estimate_edge_growth): Update.
8029 * ipa-profile.c (ipa_propagate_frequency_1): Update.
8030 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
8031 * ipa-split.c (execute_split_functions): Update.
8032 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
8034 2017-05-19 Richard Biener <rguenther@suse.de>
8037 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
8039 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
8041 * config/rs6000/rs6000.c (struct machine_function): Add field
8042 fpr_is_wrapped_separately.
8043 (rs6000_get_separate_components): Use 64 components. Handle the
8045 (rs6000_components_for_bb): Handle the FPR components.
8046 (rs6000_emit_prologue_components): Handle the FPR components.
8047 (rs6000_emit_epilogue_components): Handle the FPR components.
8048 (rs6000_set_handled_components): Handle the FPR components.
8049 (rs6000_emit_prologue): Don't output prologue code for those FPRs
8050 that are already separately shrink-wrapped.
8051 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
8052 that are already separately shrink-wrapped.
8054 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
8057 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
8060 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
8061 (define_peephole2 for Altivec d-form load): Add peepholes to catch
8062 cases where the register allocator uses a move and an offsettable
8063 memory operation to/from a FPR register on ISA 2.06/2.07.
8064 (define_peephole2 for Altivec d-form store): Likewise.
8066 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
8069 * config/i386/mmx.md (*mov<mode>_internal): Enable
8070 alternatives 11, 12, 13 and 14 also for 32bit targets.
8071 Remove alternatives 15, 16, 17 and 18.
8072 * config/i386/sse.md (vec_concatv2di): Change
8073 alternative (!x, *y) to (x, ?!*Yn).
8075 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
8077 * dumpfile.h (enum dump_kind): Remove stray comma.
8079 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
8081 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
8082 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
8083 predicate::num_conditions
8084 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
8085 (CHANGED): turn into predicate::changed.
8086 (agg_position_info): Move to ipa-predicate.h
8087 (add_condition, predicate::add_clause, predicate::operator &=,
8088 predicate::or_with, predicate::evaluate, predicate::probability,
8089 dump_condition, dump_clause, predicate::dump,
8090 predicate::remap_after_duplication, predicate::remap_after_inlining,
8091 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
8092 (evaluate_conditions_for_known_args): Update.
8093 (set_cond_stmt_execution_predicate): Update.
8094 * ipa-inline.h: Include ipa-predicate.h
8095 (condition, inline_param_summary, conditions, agg_position_info,
8096 predicate): Move to ipa-predicate.h
8097 * ipa-predicate.c: New file.
8098 * ipa-predicate.h: New file.
8100 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
8102 * final.c (leaf_function_p): Check we are not in a sequence.
8104 2017-05-18 Martin Liska <mliska@suse.cz>
8106 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
8107 * dumpfile.c (dump_register): Use new enum dump_kind.
8108 (get_dump_file_name): Likewise.
8109 (dump_enable_all): Likewise.
8110 (dump_switch_p_1): Likewise.
8111 (enable_rtl_dump_file): Remove usage of TDF_RTL.
8112 * dumpfile.h (enum dump_kind): New enum type.
8113 (struct dump_file_info): Create constructor and
8114 format fields and comments.
8115 * passes.c (pass_manager::register_one_dump_file):
8117 * statistics.c (statistics_early_init): Likewise.
8118 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
8119 TDF_TREE with TDF_SLIM.
8120 (gather_memory_references_ref): Likewise.
8122 2017-05-18 Martin Liska <mliska@suse.cz>
8124 * vec.h (struct vnull): Use it.
8126 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
8128 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
8129 (true_predicate, false_predicate, true_predicate_p,
8130 false_predicate_p): Remove.
8131 (single_cond_predicate, not_inlined_predicate): Turn to member function
8133 (add_condition): Update.
8134 (add_clause): Turn to...
8135 (predicate::add_clause): ... this one; update; allow passing NULL
8137 (and_predicates): Turn to ...
8138 (predicate::operator &=): ... this one.
8139 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
8140 (or_predicates): Turn to ...
8141 (predicate::or_with): ... this one.
8142 (evaluate_predicate): Turn to ...
8143 (predicate::evaluate): ... this one.
8144 (predicate_probability): Turn to ...
8145 (predicate::probability): ... this one.
8146 (dump_condition): Update.
8147 (dump_predicate): Turn to ...
8148 (predicate::dump): ... this one.
8149 (account_size_time): Update.
8150 (edge_set_predicate): Update.
8151 (set_hint_predicate): UPdate.
8152 (evaluate_conditions_for_known_args): Update.
8153 (evaluate_properties_for_edge): Update.
8154 (remap_predicate_after_duplication): Turn to...
8155 (predicate::remap_after_duplication): ... this one.
8156 (remap_hint_predicate_after_duplication): Update.
8157 (inline_summary_t::duplicate): UPdate.
8158 (dump_inline_edge_summary): Update.
8159 (dump_inline_summary): Update.
8160 (set_cond_stmt_execution_predicate): Update.
8161 (set_switch_stmt_execution_predicate): Update.
8162 (compute_bb_predicates): Update.
8163 (will_be_nonconstant_expr_predicate): Update.
8164 (will_be_nonconstant_predicate): Update.
8165 (phi_result_unknown_predicate): Update.
8166 (predicate_for_phi_result): Update.
8167 (array_index_predicate): Update.
8168 (estimate_function_body_sizes): Update.
8169 (estimate_node_size_and_time): Update.
8170 (estimate_ipcp_clone_size_and_time): Update.
8171 (remap_predicate): Rename to ...
8172 (predicate::remap_after_inlining): ... this one.
8173 (remap_hint_predicate): Update.
8174 (inline_merge_summary): Update.
8175 (inline_update_overall_summary): Update.
8176 (estimate_size_after_inlining): Update.
8177 (read_predicate): Rename to ...
8178 (predicate::stream_in): ... this one.
8179 (read_inline_edge_summary): Update.
8180 (write_predicate): Rename to ...
8181 (predicate::stream_out): ... this one.
8182 (write_inline_edge_summary): Update.
8183 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
8184 (clause_t): Turn to uint32_t
8185 (predicate): Turn to class; implement constructor and operators
8187 (size_time_entry): Update.
8188 (inline_summary): Update.
8189 (inline_edge_summary): Update.
8191 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
8193 * fold-const.c (fold_binary_loc): Move transformation...
8194 * match.pd (C - X CMP X): ... here.
8196 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
8198 * config/sparc/sparc.c (sparc_option_override): Set function
8199 alignment for -mcpu=niagara7 to 64 to match the I$ line.
8200 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
8202 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
8204 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
8206 2017-05-18 Marek Polacek <polacek@redhat.com>
8209 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
8210 (pass_ubsan::execute): Call gimple_assign_single_p instead of
8211 gimple_assign_load_p.
8213 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
8216 * real.c (do_compare): Give decimal_do_compare preference over
8217 comparing just the signs.
8219 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
8221 * doc/md.texi (Canonicalization of Instructions): Describe the
8222 canonical form of instructions that inherently set a condition
8225 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
8228 * tree-cfg.c: Move deletion of unreachable case statements to after
8229 the merging of consecutive case labels.
8231 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
8233 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
8234 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
8235 restoring of callee-saved registers.
8237 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
8239 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
8240 * config/visium/visium.c (single_set_and_flags): Likewise.
8241 * config/visium/visium.md (Substitutions): Likewise.
8243 2017-05-17 Martin Liska <mliska@suse.cz>
8245 * cfg.c: Introduce dump_flags_t type and
8246 use it instead of int type.
8248 * cfghooks.c: Likewise.
8249 * cfghooks.h (struct cfg_hooks): Likewise.
8250 * cfgrtl.c: Likewise.
8251 * cfgrtl.h: Likewise.
8252 * cgraph.c (cgraph_node::get_body): Likewise.
8253 * coretypes.h: Likewise.
8254 * domwalk.c: Likewise.
8255 * domwalk.h: Likewise.
8256 * dumpfile.c (struct dump_option_value_info): Likewise.
8257 (dump_enable_all): Likewise.
8258 (dump_switch_p_1): Likewise.
8259 (opt_info_switch_p): Likewise.
8260 * dumpfile.h (enum tree_dump_index): Likewise.
8261 (struct dump_file_info): Likewise.
8262 * genemit.c: Likewise.
8263 * generic-match-head.c: Likewise.
8264 * gengtype.c (open_base_files): Likewise.
8265 * gimple-pretty-print.c: Likewise.
8266 * gimple-pretty-print.h: Likewise.
8267 * graph.c (print_graph_cfg): Likewise.
8268 * graphite-scop-detection.c (dot_all_sese): Likewise.
8269 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
8270 * loop-unroll.c (report_unroll): Likewise.
8271 * passes.c (pass_manager::register_one_dump_file): Likewise.
8272 * print-tree.c: Likewise.
8273 * statistics.c: Likewise.
8274 * tree-cfg.c: Likewise.
8275 * tree-cfg.h: Likewise.
8276 * tree-dfa.c: Likewise.
8277 * tree-dfa.h: Likewise.
8278 * tree-dump.c (dump_function): Likewise.
8279 * tree-dump.h (struct dump_info): Likewise.
8280 * tree-pretty-print.c: Likewise.
8281 * tree-pretty-print.h: Likewise.
8282 * tree-ssa-live.c: Likewise.
8283 * tree-ssa-live.h: Likewise.
8284 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
8285 * tree-vect-loop.c: Likewise.
8286 * tree-vect-slp.c: Likewise.
8288 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
8289 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8291 PR tree-optimization/80457
8292 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
8293 of all arguments to a statement as scalar_to_vec operations.
8294 (vectorizable_call): Adjust call to vect_model_simple_cost for
8296 (vectorizable_conversion): Likewise.
8297 (vectorizable_assignment): Likewise.
8298 (vectorizable_shift): Likewise.
8299 (vectorizable_operation): Likewise.
8300 (vectorizable_comparison): Likewise.
8301 (vect_is_simple_cond): Record the def types for operands.
8302 (vectorizable_condition): Likewise, call vect_model_simple_cost.
8303 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
8304 for statement argument count.
8306 2017-05-16 Carl Love <cel@us.ibm.com>
8308 * config/rs6000/rs6000-c: Add support for built-in functions
8309 vector unsigned long long vec_bperm (vector unsigned long long,
8310 vector unsigned char)
8311 vector signed long long vec_mule (vector signed int,
8313 vector unsigned long long vec_mule (vector unsigned int,
8314 vector unsigned int)
8315 vector signed long long vec_mulo (vector signed int,
8317 vector unsigned long long vec_mulo (vector unsigned int,
8318 vector unsigned int)
8319 vector signed char vec_sldw (vector signed char,
8322 vector unsigned char vec_sldw (vector unsigned char,
8323 vector unsigned char,
8325 vector signed short vec_sldw (vector signed short,
8326 vector signed short,
8328 vector unsigned short vec_sldw (vector unsigned short,
8329 vector unsigned short,
8331 vector signed int vec_sldw (vector signed int,
8334 vector unsigned int vec_sldw (vector unsigned int,
8335 vector unsigned int,
8337 vector signed long long vec_sldw (vector signed long long,
8338 vector signed long long,
8340 vector unsigned long long vec_sldw (vector unsigned long long,
8341 vector unsigned long long,
8343 * config/rs6000/rs6000-c: Add support for built-in functions
8344 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
8345 * config/rs6000/altivec.h: Add defintion for vec_sldw.
8346 * doc/extend.texi: Update the built-in documentation for the
8347 new built-in functions.
8349 2017-05-16 Marek Polacek <polacek@redhat.com>
8353 * tree.c (save_expr): Don't fold the expression.
8355 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
8357 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
8358 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
8359 and (?*y,m). Update insn attributes.
8361 2017-05-16 Martin Liska <mliska@suse.cz>
8363 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
8364 flags argument of print_gimple_stmt, print_gimple_expr,
8365 print_generic_stmt and print_generic_expr.
8366 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
8367 * coretypes.h: Likewise.
8368 * except.c (dump_eh_tree): Likewise.
8369 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8370 * gimple-pretty-print.h: Likewise.
8371 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
8372 (backprop::push_to_worklist): Likewise.
8373 (backprop::pop_from_worklist): Likewise.
8374 (backprop::process_use): Likewise.
8375 (backprop::intersect_uses): Likewise.
8376 (note_replacement): Likewise.
8377 * gimple-ssa-store-merging.c
8378 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
8379 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
8380 (pass_store_merging::execute): Likewise.
8381 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
8382 (ssa_base_cand_dump_callback): Likewise.
8383 (dump_incr_vec): Likewise.
8384 (replace_refs): Likewise.
8385 (replace_mult_candidate): Likewise.
8386 (create_add_on_incoming_edge): Likewise.
8387 (create_phi_basis): Likewise.
8388 (insert_initializers): Likewise.
8389 (all_phi_incrs_profitable): Likewise.
8390 (introduce_cast_before_cand): Likewise.
8391 (replace_one_candidate): Likewise.
8392 * gimplify.c (gimplify_expr): Likewise.
8393 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
8394 (set_rename): Likewise.
8395 (rename_uses): Likewise.
8396 (copy_loop_phi_nodes): Likewise.
8397 (add_close_phis_to_merge_points): Likewise.
8398 (copy_loop_close_phi_args): Likewise.
8399 (copy_cond_phi_args): Likewise.
8400 (graphite_copy_stmts_from_block): Likewise.
8401 (translate_pending_phi_nodes): Likewise.
8402 * graphite-poly.c (print_pdr): Likewise.
8403 (dump_gbb_cases): Likewise.
8404 (dump_gbb_conditions): Likewise.
8405 (print_scop_params): Likewise.
8406 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
8407 (build_cross_bb_scalars_use): Likewise.
8408 (gather_bbs::before_dom_children): Likewise.
8409 * hsa-dump.c (dump_hsa_immed): Likewise.
8410 * ipa-cp.c (print_ipcp_constant_value): Likewise.
8411 (get_replacement_map): Likewise.
8412 * ipa-inline-analysis.c (dump_condition): Likewise.
8413 (estimate_function_body_sizes): Likewise.
8414 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
8415 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
8416 * ipa-prop.c (ipa_dump_param): Likewise.
8417 (ipa_print_node_jump_functions_for_edge): Likewise.
8418 (ipa_modify_call_arguments): Likewise.
8419 (ipa_modify_expr): Likewise.
8420 (ipa_dump_param_adjustments): Likewise.
8421 (ipa_dump_agg_replacement_values): Likewise.
8422 (ipcp_modif_dom_walker::before_dom_children): Likewise.
8423 * ipa-pure-const.c (check_stmt): Likewise.
8424 (pass_nothrow::execute): Likewise.
8425 * ipa-split.c (execute_split_functions): Likewise.
8426 * omp-offload.c (dump_oacc_loop_part): Likewise.
8427 (dump_oacc_loop): Likewise.
8428 * trans-mem.c (tm_log_emit): Likewise.
8429 (tm_memopt_accumulate_memops): Likewise.
8430 (dump_tm_memopt_set): Likewise.
8431 (dump_tm_memopt_transform): Likewise.
8432 * tree-cfg.c (gimple_verify_flow_info): Likewise.
8433 (print_loop): Likewise.
8434 * tree-chkp-opt.c (chkp_print_addr): Likewise.
8435 (chkp_gather_checks_info): Likewise.
8436 (chkp_get_check_result): Likewise.
8437 (chkp_remove_check_if_pass): Likewise.
8438 (chkp_use_outer_bounds_if_possible): Likewise.
8439 (chkp_reduce_bounds_lifetime): Likewise.
8440 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
8441 (chkp_mark_completed_bounds): Likewise.
8442 (chkp_register_incomplete_bounds): Likewise.
8443 (chkp_mark_invalid_bounds): Likewise.
8444 (chkp_maybe_copy_and_register_bounds): Likewise.
8445 (chkp_build_returned_bound): Likewise.
8446 (chkp_get_bound_for_parm): Likewise.
8447 (chkp_build_bndldx): Likewise.
8448 (chkp_get_bounds_by_definition): Likewise.
8449 (chkp_generate_extern_var_bounds): Likewise.
8450 (chkp_get_bounds_for_decl_addr): Likewise.
8451 * tree-chrec.c (chrec_apply): Likewise.
8452 * tree-data-ref.c (dump_data_reference): Likewise.
8453 (dump_subscript): Likewise.
8454 (dump_data_dependence_relation): Likewise.
8455 (analyze_overlapping_iterations): Likewise.
8456 * tree-inline.c (expand_call_inline): Likewise.
8457 (tree_function_versioning): Likewise.
8458 * tree-into-ssa.c (dump_defs_stack): Likewise.
8459 (dump_currdefs): Likewise.
8460 (dump_names_replaced_by): Likewise.
8461 (dump_update_ssa): Likewise.
8462 (update_ssa): Likewise.
8463 * tree-object-size.c (pass_object_sizes::execute): Likewise.
8464 * tree-parloops.c (build_new_reduction): Likewise.
8465 (try_create_reduction_list): Likewise.
8466 (ref_conflicts_with_region): Likewise.
8467 (oacc_entry_exit_ok_1): Likewise.
8468 (oacc_entry_exit_single_gang): Likewise.
8469 * tree-pretty-print.h: Likewise.
8470 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
8471 (get_scalar_evolution): Likewise.
8472 (add_to_evolution): Likewise.
8473 (get_loop_exit_condition): Likewise.
8474 (analyze_evolution_in_loop): Likewise.
8475 (analyze_initial_condition): Likewise.
8476 (analyze_scalar_evolution): Likewise.
8477 (instantiate_scev): Likewise.
8478 (number_of_latch_executions): Likewise.
8479 (gather_chrec_stats): Likewise.
8480 (final_value_replacement_loop): Likewise.
8481 (scev_const_prop): Likewise.
8482 * tree-sra.c (dump_access): Likewise.
8483 (disqualify_candidate): Likewise.
8484 (create_access): Likewise.
8486 (maybe_add_sra_candidate): Likewise.
8487 (create_access_replacement): Likewise.
8488 (analyze_access_subtree): Likewise.
8489 (analyze_all_variable_accesses): Likewise.
8490 (sra_modify_assign): Likewise.
8491 (initialize_constant_pool_replacements): Likewise.
8492 (find_param_candidates): Likewise.
8493 (decide_one_param_reduction): Likewise.
8494 (replace_removed_params_ssa_names): Likewise.
8495 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
8496 * tree-ssa-copy.c (dump_copy_of): Likewise.
8497 (copy_prop_visit_cond_stmt): Likewise.
8498 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
8499 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
8500 (record_equivalences_from_stmt): Likewise.
8501 * tree-ssa-dse.c (compute_trims): Likewise.
8502 (delete_dead_call): Likewise.
8503 (delete_dead_assignment): Likewise.
8504 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
8505 (forward_propagate_into_cond): Likewise.
8506 (pass_forwprop::execute): Likewise.
8507 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8508 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
8510 (move_computations_worker): Likewise.
8511 (execute_sm): Likewise.
8512 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
8513 (remove_exits_and_undefined_stmts): Likewise.
8514 (remove_redundant_iv_tests): Likewise.
8515 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
8516 (adjust_iv_update_pos): Likewise.
8517 * tree-ssa-math-opts.c (bswap_replace): Likewise.
8518 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
8519 (value_replacement): Likewise.
8520 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
8521 * tree-ssa-pre.c (print_pre_expr): Likewise.
8522 (get_representative_for): Likewise.
8523 (create_expression_by_pieces): Likewise.
8524 (insert_into_preds_of_block): Likewise.
8525 (eliminate_insert): Likewise.
8526 (eliminate_dom_walker::before_dom_children): Likewise.
8527 (eliminate): Likewise.
8528 (remove_dead_inserted_code): Likewise.
8529 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
8530 * tree-ssa-reassoc.c (get_rank): Likewise.
8531 (eliminate_duplicate_pair): Likewise.
8532 (eliminate_plus_minus_pair): Likewise.
8533 (eliminate_not_pairs): Likewise.
8534 (undistribute_ops_list): Likewise.
8535 (eliminate_redundant_comparison): Likewise.
8536 (update_range_test): Likewise.
8537 (optimize_range_tests_var_bound): Likewise.
8538 (optimize_vec_cond_expr): Likewise.
8539 (rewrite_expr_tree): Likewise.
8540 (rewrite_expr_tree_parallel): Likewise.
8541 (linearize_expr): Likewise.
8542 (break_up_subtract): Likewise.
8543 (linearize_expr_tree): Likewise.
8544 (attempt_builtin_powi): Likewise.
8545 (attempt_builtin_copysign): Likewise.
8546 (transform_stmt_to_copy): Likewise.
8547 (transform_stmt_to_multiply): Likewise.
8548 (dump_ops_vector): Likewise.
8549 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
8550 (print_scc): Likewise.
8551 (set_ssa_val_to): Likewise.
8552 (visit_reference_op_store): Likewise.
8553 (visit_use): Likewise.
8554 (sccvn_dom_walker::before_dom_children): Likewise.
8555 (run_scc_vn): Likewise.
8556 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
8558 (expr_hash_elt::print): Likewise.
8559 (const_and_copies::pop_to_marker): Likewise.
8560 (const_and_copies::record_const_or_copy_raw): Likewise.
8561 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
8562 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
8563 (dump_predicates): Likewise.
8564 (find_uninit_use): Likewise.
8565 (warn_uninitialized_phi): Likewise.
8566 (pass_late_warn_uninitialized::execute): Likewise.
8567 * tree-ssa.c (verify_vssa): Likewise.
8568 (verify_ssa): Likewise.
8569 (maybe_optimize_var): Likewise.
8570 * tree-vrp.c (dump_value_range): Likewise.
8571 (dump_all_value_ranges): Likewise.
8572 (dump_asserts_for): Likewise.
8573 (register_edge_assert_for_2): Likewise.
8574 (vrp_visit_cond_stmt): Likewise.
8575 (vrp_visit_switch_stmt): Likewise.
8576 (vrp_visit_stmt): Likewise.
8577 (vrp_visit_phi_node): Likewise.
8578 (simplify_cond_using_ranges_1): Likewise.
8579 (fold_predicate_in): Likewise.
8580 (evrp_dom_walker::before_dom_children): Likewise.
8581 (evrp_dom_walker::push_value_range): Likewise.
8582 (evrp_dom_walker::pop_value_range): Likewise.
8583 (execute_early_vrp): Likewise.
8585 2017-05-16 Richard Biener <rguenther@suse.de>
8587 * dwarf2out.c (loc_list_from_tree_1): Do not create
8588 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
8590 2017-05-16 Richard Biener <rguenther@suse.de>
8592 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
8594 (note_variable_value_in_expr): If we resolved the decl ref
8595 do not push to the stack.
8597 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
8599 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
8600 operations in fast-math mode.
8601 (vaddq_f16): Likewise.
8602 (vmul_f16): Likewise.
8603 (vmulq_f16): Likewise.
8604 (vsub_f16): Likewise.
8605 (vsubq_f16): Likewise.
8606 * config/arm/neon.md (add<mode>3): New.
8608 (fma:<VH:mode>3): New. Also remove outdated comment.
8611 2017-05-16 Martin Liska <mliska@suse.cz>
8615 * ipa-devirt.c (warn_types_mismatch): Fix typo.
8616 (odr_types_equivalent_p): Likewise.
8618 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
8620 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
8622 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
8625 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
8626 non-interunit SSE move alternatives with '?'.
8627 (zero-extendsidi peephole2): New peephole to skip intermediate
8628 general register in SSE zero-extend sequence.
8630 2017-05-15 Jeff Law <law@redhat.com>
8632 * reorg.c (relax_delay_slots): Create a new variable to hold
8633 the temporary target rather than clobbering TARGET_LABEL.
8635 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
8636 missing argument to extract_bit_field call.
8637 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
8639 2017-05-15 Martin Liska <mliska@suse.cz>
8642 * gcc.c (process_command): Do not allow empty argument of -o option.
8644 2017-05-15 Renlin Li <renlin.li@arm.com>
8646 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
8647 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
8648 * config/aarch64/constraints.md (Usf): Add long call check.
8649 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
8650 (call_value): Likewise.
8651 (sibcall): Likewise.
8652 (sibcall_value): Likewise.
8654 (call_value_insn): New.
8655 (sibcall_insn): Update rtx pattern.
8656 (sibcall_value_insn): Likewise.
8657 (call_internal): Remove.
8658 (call_value_internal): Likewise.
8659 (sibcall_internal): Likewise.
8660 (sibcall_value_internal): Likewise.
8661 (call_reg): Likewise.
8662 (call_symbol): Likewise.
8663 (call_value_reg): Likewise.
8664 (call_value_symbol): Likewise.
8666 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
8669 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
8671 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
8673 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
8674 compatible with CCGOCmode and with CCZmode.
8676 2017-05-14 Martin Sebor <msebor@redhat.com>
8679 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
8680 (gimple_fold_builtin_snprintf): Same.
8681 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
8682 (gimple_fold_builtin_snprintf): Same.
8683 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
8685 (is_call_safe): New function.
8686 (try_substitute_return_value): Call it.
8687 (try_simplify_call): New function.
8688 (pass_sprintf_length::handle_gimple_call): Call it.
8690 2017-05-14 Martin Sebor <msebor@redhat.com>
8693 * builtins.c (expand_builtin_stpncpy): Simplify.
8695 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
8697 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
8698 * config/i386/i386.h
8699 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
8700 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
8701 (struct machine_function): Add new members call_ms2sysv,
8702 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
8703 (struct machine_frame_state): New fields sp_realigned and
8704 sp_realigned_offset.
8705 * config/i386/i386.c
8706 (enum xlogue_stub): New enum.
8707 (enum xlogue_stub_sets): New enum.
8708 (class xlogue_layout): New class.
8709 (struct ix86_frame): New fields stack_realign_allocate_offset,
8710 stack_realign_offset and outlined_save_offset. Modify comments to
8711 detail stack layout when using out-of-line stubs.
8712 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
8713 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
8714 -mcall-ms2sysv-xlogues.
8715 (stub_managed_regs): New static variable.
8716 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
8717 registers managed by out-of-line stub.
8718 (disable_call_ms2sysv_xlogues): New function.
8719 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
8720 m->call_ms2sysv when appropriate and compute frame layout for
8722 (sp_valid_at, fp_valid_at): New inline functions.
8723 (choose_basereg): New function.
8724 (choose_baseaddr): Add align parameter, use choose_basereg and modify
8726 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
8727 Use align parameter of choose_baseaddr to generated aligned SSE movs
8729 (pro_epilogue_adjust_stack): Modify to track
8730 machine_frame_state::sp_realigned.
8731 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
8732 (ix86_nsaved_sseregs): Likewise.
8733 (ix86_emit_save_regs): Likewise.
8734 (ix86_emit_save_regs_using_mov): Likewise.
8735 (ix86_emit_save_sse_regs_using_mov): Likewise.
8736 (get_scratch_register_on_entry): Likewise.
8737 (gen_frame_set): New function.
8738 (gen_frame_load): Likewise.
8739 (gen_frame_store): Likewise.
8740 (emit_outlined_ms2sysv_save): Likewise.
8741 (emit_outlined_ms2sysv_restore): Likewise.
8742 (ix86_expand_prologue): Modify stack re-alignment code and call
8743 emit_outlined_ms2sysv_save when appropriate.
8744 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
8745 parameter rtx_insn *insn, which allows the function to be used to only
8747 (ix86_expand_epilogue): Modify validity checks of frame and stack
8748 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
8749 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
8750 * config/i386/predicates.md
8751 (save_multiple): New predicate.
8752 (restore_multiple): Likewise.
8753 * config/i386/sse.md
8754 (save_multiple<mode>): New pattern.
8755 (save_multiple_realign<mode>): Likewise.
8756 (restore_multiple<mode>): Likewise.
8757 (restore_multiple_and_return<mode>): Likewise.
8758 (restore_multiple_leave_return<mode>): Likewise.
8759 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
8761 2017-05-14 Julia Koval <julia.koval@intel.com>
8763 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
8764 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
8765 (__builtin_ia32_xsetbv): New builtins.
8766 * config/i386/i386.c (ix86_expand_special_args_builtin):
8768 (ix86_expand_builtin): Special expand for new intrinsics.
8769 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
8770 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
8771 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
8773 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8775 * cfganal.c (inverted_post_order_compute): Change argument type
8777 * cfganal.h (inverted_post_order_compute): Adjust prototype.
8778 * df-core.c (rest_of_handle_df_initialize): Adjust.
8779 (rest_of_handle_df_finish): Likewise.
8780 (df_analyze_1): Likewise.
8781 (df_analyze): Likewise.
8782 (loop_inverted_post_order_compute): Change argument to be a vec *.
8783 (df_analyze_loop): Adjust.
8784 (df_get_n_blocks): Likewise.
8785 (df_get_postorder): Likewise.
8786 * df.h (struct df_d): Change field to be a vec.
8787 * lcm.c (compute_laterin): Adjust.
8788 (compute_available): Likewise.
8789 * lra-lives.c (lra_create_live_ranges_1): Likewise.
8790 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
8791 * tree-ssa-pre.c (compute_antic): Likewise.
8793 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8795 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
8796 (depth_first_search::depth_first_search): Change structure init
8797 function to this constructor.
8798 (depth_first_search::add_bb): Rename function to this member.
8799 (depth_first_search::execute): Likewise.
8800 (flow_dfs_compute_reverse_finish): Adjust.
8802 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8804 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
8805 (longest_simple_path): Likewise.
8806 * shrink-wrap.c (spread_components): Likewise.
8807 (disqualify_problematic_components): Likewise.
8808 (emit_common_heads_for_components): Likewise.
8809 (emit_common_tails_for_components): Likewise.
8810 (insert_prologue_epilogue_for_components): Likewise.
8812 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8814 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
8817 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8819 * df-core.c (df_set_blocks): Start using auto_bitmap.
8820 (df_compact_blocks): Likewise.
8821 * df-problems.c (df_rd_confluence_n): Likewise.
8822 * df-scan.c (df_insn_rescan_all): Likewise.
8823 (df_process_deferred_rescans): Likewise.
8824 (df_update_entry_block_defs): Likewise.
8825 (df_update_exit_block_uses): Likewise.
8826 (df_entry_block_bitmap_verify): Likewise.
8827 (df_exit_block_bitmap_verify): Likewise.
8828 (df_scan_verify): Likewise.
8829 * lra-constraints.c (lra_constraints): Likewise.
8830 (undo_optional_reloads): Likewise.
8831 (lra_undo_inheritance): Likewise.
8832 * lra-remat.c (calculate_gen_cands): Likewise.
8833 (do_remat): Likewise.
8834 * lra-spills.c (assign_spill_hard_regs): Likewise.
8835 (spill_pseudos): Likewise.
8836 * tree-ssa-pre.c (bitmap_set_and): Likewise.
8837 (bitmap_set_subtract_values): Likewise.
8839 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8841 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
8842 management with auto_bitmap.
8843 (fix_inter_tick): Likewise.
8844 (fix_recovery_deps): Likewise.
8845 * ira.c (add_store_equivs): Likewise.
8846 (find_moveable_pseudos): Likewise.
8847 (split_live_ranges_for_shrink_wrap): Likewise.
8848 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
8849 (rtx_reuse_manager::seen_def_p): Likewise.
8850 (rtx_reuse_manager::set_seen_def): Likewise.
8851 * print-rtl.h (class rtx_reuse_manager): Likewise.
8853 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8855 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
8857 (migrate_btr_def): Likewise.
8858 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
8859 * df-core.c (loop_post_order_compute): Likewise.
8860 (loop_inverted_post_order_compute): Likewise.
8861 * hsa-common.h: Likewise.
8862 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
8863 * init-regs.c (initialize_uninitialized_regs): Likewise.
8864 * ipa-inline.c (resolve_noninline_speculation): Likewise.
8865 (inline_small_functions): Likewise.
8866 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
8867 * ira.c (combine_and_move_insns): Likewise.
8868 (build_insn_chain): Likewise.
8869 * loop-invariant.c (find_invariants): Likewise.
8870 * lower-subreg.c (propagate_pseudo_copies): Likewise.
8871 * predict.c (tree_predict_by_opcode): Likewise.
8872 (predict_paths_leading_to): Likewise.
8873 (predict_paths_leading_to_edge): Likewise.
8874 (estimate_loops_at_level): Likewise.
8875 (estimate_loops): Likewise.
8876 * shrink-wrap.c (try_shrink_wrapping): Likewise.
8877 (spread_components): Likewise.
8878 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
8879 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
8880 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
8881 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
8882 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
8883 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
8884 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
8885 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
8886 (mark_threaded_blocks): Likewise.
8887 (thread_through_all_blocks): Likewise.
8888 * tree-ssa.c (verify_ssa): Likewise.
8889 (execute_update_addresses_taken): Likewise.
8890 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
8892 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8894 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
8896 (post_order_compute): Likewise.
8897 (inverted_post_order_compute): Likewise.
8898 (pre_and_rev_post_order_compute_fn): Likewise.
8900 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8902 * genrecog.c (int_set::int_set): Explicitly construct our
8903 auto_vec base class.
8904 * vec.h (auto_vec::auto_vec): New constructor.
8906 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8908 * bitmap.h (class auto_bitmap): New constructor taking
8909 bitmap_obstack * argument.
8911 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8913 * bitmap.h (class auto_bitmap): Change type of m_bits to
8914 bitmap_head, and adjust ctor / dtor and member operators.
8916 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
8918 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
8919 when returned register mode doesn't match original mode.
8921 2017-05-12 Jeff Law <law@redhat.com>
8922 Jakub Jelinek <jakub@redhat.com>
8924 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
8925 we look for cc setter after the compare-elim changes.
8926 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
8927 within the vector to match what compare-elim now expects.
8928 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
8929 (xorsi3_flags, one_cmplsi2_flags): Likewise.
8931 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
8932 after the compare-elim changes.
8933 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
8934 the vector to match what compare-elim now expects.
8935 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
8936 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
8937 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
8938 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
8939 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
8941 * config/visium/visium.c (single_set_and_flags): Fix where
8942 we look for cc setter after the compare-elim changes.
8943 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
8944 with the vector to match what compare-elim now expects.
8945 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
8946 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
8947 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
8948 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
8949 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
8950 (neg<mode>2_insn_set_overflow): Likewise.
8952 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
8955 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
8956 maybe_expand_insn call, set ops[0].target. If still set after call,
8957 set alt_rtl. Add extra arg to recursive calls.
8958 (extract_bit_field): Add alt_rtl argument. Pass to
8960 * expmed.h (extract_bit_field): Fix prototype.
8961 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
8962 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
8963 to extract_bit_field_calls.
8964 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
8965 Pass alt_rtl to extract_bit_field calls.
8966 * calls.c (store_unaligned_arguments_into_psuedos)
8967 load_register_parameters): Pass extra NULL to extract_bit_field calls.
8968 * optabs.c (maybe_legitimize_operand): Clear op->target when call
8970 * optabs.h (struct expand_operand): Add target bitfield.
8972 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
8974 * compare-elim.c (try_eliminate_compare): Canonicalize
8975 operation with embedded compare to
8976 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
8977 (set (reg) (operation)].
8979 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
8981 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
8984 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
8985 cost of adding a carry flag for ADC instruction.
8986 [case MINUS]: Ignore the cost of subtracting a carry flag
8987 for SBB instruction.
8989 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
8991 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
8993 * config/rs6000/bmiintrin.h: New file.
8994 * config/rs6000/bmi2intrin.h: New file.
8995 * config/rs6000/x86intrin.h: New file.
8997 2017-05-12 Jeff Law <law@redhat.com>
8999 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
9002 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
9005 * tree-cfg.c: Remove cfg edges of unreachable case statements.
9007 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9009 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
9010 early expansion of vector divide builtins.
9011 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
9012 builtins identified as having unsigned arguments.
9014 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9016 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
9017 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
9018 expansion of vector logical operations (and, andc, or, xor,
9021 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9023 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
9024 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
9026 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9028 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
9029 early GIMPLE expansion of vector multiplies.
9031 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9033 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
9034 TARGET_HAVE_MOVT conditional.
9035 (movt splitter): Likewise.
9037 2017-05-12 Richard Biener <rguenther@suse.de>
9039 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
9040 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9041 Fold all stmts not inplace.
9043 2017-05-12 Richard Biener <rguenther@suse.de>
9045 PR tree-optimization/80713
9046 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
9047 inserted_exprs bit for not removed stmts.
9049 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
9052 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
9053 parallelized" attribute for parallelized OpenACC kernels.
9054 * omp-offload.c (execute_oacc_device_lower): Use it.
9056 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
9057 Set "oacc kernels" attribute.
9058 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
9059 parameter. Adjust all users.
9060 (oacc_fn_attrib_kernels_p): Remove function.
9061 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
9062 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
9063 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
9064 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
9065 assert "oacc kernels" attribute is set.
9067 2017-05-11 Carl Love <cel@us.ibm.com>
9069 * config/rs6000/rs6000-c: Add support for built-in functions
9070 vector unsigned char vec_popcnt (vector signed char)
9071 vector unsigned char vec_popcnt (vector unsigned char)
9072 vector unsigned short vec_popcnt (vector signed short)
9073 vector unsigned short vec_popcnt (vector unsigned short)
9074 vector unsigned int vec_popcnt (vector signed int)
9075 vector unsigned int vec_popcnt (vector unsigned int)
9076 vector unsigned long long vec_popcnt (vector signed long long)
9077 vector unsigned long long vec_popcnt (vector unsigned long long)
9078 vector signed long long vec_slo (vector signed long long,
9080 vector signed long long vec_slo (vector signed long long,
9081 vector unsigned char)
9082 vector unsigned long long vec_slo (vector unsigned long long,
9084 vector unsigned long long vec_slo (vector unsigned long long,
9085 vector unsigned char)
9086 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
9087 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
9088 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
9089 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
9090 * doc/extend.texi: Update the built-in documentation file for the
9091 new built-in functions.
9093 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
9095 * attribs.h (sorted_attr_string): Move machine independent
9096 functions for target clone support from the i386 port to common
9097 code. Rename ix86_function_versions to common_function_versions.
9098 Rename make_name to make_unique_name.
9099 (common_function_versions): Likewise.
9100 (make_unique_name): Likewise.
9101 (make_dispatcher_decl): Likewise.
9102 (is_function_default_version): Likewise.
9103 * attribs.c (attr_strcmp): Likewise.
9104 (sorted_attr_string): Likewise.
9105 (common_function_versions): Likewise.
9106 (make_unique_name): Likewise.
9107 (make_dispatcher_decl): Likewise.
9108 (is_function_default_version): Likewise.
9109 * config/i386/i386.c (attr_strcmp): Likewise.
9110 (sorted_attr_string): Likewise.
9111 (ix86_function_versions): Likewise.
9112 (make_name): Likewise.
9113 (make_dispatcher_decl): Likewise.
9114 (is_function_default_version): Likewise.
9115 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
9117 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9120 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
9121 Account for direct move costs for vec_construct of integer
9124 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
9127 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
9128 (UNSPEC_STX_ATOMIC): Ditto.
9129 (loaddi_via_sse): New insn.
9130 (storedi_via_sse): Ditto.
9131 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
9132 Update corresponding peephole2 patterns.
9133 (atomic_storedi_fpu): Ditto.
9135 2017-05-11 Julia Koval <julia.koval@intel.com>
9137 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
9138 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
9140 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
9141 (__builtin_ia32_rsqrt14ss_mask): New builtins.
9142 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
9144 2017-05-11 Nathan Sidwell <nathan@acm.org>
9146 * graphite-poly.c: Include dumpfile.h.
9148 * dumpfle.h (dump_function): Declare here ...
9149 * tree-dump.h (dump_function): ... not here.
9150 * dumpfile.c: #include tree-cfg.h.
9151 (dump_function): Move here from ...
9152 * tree-dump.c (dump_function): ... here.
9153 * gimplify.c: #include splay-tree.h, not tree-dump.h.
9154 * graphite-poly.c: Don't include tree-dump.h.
9155 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
9156 * print-tree.c: Likewise.
9157 * stor-layout.c: Likewise.
9158 * tree-nested.c: Likewise.
9160 * dumpfile.c (dump_start): Use TDF_FLAGS.
9161 (dump_enable_all): Fix TDF_KIND check thinko.
9163 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
9165 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9166 array entries to represent two legal parameterizations of the
9167 overloaded __builtin_cmpb function, as represented by the
9168 P6_OV_BUILTIN_CMPB constant.
9169 (altivec_resolve_overloaded_builtin): Add special case handling
9170 for the __builtin_cmpb function, as represented by the
9171 P6_OV_BUILTIN_CMPB constant.
9172 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
9173 (BU_P6_64BIT_2): New macro.
9174 (BU_P6_OVERLOAD_2): New macro
9175 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
9176 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
9177 (CMPB): Add overload support to represent both 32-bit and 64-bit
9178 compare-bytes function.
9179 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
9180 support for TARGET_CMPB.
9181 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
9182 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
9183 documentation of the __builtin_cmpb overloaded built-in function.
9185 2017-05-11 Richard Biener <rguenther@suse.de>
9187 PR tree-optimization/80705
9188 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
9189 bases are not vectorizable.
9191 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9193 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
9194 when counting register pressure.
9196 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9198 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
9199 (iv_ca_more_deps): Renamed to ...
9200 (iv_ca_compare_deps): ... this.
9201 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
9203 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9205 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
9207 (determine_group_iv_costs): ... here.
9208 (find_inv_vars_cb): Record inv var if it's not recorded before.
9210 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9212 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
9213 (get_shiftadd_cost): Ditto.
9215 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9217 * tree-ssa-address.c: Include header file.
9218 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
9220 (add_to_parts): Refactor.
9221 (addr_to_parts): New parameter. Update use of move_hint_to_base.
9222 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
9225 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9227 PR tree-optimization/53090
9228 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
9230 (extract_cond_operands): Detect condition with IV on both sides
9231 and return COMP_IV_EXPR_2.
9232 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
9233 (rewrite_use_compare): Simplify by removing call to function
9234 extract_cond_operands.
9236 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9238 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
9239 (extract_cond_operands): Detect condition comparing against non-
9240 invariant bound and return appropriate enum value.
9241 (find_interesting_uses_cond): Update use of extract_cond_operands.
9242 Handle its return value accordingly.
9243 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
9245 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9247 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
9248 nonlinear iv_use computation in loop invariant sensitive way.
9250 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9252 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
9253 (find_iv_candidates): Call relate_compare_use_with_all_cands.
9255 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9257 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
9258 (dump_cand): Support iv_cand.inv_exprs.
9259 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
9261 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
9264 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9266 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
9268 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
9269 as local function. Include necessary header files.
9270 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
9272 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9274 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
9276 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9278 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
9279 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
9280 RSHIFT_EXPR and BIT_NOT_EXPR.
9282 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9284 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
9285 (adjust_setup_cost): New parameter supporting round up adjustment.
9286 (struct address_cost_data): Delete.
9287 (force_expr_to_var_cost): Don't bound cost with spill_cost.
9288 (split_address_cost, ptr_difference_cost): Delete.
9289 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
9290 (struct ainc_cost_data): New struct.
9291 (get_address_cost_ainc): New function.
9292 (get_address_cost, get_computation_cost): Reimplement.
9293 (determine_group_iv_cost_address): Record inv_expr for all uses of
9295 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
9296 (iv_ca_has_deps): Reimplemented to ...
9297 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
9299 (iv_ca_extend): Call iv_ca_more_deps.
9301 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9303 * tree-ssa-address.c (struct mem_address): Move to header file.
9304 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
9305 * tree-ssa-address.h (struct mem_address): Move from C file.
9306 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
9308 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9310 * tree-affine.h (aff_combination_type): New interface.
9311 (aff_combination_zero_p): Remove static.
9312 (aff_combination_const_p): New interface.
9313 (aff_combination_singleton_var_p): New interfaces.
9315 2017-05-11 Richard Biener <rguenther@suse.de>
9317 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9318 Skip unreachable blocks and destinations.
9319 (eliminate): Move stmt removal and fixup ...
9320 (fini_eliminate): ... here. Skip inserted exprs.
9321 (pass_pre::execute): Move fini_pre after fini_eliminate.
9322 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
9323 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
9324 PRE to get rid of dead code that has invalid SSA form and
9325 split critical edges again.
9327 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9329 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
9331 2017-05-11 Richard Biener <rguenther@suse.de>
9333 * passes.c (execute_function_todo): Verify loops if they are
9334 said to be up-to-date.
9335 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
9336 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
9338 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
9341 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
9342 handle calling assemble_external ourself.
9345 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
9346 modes with zero size. Enhance comment.
9348 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9350 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
9351 built-ins for vec_xl and vec_xst with short and char pointer
9354 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
9356 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
9357 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
9358 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
9359 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
9360 (_mm_maskz_min_round_ss): New intrinsics.
9361 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
9362 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
9363 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
9364 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
9365 (__builtin_ia32_minss_mask_round): New builtins.
9366 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9367 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9368 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
9370 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
9371 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
9373 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
9376 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
9378 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
9379 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
9380 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
9381 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
9382 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
9383 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
9384 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
9385 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9386 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
9387 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
9388 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
9389 (__builtin_ia32_mulss_mask_round): New builtins.
9390 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9391 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9392 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
9394 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
9395 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
9397 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
9400 2017-05-10 Julia Koval <julia.koval@intel.com>
9402 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
9403 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
9404 (_mm256_setr_m128i): New intrinsics.
9406 2017-05-10 Julia Koval <julia.koval@intel.com>
9408 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
9409 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
9410 (_mm_maskz_rcp14_ss): New intrinsics.
9411 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
9412 (__builtin_ia32_rcp14ss_mask): New builtins.
9413 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
9415 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
9417 PR tree-optimization/51513
9418 * tree-cfg.c (gimple_seq_unreachable_p): New function.
9419 (assert_unreachable_fallthru_edge_p): Use it.
9420 (group_case_labels_stmt): Likewise.
9421 * tree-cfg.h: Prototype it.
9422 * stmt.c: Include cfghooks.h and tree-cfg.h.
9423 (emit_case_dispatch_table) <gap_label>: New local variable.
9424 Use it to fill dispatch table gaps.
9425 Test for default_label before updating probabilities.
9426 (expand_case) <default_label>: Remove unneeded initialization.
9427 Test for unreachable default case statement and remove its edge.
9428 Set default_label accordingly.
9429 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
9431 2017-05-10 Carl Love <cel@us.ibm.com>
9433 * config/rs6000/rs6000-c: Add support for built-in functions
9434 vector signed char vec_neg (vector signed char)
9435 vector signed short int vec_neg (vector short int)
9436 vector signed int vec_neg (vector signed int)
9437 vector signed long long vec_neg (vector signed long long)
9438 vector float vec_neg (vector float)
9439 vector double vec_neg (vector double)
9440 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
9442 * config/rs6000/altivec.h: Add define for vec_neg
9443 * doc/extend.texi: Update the built-in documentation for the
9444 new built-in functions.
9446 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9448 PR tree-optimization/77644
9449 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
9451 2017-05-10 Nathan Sidwell <nathan@acm.org>
9453 * dumpfile.h (TDI_lang_all): New.
9454 (TDF_KIND): New. Renumber others
9455 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
9457 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
9459 (get_dump_file_name): Adjust suffix generation.
9460 (dump_enable_all): Use TDF_KIND.
9461 * doc/invoke.texi (-fdump-lang-all): Document.
9463 * dumpfile.h: Tabify.
9465 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
9468 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
9469 Move member access before delete.
9471 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
9473 * tree-inline.c (expand_call_inline): Split block at stmt
9476 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9479 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
9480 are now unused after splitting mov{sf,sd}_hardfloat.
9481 (f32_lr2): Likewise.
9483 (f32_lm2): Likewise.
9485 (f32_li2): Likewise.
9488 (f32_sr2): Likewise.
9490 (f32_sm2): Likewise.
9492 (f32_si2): Likewise.
9495 (f32_vsx): Likewise.
9497 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
9498 For movsf, order stores so the VSX stores occur before the GPR
9499 store which encourages the register allocator to use a traditional
9500 FPR instead of a GPR. For movsd, order the stores so that the GPR
9501 store comes before the VSX stores to allow the power6 to work.
9502 This is due to the power6 not having a 32-bit integer store
9503 instruction from a FPR.
9504 (movsf_hardfloat): Likewise.
9505 (movsd_hardfloat): Likewise.
9507 2017-05-09 Martin Sebor <msebor@redhat.com>
9509 PR translation/80280
9510 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
9513 PR translation/80280
9514 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
9515 data member added in r247778.
9516 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
9518 2017-05-09 Nathan Sidwell <nathan@acm.org>
9520 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
9522 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
9525 2017-05-09 Marek Polacek <polacek@redhat.com>
9527 * doc/invoke.texi: Fix typo.
9529 2017-05-09 Richard Biener <rguenther@suse.de>
9531 * tree-vrp.c (vrp_val_is_max): Adjust comment.
9532 (vrp_val_is_min): Likewise.
9533 (set_value_range_to_value): Likewise.
9534 (set_value_range_to_nonnegative): Likewise.
9535 (gimple_assign_nonzero_p): Likewise.
9536 (gimple_stmt_nonzero_p): Likewise.
9537 (vrp_int_const_binop): Likewise. Remove unreachable case.
9538 (adjust_range_with_scev): Adjust comments.
9539 (compare_range_with_value): Likewise.
9540 (extract_range_from_phi_node): Likewise.
9541 (test_for_singularity): Likewise.
9543 2017-05-09 Richard Biener <rguenther@suse.de>
9545 * tree-vrp.c (get_single_symbol): Add assert that we don't
9546 get overflowed constants as invariant part.
9547 (compare_values_warnv): Add comment before the TREE_NO_WARNING
9548 checks. Use wi::cmp instead of recursing for integer constants.
9549 (compare_values): Just ignore whether we assumed undefined
9550 overflow instead of failing the compare.
9551 (extract_range_for_var_from_comparison_expr): Add comment before the
9552 TREE_NO_WARNING sets.
9553 (test_for_singularity): Likewise.
9554 (extract_range_from_comparison): Do not disable optimization
9555 when we assumed undefined overflow.
9556 (extract_range_basic): Remove init of unused var.
9558 2017-05-09 Richard Biener <rguenther@suse.de>
9560 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
9561 (extract_range_from_multiplicative_op_1): Adjust.
9562 (extract_range_from_binary_expr_1): Use int_const_binop.
9564 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
9567 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
9568 rs6000_store_data_bypass_p in seven define_bypass directives and
9569 in several comments.
9570 * config/rs6000/rs6000-protos.h: Add prototype for
9571 rs6000_store_data_bypass_p function.
9572 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
9573 function implements slightly different (rs6000-specific) semantics
9574 than store_data_bypass_p, returning false rather than aborting
9575 with assertion error when arguments do not satisfy the
9576 requirements of store data bypass.
9577 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
9578 rs6000_store_data_bypass_p.
9580 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
9582 * config/xtensa/xtensa-protos.h
9583 (xtensa_initial_elimination_offset): New declaration.
9584 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
9585 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
9586 macro definition, add case for FRAME_POINTER_REGNUM when
9587 FRAME_GROWS_DOWNWARD.
9588 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
9589 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
9590 xtensa_initial_elimination_offset.
9592 2017-05-08 Nathan Sidwell <nathan@acm.org>
9594 * doc/invoke.texi: Alphabetize -fdump options.
9596 2017-05-08 Martin Sebor <msebor@redhat.com>
9598 PR translation/80280
9599 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
9601 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
9603 * target.def (compute_frame_layout): New optional target hook.
9604 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
9605 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
9606 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
9608 * reload1.c (verify_initial_elim_offsets): Likewise.
9609 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
9610 (use_simple_return_p): Call arm_compute_frame_layout if needed.
9611 (arm_get_frame_offsets): Split up into this ...
9612 (arm_compute_frame_layout): ... and this function.
9614 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
9616 * config/aarch64/constraints.md (Usa): New constraint.
9617 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
9619 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
9621 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
9622 with_multilib_list after it has been checked.
9624 2017-05-08 Richard Biener <rguenther@suse.de>
9626 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
9627 (bitmap_set_subtract_values): Likewise.
9629 2017-05-08 Richard Biener <rguenther@suse.de>
9631 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
9632 (gimple_assign_nonzero): ... this and remove strict_overflow_p
9634 (gimple_stmt_nonzero_warnv_p): Rename to ...
9635 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
9637 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
9638 (extract_range_basic): Adjust, do not disable propagation on
9639 strict overflow sensitive simplification.
9640 (vrp_visit_cond_stmt): Likewise.
9642 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
9644 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
9645 body size unconditionally.
9647 2017-05-07 Jeff Law <law@redhat.com>
9650 2017-05-06 Jeff Law <law@redhat.com>
9651 PR tree-optimization/78496
9652 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9655 PR tree-optimization/78496
9656 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9657 (simplify_stmt_using_ranges): Call it.
9658 (vrp_dom_walker::before_dom_children): Extract equivalences
9659 from an ASSERT_EXPR with an equality comparison against a
9662 2017-05-06 Jeff Law <law@redhat.com>
9664 PR tree-optimization/78496
9665 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9668 PR tree-optimization/78496
9669 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9670 (simplify_stmt_using_ranges): Call it.
9671 (vrp_dom_walker::before_dom_children): Extract equivalences
9672 from an ASSERT_EXPR with an equality comparison against a
9675 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9677 * lra-constraints.c (lra_copy_reg_equiv): New function.
9678 (split_reg): Use it to copy equivalence information from the
9679 original register to the spill register.
9681 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9683 PR rtl-optimization/75964
9684 * simplify-rtx.c (simplify_const_relational_operation): Remove
9685 invalid handling of comparisons of integer ABS.
9687 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
9689 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
9691 (init_regs): Remove declaration.
9692 (function_arg_advance_32): Initialize error_p as boolean variable.
9694 2017-05-05 Nathan Sidwell <nathan@acm.org>
9696 * store-motion.c (remove_reachable_equiv_notes): Reformat long
9697 lines. Use for (;;).
9699 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9701 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
9702 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
9703 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
9704 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
9705 VF=2 that require versioning.
9707 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9709 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
9712 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9714 * diagnostic.h (diagnostic_override_option_index): Convert from
9715 macro to inline function.
9717 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9719 * diagnostic.c (last_module_changed_p): New function.
9720 (set_last_module): New function.
9721 (diagnostic_report_current_module): Convert macro usage to
9722 the above functions.
9723 * diagnostic.h (diagnostic_context::last_module): Strengthen
9724 from const line_map * to const line_map_ordinary *.
9725 (diagnostic_last_module_changed): Delete macro.
9726 (diagnostic_set_last_module): Delete macro.
9728 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9730 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
9731 with diagnostic_report_diagnostic.
9732 (diagnostic_n_impl_richloc): Likewise.
9733 * diagnostic.h (report_diagnostic): Delete macro.
9734 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
9735 with diagnostic_report_diagnostic.
9736 * substring-locations.c (format_warning_va): Likewise.
9738 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9740 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
9741 save/restor of format_spec. Move option-printing code to...
9742 (print_option_information): ...this new function, and
9743 reimplement by simply printing to the pretty_printer,
9744 rather than appending to the format string.
9746 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9748 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
9749 handling logic into...
9750 (update_effective_level_from_pragmas): ...this new function.
9752 2017-05-04 Andrew Waterman <andrew@sifive.com>
9754 * config/riscv/riscv.opt (mstrict-align): New option.
9755 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
9756 (SLOW_UNALIGNED_ACCESS): Define.
9757 (riscv_slow_unaligned_access): Declare.
9758 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
9760 (riscv_slow_unaligned_access): New variable.
9761 (rocket_tune_info): Set slow_unaligned_access to true.
9762 (optimize_size_tune_info): Set slow_unaligned_access to false.
9763 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
9764 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
9765 (riscv_option_override): Set riscv_slow_unaligned_access.
9766 * doc/invoke.texi: Add -mstrict-align to RISC-V.
9768 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
9770 * config/riscv/riscv.md: Unify indentation.
9772 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
9777 * config/rs6000/rs6000.md (u code attribute): Add FIX and
9779 (extendsi<mode>2): Add support for doing sign extension via
9780 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
9781 don't have ISA 3.0 instructions.
9782 (extendsi<mode>2 splitter): Likewise.
9783 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
9784 generate the normal insns since SImode can now go in vector
9785 registers. Disallow the special UNSPECs needed for previous
9786 machines to hide SImode being used. Add new insns
9787 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
9788 (fix_trunc<mode>si2_stfiwx): Likewise.
9789 (fix_trunc<mode>si2_internal): Likewise.
9790 (fixuns_trunc<mode>si2): Likewise.
9791 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9792 (fctiw<u>z_<mode>_smallint): Likewise.
9793 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
9794 of floating point to 32-bit integer from doing a direct move to
9795 the GPR registers to do a store.
9796 (fctiwz_<mode>): Break long line.
9798 2017-05-05 Bin Cheng <bin.cheng@arm.com>
9800 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
9801 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
9802 (addr_list, addr_offset_valid_p): New.
9803 (split_address_groups): Check offset validity with above function.
9804 (gt-tree-ssa-loop-ivopts.h): Include header file.
9806 2017-05-05 Nathan Sidwell <nathan@acm.org>
9808 * config.gcc (arm*-*-*): Add missing 'fi'.
9810 2017-05-05 Steve Ellcey <sellcey@cavium.com>
9812 * doc/invoke.texi (-fopt-info): Explicitly say order of options
9813 included in -fopt-info does not matter.
9814 * doc/optinfo.texi (-fopt-info): Fix description of default
9815 behavour. Explicitly say order of options included in -fopt-info
9818 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
9820 * config.gcc: Allow combinations of aprofile and rmprofile values for
9821 --with-multilib-list.
9822 * config/arm/t-multilib: New file.
9823 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
9824 variables. Remove setting of ISA and floating-point ABI in
9825 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
9826 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
9827 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
9828 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
9830 * config/arm/t-rmprofile: Likewise except for the matches changes.
9831 * doc/install.texi (--with-multilib-list): Document the combination of
9832 aprofile and rmprofile values and warn about pitfalls in doing that.
9834 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9836 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
9837 (movdi_aarch64): Likewise.
9839 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9841 PR tree-optimization/80632
9842 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
9844 (build_arrays): Initialize it for virtual phis.
9845 (fix_phi_nodes): Use it for virtual phis.
9847 PR tree-optimization/80558
9848 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
9849 [x, y] op z into [x op, y op z] for op & or | if conditions
9852 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
9853 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9856 * config/arm/arm.md (use_literal_pool): Remove.
9857 (64-bit immediate split): No longer takes cost into consideration
9858 if arm_disable_literal_pool is enabled.
9859 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
9860 used when arm_disable_literal_pool is enabled.
9861 (arm_max_const_double_inline_cost): Remove use of
9862 arm_disable_literal_pool.
9863 (push_minipool_fix): Add assert.
9864 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
9865 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
9866 (no_literal_pool_sf_immediate): New.
9868 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9870 PR tree-optimization/80613
9871 * tree-ssa-dce.c (propagate_necessity): Remove cases for
9872 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
9874 2017-05-05 Richard Biener <rguenther@suse.de>
9876 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
9878 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
9880 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
9881 of this flag from insn conditions due to removal from r247495.
9883 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9885 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
9887 (arm_early_store_addr_dep_ptr): Likewise.
9888 * config/arm/aarch-common-protos.h
9889 (arm_early_load_addr_dep_ptr): Add prototype.
9890 (arm_early_store_addr_dep_ptr): Likewise.
9891 * config/arm/cortex-a53.md: Add new bypasses.
9893 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9895 * tree.c (next_type_uid): Change type to unsigned.
9896 (type_hash_canon): Decrement back next_type_uid if
9897 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
9898 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
9901 2017-05-04 Martin Sebor <msebor@redhat.com>
9903 * builtins.c: Fix a trivial typo in a comment.
9906 * builtins.c (check_sizes): Adjust to handle reading past the end.
9907 Avoid printing excessive upper bound of ranges. Use %E to print
9908 tree nodes instead of converting them to %wu.
9909 (expand_builtin_memchr): New function.
9910 (compute_dest_size): Rename...
9911 (compute_objsize): ...to this.
9912 (expand_builtin_memcpy): Adjust.
9913 (expand_builtin_mempcpy): Adjust.
9914 (expand_builtin_strcat): Adjust.
9915 (expand_builtin_strcpy): Adjust.
9916 (check_strncat_sizes): Adjust.
9917 (expand_builtin_strncat): Adjust.
9918 (expand_builtin_strncpy): Adjust and simplify.
9919 (expand_builtin_memset): Adjust.
9920 (expand_builtin_bzero): Adjust.
9921 (expand_builtin_memcmp): Adjust.
9922 (expand_builtin): Handle memcmp.
9923 (maybe_emit_chk_warning): Check strncat just once.
9925 2017-05-04 Martin Sebor <msebor@redhat.com>
9927 PR preprocessor/79214
9930 * builtins.c (check_sizes): Add inlining context and issue
9931 warnings even when -Wno-system-headers is set.
9932 (check_strncat_sizes): Same.
9933 (expand_builtin_strncat): Same.
9934 (expand_builtin_memmove): New function.
9935 (expand_builtin_stpncpy): Same.
9936 (expand_builtin): Handle memmove and stpncpy.
9938 2017-05-04 Bin Cheng <bin.cheng@arm.com>
9940 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
9941 which is not used any more.
9943 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9945 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
9947 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9949 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
9950 (cortexa53_tunings): Likewise.
9951 (cortexa57_tunings): Likewise.
9952 (cortexa72_tunings): Likewise.
9953 (cortexa73_tunings): Likewise.
9955 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9957 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
9958 Set loop alignment to 8.
9960 2017-05-04 Martin Sebor <msebor@redhat.com>
9962 PR translation/80280
9963 * builtins.c (expand_builtin_object_size): Add missing quoting to
9964 %D and like directives.
9965 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
9966 (hsa_type_for_tree_type): Same.
9967 (verify_function_arguments): Same.
9968 * symtab.c (symbol_table::change_decl_assembler_name): Same.
9969 * varasm.c (get_section): Same.
9972 2017-05-04 Martin Sebor <msebor@redhat.com>
9974 PR translation/80280
9975 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
9977 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9979 * config/aarch64/aarch64.c (generic_addrcost_table):
9980 Change HI/TI mode setting.
9982 2017-05-04 Martin Jambor <mjambor@suse.cz>
9984 PR tree-optimization/80622
9985 * tree-sra.c (comes_initialized_p): New function.
9986 (build_accesses_from_assign): Only set write lazily when
9987 comes_initialized_p is false.
9988 (analyze_access_subtree): Use comes_initialized_p.
9989 (propagate_subaccesses_across_link): Assert !comes_initialized_p
9990 instead of testing for PARM_DECL.
9992 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9994 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
9995 constraint on operand 0 to allow more general addressing modes.
9996 Adjust output template.
9997 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
9999 * config/aarch64/aarch64-protos.h
10000 (aarch64_address_valid_for_prefetch_p): Declare prototype.
10001 * config/aarch64/constraints.md (Dp): New address constraint.
10002 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
10005 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
10007 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
10008 update use of estimate_ipcp_clone_size_and_time.
10009 (estimate_local_effects): Update use of
10010 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
10011 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
10012 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
10013 Return nonspecialized time.
10015 2017-05-04 Richard Biener <rguenther@suse.de>
10017 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
10018 for the last VUSE which def dominates the PHI. Directly call
10020 (get_continuation_for_phi_1): Remove.
10022 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
10024 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
10025 to explain the use of truncating division. Cap the number of
10026 iterations to the maximum given by nb_iterations_upper_bound,
10029 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
10031 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
10032 * configure: Regenerate.
10033 * config.in: Regenerate.
10034 * config/i386/driver-mingw32.c: new file.
10035 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
10036 * config.host: Link driver-mingw32.o on MinGW host.
10037 * doc/install.texi: Document new --enable-mingw-wildcard configure
10040 2017-05-04 Marek Polacek <polacek@redhat.com>
10042 PR tree-optimization/80612
10043 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
10045 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10046 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
10048 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
10049 (movt splitter): Likewise.
10050 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
10051 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
10052 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
10053 block for Thumb-1 with MOVT.
10054 (thumb2_legitimate_address_p): Move code block ...
10055 (can_avoid_literal_pool_for_label_p): ... into this new function.
10056 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
10058 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
10059 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
10060 "M-profile targets with the MOVT instruction".
10062 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10064 * config/arm/arm-builtins.c (arm_init_builtins): Rename
10065 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
10066 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
10068 2017-05-04 Martin Liska <mliska@suse.cz>
10070 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
10071 variable cond_code.
10073 2017-05-04 Richard Biener <rguenther@suse.de>
10075 * tree.c (array_at_struct_end_p): Handle arrays at struct
10076 end with flexarrays more conservatively. Refactor and treat
10077 arrays of arrays or aggregates more strict. Fix
10078 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
10079 * tree.c (array_at_struct_end_p): Adjust prototype.
10080 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
10081 * gimple-fold.c (get_range_strlen): Likewise.
10082 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
10084 2017-05-04 Richard Biener <rguenther@suse.de>
10086 PR tree-optimization/31130
10087 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
10089 (supports_overflow_infinity): Likewise.
10090 (is_negative_overflow_infinity): Likewise.
10091 (is_positive_overflow_infinity): Likewise.
10092 (is_overflow_infinity): Likewise.
10093 (stmt_overflow_infinity): Likewise.
10094 (overflow_infinity_range_p): Likewise.
10095 (usable_range_p): Remove as always returning true.
10096 (make_overflow_infinity): Remove.
10097 (negative_overflow_infinity): Likewise.
10098 (positive_overflow_infinity): Likewise.
10099 (avoid_overflow_infinity): Likewise.
10100 (set_value_range): Adjust accordingly.
10101 (set_value_range_to_nonnegative): Likewise, remove now unused
10102 overflow_infinity arg.
10103 (vrp_operand_equal_p): Adjust.
10104 (update_value_range): Likewise.
10105 (range_int_cst_singleton_p): Likewise.
10106 (operand_less_p): Likewise.
10107 (compare_values_warnv): Likewise.
10108 (extract_range_for_var_from_comparison_expr): Likewise.
10109 (vrp_int_const_binop): Likewise.
10110 (zero_nonzero_bits_from_vr): Likewise.
10111 (extract_range_from_multiplicative_op_1): Likewise.
10112 (extract_range_from_binary_expr_1): Likewise.
10113 (extract_range_from_unary_expr): Likewise.
10114 (extract_range_from_comparison): Likewise.
10115 (extract_range_basic): Likewise.
10116 (adjust_range_with_scev): Likewise.
10117 (compare_ranges): Likewise.
10118 (compare_range_with_value): Likewise.
10119 (dump_value_range): Likewise.
10120 (test_for_singularity): Likewise, remove strict_overflow_p parameter
10122 (simplify_cond_using_ranges): Adjust.
10124 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
10126 * brig-builtins.def: Added a builtin for class_f64.
10127 * builtin-types.def: Added a builtin type needed by class_f64.
10129 2017-05-03 Jason Merrill <jason@redhat.com>
10131 * timevar.def: Add TV_CONSTEXPR.
10133 2017-05-03 David Malcolm <dmalcolm@redhat.com>
10135 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
10137 2017-05-03 Martin Jambor <mjambor@suse.cz>
10139 * ipa-prop.c (ipa_update_after_lto_read): Removed.
10140 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
10141 * ipa-cp.c (ipcp_propagate_stage): Do not call
10142 ipa_update_after_lto_read.
10143 * ipa-inline.c (ipa_inline): Likewise.
10145 2017-05-03 Martin Jambor <mjambor@suse.cz>
10147 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
10148 tag. Added a default constructor and a destructor.
10149 (ipa_edge_args_sum_t): New class;
10150 (ipa_edge_args_sum): Declare.
10151 (ipa_edge_args_vector): Remove declaration.
10152 (IPA_EDGE_REF): Use ipa_edge_args_sum.
10153 (ipa_free_edge_args_substructures): Remove declaration.
10154 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
10155 (ipa_edge_args_info_available_for_edge_p): Likewise.
10156 * ipa-prop.c (ipa_edge_args_vector): Removed.
10157 (edge_removal_hook_holder): Likewise.
10158 (edge_duplication_hook_holder): Likewise.
10159 (ipa_edge_args_sum): New variable.
10160 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
10161 ipa_edge_args_vector.
10162 (ipa_free_edge_args_substructures): Likewise.
10163 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
10164 ipa_edge_args_vector.
10165 (ipa_edge_removal_hook): Turned into method
10166 ipa_edge_args_sum_t::remove.
10167 (ipa_edge_duplication_hook): Turned into method
10168 ipa_edge_args_sum_t::duplicate.
10169 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
10170 registering edge hooks.
10171 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
10172 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
10173 ipa_edge_args_sum instead of ipa_edge_args_vector.
10174 * ipa-profile.c (ipa_profile): Likewise.
10176 2017-05-03 Martin Jambor <mjambor@suse.cz>
10178 * symbol-summary.h (function_summary): New method exists.
10179 (function_summary::symtab_removal): Deallocate through release.
10180 (call_summary): New class.
10181 (gt_ggc_mx): New overload.
10182 (gt_pch_nx): Likewise.
10183 (gt_pch_nx): Likewise.
10185 2017-05-03 Jeff Law <law@redhat.com>
10187 PR tree-optimization/78496
10188 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
10189 from simplify_cond_using_ranges. Split off code to walk
10190 backwards through casts into ...
10191 (simplify_cond_using_ranges_2): New function.
10192 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
10193 (execute_vrp): After identifying jump threads, call
10194 simplify_cond_using_ranges_2.
10196 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
10199 * ipa-inline.h (inline_summary): Add ctor.
10200 (create_ggc): Do not use ggc_cleared_alloc.
10202 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
10203 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10205 * gcc.c (handle_braces): Support escaping in switch matching
10207 * doc/invoke.texi (Spec Files): Document it.
10208 Remove superfluous @code markup in items.
10210 2017-05-03 David Malcolm <dmalcolm@redhat.com>
10212 * diagnostic-show-locus.c (struct column_range): New struct.
10213 (get_affected_columns): New function.
10214 (get_printed_columns): New function.
10215 (struct correction): New struct.
10216 (correction::ensure_capacity): New function.
10217 (correction::ensure_terminated): New function.
10218 (struct line_corrections): New struct.
10219 (line_corrections::~line_corrections): New dtor.
10220 (line_corrections::add_hint): New function.
10221 (layout::print_trailing_fixits): Reimplement in terms of the new
10223 (selftest::test_overlapped_fixit_printing): New function.
10224 (selftest::diagnostic_show_locus_c_tests): Call it.
10226 2017-05-03 Nathan Sidwell <nathan@acm.org>
10228 Canonicalize canonical type hashing
10229 * tree.h (type_hash_canon_hash): Declare.
10230 * tree.c (type_hash_list, attribute_hash_list): Move into
10231 type_hash_canon_hash.
10232 (build_type_attribute_qual_variant): Break out hash code calc into
10233 type_hash_canon_hash.
10234 (type_hash_canon_hash): New. Generic type hash computation.
10235 (build_range_type_1, build_array_type_1, build_function_type,
10236 build_method_type_directly, build_offset_type, build_complex_type,
10237 make_vector_type): Call it.
10239 2017-05-03 Richard Biener <rguenther@suse.de>
10241 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
10242 When all DRs have unknown misaligned do not always peel
10243 when there is a store but apply the same costing model as if
10244 there were only loads.
10246 2017-05-03 Richard Biener <rguenther@suse.de>
10249 PR tree-optimization/80492
10250 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
10251 compare_base_decls returning dont-know properly.
10253 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
10255 * config/arm/iterators.md (CCSI): New mode iterator.
10256 (arch): New mode attribute.
10257 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
10258 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
10259 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
10260 code iterator for success result mode.
10261 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
10262 the corresponding new insn generators.
10264 2017-05-03 Bin Cheng <bin.cheng@arm.com>
10267 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10268 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
10270 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
10272 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
10273 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
10274 (DDR_A): Wrap DDR argument in brackets.
10275 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
10276 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
10277 (DDR_REVERSED_P): Likewise.
10279 2017-05-03 Jakub Jelinek <jakub@redhat.com>
10281 PR tree-optimization/79472
10282 * tree-switch-conversion.c (struct switch_conv_info): Add
10283 contiguous_range and default_case_nonstandard fields.
10284 (collect_switch_conv_info): Compute contiguous_range and
10285 default_case_nonstandard fields, don't clear final_bb if
10286 contiguous_range and only the default case doesn't have the required
10288 (check_all_empty_except_final): Set default_case_nonstandard instead
10289 of failing if contiguous_range and the default case doesn't have empty
10291 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
10292 and only the default case doesn't have the required constants. Skip
10294 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
10295 if default_case_nonstandard.
10296 (build_constructors): Build constant 1 just once. Assert that default
10297 values aren't inserted in between cases if contiguous_range. Skip
10299 (build_arrays): Skip virtual phis.
10300 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
10301 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
10302 Handle virtual phis.
10303 (gen_inbound_check): Handle default_case_nonstandard case.
10304 (process_switch): Adjust check_final_bb caller. Call
10305 gather_default_values with the first non-default case instead of
10306 default case if default_case_nonstandard.
10308 2017-05-02 Nathan Sidwell <nathan@acm.org>
10310 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
10311 check. Fix formatting.
10313 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
10315 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
10316 errors when comparing specialized and unspecialized times.
10318 2017-05-02 David Malcolm <dmalcolm@redhat.com>
10320 * diagnostic-show-locus.c
10321 (layout::should_print_annotation_line_p): Make private.
10322 (layout::print_annotation_line): Make private.
10323 (layout::annotation_line_showed_range_p): Make private.
10324 (layout::show_ruler): Make private.
10325 (layout::print_source_line): Make private. Pass in line and
10326 line_width, rather than calling location_get_source_line. Drop
10328 (layout::print_leading_fixits): New method.
10329 (layout::print_any_fixits): Rename to...
10330 (layout::print_trailing_fixits): ...this, and make private.
10331 Don't print newline fixits.
10332 (diagnostic_show_locus): Move logic for printing one row into...
10333 (layout::print_line): ...this new function. Move the
10334 location_get_source_line call and error-handling from
10335 print_source_line to here. Call print_leading_fixits, and rename
10336 print_any_fixits to print_trailing_fixits.
10337 (selftest::test_fixit_insert_containing_newline): Update now that
10338 newlines are partially supported.
10339 (selftest::test_fixit_insert_containing_newline_2): New test.
10340 (selftest::test_fixit_replace_containing_newline): Update comments.
10341 (selftest::diagnostic_show_locus_c_tests): Call the new test.
10342 * edit-context.c (class added_line): New class.
10343 (class edited_line): Describe newline handling in comment.
10344 (edited_line::actually_edited_p): New method.
10345 (edited_line::print_content): Delete redundant decl.
10346 (edited_line::m_predecessors): New field.
10347 (edited_file::print_content): Call edited_line::print_content.
10348 (edited_file::print_diff): Update to support newlines.
10349 (edited_file::print_diff_hunk): Likewise.
10350 (edited_file::print_run_of_changed_lines): New function.
10351 (edited_file::print_diff_line): Convert to...
10352 (print_diff_line): ...this.
10353 (edited_file::get_effective_line_count): New function.
10354 (edited_line::edited_line): Initialize new field m_predecessors.
10355 (edited_line::~edited_line): Clean up m_predecessors.
10356 (edited_line::apply_fixit): Handle newlines.
10357 (edited_line::get_effective_line_count): New function.
10358 (edited_line::print_content): New function.
10359 (edited_line::print_diff_lines): New function.
10360 (selftest::test_applying_fixits_insert_containing_newline): New
10362 (selftest::test_applying_fixits_replace_containing_newline): New
10364 (selftest::insert_line): New function.
10365 (selftest::test_applying_fixits_multiple_lines): Add example of
10367 (selftest::edit_context_c_tests): Call the new tests.
10369 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10371 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
10372 parameter cand. Update dump information.
10373 (get_computation_cost): Update uses.
10375 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10377 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
10378 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
10379 (get_computation_at, rewrite_use_address): Update use of
10380 get_computation_aff.
10382 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10384 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
10385 (get_computation): Delete.
10386 (get_computation_cost): Implement like get_computation_cost_at.
10387 Use get_computation_at.
10388 (get_computation_cost_at): Delete.
10389 (rewrite_use_nonlinear_expr): Use get_computation_at.
10390 (rewrite_use_compare, remove_unused_ivs): Ditto.
10392 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10394 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
10396 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10398 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
10399 (ivopts_global_cost_for_size): Rename parameter and update uses.
10400 (iv_ca_recount_cost): Update uses.
10401 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
10402 candidates seperately in n_invs and n_cands.
10403 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
10405 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10407 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
10408 (find_inv_vars_cb): New.
10409 (find_depends): Renamed to ...
10410 (find_inv_vars): ... this.
10411 (add_candidate_1, force_var_cost): Call find_inv_vars.
10412 (split_address_cost, determine_group_iv_cost_cond): Ditto.
10414 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10416 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
10417 inv_vars. Add inv_exprs.
10418 (struct iv_cand): Rename depends_on to inv_vars.
10419 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
10420 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
10421 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
10422 (dump_cand): Dump inv_vars.
10423 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
10424 (record_invariant, find_depends, add_candidate_1): Ditto.
10425 (set_group_iv_cost, force_var_cost): Ditto.
10426 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
10427 (get_computation_cost_at, get_computation_cost): Ditto.
10428 (determine_group_iv_cost_generic): Ditto.
10429 (determine_group_iv_cost_address): Ditto.
10430 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
10431 (determine_group_iv_costs): Ditto.
10432 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
10433 (iv_ca_set_remove_invariants): Renamed to ...
10434 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
10435 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
10436 (iv_ca_set_add_invariants): Renamed to ...
10437 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
10438 (iv_ca_set_cp): Use iv_ca_set_add_invs.
10439 (iv_ca_has_deps): Support inv_vars and inv_exprs.
10440 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
10441 (create_new_ivs): Remove useless dump.
10443 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10445 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
10447 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
10448 (iv_ca_set_no_cp, create_new_iv): Ditto.
10450 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10452 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
10454 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10456 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
10457 function tree_check2.
10459 2017-05-02 Martin Liska <mliska@suse.cz>
10461 * doc/gcov.texi: Add missing preposition.
10462 * gcov.c (function_info::function_info): Properly fill up
10463 all member variables.
10465 2017-05-02 Tamar Christina <tamar.christina@arm.com>
10467 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
10469 2017-05-02 Tamar Christina <tamar.christina@arm.com>
10471 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
10473 2017-05-02 Martin Liska <mliska@suse.cz>
10476 * lto-streamer-in.c (lto_read_tree_1): Remove
10477 LTO_STREAMER_DEBUG.
10478 * lto-streamer.c (struct tree_hash_entry): Likewise.
10479 (struct tree_entry_hasher): Likewise.
10480 (tree_entry_hasher::hash): Likewise.
10481 (tree_entry_hasher::equal): Likewise.
10482 (lto_streamer_init): Likewise.
10483 (lto_orig_address_map): Likewise.
10484 (lto_orig_address_get): Likewise.
10485 (lto_orig_address_remove): Likewise.
10486 * lto-streamer.h: Likewise.
10487 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
10488 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
10490 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
10492 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
10493 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
10494 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
10495 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
10496 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
10497 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
10498 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
10499 (mm_maskz_sub_ss): New intrinsics.
10500 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
10501 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
10502 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
10503 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
10504 (__builtin_ia32_subss_mask_round): New builtins.
10505 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
10506 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
10507 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
10509 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
10510 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
10512 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
10515 2017-05-02 Martin Jambor <mjambor@suse.cz>
10517 PR tree-optimization/78687
10518 * tree-sra.c (access): New field parent.
10519 (process_subtree_disqualification): New function.
10520 (disqualify_candidate): Call it.
10521 (build_accesses_from_assign): Reset write flag if creating an
10523 (build_access_subtree): Fill in parent field and also prpagate
10524 down grp_write flag.
10525 (create_artificial_child_access): New parameter set_grp_write, set
10526 grp_write to its value.
10527 (propagate_subaccesses_across_link): Also propagate grp_write flag
10529 (propagate_all_subaccesses): Push the closest parent back to work
10530 queue if add_access_to_work_queue returned true.
10532 2017-05-02 Richard Biener <rguenther@suse.de>
10534 * common.opt (fstrict-overflow): Alias negative to fwrapv.
10535 * doc/invoke.texi (fstrict-overflow): Remove all traces of
10536 -fstrict-overflow documentation.
10537 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
10538 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
10539 flag_strict_overflow.
10540 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
10541 * lto-opts.c (lto_write_options): Do not stream it.
10542 * lto-wrapper.c (merge_and_complain): Do not handle it.
10543 * opts.c (default_options_table): Do not set -fstrict-overflow.
10544 (finish_options): Likewise do not clear it when sanitizing.
10545 * simplify-rtx.c (simplify_const_relational_operation): Do not
10546 test flag_strict_overflow.
10548 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10550 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
10551 using enabled attribute.
10552 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
10553 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
10554 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
10555 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
10556 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
10557 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
10558 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
10559 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
10560 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
10561 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
10563 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10565 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
10567 2017-05-02 Richard Biener <rguenther@suse.de>
10569 PR tree-optimization/80591
10571 2017-04-10 Richard Biener <rguenther@suse.de>
10573 * tree-ssa-structalias.c (find_func_aliases): Properly handle
10576 2017-05-02 Richard Biener <rguenther@suse.de>
10578 PR tree-optimization/80549
10579 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
10580 (cleanup_tree_cfg_noloop): Create forwarders to known loop
10581 headers if they do not have a preheader.
10583 2017-05-02 Martin Liska <mliska@suse.cz>
10586 * common.opt: Fix typo.
10587 * doc/invoke.texi: Likewise.
10589 2017-05-01 Jan Beulich <jbeulich@suse.com>
10591 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
10592 swapping, add (x,x,m,x,n) alternative.
10594 2017-05-01 Nathan Sidwell <nathan@acm.org>
10596 * calls.c (combine_pending_stack_adjustment_and_call): Remove
10597 unnecessary unadjusted_alignment check.
10599 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10602 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
10604 * gimplify.c (gimplify_cilk_detach): New function.
10605 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
10606 * tree-core.h: Document EXPR_CILK_SPAWN.
10607 * tree.h (EXPR_CILK_SPAWN): Define.
10609 2017-05-01 David Malcolm <dmalcolm@redhat.com>
10611 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
10612 to use new fixit_hint representation, using the "replace" logic.
10613 (get_line_span_for_fixit_hint): Likewise.
10614 (layout::print_any_fixits): Likewise.
10615 (selftest::test_one_liner_many_fixits): Rename to...
10616 (selftest::test_one_liner_many_fixits_1): ...this, and update
10617 comment and expected output to reflect that the multiple fix-it
10618 hints are now consolidated into one insertion.
10619 (selftest::test_one_liner_many_fixits_2): New test.
10620 (selftest::test_diagnostic_show_locus_one_liner): Update for
10622 (selftest::test_fixit_consolidation): Update for fix-it API
10624 * diagnostic.c (print_parseable_fixits): Likewise.
10625 * edit-context.c (edited_line::m_line_events): Convert from
10626 auto_vec <line_event *> to auto_vec <line_event>.
10627 (class line_event): Convert from abstract base class to a concrete
10628 class, taking over the role of replace_event.
10629 (class insert_event): Delete.
10630 (class replace_event): Rename to class line_event. Convert to
10632 (edit_context::add_fixits): Reimplement.
10633 (edit_context::apply_insert): Delete.
10634 (edit_context::apply_replace): Rename to...
10635 (edit_context::apply_fixit): ...this. Convert to half-open range.
10636 (edited_file::apply_insert): Delete.
10637 (edited_file::apply_replace): Rename to...
10638 (edited_file::apply_fixit): ...this.
10639 (edited_line::~edited_line): Drop deletion of events.
10640 (edited_line::apply_insert): Delete.
10641 (edited_line::apply_replace): Rename to...
10642 (edited_line::apply_fixit): ...this. Convert to half-open range.
10643 Update for change to type of m_line_events.
10644 * edit-context.h (edit_context::apply_insert): Delete.
10645 (edit_context::apply_replace): Rename to...
10646 (edit_context::apply_fixit): ...this.
10648 2017-05-01 Martin Sebor <msebor@redhat.com>
10650 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
10653 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
10656 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
10657 __get_cpuid_max returns 0.
10658 (__get_cpuid_count): Ditto.
10660 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
10662 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
10663 replacement expression is another instance of one of its arguments.
10665 2017-05-01 Jakub Jelinek <jakub@redhat.com>
10668 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
10669 check for stack push/pop autoinc.
10670 * config/i386/i386.c (ix86_agi_dependent): Return false
10671 if the only reason why modified_in_p returned true is that
10672 addr is SP based and set_insn is a push or pop.
10674 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10676 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
10679 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10682 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
10683 (account_size_time): Use two predicates - exec_pred and
10685 (evaluate_conditions_for_known_args): Compute both clause and
10687 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
10688 (inline_summary_t::duplicate): Update.
10689 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
10691 (compute_inline_parameters): Likewise.
10692 (estimate_edge_size_and_time): Update caluclation of time.
10693 (estimate_node_size_and_time): Compute both time and nonspecialized
10695 (estimate_ipcp_clone_size_and_time): Update.
10696 (inline_merge_summary): Update.
10697 (do_estimate_edge_time): Update.
10698 (do_estimate_edge_size): Update.
10699 (do_estimate_edge_hints): Update.
10700 (inline_read_section, inline_write_summary): Stream both new predicates.
10701 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
10703 (compute_inlined_call_time): Cleanup.
10704 (big_speedup_p): Update.
10705 (edge_badness): Update.
10706 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
10707 (size_time_entry): Replace predicate by exec_predicate and
10708 nonconst_predicate.
10709 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
10710 (estimate_edge_time): Return also nonspec_time.
10711 (reset_edge_growth_cache): Update.
10713 2017-04-29 Jakub Jelinek <jakub@redhat.com>
10715 PR rtl-optimization/80491
10716 * ifcvt.c (noce_process_if_block): When looking for x setter
10717 with missing else_bb, don't check only the insn right before
10718 cond_earliest, but look for the last insn that x is modified in
10719 within the same bb.
10721 PR rtl-optimization/80491
10722 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
10724 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
10726 PR tree-optimization/80487
10727 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
10729 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10731 PR tree-optimization/79697
10732 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
10733 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
10734 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
10736 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
10737 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
10739 2017-04-28 Martin Sebor <msebor@redhat.com>
10741 PR tree-optimization/80523
10742 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
10743 (init_target_to_host_charmap, target_to_host, target_strtol10): New
10745 (maybe_warn, format_directive, parse_directive): Use new functions.
10746 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
10748 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
10750 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
10752 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
10754 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
10755 target_header_dir): Set correctly.
10756 * configure: Regenerated.
10757 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
10758 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
10759 instead of SYSTEM_HEADER_DIR.
10761 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
10763 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
10764 (estimate_local_effects): Likewise.
10765 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
10766 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
10767 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
10768 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
10769 do_estimate_edge_time, estimate_edge_time): Likewise.
10770 * ipa-inline-analysis.c (estimate_node_size_and_time,
10771 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
10772 (estimate_time_after_inlining): Remove.
10774 2017-04-28 Martin Liska <mliska@suse.cz>
10776 * doc/gcov.texi: Enhance documentation of gcov.
10778 2017-04-28 Martin Liska <mliska@suse.cz>
10780 * doc/gcov.texi: Sort options in alphabetic order.
10781 * doc/gcov-dump.texi: Likewise.
10782 * doc/gcov-tool.texi: Likewise.
10783 * gcov.c (print_usage): Likewise.
10784 * gcov-dump.c (print_usage): Likewise.
10785 * gcov-tool.c (print_merge_usage_message): Likewise.
10786 (print_rewrite_usage_message): Likewise.
10787 (print_overlap_usage_message): Likewise.
10789 2017-04-28 Martin Liska <mliska@suse.cz>
10791 PR gcov-profile/53915
10792 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
10794 2017-04-28 Martin Liska <mliska@suse.cz>
10796 PR gcov-profile/79891
10797 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
10798 is marked by compiler as living on a line.
10799 (get_cycles_count): Remove usage of the union.
10800 (output_intermediate_file): Likewise.
10801 (find_source): Fix GNU coding style.
10802 (accumulate_line_counts): Remove old non-all block mode.
10803 (output_lines): Remove usage of the union.
10804 * profile.c (output_location): Include all BBs, even if
10805 belonging to a same line (and file) as a previous BB.
10807 2017-04-28 Martin Liska <mliska@suse.cz>
10809 * gcov.c (process_args): Handle new argument 'w'.
10810 (read_graph_file): Assign ID to BBs.
10811 (output_branch_count): Display BB # if verbose flag is set.
10812 (output_lines): Likewise for arcs.
10813 (print_usage): Add '--verbose' option help.
10814 * doc/gcov.texi: Document --verbose (-w) option.
10816 2017-04-28 Martin Liska <mliska@suse.cz>
10818 * gcov.c (struct block_location_info): New struct.
10819 (process_file): Fill up the new structure.
10820 (read_graph_file): Replace usage of encoding by the newly added
10822 (add_line_counts): Likewise.
10823 (accumulate_line_counts): Remove usage of the union.
10824 (function_info::function_info): New function.
10825 (function_info::~function_info): Likewise.
10826 (process_file): Call delete instead of release_function.
10827 (release_function): Release the function.
10828 (release_structures): Call delete instead of release_function.
10829 (solve_flow_graph): Replace usage of num_blocks.
10830 (find_exception_blocks): Likewise.
10831 (output_lines): Fix GNU coding style.
10833 2017-04-28 Martin Liska <mliska@suse.cz>
10836 * coverage.c (coverage_remove_note_file): New function.
10837 * coverage.h: Declare the function.
10838 * toplev.c (finalize): Clean if an error has been seen.
10840 2017-04-28 Martin Liska <mliska@suse.cz>
10842 PR gcov-profile/80031
10843 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
10844 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
10845 * gcov.c (read_graph_file): Read just number of blocks.
10846 * profile.c (branch_prob): Do not stream 0 flags per a basic
10849 2017-04-28 Martin Liska <mliska@suse.cz>
10851 * gcov-dump.c (tag_*): Add new argument to declarations.
10852 (dump_gcov_file): Likewise.
10853 (tag_blocks): Add and use new argument depth.
10854 (tag_arcs): Likewise.
10855 (tag_lines): Likewise.
10856 (tag_counters): Likewise.
10857 (tag_summary): Likewise.
10858 (dump_working_sets): Use depth to do a proper indentation.
10860 2017-04-28 Jakub Jelinek <jakub@redhat.com>
10863 * cgraph.h (symtab_node::debug_symtab): No longer inline.
10864 * symtab.c (symtab_node::debug_symtab): Move definition here.
10866 2017-04-28 Richard Biener <rguenther@suse.de>
10868 * lto-streamer.h (LTO_major_version): Bump to 7.
10870 2017-04-28 Richard Biener <rguenther@suse.de>
10872 * tree-vrp.c (assert_info): New struct.
10873 (add_assert_info): New helper.
10874 (register_edge_assert_for_2): Refactor to add asserts to a vector
10876 (register_edge_assert_for_1): Likewise.
10877 (register_edge_assert_for): Likewise.
10878 (finish_register_edge_assert_for): New helper actually registering
10879 asserts where live on edge.
10880 (find_conditional_asserts): Adjust.
10881 (find_switch_asserts): Likewise.
10882 (evrp_dom_walker::try_find_new_range): Generalize.
10883 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
10885 2017-04-27 Marek Polacek <polacek@redhat.com>
10888 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
10889 arg10 and arg11 to itype.
10891 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
10893 * doc/extend.texi (Object Size Checking): Improve grammar.
10895 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
10898 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
10899 that the logic for permitting reciprocal estimates matches that
10902 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10905 * tree.c (type_cache_hasher::equal): Only compare
10906 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
10907 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
10908 non-aggregate element types.
10909 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
10910 about the flag on ARRAY_TYPEs in the comment, formatting fix.
10912 2017-04-27 Richard Biener <rguenther@suse.de>
10914 PR middle-end/80533
10915 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
10916 stripping ARRAY_REFs from MEM_EXPR make sure we're not
10917 keeping a reference to a trailing array.
10919 2017-04-27 Richard Biener <rguenther@suse.de>
10921 PR middle-end/80539
10922 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
10923 being in loop-closed SSA form conservatively.
10924 (chrec_fold_multiply_poly_poly): Likewise.
10926 2017-04-27 Tamar Christina <tamar.christina@arm.com>
10928 PR middle-end/79665
10929 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
10930 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
10932 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10935 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
10936 (aarch64_function_arg_alignment): Return unsigned int again, but still
10937 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
10938 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
10939 Don't emit -Wpsabi note.
10940 (aarch64_function_arg_boundary): Likewise.
10941 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
10944 2017-04-26 Nathan Sidwell <nathan@acm.org>
10946 * tree.h (crc32_unsigned_n): Declare.
10947 (crc32_unsigned, crc32_unsigned): Make inline.
10948 * tree.c (crc32_unsigned_bits): Replace with ...
10949 (crc32_unsigned_n): ... this.
10950 (crc32_unsigned, crc32_byte): Remove.
10951 (crc32_string): Remove unnecessary braces.
10953 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
10955 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
10956 * ipa-inline-analysis.c (MAX_TIME): Remove.
10957 (account_size_time): Use sreal for time.
10958 (dump_inline_summary): Update.
10959 (estimate_function_body_sizes): Update.
10960 (estimate_edge_size_and_time): Update.
10961 (estimate_calls_size_and_time): Update.
10962 (estimate_node_size_and_time): Update.
10963 (inline_merge_summary): Update.
10964 (inline_update_overall_summary): Update.
10965 (estimate_time_after_inlining): Update.
10966 (inline_read_section): Update.
10967 (inline_write_summary): Update.
10968 * ipa-inline.c (compute_uninlined_call_time): Update.
10969 (compute_inlined_call_time): Update.
10970 (recursive_inlining): Update.
10971 (inline_small_functions): Update.
10972 (dump_overall_stats): Update.
10973 * ipa-inline.h: Include sreal.h.
10974 (size_time_entry): Turn time to sreal.
10975 (inline_summary): Turn self_time nad time to sreal.
10977 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
10979 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
10981 (sreal::stream_out, sreal::stream_in): New.
10982 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
10984 2017-04-25 Jakub Jelinek <jakub@redhat.com>
10986 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
10989 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
10992 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
10994 (dimode_scalar_chain::compute_convert_gain): Ditto.
10995 (dimode_scalar_chain::make_vector_copies): Ditto.
10996 (dimode_scalar_chain::convert_reg): Ditto.
10997 (dimode_scalar_chain::convert_insn): Ditto.
10998 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
10999 (VI248_AVX512BW_1): New mode iterator.
11000 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
11001 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
11004 2017-04-25 Martin Sebor <msebor@redhat.com>
11006 PR tree-optimization/80497
11007 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
11008 constants are representable in HOST_WIDE_INT.
11009 (parse_directive): Ditto.
11011 2017-04-25 Martin Sebor <msebor@redhat.com>
11014 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
11015 (new_zero_array): Adjust signature.
11016 (dom_info::dom_init): Used unsigned rather that size_t.
11017 (dom_info::dom_info): Same.
11019 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11020 Jakub Jelinek <jakub@redhat.com>
11023 * config/arm/arm.c: Include gimple.h.
11024 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
11025 returns negative, increment ncrn only if it returned positive.
11026 (arm_needs_doubleword_align): Return int instead of bool,
11027 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
11028 members, but if there is any such non-FIELD_DECL
11029 > PARM_BOUNDARY aligned decl, return -1 instead of false.
11030 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
11031 returns negative, increment nregs only if it returned positive.
11032 (arm_setup_incoming_varargs): Likewise.
11033 (arm_function_arg_boundary): Emit -Wpsabi note if
11034 arm_needs_doubleword_align returns negative, return
11035 DOUBLEWORD_ALIGNMENT only if it returned positive.
11037 2017-04-25 Marek Polacek <polacek@redhat.com>
11040 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
11041 first argument to type.
11043 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
11046 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
11047 type checks to test for compatibility instead of equality.
11049 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11050 Jakub Jelinek <jakub@redhat.com>
11053 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
11055 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
11056 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
11057 the alignment computation, but return their maximum in warn_alignment.
11058 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
11059 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
11061 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
11062 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
11065 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11067 * config/arc/simdext.md (dmpyh): Fix typo.
11069 2017-04-25 Richard Biener <rguenther@suse.de>
11071 PR tree-optimization/80492
11072 * alias.c (compare_base_decls): Handle registers with asm
11073 specification conservatively.
11074 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
11075 compare_base_decls returning dont-know properly.
11077 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11079 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
11080 (legitimate_offset_address_p): New function.
11081 (arc_legitimate_address_p): Use above function.
11083 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11085 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
11087 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11089 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
11090 ACCH registers whenever they are available.
11092 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11094 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
11095 double regs fix when not used.
11097 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11099 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
11101 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
11102 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
11104 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11106 * config/arc/arc.c (arc_output_addsi): Check for h-register class
11107 when emitting short ADD instructions.
11109 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11111 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
11113 (cmpsi_cc_c_insn): Likewise.
11114 (cbranchsi4_scratch): Compute proper instruction length using
11115 compact_hreg_operand.
11116 * config/arc/predicates.md (compact_hreg_operand): New predicate.
11118 2017-04-25 Richard Biener <rguenther@suse.de>
11120 PR middle-end/80509
11121 * passes.c (pass_manager::pass_manager): Initialize
11122 m_name_to_pass_map.
11124 2017-04-25 Richard Biener <rguenther@suse.de>
11126 PR tree-optimization/79201
11127 * tree-ssa-sink.c (statement_sink_location): Handle calls.
11129 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11132 * config/s390/vector.md: Split MEM->GPR vector moves for
11133 non-s_operand addresses.
11135 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11138 * config/s390/predicates.md (reload_const_wide_int_operand): New
11140 * config/s390/s390.md ("movti"): Remove d/P alternative.
11141 ("movti_bigconst"): New pattern definition.
11143 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11146 * s390-protos.h (s390_expand_cs_hqi): Removed.
11147 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
11148 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
11149 modes as well as CCZ1mode and CCZmode.
11150 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
11151 signature of s390_emit_compare_and_swap.
11152 (s390_expand_cs_hqi): Likewise, make static.
11153 (s390_expand_cs_tdsi): Generate an explicit compare before trying
11154 compare-and-swap, in some cases.
11155 (s390_expand_cs): Wrapper function.
11156 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
11158 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
11159 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
11160 patterns for small and large integers. Forbid symref memory operands.
11161 Move expander to s390.c. Require cc register.
11162 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
11163 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
11164 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
11165 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
11166 symref memory operands. Remove CC mode and call s390_match_ccmode
11168 ("atomic_exchange<mode>"): Allow and implement all integer modes.
11170 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11172 * config/s390/s390.md (define_peephole2): New peephole to help
11173 combining the load-and-test pattern with volatile memory.
11175 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11177 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
11178 with CCZmode for TARGET_Z196.
11180 2017-04-25 Jakub Jelinek <jakub@redhat.com>
11182 PR rtl-optimization/80501
11183 * combine.c (make_compound_operation_int): Set subreg_code to SET
11184 even for AND with mask of the sign bit of mode.
11186 PR rtl-optimization/80500
11187 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
11188 sum's initial value.
11190 2017-04-25 Julian Brown <julian@codesourcery.com>
11191 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11193 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
11195 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
11197 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
11199 2017-04-25 Julian Brown <julian@codesourcery.com>
11200 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11202 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
11203 (thunderx2t99_sha): New Reservation.
11205 2017-04-25 Julian Brown <julian@codesourcery.com>
11206 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11208 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
11209 type for 1-element load.
11211 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
11213 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
11215 2017-04-24 Martin Jambor <mjambor@suse.cz>
11217 PR tree-optimization/80293
11218 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
11219 char arrays not totally scalarizable if it is false.
11220 (analyze_all_variable_accesses): Pass correct value in the new
11221 parameter. Add a statistics counter.
11223 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
11225 PR middle-end/79931
11226 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
11228 2017-04-24 Richard Biener <rguenther@suse.de>
11230 PR tree-optimization/80494
11231 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
11232 out for complex types.
11234 2017-04-24 Richard Biener <rguenther@suse.de>
11236 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
11237 * tree-ssa-sccvn.c (print_scc): Print SCC size.
11238 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
11239 (DFS): Adjust and never fail.
11240 (sccvn_dom_walker::fail): Remove.
11241 (sccvn_dom_walker::before_dom_children): Adjust.
11242 (run_scc_vn): Likewise and never fail.
11243 * tree-ssa-pre.c (pass_pre::execute): Adjust.
11244 (pass_fre::execute): Likewise.
11246 2017-04-24 Richard Biener <rguenther@suse.de>
11248 PR tree-optimization/79725
11249 * tree-ssa-sink.c (statement_sink_location): Return whether
11250 failure reason was zero uses. Move that check later.
11251 (sink_code_in_bb): Deal with zero uses by removing the stmt
11254 2017-04-24 Richard Biener <rguenther@suse.de>
11257 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
11258 pointer-based references.
11260 2017-04-24 Richard Biener <rguenther@suse.de>
11263 * pass_manager.h (pass_manager::operator new): Remove.
11264 (pass_manager::operator delete): Likewise.
11265 * passes.c (pass_manager::operator new): Remove.
11266 (pass_manager::operator delete): Likewise.
11267 (pass_manager::pass_manager): Zero individual pass members.
11269 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
11272 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
11273 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
11274 Check "XEXP (src, 1)" operand here.
11275 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
11276 Check "XEXP (src, 1)" operand here.
11277 (dimode_scalar_chain::make_vector_copies): Detect count register
11278 of a shift instruction. Zero extend count register from QImode
11279 to DImode to satisfy vector shift pattern count operand predicate.
11280 Substitute vector shift count operand with a DImode copy.
11281 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
11284 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
11286 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
11287 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
11288 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
11289 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
11290 (UNSPEC_NOREX_MEM): Remove definition.
11292 2017-04-21 Richard Biener <rguenther@suse.de>
11294 PR tree-optimization/79547
11295 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
11296 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
11297 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
11298 without any constraints.
11300 2017-04-21 Richard Biener <rguenther@suse.de>
11302 PR tree-optimization/78847
11303 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
11305 2017-04-21 Richard Biener <rguenther@suse.de>
11307 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
11308 (build_distinct_type_copy): Likewise.
11309 (build_variant_type_copy): Likewise.
11310 * tree.c (build_qualified_type): Pass down mem-stat info.
11311 (build_distinct_type_copy): Likewise.
11312 (build_variant_type_copy): Likewise.
11314 2017-04-21 Richard Biener <rguenther@suse.de>
11316 PR tree-optimization/80237
11317 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
11319 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
11320 for a simplified result.
11322 2016-04-21 Richard Biener <rguenther@suse.de>
11324 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
11325 sth as strict as a simple_iv but a chrec without symbols and an
11326 operand defined in the loop we are peeling (and not some subloop).
11327 (propagate_constants_for_unrolling): Propagate all constants.
11329 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
11332 * config/i386/i386.c (print_reg): Remove assert for disalowed
11333 regno values, call output_operand_lossage instead.
11335 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
11338 * config/i386/constraints.md (Yc): New register constraint.
11339 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
11340 Use Yc constraint for alternative 2 of operand 0. Remove
11341 preferred_for_speed attribute.
11343 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
11345 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
11346 lastprivate clauses in SIMT case.
11348 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
11350 * doc/invoke.texi (-Wextra-semi): Document new warning option.
11352 2017-04-20 Richard Biener <rguenther@suse.de>
11354 PR tree-optimization/57796
11355 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
11356 as N scalar stores.
11357 (vect_model_load_cost): Cost gathers as N scalar loads.
11359 2017-04-20 Richard Biener <rguenther@suse.de>
11361 * ggc-page.c (ggc_allocated_p): Rename to ...
11362 (safe_lookup_page_table_entry): ... this and return the lookup
11364 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
11366 2017-04-20 Richard Biener <rguenther@suse.de>
11368 PR tree-optimization/80453
11369 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
11370 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
11371 from the conditions.
11372 (vn_phi_eq): Pass them down.
11373 (vn_phi_lookup): Record them.
11374 (vn_phi_insert): Likewise.
11376 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
11378 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
11379 uninitialized variable warning to avoid buffer overrun.
11381 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
11384 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
11385 is suppressed for '{ 0 }' in C.
11387 2017-04-20 Jakub Jelinek <jakub@redhat.com>
11389 * BASE-VER: Set to 8.0.0.
11391 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
11393 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
11394 priority .init_array and .fini_array section with SECTION_NOTYPE
11397 2017-04-20 Jakub Jelinek <jakub@redhat.com>
11399 PR middle-end/80423
11400 * tree.h (build_array_type): Add typeless_storage default argument.
11401 * tree.c (type_cache_hasher::equal): Also compare
11402 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
11403 (build_array_type): Add typeless_storage argument, set
11404 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
11406 (build_nonshared_array_type): Adjust build_array_type_1 caller.
11407 (build_array_type): Likewise. Add typeless_storage argument.
11409 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
11410 Jakub Jelinek <jakub@redhat.com>
11412 PR tree-optimization/80426
11413 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
11414 operation on symbolic operands, also compute the overflow for the
11415 invariant part when the operation degenerates into a negation.
11417 2017-04-19 Jakub Jelinek <jakub@redhat.com>
11420 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
11421 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
11424 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
11426 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
11429 * config/avr/avr.c (tree.h): Include it.
11430 (cgraph.h): Include it.
11431 (avr_encode_section_info): Don't warn for uninitialized progmem
11432 variable if it's just an alias.
11434 2017-04-19 Richard Biener <rguenther@suse.de>
11437 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
11438 when needed by AutoPGO.
11440 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
11443 * doc/lto.texi: Remove an extra 'that'.
11445 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
11447 PR rtl-optimization/80429
11448 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
11449 are only used in debug insns.
11451 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
11452 Vladimir Makarov <vmakarov@redhat.com>
11454 * config/sparc/predicates.md (input_operand): Add comment. Return
11455 true for any memory operand when LRA is in progress.
11456 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
11458 2017-04-18 Jeff Law <law@redhat.com>
11461 * mips.md ({return,simple_return}_internal): Do not overwrite
11464 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11466 PR tree-optimization/80443
11467 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
11468 instead of adding 1, subtract -1 and similarly instead of subtracting
11471 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
11473 PR rtl-optimization/80357
11474 * haifa-sched.c (tmp_bitmap): New variable.
11475 (model_recompute): Handle duplicate use records.
11476 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
11477 (free_global_sched_pressure_data): Free it.
11479 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
11482 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
11483 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
11484 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
11485 instead of SYSTEM_HEADER_DIR.
11487 2017-04-18 Jeff Law <law@redhat.com>
11489 PR middle-end/80422
11490 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
11491 predecessors after walking up the insn chain.
11493 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11496 * dwarf2out.c (modified_type_die): Try harder not to emit internal
11497 sizetype type into debug info.
11499 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
11502 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
11503 unneeded test for TARGET_UPPER_REGS_SF.
11504 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
11506 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11509 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
11510 instead of gsi_after_labels.
11512 2017-04-18 Jeff Law <law@redhat.com>
11514 * regcprop.c (maybe_mode_change): Avoid creating copies of the
11518 2017-04-13 Jeff Law <law@redhat.com>
11519 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11520 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11522 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
11525 * config/avr/avr.c (intl.h): Include it.
11526 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
11528 2017-04-18 Martin Liska <mliska@suse.cz>
11530 PR gcov-profile/78783
11531 * gcov-tool.c (gcov_output_files): Validate that destination
11532 file is either removed by the tool or by a user.
11534 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
11535 Guy Benyei <guybe@mellanox.com>
11537 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
11538 block, and do not negate it, the stored id is already negative.
11540 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
11542 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
11544 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
11547 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
11548 masks of options that should be turned off if the VSX vector
11549 options are turned off.
11550 (OTHER_P8_VECTOR_MASKS): Likewise.
11551 (OTHER_VSX_VECTOR_MASKS): Likewise.
11552 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
11553 rs6000_disable_incompatible_switches to validate no type switches
11555 (rs6000_incompatible_switch): New function to disallow turning on
11556 other vector options if -mno-vsx, -mno-power8-vector, or
11557 -mno-power9-vector are specified.
11559 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11561 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
11563 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11565 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
11566 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
11567 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
11568 (ARG_POINTER_CFA_OFFSET): Likewise.
11570 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11572 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
11573 conditions to take advantage of various optimizations.
11575 2017-04-13 Jeff Law <law@redhat.com>
11577 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11578 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11579 (zero_extendsidi2_dext): Likewise.
11581 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11584 * fold-const.c (fold_ternary_loc): Revert
11585 use op0 instead of fold_convert_loc (loc, type, arg0) part of
11588 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
11590 PR rtl-optimization/80343
11591 * lra-remat.c (update_scratch_ops): Assign original hard reg to
11592 new scratch pseudo.
11594 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
11597 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
11598 to ubsan_encode_value.
11600 2017-04-13 Jeff Law <law@redhat.com>
11602 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
11603 appearing in DEBUG_INSNs.
11605 2017-04-13 Martin Liska <mliska@suse.cz>
11607 PR gcov-profile/80413
11608 * gcov-io.c (gcov_write_string): Copy to buffer just when
11609 allocated size is greater than zero.
11611 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11614 * dwarf2out.c (decls_for_scope): Ignore declarations of
11615 current_function_decl in BLOCK_NONLOCALIZED_VARS.
11617 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
11620 * ipa-visibility.c (non_local_p): Fix typos.
11621 (localize_node): When localizing symbol in same comdat group,
11622 dissolve the group only when we know external symbols are going
11624 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
11626 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11628 PR tree-optimization/79390
11629 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
11630 order does not result in usable sequence, retry with reversed operand
11636 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
11637 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
11638 op0 instead of fold_convert_loc (loc, type, arg0).
11640 2017-04-12 Jeff Law <law@redhat.com>
11642 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
11643 has a delay slot in the generated code.
11645 * config/cris/cris.md (cris_preferred_reload_class): Return
11646 GENNONACR_REGS rather than GENERAL_REGS.
11648 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11651 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
11652 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
11653 signedness of the result type.
11655 2017-04-12 Richard Biener <rguenther@suse.de>
11656 Jeff Law <law@redhat.com>
11658 PR tree-optimization/80359
11659 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
11660 trim stores to TARGET_MEM_REFs.
11662 2017-04-12 Richard Biener <rguenther@suse.de>
11664 PR tree-optimization/79390
11665 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
11666 threading case even more.
11668 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
11671 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
11672 for quad_address_p for TImode, instead of just not indexed_address.
11674 2017-04-12 Richard Biener <rguenther@suse.de>
11675 Bernd Edlinger <bernd.edlinger@hotmail.de>
11677 PR middle-end/79671
11678 * alias.c (component_uses_parent_alias_set_from): Handle
11679 TYPE_TYPELESS_STORAGE.
11680 (get_alias_set): Likewise.
11681 * tree-core.h (tree_type_common): Add typeless_storage flag.
11682 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
11683 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
11684 for types containing members with TYPE_TYPELESS_STORAGE.
11685 (place_field): Likewise.
11686 (layout_type): Likewise for ARRAY_TYPE.
11687 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
11688 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
11689 TYPE_TYPELESS_STORAGE.
11690 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
11692 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11695 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
11696 first argument to type.
11698 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11702 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
11703 CONST0_RTX (mode) rather than const0_rtx where appropriate.
11704 (rs6000_expand_binop_builtin): Likewise.
11705 (rs6000_expand_ternop_builtin): Likewise; also add missing
11706 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
11707 vshasigma built-ins.
11708 * doc/extend.texi: Document that vec_xxpermdi's third argument
11709 must be a constant.
11711 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
11713 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
11714 Use shift_const cost parameter when calculating gain of STV shifts.
11716 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
11718 PR rtl-optimization/70478
11719 * lra-constraints.c (process_alt_operands): Check memory for
11720 disfavoring memory insn operand.
11722 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11724 PR middle-end/80100
11725 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
11726 left shift in unsigned HOST_WIDE_INT type.
11728 PR rtl-optimization/80385
11729 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
11730 (not (neg X)) into (plus X -1) for complex or non-integral modes.
11733 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
11734 if they have any depend clauses.
11736 2017-04-11 Martin Liska <mliska@suse.cz>
11739 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
11740 * ipa-split.c (split_function): Create a local comdat symbol
11741 if caller is in a comdat group.
11743 2017-04-11 Martin Liska <mliska@suse.cz>
11746 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
11749 2017-04-11 Martin Sebor <msebor@redhat.com>
11751 PR middle-end/80364
11752 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
11753 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
11755 (directive::set_width, directive::set_precision, format_character):
11757 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
11760 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
11763 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
11764 conflict, set target->arch_name instead of target->cpu_name.
11766 2017-04-11 Richard Biener <rguenther@suse.de>
11768 PR tree-optimization/80374
11769 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
11770 build_zero_cst, remove fold_convertible_p check again.
11772 2017-04-11 Martin Liska <mliska@suse.cz>
11775 * ubsan.c (instrument_object_size): Do not instrument register
11778 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11781 * config/i386/i386-builtin-types.def
11782 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
11783 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
11784 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
11785 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
11786 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
11787 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
11788 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
11789 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
11790 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
11791 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
11792 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
11793 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
11794 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
11795 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
11796 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
11797 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
11798 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
11799 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
11800 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
11801 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
11802 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
11803 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
11804 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
11805 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
11806 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
11807 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
11808 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
11809 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
11810 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
11811 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
11812 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
11813 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
11814 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
11815 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
11816 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
11817 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
11818 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
11819 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
11820 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
11821 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
11822 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
11823 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
11824 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
11825 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
11826 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
11827 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
11829 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
11830 flag to second_arg_count, handle 4 argument function type _COUNT
11831 aliases, handle second_arg_count on second argument rather than last.
11833 2017-04-10 Jeff Law <law@redhat.com>
11835 PR tree-optimization/80374
11836 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
11837 record anything if we can not convert integer_zero_node to the
11840 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
11843 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11844 Enhance special handling given to the TARGET_P9_MINMAX option in
11845 relation to certain other options.
11847 2017-04-10 Bin Cheng <bin.cheng@arm.com>
11849 PR tree-optimization/80153
11850 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
11851 remove POINTER_PLUS_EXPR's base part directly, rather than through
11854 2017-04-10 Richard Biener <rguenther@suse.de>
11855 Bin Cheng <bin.cheng@arm.com>
11857 PR tree-optimization/80153
11858 * tree-affine.c (aff_combination_to_tree): Get base pointer from
11859 the first element of pointer type aff_tree. Build result expr in
11861 (add_elt_to_tree): Convert to type unconditionally. Remove other
11862 fold_convert calls.
11863 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
11864 (rewrite_use_nonlinear_expr): Check invariant using iv information.
11866 2017-04-10 Richard Biener <rguenther@suse.de>
11868 * tree-ssa-structalias.c (find_func_aliases): Properly handle
11871 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
11873 PR rtl-optimization/70478
11874 * lra-constraints.c (curr_small_class_check): New.
11875 (update_and_check_small_class_inputs): New.
11876 (process_alt_operands): Update curr_small_class_check. Disfavor
11877 alternative insn memory operands. Check available regs for small
11880 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
11883 * config/mips/mips.opt (-mvirt): Update description.
11884 * doc/invoke.texi (-mvirt): Likewise.
11886 2017-04-10 Richard Biener <rguenther@suse.de>
11888 PR middle-end/80362
11889 * fold-const.c (fold_binary_loc): Look at unstripped ops when
11890 looking for NEGATE_EXPR in -A / -B to A / B folding.
11892 2017-04-10 Martin Liska <mliska@suse.cz>
11894 PR gcov-profile/80224
11895 * gcov.c (print_usage): Fix usage string.
11896 (get_gcov_intermediate_filename): Remove.
11897 (output_gcov_file): Use both for normal and intermediate format.
11898 (generate_results): Do not initialize special file for
11899 intermediate format.
11901 2017-04-10 Richard Biener <rguenther@suse.de>
11903 PR tree-optimization/80304
11904 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
11907 2017-04-10 Nathan Sidwell <nathan@acm.org>
11910 * config/rs6000/rs6000.c (rs6000_vector_type): New.
11911 (rs6000_init_builtins): Use it.
11913 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11915 * config/arm/arm.md (<mrc>): Add mode to SET source.
11916 (<mrrc>): Likewise.
11918 2017-04-10 Richard Biener <rguenther@suse.de>
11920 PR middle-end/80344
11921 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
11923 2017-04-10 Jakub Jelinek <jakub@redhat.com>
11926 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
11927 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
11928 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
11929 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
11930 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
11931 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
11932 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
11933 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
11934 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
11935 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
11936 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
11937 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
11938 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
11939 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
11940 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
11941 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
11942 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
11943 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
11944 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
11945 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
11946 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
11947 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
11948 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
11950 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
11952 PR rtl-optimization/70478
11953 * lra-constraints.c: Reverse the last patch.
11955 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
11957 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
11958 Add comment for WCHAR_T.
11960 2017-04-08 Martin Liska <mliska@suse.cz>
11963 2017-04-07 Martin Liska <mliska@suse.cz>
11966 * ipa-split.c (split_function): Add function part to a same comdat
11969 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11972 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
11974 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
11976 * rs6000/rs6000.c (vec_load_pendulum): Rename...
11977 (vec_pairing): ...to this.
11978 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
11979 (rs6000_sched_init): Adjust for name change.
11980 (struct rs6000_sched_context): Likewise.
11981 (rs6000_init_sched_context): Likewise.
11982 (rs6000_set_sched_context): Likewise.
11984 2017-04-07 Jakub Jelinek <jakub@redhat.com>
11990 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
11992 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
11993 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
11994 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
11996 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
11998 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
12000 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
12002 PR rtl-optimization/70703
12003 * ira-color.c (update_conflict_hard_regno_costs): Use
12004 int64_t instead of HOST_WIDE_INT.
12006 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
12008 PR rtl-optimization/70478
12009 * lra-constraints.c (process_alt_operands): Disfavor alternative
12010 insn memory operands.
12012 2017-04-07 Jeff Law <law@redhat.com>
12014 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
12015 CALL and NOTE_INSN_CALL_ARG_LOCATION.
12017 2017-04-07 Martin Liska <mliska@suse.cz>
12020 * config/aarch64/aarch64.c (aarch64_process_target_attr):
12021 Show error message instead of an ICE.
12023 2017-04-07 Martin Liska <mliska@suse.cz>
12026 * ipa-split.c (split_function): Add function part to a same comdat
12029 2017-04-07 Richard Biener <rguenther@suse.de>
12031 PR middle-end/80341
12032 * tree.c (get_unwidened): Also handle ! for_type case for
12034 * convert.c (do_narrow): Split out from ...
12035 (convert_to_integer_1): ... here. Do not pass final truncation
12036 type to get_unwidened for TRUNC_DIV_EXPR.
12038 2017-04-07 Richard Biener <rguenther@suse.de>
12040 * tree-affine.c (wide_int_ext_for_comb): Take type rather
12042 (aff_combination_const): Adjust.
12043 (aff_combination_scale): Likewise.
12044 (aff_combination_add_elt): Likewise.
12045 (aff_combination_add_cst): Likewise.
12046 (aff_combination_convert): Likewise.
12047 (add_elt_to_tree): Likewise. Remove unused argument.
12048 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
12050 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
12052 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
12054 * config/arm/arm.c (arm_default_short_enums): Use
12055 ARM_DEFAULT_SHORT_ENUMS.
12056 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
12058 2017-04-06 Jakub Jelinek <jakub@redhat.com>
12061 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
12062 members with redundant out-of-class redeclaration.
12064 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
12067 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
12068 * config/i386/i386.md (*zero_extendsidi2):
12069 Add (?*x,*x) and (?*v,*v) alternatives.
12071 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
12074 * config/i386/i386.c (ix86_expand_builtin)
12075 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
12076 mode from insn data. Convert operands to insn operand mode.
12077 Copy operands that don't satisfy insn predicate to a register.
12079 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
12081 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
12084 2017-04-06 Richard Biener <rguenther@suse.de>
12086 PR tree-optimization/80334
12087 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
12088 preserve alignment of accesses.
12090 2017-04-06 Richard Biener <rguenther@suse.de>
12092 PR tree-optimization/80262
12093 * tree-sra.c (build_ref_for_offset): Preserve address-space
12095 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
12096 Drop useless address-space information on MEM_REF offsets.
12098 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
12100 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
12102 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
12104 PR rtl-optimization/70703
12105 * ira-color.c (update_conflict_hard_regno_costs): Use
12106 HOST_WIDE_INT instead of long.
12108 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
12111 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
12112 not defined for x86_64 target. Add -mmmx target option when __SSE2__
12114 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
12115 for x86_64 target. Handle -m3dnowa option.
12117 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
12119 PR rtl-optimization/70703
12120 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
12121 (update_conflict_hard_regno_costs): Use long instead of unsigned
12122 arithmetic for cost calculation.
12124 2017-04-05 Jakub Jelinek <jakub@redhat.com>
12125 Bernd Edlinger <bernd.edlinger@hotmail.de>
12128 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
12131 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
12134 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
12135 ptr_mode with Pmode throughout.
12136 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
12137 into probe_stack_range and use DImode.
12139 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
12142 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
12143 call_eh_return is true.
12145 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12147 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
12148 Initialize last_match_fntype_index.
12150 2017-04-05 Jakub Jelinek <jakub@redhat.com>
12153 * tree-nvr.c: Include internal-fn.h.
12154 (pass_return_slot::execute): Ignore internal calls without
12157 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12158 Richard Biener <rguenther@suse.de>
12161 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
12162 captures used multiple times, except for the last use.
12163 * generic-match-head.c: Include gimplify.h.
12165 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12167 PR tree-optimization/79390
12168 * target.h (struct noce_if_info): Declare.
12169 * targhooks.h (default_noce_conversion_profitable_p): Declare.
12170 * target.def (noce_conversion_profitable_p): New target hook.
12171 * ifcvt.h (struct noce_if_info): New type, moved from ...
12172 * ifcvt.c (struct noce_if_info): ... here.
12173 (noce_conversion_profitable_p): Renamed to ...
12174 (default_noce_conversion_profitable_p): ... this. No longer
12176 (noce_try_store_flag_constants, noce_try_addcc,
12177 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
12178 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
12179 instead of noce_conversion_profitable_p.
12180 * config/i386/i386.c: Include ifcvt.h.
12181 (ix86_option_override_internal): Don't override
12182 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
12183 (ix86_noce_conversion_profitable_p): New function.
12184 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
12185 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
12186 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
12187 * doc/tm.texi: Regenerated.
12189 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12191 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
12194 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
12197 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
12198 instructions for small multiply cores.
12200 2017-04-04 Jeff Law <law@redhat.com>
12202 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
12204 (mips_expand_vec_perm_const): Initialize elements in orig_perm
12205 that are not set by the loop over the elements.
12207 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12210 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
12211 int mode, convert_modes it to mode as unsigned, otherwise use
12212 lowpart_subreg to mode rather than SImode.
12213 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
12214 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
12215 Use DImode instead of SImode for the shift count operand.
12216 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
12219 2017-04-04 Richard Biener <rguenther@suse.de>
12221 PR middle-end/80281
12222 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
12223 arithmetic done for the negate or the plus. Simplify.
12224 (A - (-B) -> A + B): Likewise.
12225 * fold-const.c (split_tree): Make sure to not negate pointers.
12227 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
12229 PR rtl-optimization/60818
12230 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
12231 a compare of comparisons with the thing compared if this results
12232 in a different machine mode.
12234 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
12236 * alias.c (base_alias_check): Fix typo in comment.
12237 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
12238 * cgraphunit.c (symbol_table::compile): Likewise.
12239 * collect2.c (maybe_run_lto_and_relink): Likewise.
12240 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
12241 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
12242 * config/avr/avr.c (avr_map_op_t): Likewise.
12243 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
12244 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
12245 * config/epiphany/epiphany.md (movcc): Likewise.
12246 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
12247 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
12249 * config/mips/mips.c (mips_save_restore_reg): Likewise.
12250 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
12251 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
12252 * config/sh/sh.c (sh_rtx_costs): Likewise.
12253 * fold-const.c (fold_truth_andor): Likewise.
12254 * genautomata.c (collapse_flag): Likewise.
12255 * gengtype.h (struct type::u::s): Likewise.
12256 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
12257 * input.c (FORMAT_AMOUNT): Likewise.
12258 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
12259 (known_aggs_to_agg_replacement_list): Likewise.
12260 * ipa-inline-analysis.c: Likewise.
12261 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
12262 * ipa-polymorphic-call.c
12263 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
12264 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
12265 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
12267 * modulo-sched.c (apply_reg_moves): Likewise.
12268 * omp-expand.c (build_omp_regions_1): Likewise.
12269 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
12270 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
12271 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
12272 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
12273 * value-prof.c: Likewise.
12274 * var-tracking.c (val_reset): Likewise.
12276 2017-04-03 Richard Biener <rguenther@suse.de>
12278 PR tree-optimization/80275
12279 * fold-const.c (split_address_to_core_and_offset): Handle
12282 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
12284 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
12285 descriptors is at least equal to that of functions.
12287 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
12289 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
12291 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
12294 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
12295 (mov<IMOD4:mode>): New expander.
12296 (*mov<IMOD4:mode>_internal): New insn and split pattern.
12298 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
12300 PR rtl-optimization/79405
12301 * fwprop.c (propagations_left): New variable.
12302 (forward_propagate_into): Decrement it.
12303 (fwprop_init): Initialize it.
12304 (fw_prop): If the variable has reached zero, stop propagating.
12305 (fwprop_addr): Ditto.
12307 2017-03-31 Jakub Jelinek <jakub@redhat.com>
12310 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
12311 a FUNCTION_DECL, pass it as decl instead of origin to
12314 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
12316 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
12319 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
12322 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
12323 TARGET_VSX_SMALL_INTEGER.
12325 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12327 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
12328 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
12330 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
12332 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
12333 extraction from odd-numbered MSA register.
12335 2017-03-31 Jakub Jelinek <jakub@redhat.com>
12337 PR middle-end/80173
12338 * expmed.c (store_bit_field_1): Don't attempt to create
12339 a word subreg out of hard registers wider than word if they
12340 have HARD_REGNO_NREGS of 1 for their mode.
12342 PR middle-end/80163
12343 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
12344 conversions to integer types wider than word and pointer.
12347 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
12348 (rtx_equal_for_cselib_p): Pass 0 to it.
12349 * cselib.c (cselib_hasher::equal): Likewise.
12350 (rtx_equal_for_cselib_1): Add depth argument. If depth
12351 is 128, don't look up VALUE locs and punt. Increment
12352 depth in recursive calls when walking VALUE locs.
12354 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
12356 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
12357 (make_gcov_file_name): Use the canonical path name for generating
12359 (read_line): Fix handling of files with ascii null bytes.
12361 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
12363 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
12364 to initialise a vector register instead
12365 of using a const_int.
12367 2017-03-30 Jakub Jelinek <jakub@redhat.com>
12369 PR translation/80189
12370 * gimplify.c (omp_default_clause): Use %qs instead of %s in
12371 diagnostic messages.
12373 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
12376 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
12377 (dfp_diex_<mode>): Update mode of operand 1.
12378 * doc/extend.texi (dxex, dxexq): Document change to return type.
12379 (diex, diexq): Document change to argument type.
12381 2017-03-30 Martin Jambor <mjambor@suse.cz>
12384 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
12385 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
12386 it reflects the signature changes performed at the callee side.
12387 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
12388 to cgraph_build_function_type_skip_args.
12389 (build_function_decl_skip_args): Adjust call to the above function.
12391 2017-03-30 Jakub Jelinek <jakub@redhat.com>
12394 * config/i386/sse.md
12395 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
12396 register as dest whenever it is a MEM not rtx_equal_p to the
12397 corresponding dup operand, and when forcing into reg move the
12398 reg into the memory afterwards.
12399 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
12400 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
12401 for the force_reg mode.
12402 (avx512vl_vextractf128<mode>): Use register as dest either
12403 always when a MEM, or when it is a MEM not rtx_equal_p to the
12404 corresponding dup operand, or even not when it is a CONST_VECTOR
12405 depending on the mode and lo vs. hi.
12406 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
12408 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
12409 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
12410 Likewise. Require that operands[2] is even.
12411 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
12412 Remove extraneous parens. Require that operands[2] is a multiple
12414 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
12415 operands[0] is a MEM if <mask_applied>, the predicates/constraints
12416 disallow memory then.
12418 2017-03-30 Richard Biener <rguenther@suse.de>
12420 PR tree-optimization/77498
12421 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
12422 to non-constants over backedges.
12424 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
12426 PR rtl-optimization/80233
12427 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
12428 as last_combined_insn. Do not test for BARRIER_P separately.
12430 2017-03-29 Andreas Schwab <schwab@suse.de>
12433 * calls.c (prepare_call_address): Convert funexp to Pmode before
12434 copying to temp reg.
12436 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12438 PR tree-optimization/80158
12439 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
12440 Handle possible future case of more than one alternate
12442 (replace_rhs_if_not_dup): Likewise.
12443 (replace_one_candidate): Likewise.
12445 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
12447 PR rtl-optimization/80193
12448 * ira.c (ira): Do not check allocation for LRA.
12450 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
12452 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
12453 (nvptx_output_simt_exit): Declare.
12454 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
12455 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
12456 (init_softstack_frame): Move initialization of crtl->is_leaf to...
12457 (nvptx_declare_function_name): ...here. Emit declaration of local
12458 memory space buffer for omp_simt_enter insn.
12459 (nvptx_output_unisimt_switch): New.
12460 (nvptx_output_softstack_switch): New.
12461 (nvptx_output_simt_enter): New.
12462 (nvptx_output_simt_exit): New.
12463 * config/nvptx/nvptx.h (struct machine_function): New fields
12464 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
12465 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
12466 (UNSPECV_SIMT_EXIT): Ditto.
12467 (omp_simt_enter_insn): New insn.
12468 (omp_simt_enter): New expansion.
12469 (omp_simt_exit): New insn.
12470 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
12472 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
12473 (expand_GOMP_SIMT_ENTER_ALLOC): New.
12474 (expand_GOMP_SIMT_EXIT): New.
12475 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
12476 (GOMP_SIMT_ENTER_ALLOC): Ditto.
12477 (GOMP_SIMT_EXIT): Ditto.
12478 * target-insns.def (omp_simt_enter): New insn.
12479 (omp_simt_exit): Ditto.
12480 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
12482 (lower_rec_simd_input_clauses): Implement SIMT privatization.
12483 (lower_rec_input_clauses): Likewise.
12484 (lower_lastprivate_clauses): Handle SIMT privatization.
12486 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
12487 (ompdevlow_adjust_simt_enter): New.
12488 (find_simtpriv_var_op): New.
12489 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
12490 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
12492 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
12493 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
12494 (copy_decl_for_dup_finish): Ditto.
12496 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
12498 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
12501 * config/i386/i386.c (ix86_option_override_internal): Always
12502 allow -mpreferred-stack-boundary=3 for 64-bit targets.
12504 2017-03-28 Bin Cheng <bin.cheng@arm.com>
12506 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
12508 2017-03-28 Bin Cheng <bin.cheng@arm.com>
12510 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
12511 mark new edge's irreducible flag accordign to it.
12512 (vect_do_peeling): Check loop preheader edge's irreducible flag
12513 and pass it to function slpeel_add_loop_guard.
12515 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
12517 PR tree-optimization/80218
12518 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
12519 Update block frequencies and counts.
12521 2017-03-28 Richard Biener <rguenther@suse.de>
12523 PR tree-optimization/78644
12524 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
12525 of a simplification result we may not use it at all.
12527 2017-03-28 Richard Biener <rguenther@suse.de>
12530 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
12531 without arguments, generate default definition of a SSA name.
12533 2017-03-28 Richard Biener <rguenther@suse.de>
12535 PR middle-end/80222
12536 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
12537 TYPE_REF_CAN_ALIAS_ALL references.
12538 * fold-const.c (fold_indirect_ref_1): Likewise.
12540 2017-03-28 Martin Liska <mliska@suse.cz>
12543 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
12544 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
12546 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12547 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
12549 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
12550 (EXTRA_SPECS): Define.
12551 (SUBTARGET_EXTRA_SPECS): Likewise.
12552 (SUBTARGET_CPP_SPEC): Likewise.
12553 * config/arc/elf.h (EXTRA_SPECS): Renamed to
12554 SUBTARGET_EXTRA_SPECS.
12555 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
12557 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12559 * config/arc/simdext.md (vst64_insn): Update pattern.
12560 (vld32wh_insn): Likewise.
12561 (vld32wl_insn): Likewise.
12562 (vld64_insn): Likewise.
12563 (vld32_insn): Likewise.
12565 2017-03-28 Marek Polacek <polacek@redhat.com>
12568 * fold-const.c (fold_comparison): Use protected_set_expr_location
12569 instead of SET_EXPR_LOCATION.
12571 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
12573 * tree.c (add_expr): Avoid name lookup warning.
12575 2017-03-27 Jeff Law <law@redhat.com>
12577 PR tree-optimization/80216
12578 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
12579 function name. Limit recursion depth.
12580 (record_temporary_equivalences): Corresponding changes.
12582 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
12584 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
12587 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12590 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
12592 * cfgcleanup.c (reg_note_cfa_p): New array.
12593 (insns_have_identical_cfa_notes): New function.
12594 (old_insns_match_p): Don't cross-jump in between /f
12595 and non-/f instructions. If both i1 and i2 are frame related,
12596 verify all CFA notes, their order and content.
12598 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
12601 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
12602 HImode and SImode with zero extend to DImode to one insn.
12603 (bswap<mode>2_extenddi): Likewise.
12604 (bswapsi2_extenddi): Likewise.
12605 (bswaphi2_extendsi): Likewise.
12606 (bswaphi2): Combine bswap HImode and SImode into one insn.
12607 Separate memory insns from swapping register.
12608 (bswapsi2): Likewise.
12609 (bswap<mode>2): Likewise.
12610 (bswaphi2_internal): Delete, no longer used.
12611 (bswapsi2_internal): Likewise.
12612 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
12613 store, and gpr<-gpr swap insns.
12614 (bswap<mode>2_store): Likewise.
12615 (bswaphi2_reg): Register only splitter, combine with the splitter.
12616 (bswaphi2 splitter): Likewise.
12617 (bswapsi2_reg): Likewise.
12618 (bswapsi2 splitter): Likewise.
12619 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
12620 the insns into load, store, and register/register insns.
12621 (bswapdi2_ldbrx): Likewise.
12622 (bswapdi2_load): Likewise.
12623 (bswapdi2_store): Likewise.
12624 (bswapdi2_reg): Likewise.
12626 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
12628 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
12629 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
12631 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12634 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
12636 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
12637 special handling for target option conflicts between dform
12638 options (-mpower9-dform, -mpower9-dform-vector,
12639 -mpower9-dform-scalar) and -mno-direct-move.
12641 2017-03-27 Richard Biener <rguenther@suse.de>
12643 PR tree-optimization/80181
12644 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
12646 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12648 * config/arc/predicates.md (move_double_src_operand): Replace the
12649 call to move_double_src_operand with a call to address_operand.
12651 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12653 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
12654 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
12655 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
12657 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12659 * config/arc/predicates.md (long_immediate_loadstore_operand):
12660 Consider scaled addresses cases.
12662 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12664 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
12665 restored when in interrupt.
12666 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
12667 doesn't have delay slot.
12669 2017-03-27 Richard Biener <rguenther@suse.de>
12672 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
12673 inlined thunk clones.
12675 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12678 * asan.c (instrument_derefs): Copy over last operand from
12679 original COMPONENT_REF to the new COMPONENT_REF with
12680 DECL_BIT_FIELD_REPRESENTATIVE.
12681 * ubsan.c (instrument_object_size): Likewise.
12683 2017-03-27 Richard Biener <rguenther@suse.de>
12685 PR tree-optimization/80170
12686 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
12687 sure DR/SCEV didnt fold in constants we do not see when looking
12688 at the reference base alignment.
12690 2017-03-27 Richard Biener <rguenther@suse.de>
12692 PR middle-end/80171
12693 * gimple-fold.c (fold_ctor_reference): Properly guard against
12694 NULL return value from canonicalize_constructor_val.
12696 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
12699 * config/i386/i386.c (ix86_expand_builtin)
12700 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
12701 flags reg setting and flags reg using instructions.
12702 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
12703 clobbering instructions to zero extend op2.
12705 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
12707 * doc/install.texi (Configuration) <--with-aix-soname>:
12708 Update link to AIX ld.
12710 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
12712 PR rtl-optimization/80160
12713 PR rtl-optimization/80159
12714 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
12715 reg_alternate_class into account.
12717 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
12720 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
12721 to consider in curr_insn_transform.
12723 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12725 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
12726 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
12727 and emit_mode_inner.
12729 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12731 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
12732 argument to the overloaded builtin variants. Use the new flag to
12733 deprecate certain builtin variants.
12734 * config/s390/s390-builtin-types.def: Add new builtin types.
12735 * config/s390/s390-builtins.h: Support new flags field for
12736 overloaded builtins.
12737 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
12738 (s390_macro_to_expand): Enable vector float data type.
12739 (s390_cpu_cpp_builtins_internal): Indicate support of the new
12740 builtins by incrementing the __VEC__ version number.
12741 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
12743 (s390_resolve_overloaded_builtin): Emit error messages depending
12744 on the builtin flags.
12745 * config/s390/s390.c (s390_expand_builtin): Support additional
12746 flags argument. Change error message to match the messages
12747 emitted in s390-c.c.
12748 * config/s390/s390.md: New UNSPEC_* constants.
12749 (op_type): Add new instruction types.
12750 * config/s390/vecintrin.h: Add new builtins and test data class
12752 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
12753 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
12754 (VEC_INEXACT, VEC_NOINEXACT): New constants.
12755 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
12756 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
12757 ("vec_mergel<mode>"): V_HW -> VEC_HW.
12759 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
12760 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
12761 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
12762 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
12764 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
12765 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
12766 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
12767 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
12769 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
12770 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
12771 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
12772 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
12773 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
12774 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
12775 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
12777 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
12778 ("vec_scatter_element<V_HW_4:mode>_DI")
12779 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
12780 ("vec_fpint<mode>", "vflls")
12781 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
12782 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
12783 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
12784 ("*vec_cmphe<mode>_cc"): ... these.
12786 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
12787 mode constant instead of magic value.
12789 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12791 * config/s390/s390.c (s390_expand_vec_compare): Support other
12792 vector floating point modes than just V2DF.
12793 (s390_expand_vcond): Likewise.
12794 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
12795 (s390_cannot_change_mode_class): Prevent mode changes between TF
12796 and V1TF in vector registers.
12797 * config/s390/s390.md (DF, SF): New mode attributes.
12798 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
12799 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
12800 SFmode support for VRs.
12801 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
12803 (VFT, VF_HW): New mode iterators.
12804 (vw, sdx): New mode attributes.
12805 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
12806 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
12807 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
12808 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
12809 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
12810 also the new vector floating point modes. Renaming to ...
12812 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
12813 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
12814 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
12815 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
12816 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
12817 ("vec_unordered<mode>"): ... these.
12819 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
12820 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
12821 ("*vec_extendv2df"): New insn definitions.
12823 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12825 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
12826 ("mulditi3_2", "*muldi3_sign"): New patterns.
12827 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
12828 rename the pattern definition.
12830 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12832 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
12834 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
12836 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12838 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
12839 instruction if possible.
12840 * config/s390/vector.md (vec_halfnumelts): New mode
12842 ("*vec_vllezlf<mode>"): New pattern.
12844 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12846 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
12847 ("popcountv4si2", "popcountv2di2"): Rename to ...
12848 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
12849 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
12851 ("popcount<mode>2_vxe"): New pattern.
12853 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12855 * common/config/s390/s390-common.c (processor_flags_table): Add
12857 * config.gcc: Add arch12.
12858 * config/s390/driver-native.c (s390_host_detect_local_cpu):
12859 Default to arch12 for unknown CPU model numbers.
12860 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
12861 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
12862 PROCESSOR_max sanity check.
12863 * config/s390/s390-opts.h (enum processor_type): Add
12865 * config/s390/s390.c (processor_table): Add arch12.
12866 (s390_expand_builtin): Add check for B_VXE flag.
12867 (s390_issue_rate): Add PROCESSOR_ARCH12.
12868 (s390_get_sched_attrmask): Likewise.
12869 (s390_get_unit_mask): Likewise.
12870 (s390_sched_score): Enable z13 scheduling for arch12.
12871 (s390_sched_reorder): Likewise.
12872 (s390_sched_variable_issue): Likewise.
12873 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
12875 (s390_tune_attr): Use z13 scheduling also for arch12.
12876 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
12877 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
12878 (TARGET_VXE_P): New macros.
12879 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
12880 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
12881 * config/s390/s390.opt: Add arch12 as processor_type.
12883 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12885 * config/s390/s390.md
12886 ("fixuns_truncdddi2", "fixuns_trunctddi2")
12887 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
12888 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
12890 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
12891 Rename expanders to ...
12893 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
12894 ("fixuns_truncdddi2_emu"): ... these.
12896 ("fixuns_trunc<mode>si2_emu"): New expander.
12898 ("*fixuns_truncdfdi2_z13"): Rename to ...
12899 ("*fixuns_truncdfdi2_vx"): ... this.
12901 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12903 * config/s390/2964.md: Remove the single element vector compare
12904 instructions which are no longer used.
12905 * config/s390/s390.c (s390_select_ccmode): Remove handling of
12907 (s390_canonicalize_comparison): Remove handling of DFmode
12909 (s390_expand_vec_compare_scalar): Remove function.
12910 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
12911 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
12913 ("*cmp<mode>_ccs"): Add wfcdb instruction.
12915 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12917 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
12919 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
12920 will anyway by matched by mov<mode>_64dfp.
12922 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12924 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
12925 vlef/vstef. Add missing operand to vleif.
12927 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12929 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
12930 pair for all vector types with 64 bit elements.
12931 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
12932 * config/s390/vector.md (V_HW_64): ... here.
12933 (V_128_NOSINGLE): New mode iterator.
12934 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
12935 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
12936 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
12937 ("*vec_load_pairv2di"): Change to ...
12938 ("*vec_load_pair<mode>"): ... this one.
12940 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12942 * config/s390/constraints.md: Add comments.
12943 (jKK): Reject element sizes > 8 bytes.
12944 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
12946 * config/s390/s390.md: Add the s_operand checks formerly in
12947 s390_split_ok_p to various splitters where they are still
12949 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
12950 for 128 bit vectors. Plus two splitters.
12952 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12954 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
12957 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12960 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
12961 error if the boundary argument is not constant.
12963 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12965 PR rtl-optimization/80112
12966 * loop-doloop.c (doloop_condition_get): Don't check condition
12967 if cmp isn't SET with IF_THEN_ELSE src.
12969 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12971 PR tree-optimization/80158
12972 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
12973 replacing a candidate statement, also replace it for the
12974 candidate's alternate interpretation.
12975 (replace_rhs_if_not_dup): Likewise.
12976 (replace_one_candidate): Likewise.
12978 2017-03-24 Richard Biener <rguenther@suse.de>
12980 PR tree-optimization/80167
12981 * graphite-isl-ast-to-gimple.c
12982 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
12984 (translate_isl_ast_to_gimple::get_rename): Likewise.
12986 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
12988 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
12989 handling of certain combinations of target options, including the
12990 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
12991 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
12993 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12996 * config/arm/arm.md (*load_multiple): Add reload_completed to
12997 matching condition.
12999 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13000 Richard Biener <rguenth@suse.de>
13002 PR tree-optimization/79908
13003 PR tree-optimization/80136
13004 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
13005 been cast away, gimplify_and_add suffices.
13007 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
13009 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
13011 2017-03-23 Richard Biener <rguenther@suse.de>
13013 PR tree-optimization/80032
13014 * gimplify.c (gimple_push_cleanup): Forced unconditional
13015 cleanups still have to go to the conditional_cleanups
13018 2017-03-22 Jakub Jelinek <jakub@redhat.com>
13020 PR tree-optimization/80072
13021 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
13023 (next_operand_entry_id): Change type to unsigned int.
13024 (sort_by_operand_rank): Make sure to return the right return value
13025 even if unsigned fields are bigger than INT_MAX.
13026 (struct oecount): Change cnt and id type to unsigned int.
13027 (oecount_hasher::equal): Formatting fix.
13028 (oecount_cmp): Make sure to return the right return value
13029 even if unsigned fields are bigger than INT_MAX.
13030 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
13033 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
13034 TREE_READONLY on result if writing it more than once.
13037 * doc/invoke.texi (-fsanitize=thread): Document that with
13038 -fnon-call-exceptions atomics are not able to throw
13042 * tsan.c: Include tree-eh.h.
13043 (instrument_builtin_call): Call maybe_clean_eh_stmt or
13044 maybe_clean_or_replace_eh_stmt where needed.
13045 (instrument_memory_accesses): Add cfg_changed argument.
13046 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
13047 if it returned true.
13048 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
13050 PR rtl-optimization/63191
13051 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
13052 wrapper function, moved the whole old content into ...
13053 (ix86_delegitimize_address_1): ... this. New inline function.
13054 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
13055 true as last argument instead of ix86_delegitimize_address.
13057 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
13059 * config/aarch64/aarch64.c (generic_branch_cost): Copy
13060 cortexa57_branch_cost.
13062 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
13064 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
13066 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13069 * doc/md.texi (Constraints): Document wA constraint.
13070 * config/rs6000/constraints.md (wA): New.
13071 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
13072 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
13073 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
13074 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
13076 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
13079 * gimplify.c (is_oacc_declared): New function.
13080 (oacc_default_clause): Use it to set default flags for acc declared
13081 variables inside parallel regions.
13082 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
13083 declared variables.
13084 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
13085 declare attribute to any decl as necessary.
13087 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
13090 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
13091 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
13092 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
13093 (arm_arch_lpae): This.
13094 * config/arm/arm.c (arm_arch7ve): Rename into ...
13095 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
13096 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
13099 2017-03-22 Martin Liska <mliska@suse.cz>
13102 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
13103 error message instead of an ICE.
13105 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13107 * doc/extend.texi (6.11 Additional Floating Types): Revise.
13109 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
13111 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
13113 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
13116 2017-03-21 Martin Sebor <msebor@redhat.com>
13118 * doc/extend.texi: Use "cannot" instead of "can't."
13119 * doc/hostconfig.texi: Same.
13120 * doc/install.texi: Same.
13121 * doc/invoke.texi: Same.
13122 * doc/loop.texi: Same.
13123 * doc/md.texi: Same.
13124 * doc/objc.texi: Same.
13125 * doc/rtl.texi: Same.
13126 * doc/tm.texi: Same.
13127 * doc/tm.texi.in: Same.
13128 * doc/trouble.texi: Same.
13130 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
13133 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
13134 (collect_checksum_attributes): Set it.
13135 (die_checksum_ordered): Use it.
13137 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13139 PR tree-optimization/79908
13140 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
13141 change: For a VA_ARG whose LHS has been cast away, use
13142 force_gimple_operand to construct the side effects.
13144 2017-03-21 David Malcolm <dmalcolm@redhat.com>
13146 PR translation/80001
13147 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
13148 more amenable to translation.
13149 (oacc_loop_auto_partitions): Likewise.
13151 2017-03-21 Marek Polacek <polacek@redhat.com>
13152 Martin Sebor <msebor@redhat.com>
13154 PR tree-optimization/80109
13155 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
13156 on INTEGRAL_TYPE_P.
13158 2017-03-21 Jakub Jelinek <jakub@redhat.com>
13159 Segher Boessenkool <segher@kernel.crashing.org>
13162 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
13163 check reg_used_between_p between insn and one of succ or succ2
13164 depending on if succ is artificial insn not inserted into insn
13167 2017-03-21 Martin Liska <mliska@suse.cz>
13169 PR gcov-profile/80081
13170 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
13171 * doc/gcc.texi: Include gcov-dump stuff.
13172 * doc/gcov-dump.texi: New file.
13174 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
13176 PR rtl-optimization/79150
13177 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
13178 conditional jump, if the jump is the last insn of the loop.
13180 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13181 Richard Biener <rguenth@suse.de>
13183 PR tree-optimization/79908
13184 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
13185 been cast away, use force_gimple_operand to construct the side
13188 2017-03-21 Martin Liska <mliska@suse.cz>
13190 PR libfortran/79956
13191 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
13194 2017-03-21 Brad Spengler <spender@grsecurity.net>
13197 * plugin.c (htab_hash_plugin): New function.
13198 (add_new_plugin): Use it and adjust.
13199 (parse_plugin_arg_opt): Adjust.
13200 (init_one_plugin): Likewise.
13202 2017-03-21 Richard Biener <rguenther@suse.de>
13204 PR tree-optimization/80032
13205 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
13206 if set force the cleanup to happen unconditionally.
13207 (gimplify_target_expr): Push inserted clobbers with force_uncond
13208 to avoid them being removed by control-dependent DCE.
13210 2017-03-21 Richard Biener <rguenther@suse.de>
13212 PR tree-optimization/80122
13213 * tree-inline.c (copy_bb): Do not expans va-arg packs or
13214 va_arg_pack_len when the inlined call stmt requires pack
13216 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
13218 2017-03-21 Jakub Jelinek <jakub@redhat.com>
13221 * tsan.c (instrument_builtin_call): If the memory model argument
13222 is not a constant, assume it is valid.
13225 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
13228 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
13230 PR rtl-optimization/79910
13231 * combine.c (can_combine_p): Do not allow combining an I0 or I1
13232 if its dest is used by an insn before I2 (other than the combined
13233 insns themselves, which are properly handled already).
13235 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
13238 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
13240 * combine.c (record_used_regs): New static function.
13241 (try_combine): Handle situations where there is an additional
13242 instruction between I2 and I3 which needs to have a LOG_LINK
13246 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
13248 * combine.c (try_combine): Delete redundant i1 test. Call
13249 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
13251 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13254 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
13255 alternatives 13/14.
13257 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13259 PR tree-optimization/80054
13260 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
13261 the optimization if a PHI or any of its arguments is not dominated
13262 by the candidate's basis. Use gphi* rather than gimple* as
13264 (replace_profitable_candidates): Clean up a gimple* variable that
13265 should be a gphi* variable.
13267 2017-03-20 Martin Sebor <msebor@redhat.com>
13270 * doc/extend.texi (attribute constructor): Document present limitation.
13272 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
13275 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
13276 __POWER9_VECTOR__ #ifdef control, change template definition to
13277 use Power9-specific built-in function.
13278 (vec_any_eq): Likewise.
13279 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
13280 to control outcomes from this test.
13281 (vector_ae_<mode>p): For VEC_F modes, likewise.
13283 2017-03-20 Ian Lance Taylor <iant@google.com>
13285 * config/i386/i386.c (ix86_function_regparm): Save an extra
13286 register for -fsplit-stack with DECL_STATIC_CHAIN.
13288 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
13291 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
13292 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
13294 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
13296 * config/riscv/riscv.c (riscv_print_operand): Use "fence
13298 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
13301 2017-03-20 Marek Polacek <polacek@redhat.com>
13304 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
13306 2017-03-20 Richard Biener <rguenther@suse.de>
13308 PR tree-optimization/80113
13309 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
13310 allocate extra SSA name for PHI def.
13311 (add_close_phis_to_outer_loops): Likewise.
13312 (add_close_phis_to_merge_points): Likewise.
13313 (copy_loop_close_phi_args): Likewise.
13314 (copy_cond_phi_nodes): Likewise.
13316 2017-03-20 Martin Liska <mliska@suse.cz>
13318 PR middle-end/79753
13319 * tree-chkp.c (chkp_build_returned_bound): Do not build
13320 returned bounds for a LHS that's not a BOUNDED_P type.
13322 2017-03-20 Martin Liska <mliska@suse.cz>
13326 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
13327 COMPLEX_CST and VECTOR_CST.
13329 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13332 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
13333 target operand. A new splitter adds the clobber statement in case
13334 the target operand is dead anyway.
13336 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
13338 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
13339 to age-old versions of binutils and glibc.
13341 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
13343 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
13345 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
13347 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
13349 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
13351 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
13352 requirement for binutils 2.13.
13354 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
13356 * combine.c (try_combine): Delete redundant i1 test. Call
13357 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
13359 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
13361 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
13362 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
13364 <riscv64-*-elf>: Re-arrange section
13365 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
13366 <riscv32-*-linux>: Likewise.
13367 <riscv64-*-elf>: Likewise
13368 <riscv64-*-linux>: Likewise.
13370 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
13373 * aarch64.opt(verbose-cost-dump): Fix typo.
13375 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
13378 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
13379 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
13381 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
13383 * reload.c (find_reloads): When reloading a nonoffsettable address,
13384 use RELOAD_OTHER for it and its address reloads.
13386 PR rtl-optimization/79910
13387 * combine.c (record_used_regs): New static function.
13388 (try_combine): Handle situations where there is an additional
13389 instruction between I2 and I3 which needs to have a LOG_LINK
13392 2017-03-17 Jeff Law <law@redhat.com>
13394 PR tree-optimization/71437
13395 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
13396 conditional in the hash table first.
13397 (vrp_dom_walker::before_dom_children): Extract condition from
13398 ASSERT_EXPR. Record condition, its inverion and any implied
13399 conditions as well.
13401 2017-03-17 Marek Polacek <polacek@redhat.com>
13402 Markus Trippelsdorf <markus@trippelsdorf.de>
13404 PR tree-optimization/80079
13405 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
13408 2017-03-17 Richard Biener <rguenther@suse.de>
13410 PR middle-end/80075
13411 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
13412 Properly verify the LHS before the RHS possibly claims to be
13414 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
13417 2017-03-17 Martin Jambor <mjambor@suse.cz>
13419 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
13420 (List of -O2 options): Likewise.
13421 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
13424 2017-03-17 Tom de Vries <tom@codesourcery.com>
13426 * gcov-dump.c (print_usage): Print bug_report_url.
13428 2017-03-17 Richard Biener <rguenther@suse.de>
13430 PR middle-end/80050
13431 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
13432 (parser::peek): Likewise.
13434 2017-03-17 Richard Biener <rguenther@suse.de>
13436 PR tree-optimization/80048
13437 * sese.c (free_sese_info): Properly release rename_map and
13438 copied_bb_map elements.
13440 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
13442 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
13443 Add linked-list forward and backlinks. Insert on
13444 construction, remove on destruction.
13445 (class pass_store_merging): Add m_stores_head field.
13446 (pass_store_merging::terminate_and_process_all_chains):
13447 Iterate over m_stores_head list.
13448 (pass_store_merging::terminate_all_aliasing_chains):
13450 (pass_store_merging::execute): Check for debug stmts first.
13451 Push new chains onto the m_stores_head stack.
13453 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
13456 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
13457 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
13458 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
13460 2017-03-16 Jeff Law <law@redhat.com>
13462 PR tree-optimization/71437
13463 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
13464 member function. Implementation moved into after_dom_children
13465 member function and into the threader's thread_outgoing_edges
13467 (dom_opt_dom_walker::after_dom_children): Simplify by moving
13468 some code into new thread_outgoing_edges.
13469 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
13470 definition. Simplify marker handling (do it here). Assume we always
13471 have the available expression and the const/copies tables.
13472 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
13474 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
13475 * tree-vrp.c (equiv_stack): No longer file scoped.
13476 (vrp_dom_walker): New class.
13477 (vrp_dom_walker::before_dom_children): New member function.
13478 (vrp_dom_walker::after_dom_children): Likewise.
13479 (identify_jump_threads): Setup domwalker. Use it rather than
13480 walking edges in a random order by hand. Simplify setup/finalization.
13481 (finalize_jump_threads): Remove.
13482 (vrp_finalize): Do not call identify_jump_threads here.
13483 (execute_vrp): Do it here instead and call thread_through_all_blocks
13486 PR tree-optimization/71437
13487 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
13489 (simplify_stmt_for_jump_threading): Add basic_block argument. All
13491 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
13492 (dom_opt_dom_walker::thread_across_edge): Remove
13493 handle_dominating_asserts argument. All callers changed.
13494 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
13495 changes. Remove calls to lhs_of_dominating_assert. Other
13496 uses of handle_dominating_asserts turn into unconditional code
13497 (simplify_control_stmt_condition_1): Likewise.
13498 (simplify_control_stmt_condition): Likewise.
13499 (thread_through_normal_block, thread_across_edge): Likewise.
13500 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
13501 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
13502 object if it is not an SSA_NAME.
13503 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
13504 before calling into the VRP specific simplifiers.
13505 (identify_jump_threads): Remove handle_dominating_asserts
13508 2017-03-16 Jakub Jelinek <jakub@redhat.com>
13511 * tree-diagnostic.c (default_tree_printer): No longer static.
13512 * tree-diagnostic.h (default_tree_printer): New prototype.
13514 2017-03-16 Tamar Christina <tamar.christina@arm.com>
13516 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
13517 Change ins into fmov.
13519 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13521 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
13522 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
13523 Use h_con constraint for operand 1.
13524 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
13525 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
13527 2017-03-15 Jeff Law <law@redhat.com>
13529 PR tree-optimization/71437
13530 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
13531 (record_temporary_equivalences): Use it.
13533 PR tree-optimization/71437
13534 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
13535 tree-ssa-scopedtables.
13536 (lookup_avail_expr, build_and_record_new_cond): Likewise.
13537 (record_conditions, record_cond, vuse_eq): Likewise.
13538 (record_edge_info): Adjust to API tweak of record_conditions.
13539 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
13540 (record_temporary_equivalences, optimize_stmt): Likewise.
13541 (eliminate_redundant_computations): Likewise.
13542 (record_equivalences_from_stmt): Likewise.
13543 * tree-ssa-scopedtables.c: Include options.h and params.h.
13544 (vuse_eq): New function, moved from tree-ssa-dom.c
13545 (build_and_record_new_cond): Likewise.
13546 (record_conditions): Likewise. Accept vector of conditions rather
13547 than edge_equivalence structure for first argument.
13548 for the first argument.
13549 (avail_exprs_stack::lookup_avail_expr): New member function, moved
13550 from tree-ssa-dom.c.
13551 (avail_exprs_stack::record_cond): Likewise.
13552 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
13553 from tree-ssa-dom.c.
13554 (avail_exprs_stack): Add new member functions lookup_avail_expr
13556 (record_conditions): Declare.
13558 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
13561 * lra-constraints.c (process_alt_operands): Increase reject for
13562 reloading an input/output operand.
13564 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
13567 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
13568 insns to convert from signed/unsigned char/short to IEEE 128-bit
13570 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
13572 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
13575 * config/i386/i386.c (ix86_vector_duplicate_value): Create
13576 subreg of inner mode for values already in registers.
13578 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
13580 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
13581 iteration reg is used after the loop.
13583 2017-03-14 Martin Sebor <msebor@redhat.com>
13585 PR tree-optimization/79800
13586 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
13587 precision in negative-positive range.
13588 (format_floating): Call non-const overload with adjusted precision.
13590 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
13593 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
13596 2017-03-14 Martin Sebor <msebor@redhat.com>
13598 PR middle-end/80020
13599 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
13600 * builtins.def (aligned_alloc): Use it.
13603 * Makefile.in (GTFILES): Add calls.c.
13604 * calls.c: Include "gt-calls.h".
13606 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
13608 PR rtl-optimization/79728
13609 * regs.h (struct target_regs): New field
13610 x_contains_allocatable_regs_of_mode.
13611 (contains_allocatable_regs_of_mode): New macro.
13612 * reginfo.c (init_reg_sets_1): Initialize it, and change
13613 contains_reg_of_mode so it includes global regs as well.
13614 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
13615 rather than contains_regs_of_mode.
13617 2017-03-14 Martin Liska <mliska@suse.cz>
13619 * doc/invoke.texi: Document options that can't be combined with
13620 -fcheck-pointer-bounds.
13622 2017-03-14 Martin Liska <mliska@suse.cz>
13624 PR middle-end/79831
13625 * doc/invoke.texi (-Wchkp): Document the option.
13627 2017-03-14 Martin Liska <mliska@suse.cz>
13629 * Makefile.in: Install gcov-dump.
13631 2017-03-14 Martin Liska <mliska@suse.cz>
13633 * multiple_target.c (expand_target_clones): Bail out for
13634 an invalid attribute.
13636 2017-03-14 Richard Biener <rguenther@suse.de>
13638 * alias.c (struct alias_set_entry): Pack properly.
13639 * cfgloop.h (struct loop): Likewise.
13640 * cse.c (struct set): Likewise.
13641 * ipa-utils.c (struct searchc_env): Likewise.
13642 * loop-invariant.c (struct invariant): Likewise.
13643 * lra-remat.c (struct cand): Likewise.
13644 * recog.c (struct change_t): Likewise.
13645 * rtl.h (struct address_info): Likewise.
13646 * symbol-summary.h (function_summary): Likewise.
13647 * tree-loop-distribution.c (struct partition): Likewise.
13648 * tree-object-size.c (struct object_size_info): Likewise.
13649 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
13650 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
13651 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
13652 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
13653 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
13654 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
13655 (struct _stmt_vec_info): Likewise.
13657 2017-03-14 Martin Liska <mliska@suse.cz>
13660 * multiple_target.c (create_dispatcher_calls): Check that
13661 a target can create a function dispatcher.
13663 2017-03-14 Martin Liska <mliska@suse.cz>
13666 * multiple_target.c (expand_target_clones): Drop local.local
13667 flag for default implementation.
13669 2017-03-14 Richard Biener <rguenther@suse.de>
13671 PR tree-optimization/80030
13672 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
13674 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
13676 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
13677 gcc_fallthrough() instead of __attribute__((fallthrough));
13679 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13681 * doc/gcc.texi: Remove "up" link to (DIR).
13682 * doc/gccint.texi: Ditto.
13684 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13686 * doc/install.texi (Specific) <avr>: Remove reference to
13689 2017-03-13 Jeff Law <law@redhat.com>
13691 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
13692 attribute rather than comments.
13694 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
13695 match_scratch operand is highest.
13697 2017-03-13 Martin Liska <mliska@suse.cz>
13699 PR middle-end/78339
13700 * ipa-pure-const.c (warn_function_noreturn): If the declarations
13701 is a CHKP clone, use original declaration.
13703 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13705 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
13706 (arc_conditional_register_usage): Use a different allocation order
13707 when optimizing for size.
13708 * common/config/arc/arc-common.c (arc_option_optimization_table):
13709 Section anchors default on when optimizing for size.
13711 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13713 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
13715 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13717 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
13718 * config/arc/arc.md (cpu_facility): Add cd variant.
13719 (*movqi_insn): Add code density variant.
13720 (*movhi_insn): Likewise.
13721 (*movqi_insn): Likewise.
13722 (*addsi3_mixed): Likewise.
13723 (subsi3_insn): Likewise.
13725 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13727 * config/arc/arc.md (movsi_cond_exec): Update constraint.
13729 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13731 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
13732 expressions with MINUS and UNARY ops.
13734 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13737 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
13739 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
13740 between vec_select and vector argument.
13741 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
13742 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
13743 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
13744 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
13745 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
13746 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
13748 2017-03-13 Richard Biener <rguenther@suse.de>
13751 * params.def (vect-max-peeling-for-alignment): Fix typo.
13753 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13755 * doc/install.texi (Specific) <mips-*-*>: Remove description of
13756 issue that only occurred with binutils below 2.18.
13758 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13760 * doc/install.texi (Specific) <cris-axis-elf>: No longer
13761 refer to binutils 2.11/2.12 minimum.
13763 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13765 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
13766 ftp.kernel.org and simplify binutils requirement.
13768 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
13770 * doc/invoke.texi (Warning Options): Fix spelling of link-time
13772 (Optimize Options): Ditto. Also remove redundancy.
13774 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13776 PR translation/79848
13777 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
13779 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
13780 to G_ to avoid double translation.
13782 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13784 PR translation/79923
13785 * auto-profile.c (get_combined_location): Convert leading
13786 character of diagnostics to lower case and remove trailing period.
13787 (read_profile): Likewise for various diagnostics.
13788 * config/arm/arm.c (arm_option_override): Remove trailing period
13789 from various diagnostics.
13790 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
13791 (msp430_expand_delay_cycles): Likewise.
13793 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13796 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
13797 full command-line argument, rather than just "str".
13798 (aarch64_validate_march): Likewise.
13799 (aarch64_validate_mtune): Likewise.
13801 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
13803 PR rtl-optimization/78911
13804 * lra-assigns.c (must_not_spill_p): New function.
13805 (spill_for): Use it.
13807 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13809 PR tree-optimization/79981
13810 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
13811 ATOMIC_COMPARE_EXCHANGE ifn result.
13812 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
13813 IFN_ATOMIC_COMPARE_EXCHANGE.
13815 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13818 * opts.c (parse_sanitizer_options): Add missing question mark to
13819 "did you mean" message.
13821 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13823 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
13825 (VMULEUH_UNS): Likewise.
13826 (VMULOUB_UNS): Likewise.
13827 (VMULOUH_UNS): Likewise.
13828 * config/rs6000/rs6000.c (builtin_function_type): Remove
13829 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
13831 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13834 * read-rtl-function.c (function_reader::read_rtx_operand): Update
13835 x with result of extra_parsing_for_operand_code_0.
13836 (function_reader::extra_parsing_for_operand_code_0): Convert
13837 return type from void to rtx, returning x. When reading
13838 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
13839 larger size containing struct block_symbol.
13841 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
13843 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
13844 -mfloat128-hardware without -m64.
13846 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
13849 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
13850 entries to the case statement that marks unsigned arguments to
13851 overloaded functions.
13853 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
13855 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
13856 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
13858 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
13861 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
13862 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
13864 2017-03-10 Martin Liska <mliska@suse.cz>
13868 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
13869 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
13872 2017-03-10 Olivier Hainque <hainque@adacore.com>
13874 * tree-switch-conversion (array_value_type): Start by resetting
13875 candidate type to it's main variant.
13877 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13879 PR rtl-optimization/79909
13880 * combine.c (try_combine): Use simplify_replace_rtx on individual
13881 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
13882 of the whole CALL_INSN_FUNCTION_USAGE.
13884 PR tree-optimization/79972
13885 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
13886 get_range_info on SSA_NAMEs. Formatting fixes.
13888 2017-03-10 Richard Biener <rguenther@suse.de>
13889 Jakub Jelinek <jakub@redhat.com>
13891 PR tree-optimization/77975
13892 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
13893 edge to be constant.
13894 (get_val_for): For constant x return it. Formatting fix.
13895 (loop_niter_by_eval): Avoid pointless looping if the next iteration
13896 would use the same bases as the current one.
13898 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13900 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
13901 instead of vec_select for V1TImode.
13902 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
13904 (VSX_LE_128): Add V1TI to this mode iterator.
13905 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
13906 (*vsx_le_perm_store_<mode>): Likewise.
13907 (pre-reload splitter for VSX stores): Likewise.
13908 (post-reload splitter for VSX stores): Likewise.
13909 (*vsx_xxpermdi2_le_<mode>): Likewise.
13910 (*vsx_lxvd2x2_le_<mode>): Likewise.
13911 (*vsx_stxvd2x2_le_<mode>): Likewise.
13913 2017-03-09 Michael Eager <eager@eagercon.com>
13915 Correct failures with --enable-checking=yes,rtl.
13917 * config/microblaze/microblaze.c (microblaze_expand_shift):
13918 Replace GET_CODE test with CONST_INT_P and INTVAL test with
13919 test for const0_rtx.
13920 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
13921 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
13923 2017-03-09 Richard Biener <rguenther@suse.de>
13925 PR tree-optimization/79977
13926 * graphite-scop-detection.c (scop_detection::merge_sese):
13927 Handle the case of extra exits to blocks dominating the entry.
13929 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
13931 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13934 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
13936 PR rtl-optimization/79949
13937 * lra-constraints.c (process_alt_operands): Check memory when
13938 trying to predict a cycle. Print about the overall increase.
13940 2017-03-09 Richard Biener <rguenther@suse.de>
13942 PR middle-end/79971
13943 * gimple-expr.c (useless_type_conversion_p): Preserve
13944 TYPE_SATURATING for fixed-point types.
13946 2017-03-09 Richard Biener <rguenther@suse.de>
13949 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
13950 alignment of BLKmode params.
13952 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13955 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
13956 (VALL_NO_V2Q): Likewise.
13958 * config/aarch64/aarch64-simd.md
13959 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
13961 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
13962 VALL_NO_V2Q mode iterator.
13963 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
13965 2017-03-09 Martin Liska <mliska@suse.cz>
13967 PR tree-optimization/79631
13968 * tree-chkp-opt.c (chkp_is_constant_addr): Call
13969 tree_int_cst_sign_bit just for INTEGER constants.
13971 2017-03-09 Martin Liska <mliska@suse.cz>
13975 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
13978 2017-03-09 Marek Polacek <polacek@redhat.com>
13981 * tree.c (inchash::add_expr): Handle TREE_VEC.
13983 2017-03-09 Martin Liska <mliska@suse.cz>
13986 (chkp_narrow_size_and_offset): New function.
13987 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
13988 (void chkp_parse_bit_field_ref): New function.
13989 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
13990 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
13992 2017-03-09 Martin Liska <mliska@suse.cz>
13995 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
13996 (chkp_find_bounds_1): Remove gcc_unreachable.
13998 2017-03-09 Jakub Jelinek <jakub@redhat.com>
14001 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
14002 BUILT_IN_SYNC*, determine the access type from the size suffix and
14003 always build a MEM_REF with that type. Handle forgotten
14004 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
14007 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
14008 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
14009 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
14010 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
14011 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
14012 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
14013 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
14014 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
14015 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
14016 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
14017 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
14018 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
14019 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
14020 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
14021 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
14022 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
14023 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
14024 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
14025 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
14026 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
14027 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
14028 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
14029 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
14030 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
14031 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
14032 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
14033 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
14034 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
14035 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
14036 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
14037 definitions outside of __OPTIMIZE__ guarded section.
14040 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
14041 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
14042 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
14043 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
14046 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14048 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
14049 ("vfenez<mode>"): Add missing constraints.
14051 2017-03-08 Martin Sebor <msebor@redhat.com>
14054 * config/nds32/nds32.c (nds32_option_override):
14055 Fix misspelled diagnostic.
14057 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14060 * gimplify.c (gimplify_omp_for): Replace index var in outer
14061 taskloop statement with an artificial variable and add
14062 OMP_CLAUSE_PRIVATE clause for it.
14064 2017-03-08 Richard Biener <rguenther@suse.de>
14066 PR tree-optimization/79955
14067 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
14068 for accesses that are completely outside of the variable.
14070 2017-03-08 Andrew Haley <aph@redhat.com>
14072 PR tree-optimization/79943
14073 * tree-ssa-loop-split.c (compute_new_first_bound): When
14074 calculating the new upper bound, (END-BEG) should be added, not
14077 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14079 * config/avr/avr.md (setmemhi): Make sure match_dup
14080 operand number comes before match_scratch.
14082 2017-03-08 Richard Biener <rguenther@suse.de>
14084 PR tree-optimization/79920
14085 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
14086 with ncopies == 1 to ...
14087 (vect_transform_slp_perm_load): ... here. Properly compute
14088 all element loads by iterating VF times over the group. Do
14089 not handle ncopies (computed in a broken way) in
14090 vect_create_mask_and_perm.
14092 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14095 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
14096 is a uniform vector, use uniform_vector_p return value instead of
14097 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
14099 2017-03-07 Marek Polacek <polacek@redhat.com>
14101 PR middle-end/79809
14102 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
14103 (alloca_call_type): Likewise.
14105 2017-03-07 Martin Liska <mliska@suse.cz>
14107 * gcov.c (process_args): Put comment to correct location.
14109 2017-03-07 Martin Liska <mliska@suse.cz>
14111 PR middle-end/68270
14112 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
14113 Use array_at_struct_end_p instead of DECL_CHAIN (field).
14114 (chkp_narrow_bounds_for_field): Likewise.
14115 (chkp_parse_array_and_component_ref): Pass one more argument to
14118 2017-03-07 Richard Biener <rguenther@suse.de>
14120 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
14123 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
14125 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
14126 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
14128 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14131 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
14132 to end of description.
14133 (PARAM_MAX_STORES_TO_MERGE): Likewise.
14135 2017-03-07 Jakub Jelinek <jakub@redhat.com>
14137 PR rtl-optimization/79901
14138 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
14140 (*avx512f_<code><mode>3<mask_name>): ... this.
14141 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
14142 iterator instead of VI8_AVX2_AVX512BW.
14144 PR rtl-optimization/79901
14145 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
14146 min/max expander, expand it using expand_vec_cond_expr.
14149 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
14152 2017-03-06 Jakub Jelinek <jakub@redhat.com>
14155 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
14156 to void * for PCH reasons.
14157 * dwarf2out.c (output_loc_operands, output_die): Cast
14158 v.val_vec.array to unsigned char *.
14160 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
14163 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
14166 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
14168 PR rtl-optimization/79571
14169 * lra-constraints.c (process_alt_operands): Calculate static
14170 reject and subtract it from overall when only addresses will be
14173 2017-03-06 Julia Koval <julia.koval@intel.com>
14176 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
14177 incoming stack boundary to 128 for 64-bit targets.
14179 2017-03-06 Richard Biener <rguenther@suse.de>
14181 PR tree-optimization/79894
14182 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
14183 to NULL after folding it.
14185 2017-03-06 Richard Biener <rguenther@suse.de>
14187 PR tree-optimization/79824
14188 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
14189 check disabling peeling for gaps.
14191 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
14193 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
14194 attributes): Document gettimeofday.
14196 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
14198 * config/s390/s390.c (s390_option_override_internal): Set
14199 PARAM_MIN_VECT_LOOP_BOUND
14201 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
14203 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
14204 * config/s390/s390.md: Likewise.
14206 2017-03-06 Jakub Jelinek <jakub@redhat.com>
14209 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
14210 (<avx2_avx512>_perm<mode>): Rename to ...
14211 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
14213 (<avx512>_perm<mode>_mask): Rename to ...
14214 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
14216 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
14217 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
14218 instead of VI8F_256_512.
14219 (avx512f_perm<mode>): New define_expand.
14220 (avx512f_perm<mode>_mask): Likewise.
14221 (avx512f_perm<mode>_1<mask_name>): New define_insn.
14222 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
14224 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14226 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
14227 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
14229 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
14231 2017-03-06 Martin Liska <mliska@suse.cz>
14234 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
14235 when having a SSA NAME w/o VAR_DECL assigned to it.
14237 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14239 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
14240 msa_dpsub_<su>_d): Fix MODE for vec_select.
14242 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14244 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
14246 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
14248 2017-03-06 Richard Biener <rguenther@suse.de>
14250 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
14251 * plugin.c (register_plugin_info): Likewise.
14252 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
14254 2017-03-05 Jakub Jelinek <jakub@redhat.com>
14256 * config/i386/sse.md (sse_storehps, sse_storelps,
14257 avx_<castmode><avxsizesuffix>_<castmode>,
14258 avx512f_<castmode><avxsizesuffix>_<castmode>,
14259 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
14260 in condition that at least one operand is not a MEM.
14262 2017-03-03 Jakub Jelinek <jakub@redhat.com>
14264 PR middle-end/79805
14265 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
14266 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
14268 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
14269 gimple_call_nothrow_p flag based on whether original builtin can throw.
14270 If it can, emit following stmts on the fallthrough edge.
14271 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
14272 don't create new bb if inserting just debug stmts on the edge, try to
14273 insert them on the fallthru bb or just reset debug stmts.
14275 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
14278 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
14279 restore recog_data (including the operand rtxes inside it) around
14280 the call to get_insn_template.
14282 2017-03-03 Martin Sebor <msebor@redhat.com>
14284 PR tree-optimization/79699
14285 * context.c (context::~context): Free MPFR caches to avoid
14286 a memory leak on program exit.
14288 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14290 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
14291 Use wide_int::ulow () instead of .elt (0).
14293 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
14295 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
14296 (*pushxf): Limit oF constraint to 32bit targets and add oC
14297 constraint for 64bit targets.
14298 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
14299 (*pushdf): Change rmF constraint to rmC.
14301 2017-03-03 Martin Liska <mliska@suse.cz>
14303 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
14304 Remove unused variable.
14306 2017-03-03 Jakub Jelinek <jakub@redhat.com>
14309 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
14310 is a memory operand, increase num_memory.
14311 (ix86_expand_args_builtin): Likewise.
14313 2017-03-03 Jan Hubicka <jh@suse.cz>
14316 * ipa-devirt.c (maybe_record_node): Properly handle
14317 __cxa_pure_virtual visibility.
14319 2017-03-03 Martin Liska <mliska@suse.cz>
14321 PR tree-optimization/79803
14322 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
14324 (pass_loop_prefetch::execute): Disabled optimization if an
14325 assumption about L1 cache size is not met.
14327 2017-03-03 Martin Liska <mliska@suse.cz>
14329 PR rtl-optimization/79574
14330 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
14331 (hash_scan_set): Likewise.
14332 (dump_hash_table): Likewise.
14333 (hoist_code): Likewise.
14335 2017-03-03 Richard Biener <rguenther@suse.de>
14337 * fixed-value.c (fixed_from_string): Restore use of elt (1)
14338 in place of uhigh ().
14339 (fixed_convert_from_real): Likewise.
14341 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
14344 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
14346 2017-03-03 Richard Biener <rguenther@suse.de>
14348 PR middle-end/79818
14349 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
14350 TYPE_OVERFLOW_UNDEFINED check.
14352 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14354 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
14356 (vector_ae_<mode>_p): Likewise.
14357 (vector_nez_<mode>_p): Likewise.
14358 (vector_ne_v2di_p): Likewise.
14359 (vector_ae_v2di_p): Likewise.
14360 (vector_ne_<mode>_p): Likewise.
14361 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
14363 (vsx_tsqrt<mode>2_fe): Likewise.
14365 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
14368 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
14370 2017-03-02 Jakub Jelinek <jakub@redhat.com>
14372 PR rtl-optimization/79780
14373 * cprop.c (one_cprop_pass): When second and further conditional trap
14374 in a single basic block is turned into an unconditional trap, turn it
14375 into a deleted note to avoid RTL verification failures.
14377 2017-03-02 Richard Biener <rguenther@suse.de>
14379 * fold-const.c (const_binop): Use ulow () instead of elt (0).
14381 2017-03-02 Richard Biener <rguenther@suse.de>
14383 PR tree-optimization/79345
14385 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
14386 param and abort the walk, returning -1 if it is hit.
14387 (walk_aliased_vdefs): Take a limit param and pass it on.
14388 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
14389 defaulting to 0 and return a signed int.
14390 * tree-ssa-uninit.c (struct check_defs_data): New struct.
14391 (check_defs): New helper.
14392 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
14393 about uninitialized memory.
14394 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
14395 bogus uninitialized warning.
14396 (fixed_convert_from_real): Likewise.
14398 2017-03-02 Bin Cheng <bin.cheng@arm.com>
14400 PR tree-optimization/66768
14401 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
14402 iv_use if base object can't be determined.
14404 2017-03-02 Jakub Jelinek <jakub@redhat.com>
14406 PR tree-optimization/79345
14407 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
14408 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
14409 (get_pattern_stats): Initialize it.
14410 * genemit.c (gen_expand): Verify match_scratch numbers come after
14411 match_operand/match_dup numbers.
14412 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
14413 match_scratch numbers.
14414 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
14416 * config/s390/s390.md (trunctdsd2): Likewise.
14418 2017-03-02 Richard Biener <rguenther@suse.de>
14420 * wide-int.h (wide_int_storage::operator=): Implement in terms
14423 2017-03-02 Richard Biener <rguenther@suse.de>
14425 PR tree-optimization/79777
14426 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
14427 the to insert expression to sth existing.
14429 2017-03-01 Martin Sebor <msebor@redhat.com>
14431 PR middle-end/79692
14432 * gimple-ssa-sprintf.c
14433 (directive::known_width_and_precision): New function.
14434 (format_integer): Use it.
14435 (get_mpfr_format_length): Consider the full range of precision
14436 when computing %g output with the # flag. Set the likely byte
14437 count to 3 rather than 1 when precision is indeterminate.
14438 (format_floating): Correct the lower bound of precision.
14440 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14442 * doc/invoke.texi: Document default code model for 64-bit Linux.
14444 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14447 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
14448 udiv rather than div since input pattern is unsigned.
14450 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
14452 * config/i386/i386.c (print_reg): Warn for values of
14453 unsupported size in integer register.
14455 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
14458 * config/rs6000/predicates.md (current_file_function_operand): Do
14459 not allow self calls to be local if the function is replaceable.
14461 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
14464 * config/rs6000/altivec.h (vec_ctz and others): Change the
14465 preprocessor macro that controls conditional compilation from
14466 _ARCH_PWR9 to __POWER9_VECTOR__.
14467 (vec_all_ne): Change parameterization of __altivec_scalar_pred
14468 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
14469 control (instead of _ARCH_PWR9 control) so that template
14470 definition uses power9-specific function.
14471 (vec_any_eq): Likewise.
14472 (vec_all_ne): Change macro definition to use a power9-specific
14473 expansion under #ifdef __POWER9_VECTOR__ control (instead of
14474 _ARCH_PWR9 control).
14475 (vec_any_eq) Likewise.
14476 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
14477 expansion for CMPNEF to remove support for xvcmpnesp instruction.
14478 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
14479 support for xvcmpnedp instruction.
14480 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
14481 macro expansion so that Power9 implementation of vec_all_ne does
14482 not use the AltiVec predicate framework.
14483 (VCMPNEH_P): Likewise.
14484 (VCMPNEW_P): Likewise.
14485 (VCMPNED_P): Likewise.
14486 (VCMPNEFP_P): Likewise.
14487 (VCMPNEDP_P): Likewise.
14488 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
14489 implementation of vec_any_eq to not use AltiVec predicate
14491 (VCMPAEH_P): Likewise.
14492 (VCMPAEW_P): Likewise.
14493 (VCMPAED_P): Likewise.
14494 (VCMPAEFP_P): Likewise.
14495 (VCMPAEDP_P): Likewise.
14496 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
14497 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
14498 not use the AltiVec predicate framework.
14499 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
14500 of vec_any_eq to not use AltiVec predicate framework.
14501 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
14502 support for predefined __POWER9_VECTOR__ macro to indicate that
14503 Power9 instruction selection is enabled.
14504 (altivec_overloaded_builtins): Remove extraneous
14505 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
14506 function argument types RS6000_BTI_bool_V16QI and
14507 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
14508 entry for overloaded function argument types RS6000_BTI_bool_V4SI
14509 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
14510 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
14511 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
14512 Power9 for implementations of vec_cmpne. Change the signature for
14513 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
14514 (representing vec_all_ne) to remove the previously described first
14515 argument of type RS6000_BTI_INTSI, as this was an artifact of
14516 reliance on the AltiVec predicate framework, which is no longer
14517 used in the implementation of these functions. Add
14518 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
14519 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
14520 since, unlike the AltiVec predicate framework implementation, we
14521 do not share function descriptors between vec_alle and vec_anyeq.
14522 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
14523 set of modes that receive special treatment even when
14524 TARGET_P9_VECTOR is true. The special treatment emits code that
14525 does not depend on Power9 instructions.
14526 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
14527 define_expand to not rely on AltiVec predicate framework.
14528 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
14530 (vector_ne_v2di_p): Change this define_expand to not rely on
14531 AltiVec predicate framework.
14532 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
14534 (vector_ne_<mode>_p): Change this define_expand to not rely on
14535 AltiVec predicate framework.
14536 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
14538 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
14539 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
14540 define_insn pattern.
14541 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
14542 define_insn pattern because the xvcmpne<VSs>. instruction is not
14544 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
14545 instruction is not supported.
14547 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14549 * config/nvptx/nvptx.c: Include intl.h.
14551 2017-03-01 Martin Jambor <mjambor@suse.cz>
14554 * ipa-prop.h (ipa_bits): Removed field known.
14555 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
14556 to pointers. Adjusted their comments to warn about their sharing.
14557 (ipcp_transformation_summary): Change bits to a vector of pointers.
14558 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
14559 (ipa_get_ipa_bits_for_value): Declare.
14560 * tree-vrp.h (value_range): Mark as GTY((for_user)).
14561 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
14562 (ipa_bits_hash_table): Likewise.
14563 (ipa_vr_ggc_hash_traits): Likewise.
14564 (ipa_vr_hash_table): Likewise.
14565 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
14566 being pointers and vr_known being removed.
14567 (ipa_set_jf_unknown): Likewise.
14568 (ipa_get_ipa_bits_for_value): New function.
14569 (ipa_set_jfunc_bits): Likewise.
14570 (ipa_get_value_range): New overloaded functions.
14571 (ipa_set_jfunc_vr): Likewise.
14572 (ipa_compute_jump_functions_for_edge): Use the above functions to
14573 construct bits and vr parts of jump functions.
14574 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
14575 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14577 (ipcp_grow_transformations_if_necessary): Also allocate
14578 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14580 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
14581 them. Fix too long lines.
14582 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
14583 vr_known being removed.
14584 (ipa_read_jump_function): Use new setter functions to construct bits
14585 and vr parts of jump functions or set them to NULL.
14586 (write_ipcp_transformation_info): Adjust for bits being pointers.
14587 (read_ipcp_transformation_info): Likewise.
14588 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
14590 Include gt-ipa-prop.h.
14591 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
14593 (ipcp_store_bits_results): Likewise.
14594 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
14595 Do not write to existing jump functions but use a temporary instead.
14597 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14600 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
14601 attempt to use its first operand as BIT_FIELD_REF base.
14603 2017-03-01 Richard Biener <rguenther@suse.de>
14605 PR middle-end/79721
14606 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
14607 interpolating formula in wrapping arithmetic.
14608 (chrec_apply): Convert chrec_evaluate return value to wanted type.
14610 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14612 PR tree-optimization/79734
14613 * tree-vect-generic.c (expand_vector_condition): Optimize
14614 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
14615 Handle VEC_COND_EXPR where comparison has different inner width from
14616 type's inner width.
14618 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
14620 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
14621 markup, and similar issues. Remove @opindex entries for things
14622 that aren't options. Add missing -mmpy-option entries.
14624 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14626 PR tree-optimization/79737
14627 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
14628 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
14629 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
14630 instead of byte_size. Formatting fix.
14631 (shift_bytes_in_array_right): Formatting fix.
14633 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
14636 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
14637 condition on optimize for the leaf function test.
14639 2017-02-28 Martin Liska <mliska@suse.cz>
14642 * read-rtl-function.c (function_reader::handle_unknown_directive):
14643 Bail out when one uses -flto.
14645 2017-02-28 Martin Liska <mliska@suse.cz>
14647 * common.opt: Replace space with tabular for options of <number>
14649 * config/i386/i386.opt: Show <number> value for
14650 -mlarge-data-threshold.
14651 * opts.c (print_filtered_help): Do not display number in hexadecimal
14654 2017-02-28 Martin Liska <mliska@suse.cz>
14656 * common.opt: Fix --help=option -Q for options which are of
14659 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
14661 * config/i386/i386.c (print_reg): Error out for values
14662 of 8-bit size in invalid integer register.
14664 2017-02-28 Martin Sebor <msebor@redhat.com>
14666 PR tree-optimization/79691
14667 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
14669 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14672 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
14673 gcc_unreachable with output_operand_lossage.
14675 2017-02-28 Richard Biener <rguenther@suse.de>
14677 PR tree-optimization/79740
14678 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
14680 (visit_nary_op): Insert the nary into the hashtable if we
14681 pattern-matched sth.
14682 * tree-ssa-pre.c (eliminate_insert): Robustify.
14684 2017-02-28 Richard Biener <rguenther@suse.de>
14686 PR middle-end/79731
14687 * fold-const.c (decode_field_reference): Reject out-of-bound
14690 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14692 * config/i386/i386.c: Include intl.h.
14693 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
14694 instead of just cond ? "..." : "...".
14695 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
14696 * coverage.c (read_counts_file): Likewise.
14697 * omp-offload.c: Include intl.h.
14698 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
14699 of just cond ? "..." : "...".
14700 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
14701 of just cond ? "..." : "...".
14703 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
14706 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
14708 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
14709 'tune for' CPU name.
14710 * config/arm/arm-cpu-data.h: Regenerated.
14712 2017-02-28 Richard Biener <rguenther@suse.de>
14714 PR tree-optimization/79732
14715 * tree-inline.c (expand_call_inline): Do not shadow var.
14717 2017-02-28 Richard Biener <rguenther@suse.de>
14719 PR tree-optimization/79723
14720 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
14721 address-space properly.
14723 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
14725 * doc/optinfo.texi (Optimization groups): Fix option used for
14727 * doc/invoke.texi (-fopt-info): Document "omp".
14728 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
14729 (OPTGROUP_ALL): Add OPTGROUP_OMP.
14730 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
14731 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
14732 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
14734 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
14736 * dumpfile.c (optgroup_options): Instead of "openmp", associate
14737 OPTGROUP_OMP with "omp".
14739 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
14742 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
14743 for arithmetic shift of unsigned V2DI.
14745 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
14747 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
14748 arc/linux.h headers.
14749 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
14750 (LINK_SPEC): Likewise.
14751 (ARC_TLS_EXTRA_START_SPEC): Likewise.
14752 (EXTRA_SPECS): Likewise.
14753 (STARTFILE_SPEC): Likewise.
14754 (ENDFILE_SPEC): Likewise.
14755 (LIB_SPEC): Likewise.
14756 (TARGET_SDATA_DEFAULT): Likewise.
14757 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
14758 (MULTILIB_DEFAULTS): Likewise.
14759 (DWARF2_UNWIND_INFO): Likewise.
14760 * config/arc/big.h: New file.
14761 * config/arc/elf.h: Likewise.
14762 * config/arc/linux.h: Likewise.
14763 * config/arc/t-uClibc: Remove.
14765 2017-02-27 Bin Cheng <bin.cheng@arm.com>
14767 PR tree-optimization/77536
14768 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
14769 (tree_transform_and_unroll_loop): Use above function to compute the
14770 estimated niter of unrolled loop and use it when scaling profile.
14771 Also use count info rather than frequency if it's non-zero.
14772 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
14773 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
14774 (vect_transform_loop): Call above function.
14776 2017-02-27 Richard Biener <rguenther@suse.de>
14778 PR tree-optimization/45397
14779 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
14780 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
14781 (visit_nary_op): Add pattern matching for CSEing sign-changed
14782 or truncated operations with wider ones.
14784 2017-02-27 Richard Biener <rguenther@suse.de>
14786 PR tree-optimization/79690
14787 * tree-vect-stmts.c (vectorizable_store): Use vector type
14788 built from the DR with address-space.
14790 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
14792 * doc/invoke.texi (Optimize Options): Refine the description
14793 of asan-use-after-return.
14795 2017-02-25 Alan Modra <amodra@gmail.com>
14797 PR rtl-optimization/79584
14798 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
14799 base, not ad->base_term, the reg within base. Remove assertion
14800 that ad->base == ad->base_term. Replace gen_int_mode using
14801 bogus mode with const0_rtx.
14803 2017-02-25 Jakub Jelinek <jakub@redhat.com>
14805 PR middle-end/79396
14806 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
14807 FMA_EXPR like tcc_binary or tcc_unary.
14809 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
14812 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
14814 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
14815 (output_loc_operands): Handle DW_OP_call_ref and
14816 DW_OP_GNU_variable_value.
14817 (struct variable_value_struct): New type.
14818 (struct variable_value_hasher): Likewise.
14819 (variable_value_hash): New variable.
14820 (string_types): Remove.
14821 (copy_loc_descr): New function.
14822 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
14823 (prepend_loc_descr_to_each): New function.
14824 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
14825 instead of add_loc_descr_to_each if the first argument is single
14826 location list and the second has multiple.
14827 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
14828 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
14829 when looking for variable value which doesn't have other location info.
14830 (loc_list_from_tree): Formatting fix.
14831 (gen_array_type_die): Simplify DW_AT_string_length handling.
14832 (adjust_string_types): Remove.
14833 (gen_subprogram_die): Don't call adjust_string_types nor test/set
14834 string_types. Call resolve_variable_values.
14835 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
14836 (resolve_addr_in_expr): Likewise. Add A argument.
14837 (copy_deref_exprloc): Remove deref argument. Adjust for the
14838 original expression being DW_OP_GNU_variable_value with optionally
14839 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
14840 optionally after it.
14841 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
14842 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
14843 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
14844 (variable_value_hasher::hash, variable_value_hasher::equal): New
14846 (resolve_variable_value_in_expr, resolve_variable_value,
14847 resolve_variable_values, note_variable_value_in_expr,
14848 note_variable_value): New functions.
14849 (dwarf2out_early_finish): Call note_variable_value on all toplevel
14852 2017-02-24 Jakub Jelinek <jakub@redhat.com>
14855 * opts.h (handle_generated_option): Add GENERATED_P argument.
14856 * opts-common.c (handle_option): Adjust function comment.
14857 (handle_generated_option): Add GENERATED_P argument, pass it to
14859 (control_warning_option): Pass false to handle_generated_option
14861 * opts.c (maybe_default_option): Pass true to handle_generated_option
14863 * optc-gen.awk: Likewise.
14865 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14867 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
14868 a REG, look at the REG it is a SUBREG of.
14869 (splitter for cmpeqsi_t): Ditto.
14871 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14873 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
14874 the special USEs with the pattern of the insn, not the insn itself.
14876 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
14879 * doc/invoke.texi: Document -mload-store-pairs.
14881 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14882 Sandra Loosemore <sandra@codesourcery.com>
14884 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
14885 argument isn't a CONST_INT.
14886 (nios2_alternate_compare_const): Assert op is a CONST_INT.
14887 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
14888 (nios2_validate_compare): Bypass alternate compare logic if *op2
14889 is not a CONST_INT.
14890 (ldstwm_operation_p): Return false if first_base is not a REG or
14891 if first_offset is not a CONST_INT.
14893 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14895 * config/cris/cris.md: Use correct operand in a define_peephole2.
14897 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14899 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
14901 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14903 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
14904 this_insn if it is an INSN or JUMP_INSN.
14905 (force_offsettable): Look at base, not at addr.
14906 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
14907 on things that aren't necessarily CONST_INTs.
14909 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
14911 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
14912 -mfpmath=sse is the default also for x86-32 targets with SSE2
14913 instruction set when @option{-ffast-math} is enabled
14915 2017-02-24 Jeff Law <law@redhat.com>
14917 PR rtl-optimizatoin/79286
14918 * ira.c (update_equiv_regs): Drop may_trap_p exception to
14921 2017-02-24 Richard Biener <rguenther@suse.de>
14923 PR tree-optimization/79389
14924 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
14927 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
14929 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
14930 function comment to reflect reality.
14931 (loop_exits_before_overflow): Fix typo in function description.
14933 2017-02-24 Richard Biener <rguenther@suse.de>
14935 PR tree-optimization/79389
14936 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
14937 properly that a threading opportunity exists. Detect conditional
14938 copy/constant propagation opportunities.
14940 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
14942 * config/visium/visium.md (type): Add trap.
14943 (b): New mode attribute.
14944 (*btst): Rename into...
14945 (*btst<mode>): ...this and adjust.
14946 (*cbranchsi4_btst_insn): Rename into...
14947 (*cbranch<mode>4_btst_insn): ...this and adjust.
14948 (trap): New define_insn.
14950 2017-02-23 Jakub Jelinek <jakub@redhat.com>
14952 PR tree-optimization/79389
14953 * ifcvt.c (struct noce_if_info): Add rev_cond field.
14954 (noce_reversed_cond_code): New function.
14955 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
14956 reversed_comparison_code. Formatting fix.
14957 (noce_try_store_flag): Test rev_cond != NULL in addition to
14958 reversed_comparison_code.
14959 (noce_try_store_flag_constants): Likewise.
14960 (noce_try_store_flag_mask): Likewise.
14961 (noce_try_addcc): Use rev_cond if non-NULL instead of
14962 reversed_comparison_code.
14963 (noce_try_cmove_arith): Likewise. Formatting fixes.
14964 (noce_try_minmax, noce_try_abs): Clear rev_cond.
14965 (noce_find_if_block): Initialize rev_cond.
14966 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
14967 instead of false as last argument never attempt to reverse it
14970 2017-02-23 Bin Cheng <bin.cheng@arm.com>
14972 PR tree-optimization/79663
14973 * tree-predcom.c (combine_chains): Process refs in reverse order
14974 only for ZERO length chains, and add explaining comment.
14976 2017-02-23 Jeff Law <law@redhat.com>
14978 PR tree-optimization/79578
14979 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
14980 in call to operand_equal_p.
14982 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
14985 * config/i386/cpuid.h: Fix another undefined behavior.
14987 2017-02-23 Richard Biener <rguenther@suse.de>
14989 PR tree-optimization/79683
14990 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
14991 vector types for data-refs.
14993 2017-02-23 Martin Liska <mliska@suse.cz>
14995 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
14997 2017-02-23 Jakub Jelinek <jakub@redhat.com>
14999 PR middle-end/79665
15000 * internal-fn.c (get_range_pos_neg): Moved to ...
15001 * tree.c (get_range_pos_neg): ... here. No longer static.
15002 * tree.h (get_range_pos_neg): New prototype.
15003 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
15004 are known to be in between 0 and signed maximum inclusive, try to
15005 expand both unsigned and signed divmod and use the cheaper one from
15008 2017-02-22 Jeff Law <law@redhat.com>
15010 PR tree-optimization/79578
15011 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
15012 to compare base operands.
15014 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
15017 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
15018 gpc_reg_operand instead of fpr_reg_operand.
15020 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
15022 * config/mips/mips.c (mips_return_in_memory): Force FP
15023 vector types to be returned in memory for o32 ABI.
15025 2017-02-22 Jakub Jelinek <jakub@redhat.com>
15027 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
15028 instead of DW_TAG_member for static data member declarations and don't
15029 set no_linkage_name for static inline data members.
15030 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
15033 2017-02-22 Martin Liska <mliska@suse.cz>
15035 * doc/invoke.texi: Replace inequality signs with square brackets
15038 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15040 PR tree-optimization/68644
15041 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
15043 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
15046 * lra-constraints.c (simplify_operand_subreg): Handle
15047 WORD_REGISTER_OPERATIONS targets.
15049 2017-02-22 Jakub Jelinek <jakub@redhat.com>
15052 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
15053 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
15054 elimination by swapping fld*.
15056 2017-02-22 Richard Biener <rguenther@suse.de>
15058 PR tree-optimization/79673
15059 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
15060 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
15061 irrelevant address-space qualifiers and avoiding a
15062 ADDR_SPACE_CONVERT_EXPR from fold_convert.
15064 2017-02-22 Richard Biener <rguenther@suse.de>
15066 PR tree-optimization/79666
15067 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
15068 to not symbolically negate if that may introduce undefined
15071 2017-02-22 Martin Liska <mliska@suse.cz>
15074 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
15075 * data-streamer-out.c (streamer_write_gcov_count_stream):
15077 * value-prof.c (stream_out_histogram_value): Make assert more
15078 precise based on type of counter.
15080 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
15083 * config/i386/i386.md (standard_x87sse_constant_load splitter):
15084 Use nonimmediate_operand instead of memory_operand for operand 1.
15085 (float-extend standard_x87sse_constant_load splitter): Ditto.
15087 2017-02-21 Jeff Law <law@redhat.com>
15089 PR tree-optimization/79621
15090 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
15091 blocks with edges to themselves.
15093 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15096 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
15097 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
15098 Use gimple_call_builtin_p.
15101 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
15102 on temporarily removed DEBUG_INSNs.
15104 PR tree-optimization/79649
15105 * tree-loop-distribution.c (classify_partition): Give up on
15106 non-generic address space loads/stores.
15108 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
15110 * doc/loop.texi (Loop manipulation): Remove nonexistent
15111 tree_ssa_loop_version from the documentation.
15112 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
15114 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15117 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
15118 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
15119 * config/rs6000/rs6000.c: Include except.h.
15120 (rs6000_expand_split_stack_prologue): Call
15121 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
15123 2017-02-21 Martin Jambor <mjambor@suse.cz>
15126 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
15127 have been analyzed.
15129 2017-02-21 Martin Jambor <mjambor@suse.cz>
15131 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
15132 for backward compatibility only.
15133 * doc/invoke.texi (Option Summary): Remove all references to
15134 -fipa-cp-alignment.
15136 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
15140 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15142 * lra-constraints.c (curr_insn_transform): Handle
15143 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
15145 2017-02-21 Martin Liska <mliska@suse.cz>
15147 * config/i386/i386.opt: Replace -masm-dialect with -masm.
15149 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
15151 PR translation/79638
15152 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
15154 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
15157 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
15158 (arm_function_ok_for_sibcall): Return false for an indirect call by
15159 descriptor if all the argument registers are used.
15160 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
15161 alignment of the function.
15163 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15165 PR tree-optimization/61441
15166 * simplify-rtx.c (simplify_const_unary_operation): For
15167 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
15168 the sNaN unmodified.
15170 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
15172 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
15173 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
15174 instead of SYSTEM_HEADER_DIR.
15176 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
15177 Martin Liška <mliska@suse.cz>
15179 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
15180 Fix typos and grammar, use active voice, and clarify.
15182 2017-02-20 Marek Polacek <polacek@redhat.com>
15184 PR middle-end/79537
15185 * gimplify.c (gimplify_expr): Handle unused *&&L;.
15188 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
15190 2017-02-20 Jakub Jelinek <jakub@redhat.com>
15193 * config/i386/i386.c (ix86_expand_builtin): Handle
15194 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
15195 ix86_builtins_isa[fcode].isa as a requirement of those
15196 flags and any other flag in the bitmask.
15197 (ix86_init_mmx_sse_builtins): Use 0 instead of
15198 ~OPTION_MASK_ISA_64BIT as mask.
15199 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
15200 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
15201 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
15202 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
15204 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15207 * lra-constraints.c (split_reg): Check requested split mode
15208 is supported by the register.
15210 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15212 * lra-constraints.c (simplify_operand_subreg): Remove early
15215 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15218 * lra-constraints.c (curr_insn_transform): Tighten condition
15219 for converting SUBREG reloads from OP_OUT to OP_INOUT.
15221 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15224 * lra-constraints.c (curr_insn_transform): Handle
15225 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
15227 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
15230 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
15232 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
15234 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
15237 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
15240 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15242 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
15245 2017-02-18 Jakub Jelinek <jakub@redhat.com>
15247 * final.c (last_columnnum, override_columnnum): New variables.
15248 (final_start_function): Set last_columnnum, pass it to begin_prologue
15249 hook and pass 0 to dwarf2out_begin_prologue.
15250 (final_scan_insn): Update override_columnnum. Pass last_columnnum
15251 to source_line debug hook.
15252 (notice_source_line): Compute last_columnnum and for debug_column_info
15253 return true on column changes.
15254 * debug.h (struct gcc_debug_hooks): Add column argument to
15255 source_line and begin_prologue hooks.
15256 (debug_nothing_int_charstar_int_bool): Remove prototype.
15257 (debug_nothing_int_int_charstar,
15258 debug_nothing_int_int_charstar_int_bool): New prototypes.
15259 (dwarf2out_begin_prologue): Add column argument.
15260 * debug.c (do_nothing_debug_hooks): Adjust source_line and
15261 begin_prologue hooks.
15262 (debug_nothing_int_charstar_int_bool): Remove.
15263 (debug_nothing_int_int_charstar,
15264 debug_nothing_int_int_charstar_int_bool): New functions.
15265 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
15266 through to dwarf2out_source_line.
15267 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
15268 (dwarf2out_source_line): Add column argument, emit it if requested.
15269 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
15271 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
15272 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
15273 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
15274 through to dwarf2out_begin_prologue.
15275 (vmsdbgout_source_line): Add column argument, pass it through to
15276 dwarf2out_source_line.
15277 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
15278 dbxout_source_line caller.
15279 (dbxout_source_line): Add column argument.
15281 * common.opt (gno-column-info, gcolumn-info): New options.
15282 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
15283 (check_die): Also test for multiple DW_AT_decl_column attributes.
15284 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
15285 DW_AT_decl_column if requested.
15286 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
15288 (gen_variable_die): Likewise.
15289 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
15290 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
15293 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
15294 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
15295 (ix86_handle_option): Handle OPT_m3dnowa.
15296 * doc/invoke.texi (-m3dnowa): Document.
15297 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
15298 -m3dnowa instead of -m3dnow -march=athlon.
15301 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
15302 instead of gcc_assert for K, r and R code checks. Formatting fixes.
15304 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15307 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
15308 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
15309 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
15310 generator for vsx_xxpermdi_<mode>_be.
15311 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
15312 force big-endian semantics.
15313 (vsx_xxpermdi_<mode>_be): New define_expand with same
15314 implementation as previous version of vsx_xxpermdi_<mode>.
15316 2017-02-17 Jakub Jelinek <jakub@redhat.com>
15318 PR tree-optimization/79327
15319 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
15320 variable, its initialization and use.
15322 2017-02-17 Julia Koval <julia.koval@intel.com>
15324 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
15325 (OPTION_MASK_ISA_PKU_UNSET): New.
15326 (ix86_handle_option): Handle -mrdpid.
15327 * config/i386/cpuid.h (bit_RDPID): New.
15328 * config/i386/driver-i386.c (host_detect_local_cpu):
15329 Detect RDPID feature.
15330 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
15331 * config/i386/i386-c.c (ix86_target_macros_internal):
15333 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
15334 (ix86_valid_target_attribute_inner_p): Add "rdpid".
15335 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
15336 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
15337 * config/i386/i386.md (define_insn "rdpid"): New.
15338 * config/i386/i386.opt Add -mrdpid.
15339 * config/i386/immintrin.h (_rdpid_u32): New.
15341 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
15343 PR rtl-optimization/79541
15344 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
15345 instead of transforming it into USE.
15347 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
15349 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
15350 If HONOR_SNANS (SFmode) force the input to a register.
15351 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
15352 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
15353 an frsp or similar insn.
15355 2017-02-17 Martin Liska <mliska@suse.cz>
15357 PR rtl-optimization/79577
15358 * params.def (selsched-max-sched-times): Increase minimum to 1.
15360 2017-02-17 Martin Liska <mliska@suse.cz>
15362 PR rtl-optimization/79574
15363 * gcse.c (want_to_gcse_p): Prevent integer overflow.
15365 2017-02-17 Martin Liska <mliska@suse.cz>
15367 PR tree-optimization/79529
15368 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
15369 ssa_defined_default_def_p to handle cases which are implicitly
15371 * tree-ssa.c (ssa_defined_default_def_p): New function.
15372 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
15373 which are implicitly defined.
15374 * tree-ssa.h (ssa_defined_default_def_p): Declare.
15376 2017-02-17 Richard Biener <rguenther@suse.de>
15378 PR middle-end/79576
15379 * params.def (max-ssa-name-query-depth): Limit to 10.
15381 2017-02-17 Richard Biener <rguenther@suse.de>
15383 PR tree-optimization/79552
15384 * tree-ssa-structalias.c (visit_loadstore): Properly verify
15387 2017-02-17 Richard Biener <rguenther@suse.de>
15390 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
15392 2017-02-17 Marek Polacek <polacek@redhat.com>
15394 PR middle-end/79536
15395 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
15396 (fold_negate_expr): New wrapper.
15398 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
15400 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
15401 Correct terminology and de-emphasize pre-standard behavior.
15403 2017-02-16 Alan Modra <amodra@gmail.com>
15405 PR rtl-optimization/79286
15406 * ira.c (def_dominates_uses): New function.
15407 (update_equiv_regs): Don't create an equivalence for insns that
15408 may trap where the register def does not dominate the use.
15410 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
15412 PR rtl-optimization/78127
15413 * lra.c (lra): Call lra_eliminate before finish the loop after
15416 2017-02-16 Richard Biener <rguenther@suse.de>
15418 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
15420 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
15421 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
15422 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
15423 (isl_val_int_from_wi): New function.
15424 (extract_affine_gmp): Rename to ...
15425 (extract_affine_wi): ... this, take a widest_int.
15426 (extract_affine_int): Just wrap extract_affine_wi.
15427 (add_param_constraints): Use isl_val_int_from_wi.
15428 (add_loop_constraints): Likewise, and extract_affine_wi.
15430 2017-02-15 Jeff Law <law@redhat.com>
15432 PR middle-end/79521
15433 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
15434 ira_init_register_move_cost_if_necessary.
15436 2017-02-15 Martin Sebor <msebor@redhat.com>
15438 PR middle-end/32003
15439 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
15440 removed in a prior commit.
15442 2017-02-15 Bin Cheng <bin.cheng@arm.com>
15444 PR tree-optimization/79347
15445 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
15446 counters during peeling.
15448 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
15450 * Makefile.in (site.exp): Remove "set ISLVER".
15452 2017-02-15 Jakub Jelinek <jakub@redhat.com>
15455 * real.c (real_from_integer): Call real_convert even for decimal.
15457 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
15460 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
15462 2017-02-14 Andrew Pinski <apinski@cavium.com>
15464 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
15465 cores and change the partno/implementer to be correct.
15466 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
15467 the 'B" as the implementer.
15468 * config/aarch64/aarch64-tune.md: Regenerate.
15470 2017-02-14 Carl Love <cel@us.ibm.com>
15472 * config/rs6000/rs6000.c: Add case statement entry to make the
15473 xvcvuxdsp built-in argument unsigned.
15474 * config/rs6000/vsx.md: Fix the source and return operand types so they
15475 match the instruction definitions from the ISA document. Fix typo
15476 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
15479 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
15482 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
15483 member early_clobber_alts.
15484 * lra-lives.c (reg_early_clobber_p): New.
15485 (process_bb_lives): Use it.
15486 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
15487 (debug_operand_data): Initialize early_clobber_alts.
15488 (setup_operand_alternative): Set up early_clobber_alts.
15489 (collect_non_operand_hard_regs): Ditto. Pass early clobber
15490 alternatives to new_insn_reg.
15491 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
15493 (lra_update_insn_regno_info): Pass the new arg.
15495 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15497 PR middle-end/79505
15498 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
15499 (new_oacc_loop_raw): Don't clear already cleared fields.
15502 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
15503 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
15504 _mm512_prefetch_i64gather_ps): New inline functions and macros.
15506 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
15509 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
15511 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
15514 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
15515 the extra instruction to the right place to store 128-bit constant
15518 2017-02-14 Martin Sebor <msebor@redhat.com>
15520 PR middle-end/79448
15521 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
15522 warning for strings of unknown length.
15524 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
15526 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
15528 2017-02-14 Jeff Law <law@redhat.com>
15531 * ira-costs.c (scan_one_insn): Initialize register move costs
15532 for pseudos seen in USE/CLOBBER insns.
15534 PR tree-optimization/79095
15535 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
15536 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
15537 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
15538 if the operands are known to be not equal, then the resulting range
15540 (intersect_ranges): If the new range is ~[0,0] and the old range is
15541 wide, then prefer ~[0,0].
15542 * tree-vrp.c (overflow_comparison_p_1): New function.
15543 (overflow_comparison_p): New function.
15544 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
15545 if NAME is used in an overflow test.
15546 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
15547 overflow check that can be expressed as an equality test, then adjust
15548 ops to be that equality test.
15550 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15552 * config/s390/s390-builtin-types.def: Remove flags argument.
15553 * config/s390/s390.c (s390_init_builtins): Likewise.
15555 2017-02-14 Martin Liska <mliska@suse.cz>
15557 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
15558 vector. Fix trailing white spaces.
15560 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
15562 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
15565 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15567 PR rtl-optimization/68664
15568 * config/arm/arm.c (arm_sched_can_speculate_insn):
15569 New function. Declare prototype.
15570 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15572 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15574 PR rtl-optimization/68664
15575 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
15577 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15579 2017-02-14 Amit Pawar <amit.pawar@amd.com>
15581 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
15582 max skip bytes for function, loop and jump.
15584 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15586 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
15587 ABS_EXPR for gimple dump.
15589 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15592 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
15594 PR tree-optimization/79408
15595 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
15596 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
15597 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
15598 also if rhs1 is INTEGER_CST.
15600 2017-02-14 Richard Biener <rguenther@suse.de>
15602 PR middle-end/79432
15603 * tree-into-ssa.c (insert_phi_nodes): When the function can
15604 have abnormal edges rewrite SSA names with broken use-def
15605 dominance out of SSA and register them for PHI insertion.
15607 2017-02-13 Martin Sebor <msebor@redhat.com>
15609 PR middle-end/79496
15610 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
15611 clearing info.nowrite flag when snprintf size argument is a range.
15613 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15615 * cprop.c (cprop_jump): Add missing space in string literal.
15616 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
15617 (get_constraint_for_component_ref): Likewise.
15618 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
15619 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
15620 * lra-constraints.c (process_alt_operands): Likewise.
15621 * ipa-inline.c (inline_small_functions): Likewise.
15622 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
15623 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15624 * trans-mem.c (diagnose_tm_1_op): Likewise.
15625 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
15626 (grid_parallel_clauses_gridifiable): Likewise.
15628 * config/nvptx/mkoffload.c (process): Add space in between
15631 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
15632 "MOD4_SSE_REGS" and "ALL_REGS".
15634 * spellcheck.c (test_data): Add , in between "foo" and "food".
15636 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15639 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
15640 boundary crossing check and subsequent code generation agree.
15642 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15644 * config/aarch64/aarch64.c (has_memory_op): Delete.
15645 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
15648 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15650 PR rtl-optimization/79388
15651 PR rtl-optimization/79450
15652 * combine.c (distribute_notes): When removing TEM_INSN for which
15653 corresponding dest has last value recorded, invalidate that last
15656 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15658 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
15659 of explicit '@'. Add missing assembly comment marker on branch costs
15662 2017-02-13 Nathan Sidwell <nathan@acm.org>
15664 * gengtype-lex.l (<in_struct>): Add '/'.
15666 2017-02-13 Martin Liska <mliska@suse.cz>
15669 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
15671 2017-02-13 Richard Biener <rguenther@suse.de>
15673 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
15675 * configure: Re-generate.
15676 * config.in: Likewise.
15677 * graphite-dependences.c: Simplify as if
15678 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
15679 * graphite-isl-ast-to-gimple.c: Likewise.
15680 * graphite-optimize-isl.c: Likewise.
15681 * graphite-poly.c: Likewise.
15682 * graphite-sese-to-poly.c: Likewise.
15683 * graphite.h: Likewise.
15684 * toplev.c: Include isl/version.h and use isl_version () for
15685 printing the ISL version.
15686 * doc/install.texi: Update ISL requirement.
15688 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15690 * doc/standards.texi (Standards): Update reference to
15693 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15695 * doc/extend.texi (Named Address Spaces): sourceware.org now
15697 * doc/install.texi (Binaries): Ditto.
15700 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15702 * doc/cpp.texi: Replace "stringify"/"stringification" with C
15703 standard terminology "stringize"/"stringizing" throughout.
15704 * doc/cppinternals.texi: Likewise.
15706 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15708 * doc/extend.texi: Fix some spelling mistakes and typos.
15709 * doc/invoke.texi: Likewise.
15711 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15714 * params.def (inline-min-speedup) Change from 10 to 8.
15716 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15718 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
15721 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15724 * ipa-inline-analysis.c (get_minimal_bb): New function.
15725 (record_modified): Use it.
15726 (remap_edge_change_prob): Handle also ancestor functions.
15728 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
15730 * doc/contrib.texi (Contributors): Remove broken link into
15731 the Mauve CVS repository.
15733 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15735 PR middle-end/79454
15736 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
15737 result computation whenever lhs doesn't have vector mode, not
15738 just when it has BLKmode.
15740 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15742 * doc/makefile.texi (profiledbootstrap): Refer to the
15743 installation instructions only in textual form.
15745 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15748 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
15750 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15752 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
15753 (Specific): Update mingw-w64 reference.
15755 (Specific): Remove broken link to Renesas RX processor.
15757 2017-02-10 Richard Biener <rguenther@suse.de>
15759 * toplev.c (process_options): Do not mention obsolete graphite
15760 options when printing sorry message about missing graphite support.
15761 Mention -floop-nest-optimize.
15763 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
15765 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
15766 (vtst_p16): Likewise.
15767 (vtstq_p8): Likewise.
15768 (vtstq_p16): Likewise.
15770 (vtstq_p64): Likewise.
15771 * config/arm/arm_neon.h (vgetq_lane_p64): New.
15772 (vset_lane_p64): New.
15773 (vsetq_lane_p64): New.
15775 2017-02-10 Jakub Jelinek <jakub@redhat.com>
15777 PR tree-optimization/79411
15778 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
15779 stmt operands are SSA_NAMEs used in abnormal phis.
15780 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
15783 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15786 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
15789 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15791 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
15793 2017-02-09 Jakub Jelinek <jakub@redhat.com>
15795 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
15799 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
15800 not arbitrary TREE_CONSTANT.
15803 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
15804 "omp declare target link" attribute unless is_global_var.
15805 * omp-offload.c (find_link_var_op): Likewise.
15807 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
15808 Chung-Lin Tang <cltang@codesourcery.com>
15810 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
15812 (gimplify_adjust_omp_clauses): Don't delete TILE.
15813 (gimplify_omp_for): Deal with TILE.
15814 * internal-fn.c (expand_GOACC_TILE): New function.
15815 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
15817 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
15818 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
15820 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
15821 avoid DIV for outermost collapse var.
15822 (expand_oacc_for): Insert tile element loop as needed. Adjust.
15823 Remove out of date comments, fix whitespace.
15824 * omp-general.c (omp_extract_for_data): Deal with tiling.
15825 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
15826 adjust OLF_DIM_BASE value.
15827 (struct omp_for_data): Add tiling field.
15828 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
15829 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
15830 for auto loops. Remove default auto determining, moved to
15831 oacc_loop_fixed_partitions.
15832 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
15833 stmts, add e_mask field.
15834 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
15835 (oacc_thread_numbers): Use oacc_dim_call.
15836 (oacc_xform_tile): New.
15837 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
15838 (finish_oacc_loop): Adjust for ifns vector.
15839 (oacc_loop_discover_walk): Append loop abstraction sites to list,
15840 add case for GOACC_TILE fns.
15841 (oacc_loop_xform_loop): Delete.
15842 (oacc_loop_process): Iterate over call list directly, and add
15843 handling for GOACC_TILE fns.
15844 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
15846 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
15847 vector partitioning to outer loops. Assign 2 partitions to loops
15848 when available. Add TILE handling.
15849 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
15850 (execite_oacc_device_lower): Process GOACC_TILE fns,
15851 ignore unknown specs.
15852 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
15853 * tree.c (omp_clause_num_ops): Adjust TILE ops.
15854 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
15856 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
15858 * configure.ac (ACX_BUGURL): Update.
15859 * configure: Regenerate.
15861 2017-02-09 Richard Biener <rguenther@suse.de>
15863 PR tree-optimization/69823
15864 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
15865 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
15867 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15869 * config/arc/arc-c.def: Add __NPS400__ definition.
15870 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
15871 (TARGET_NPS400): Define.
15873 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15875 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
15877 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
15878 pointer, arch_info.
15879 (arc_cpu_types): Fill the arch_info field with a pointer into the
15880 arc_arch_types table.
15881 (arc_selected_cpu): Declare.
15882 * config/arc/arc.c (arc_selected_cpu): Make global.
15883 (arc_selected_arch): Delete.
15884 (arc_base_cpu): Delete.
15885 (arc_override_options): Remove references to deleted variables,
15886 update access to arch information.
15887 (ARC_OPT): Update access to arch information.
15888 (ARC_OPTX): Likewise.
15889 * config/arc/arc.h (arc_base_cpu): Remove declaration.
15890 (TARGET_ARC600): Update access to arch information.
15891 (TARGET_ARC601): Likewise.
15892 (TARGET_ARC700): Likewise.
15893 (TARGET_EM): Likewise.
15894 (TARGET_HS): Likewise.
15895 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
15898 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
15901 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
15902 condition/operands for integer GE/LE/GEU/LEU operations.
15904 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
15906 PR translation/79397
15907 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
15910 2017-02-08 Martin Jambor <mjambor@suse.cz>
15913 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
15914 whether allocation happened.
15915 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
15916 nothing was allocated.
15918 2017-02-08 Jakub Jelinek <jakub@redhat.com>
15920 PR tree-optimization/79408
15921 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
15922 constant, but SSA_NAME with a known integer range, use the minimum
15923 of that range instead of op1 to determine if modulo can be replaced
15924 with its first operand.
15926 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15928 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
15930 2017-02-08 Richard Biener <rguenther@suse.de>
15932 PR tree-optimization/71824
15933 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
15934 Check all loops contained in the merged region.
15936 2017-02-07 Andrew Pinski <apinski@cavium.com>
15938 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
15940 2017-02-07 Andrew Pinski <apinski@cavium.com>
15942 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
15943 (thunderxt88): Likewise.
15944 (thunderxt81): Disable LSE and change v8.1 to v8.
15945 (thunderxt83): Likewise.
15947 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15948 Richard Biener <rguenther@suse.de>
15950 PR middle-end/79399
15951 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
15952 type from int to size_t.
15953 * ira-costs.c (struct_costs_size): Change type from int to size_t.
15955 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15957 PR rtl-optimization/79386
15958 * cprop.c (bypass_conditional_jumps): Initialize
15959 bypass_last_basic_block already before splitting bbs after
15960 unconditional traps...
15961 (bypass_conditional_jumps): ... rather than here.
15964 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
15965 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
15966 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
15967 fix -masm=intel patterns.
15969 2017-02-07 Richard Biener <rguenther@suse.de>
15971 PR tree-optimization/79256
15972 PR middle-end/79278
15973 * builtins.c (get_object_alignment_2): Use min_align_of_type
15974 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
15975 and ADJUST_FIELD_ALIGN.
15977 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
15979 * doc/tm.texi: Regenerate.
15980 * stor-layout.c (layout_decl): Adjust.
15981 (update_alignment_for_field): Likewise.
15982 (place_field): Likewise.
15983 (min_align_of_type): Likewise.
15984 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
15985 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
15986 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
15987 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
15988 * config/frv/frv.c (frv_adjust_field_align): Likewise.
15989 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
15990 * config/i386/i386.c (x86_field_alignment): Likewise.
15991 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
15992 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
15993 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
15994 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
15995 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
15996 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
16000 2017-01-30 Richard Biener <rguenther@suse.de>
16002 PR tree-optimization/79256
16003 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
16004 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
16007 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
16009 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
16010 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
16011 builtins to SImode and emit a zero-extend, if necessary.
16013 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
16015 * docs/invoke.texi (RISC-V Options): Alphabetize.
16017 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
16019 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
16022 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
16024 * config/riscv/riscv.c: New file.
16025 * common/config/riscv/riscv-common.c: Likewise.
16026 * config.gcc: Likewise.
16027 * config/riscv/constraints.md: Likewise.
16028 * config/riscv/elf.h: Likewise.
16029 * config/riscv/generic.md: Likewise.
16030 * config/riscv/linux.h: Likewise.
16031 * config/riscv/multilib-generator: Likewise.
16032 * config/riscv/peephole.md: Likewise.
16033 * config/riscv/pic.md: Likewise.
16034 * config/riscv/predicates.md: Likewise.
16035 * config/riscv/riscv-builtins.c: Likewise.
16036 * config/riscv/riscv-c.c: Likewise.
16037 * config/riscv/riscv-ftypes.def: Likewise.
16038 * config/riscv/riscv-modes.def: Likewise.
16039 * config/riscv/riscv-opts.h: Likewise.
16040 * config/riscv/riscv-protos.h: Likewise.
16041 * config/riscv/riscv.h: Likewise.
16042 * config/riscv/riscv.md: Likewise.
16043 * config/riscv/riscv.opt: Likewise.
16044 * config/riscv/sync.md: Likewise.
16045 * config/riscv/t-elf-multilib: Likewise.
16046 * config/riscv/t-linux: Likewise.
16047 * config/riscv/t-linux-multilib: Likewise.
16048 * config/riscv/t-riscv: Likewise.
16049 * configure.ac: Likewise.
16050 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
16051 Waterman as RISC-V maintainers.
16052 * doc/install.texi: Add RISC-V entries.
16053 * doc/invoke.texi: Add RISC-V options section.
16054 * doc/md.texi: Add RISC-V constraints section.
16055 * configure: Regenerated.
16057 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
16060 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
16061 false values to be constant vectors with all 0 or all 1 bits set.
16062 (vcondu<mode><mode>): Likewise.
16063 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
16065 (fpmask_comparison_operator): Update comment.
16066 (vecint_comparison_operator): New predicate.
16067 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
16068 vector conditionals when the true and false values are constant
16069 vectors with all 0 bits or all 1 bits set.
16071 2017-02-06 Martin Sebor <msebor@redhat.com>
16073 PR tree-optimization/79376
16074 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
16076 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
16078 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
16079 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
16080 to simplify split condition.
16082 2017-02-06 Jakub Jelinek <jakub@redhat.com>
16084 * omp-expand.c (oxpand_omp_atomic_fetch_op,
16085 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
16088 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
16090 PR rtl-optimization/68664
16091 * target.def (can_speculate_insn): New hook.
16092 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
16093 * doc/tm.texi: Regenerate.
16094 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
16095 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
16096 (rs6000_sched_can_speculate_insn): New function.
16098 2017-02-06 Jakub Jelinek <jakub@redhat.com>
16100 PR tree-optimization/79284
16101 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
16102 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
16103 vectorizable_mask_load_store, vectorizable_operation,
16104 vect_is_simple_cond, get_same_sized_vectype): Use it instead
16105 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
16106 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
16107 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
16108 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
16109 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
16110 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
16111 is_gimple_assign (stmt). Replace another such test with
16112 is_gimple_assign (stmt).
16114 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
16117 * config/avr/avr.c (rtl-iter.h): Include it.
16118 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
16119 (avr_legitimate_combined_insn): ...and implementation.
16121 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
16123 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
16124 * config/s390/s390.c (s390_const_operand_ok)
16125 (s390_canonicalize_comparison, s390_extract_part)
16126 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
16127 (s390_contiguous_bitmask_p, s390_rtx_costs)
16128 (legitimize_pic_address): Likewise.
16129 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
16130 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
16131 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
16132 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
16133 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
16135 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
16137 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
16138 REGNO($0) == REGNO($1).
16140 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16142 * config/s390/linux.h(SIZE_TYPE): Add comment.
16144 2017-02-06 Julian Brown <julian@codesourcery.com>
16145 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16146 Virendra Pathak <virendra.pathak@broadcom.com>
16148 * config/aarch64/aarch64-cores.def: Change the scheduler
16150 * config/aarch64/aarch64.md: Include thunderx2t99.md.
16151 * config/aarch64/thunderx2t99.md: New file.
16153 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16155 * doc/standards.texi (Go Language): Update link to language
16158 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
16160 * tree-eh.c (lower_resx): Sanitize profile.
16161 (cleanup_empty_eh_move_lp): Likewise.
16163 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
16166 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
16168 * cfgloopmanip.h (loop_version): Update prototype.
16169 * modulo-sched.c (sms_schedule): Update call of loop_version.
16170 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
16171 * tree-parloops.c (gen_parallel_loop): Likewise.
16172 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
16173 * tree-ssa-loop-split.c (split_loop): Likewise.
16174 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
16175 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
16177 2017-02-05 Martin Liska <mliska@suse.cz>
16180 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
16182 (print_operand_address): Initialize a struct to zero.
16184 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16186 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
16187 garbage collector only in textual form.
16189 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16191 * doc/extend.texi (x86 specific memory model extensions for
16192 transactional memory): Simplify a phrase.
16194 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
16197 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
16198 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
16199 (atomic_storedi_1): Likewise.
16201 2017-02-04 Jakub Jelinek <jakub@redhat.com>
16203 PR tree-optimization/79338
16204 * tree-parloops.c (gather_scalar_reductions): Don't call
16205 vect_analyze_loop_form for loop->inner before destroying loop's
16208 2017-02-03 Martin Sebor <msebor@redhat.com>
16210 PR tree-optimization/79327
16211 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
16212 when precision has resulted in leading zeros.
16213 (format_integer): Adjust the likely counter to assume an unknown
16214 argument that may be zero is non-zero.
16216 2017-02-03 Jason Merrill <jason@redhat.com>
16219 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
16220 avoid copying non-taken branch.
16222 2017-02-03 Jakub Jelinek <jakub@redhat.com>
16224 PR tree-optimization/79340
16225 * tree-vect-loop.c (vectorizable_reduction): Release
16226 vec_defs elements after safe_splicing them into other vectors.
16229 PR tree-optimization/79327
16230 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
16231 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
16233 (format_integer): Use wide_int_to_tree instead of build_int_cst
16234 + to_?hwi. If argmin is NULL, just set argmin and argmax to
16235 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
16236 of shortest and longest sequence.
16238 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
16240 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
16241 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
16243 2017-02-03 Walter Lee <walt@tilera.com>
16246 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
16247 after initial stackframe link reg save.
16248 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
16250 2017-02-03 Jakub Jelinek <jakub@redhat.com>
16253 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
16254 wu for stxssp alternative.
16256 2017-02-03 Martin Sebor <msebor@redhat.com>
16258 PR tree-optimization/79352
16259 * gimple-fold.c (get_range_strlen): Add argument.
16260 (get_range_strlen): Change return type to bool.
16261 (get_maxval_strlen): Pass in a dummy argument.
16262 * gimple-fold.h (get_range_strlen): Change return type to bool.
16263 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
16264 * tree.h (array_at_struct_end_p): Add argument.
16265 * tree.c (array_at_struct_end_p): Handle it.
16267 2017-02-03 Martin Liska <mliska@suse.cz>
16270 * multiple_target.c (create_dispatcher_calls): Redirect edge
16271 from a caller of a dispatcher.
16272 (expand_target_clones): Make the clones local.
16273 (ipa_target_clone): Do both target clones and resolvers.
16274 (ipa_dispatcher_calls): Remove the pass.
16275 (pass_dispatcher_calls::gate): Likewise.
16276 (make_pass_dispatcher_calls): Likewise.
16277 * passes.def (pass_target_clone): Put as very first IPA early
16280 2017-02-03 Martin Liska <mliska@suse.cz>
16282 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
16283 in case of a function with ifunc attribute.
16285 2017-02-03 Martin Liska <mliska@suse.cz>
16287 * cgraph.c (cgraph_node::dump): Dump function version info.
16288 * symtab.c (symtab_node::dump_base): Add missing new line.
16290 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
16292 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
16293 (ifcombine_ifandif): Use it.
16295 2017-02-03 Martin Liska <mliska@suse.cz>
16297 * doc/invoke.texi: Document default value for
16298 use-after-scope-direct-emission-threshold.
16300 2017-02-03 Martin Liska <mliska@suse.cz>
16302 PR tree-optimization/79339
16303 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
16304 (format_floating): Likewise.
16306 2017-02-03 Martin Liska <mliska@suse.cz>
16309 * ipa-prop.c (ipa_node_params_t::insert): Remove current
16311 (ipa_node_params_t::remove): Likewise.
16312 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
16313 initialization from removed ipa_node_params_t::insert.
16314 (ipa_node_params::~ipa_node_params): Move from removed
16315 ipa_node_params_t::release.
16316 * symbol-summary.h (symbol_summary::m_released): New member.
16317 Do not release a summary twice. Do not allow to call finalizer
16318 for types of a summary that live in GGC memory.
16320 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16322 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
16325 2017-02-02 Martin Sebor <msebor@redhat.com>
16327 PR middle-end/79275
16328 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
16329 (format_string): Tighten up the range of output for non-constant
16330 strings and correct the expected range for wide non-constant strings.
16332 2017-02-02 Martin Sebor <msebor@redhat.com>
16334 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
16336 PR middle-end/32003
16337 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
16339 (-fdump-tree-@var): Add to index and document how to come up
16340 with pass-specific option and dump file names.
16341 (-fdump-passes): Clarify where to look for output.
16343 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
16345 PR middle-end/77445
16346 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
16347 statistics of the analyzed path; allow threading for speed when
16348 any of BBs along the path are optimized for speed.
16350 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
16352 PR middle-end/78468
16353 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
16354 settings of the virtual registers.
16357 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
16359 * explow.c (get_dynamic_stack_size): Take known alignment of stack
16360 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
16363 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16365 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
16366 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
16368 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16370 * config/s390/s390.md: Add missing comments with the expanded
16372 * config/s390/vector.md: Likewise.
16373 * config/s390/vx-builtins.md: Likewise.
16375 2017-02-02 Jakub Jelinek <jakub@redhat.com>
16378 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
16379 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
16380 conditions on a single line.
16382 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16384 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
16385 __S390_VX__ to __VX__.
16387 2017-02-01 Andrew Pinski <apinski@cavium.com>
16389 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
16390 stmt_info to record_stmt_cost.
16391 (vect_get_known_peeling_cost): Pass stmt_info if known to
16393 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
16394 cpu_vector_cost field into
16395 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
16396 field into vec_int_stmt_cost and vec_fp_stmt_cost.
16397 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
16398 splitting of scalar_stmt_cost and vec_stmt_cost.
16399 (thunderx_vector_cost): Likewise.
16400 (cortexa57_vector_cost): LIkewise.
16401 (exynosm1_vector_cost): Likewise.
16402 (xgene1_vector_cost): Likewise.
16403 (thunderx2t99_vector_cost): Improve after the splitting of the two
16405 (aarch64_builtin_vectorization_cost): Update for the splitting of
16406 scalar_stmt_cost and vec_stmt_cost.
16408 2017-02-01 Torvald Riegel <triegel@redhat.com>
16409 Richard Henderson <rth@redhat.com>
16411 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
16412 conditional on existance of a fast atomic load.
16413 * optabs-query.c (can_atomic_load_p): New function.
16414 * optabs-query.h (can_atomic_load_p): Declare it.
16415 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
16416 no fast atomic load is available for the particular size of access.
16417 (expand_atomic_compare_and_swap): Likewise.
16418 (expand_atomic_load): Likewise.
16419 (expand_atomic_store): Likewise.
16420 (expand_atomic_fetch_op): Likewise.
16421 * testsuite/lib/target-supports.exp
16422 (check_effective_target_sync_int_128): Remove x86 because it provides
16423 no fast atomic load.
16424 (check_effective_target_sync_int_128_runtime): Likewise.
16426 2017-02-01 Richard Biener <rguenther@suse.de>
16428 * graphite.c: Include tree-vectorizer.h for find_loop_location.
16429 (graphite_transform_loops): Provide opt-info for optimized nests.
16430 * tree-parloop.c (parallelize_loops): Provide opt-info for
16431 parallelized loops.
16433 2017-02-01 Richard Biener <rguenther@suse.de>
16435 PR middle-end/79315
16436 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
16437 was not set before.
16439 2017-02-01 Richard Biener <rguenther@suse.de>
16441 PR tree-optimization/71824
16442 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
16443 Verify the loops are valid in the merged SESE region.
16444 (scop_detection::can_represent_loop_1): Check analyzing the
16445 evolution of the number of iterations in the region succeeds.
16447 2017-01-31 Ian Lance Taylor <iant@golang.org>
16449 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
16450 REG_ARGS_SIZE note to 32-bit push insns and call insn.
16452 2017-01-31 David Malcolm <dmalcolm@redhat.com>
16454 PR preprocessor/79210
16455 * input.c (get_substring_ranges_for_loc): Replace line_width
16456 assertion with error-handling.
16458 2017-01-31 Richard Biener <rguenther@suse.de>
16460 PR tree-optimization/77318
16461 * graphite-sese-to-poly.c (extract_affine): Fix assert.
16462 (create_pw_aff_from_tree): Take loop parameter.
16463 (add_condition_to_pbb): Pass loop of the condition to
16464 create_pw_aff_from_tree.
16466 2017-01-31 Jakub Jelinek <jakub@redhat.com>
16468 * config/s390/s390.c (s390_asan_shadow_offset): New function.
16469 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
16471 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
16475 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
16477 (convert_int_to_float128): Likewise.
16478 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
16479 (convert_int_to_float128): Likewise.
16480 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
16481 (UNSPEC_IEEE128_CONVERT): Likewise.
16482 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
16483 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
16484 Use local variables for IBM extended format.
16485 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
16486 (fix_trunc<mode>si2_fprs): Likewise.
16487 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
16488 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
16489 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
16490 to know that we can now have integers of all sizes in vector
16492 (fix<uns>_<mode>di2_hw): Likewise.
16493 (float<uns>_<mode>si2_hw): Likewise.
16494 (fix_<mode>si2_hw): Likewise.
16495 (fixuns_<mode>si2_hw): Likewise.
16496 (float<uns>_<mode>di2_hw): Likewise.
16497 (float_<mode>di2_hw): Likewise.
16498 (float_<mode>si2_hw): Likewise.
16499 (floatuns_<mode>di2_hw): Likewise.
16500 (floatuns_<mode>si2_hw): Likewise.
16501 (xscvqp<su>wz_<mode>): Delete, no longer used.
16502 (xscvqp<su>dz_<mode>): Likewise.
16503 (xscv<su>dqp_<mode>): Likewise.
16504 (ieee128_mfvsrd_64bit): Likewise.
16505 (ieee128_mfvsrd_32bit): Likewise.
16506 (ieee128_mfvsrwz): Likewise.
16507 (ieee128_mtvsrw): Likewise.
16508 (ieee128_mtvsrd_64bit): Likewise.
16509 (ieee128_mtvsrd_32bit): Likewise.
16511 2017-01-31 Martin Liska <mliska@suse.cz>
16514 * ipa-prop.c (ipa_free_all_node_params): Call release method
16515 instead of ~sumbol_summary to not to trigger double times
16518 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
16520 PR tree-optimization/71691
16521 * bitmap.h (class auto_bitmap): New.
16522 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
16523 is_maybe_undefined instead of ssa_undefined_value_p.
16525 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16527 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
16528 __S390_ARCH_LEVEL__ to __ARCH__.
16530 2017-01-31 Jakub Jelinek <jakub@redhat.com>
16532 PR tree-optimization/79267
16533 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
16534 if should_remove_lhs_p is true.
16536 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
16539 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
16540 (add_alignment_attribute): New.
16541 (base_type_die): Add alignment attribute.
16542 (subrange_type_die): Likewise.
16543 (modified_type_die): Likewise.
16544 (gen_array_type_die): Likewise.
16545 (gen_descr_array_type_die: Likewise.
16546 (gen_enumeration_type_die): Likewise.
16547 (gen_subprogram_die): Likewise.
16548 (gen_variable_die): Likewise.
16549 (gen_field_die): Likewise.
16550 (gen_ptr_to_mbr_type_die): Likewise.
16551 (gen_struct_or_union_type_die): Likewise.
16552 (gen_subroutine_type_die): Likewise.
16553 (gen_typedef_die): Likewise.
16554 (base_type_cmp): Compare alignment attribute.
16556 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16559 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
16560 (setb_unsigned) New pattern for setb with CCUNS.
16561 * config/rs6000/rs6000.c (expand_block_compare): Use a different
16562 subfc./subfe sequence to avoid overflow problems. Generate a
16563 shorter sequence with cmpld/setb for power9.
16564 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
16565 for generating subfc. instruction.
16566 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
16567 now uses this instruction.
16569 2017-01-30 Ian Lance Taylor <iant@google.com>
16572 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
16573 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
16575 2017-01-30 Martin Sebor <msebor@redhat.com>
16577 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
16578 Move constant to the right of a relational operator.
16579 (get_mpfr_format_length, format_character, format_string): Ditto.
16580 (should_warn_p, maybe_warn): Same.
16582 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
16584 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
16587 * asan.c (get_translation_unit_decl): Remove function.
16588 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
16590 2017-01-30 Martin Liska <mliska@suse.cz>
16592 PR gcov-profile/79259
16593 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
16594 -fprofile-generate.
16596 2017-01-30 Martin Liska <mliska@suse.cz>
16599 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
16600 Initialize variables with NULL value.
16602 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
16605 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
16607 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
16609 2017-01-30 Richard Biener <rguenther@suse.de>
16611 PR tree-optimization/79276
16612 * tree-vrp.c (process_assert_insertions): Properly adjust common
16613 when removing a duplicate.
16615 * gcc.dg/torture/pr79276.c: New testcase.
16617 2017-01-30 Richard Biener <rguenther@suse.de>
16619 PR tree-optimization/79256
16620 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
16621 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
16623 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
16625 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
16628 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
16629 ("*r<noxa>sbg_<mode>_sll_bitmask")
16630 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
16631 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
16632 Use contiguous_bitmask_nowrap_operand.
16634 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16637 * config/rs6000/altivec.h (vec_xl): Revise #define.
16638 (vec_xst): Likewise.
16640 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
16642 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
16644 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
16646 PR rtl-optimization/79194
16647 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
16648 traps before call to bypass_conditional_jumps.
16650 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16652 PR tree-optimization/71374
16653 * lra-constraints.c (check_conflict_input_operands): New.
16654 (match_reload): Use it.
16656 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16659 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
16660 account to calculate conflict_set.
16662 2017-01-27 Bin Cheng <bin.cheng@arm.com>
16664 PR rtl-optimization/78559
16665 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
16666 other_insn in combine.
16668 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16670 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
16671 uint16_type_node for BT_UINT16.
16673 2017-01-27 David Malcolm <dmalcolm@redhat.com>
16675 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
16676 "RTL Tests" to menu.
16677 (GIMPLE Tests): New node.
16678 (RTL Tests): New node.
16680 2017-01-27 Richard Biener <rguenther@suse.de>
16682 PR tree-optimization/79245
16683 * tree-loop-distribution.c (distribute_loop): Apply cost
16684 modeling also to detected patterns.
16686 2017-01-27 Richard Biener <rguenther@suse.de>
16688 PR tree-optimization/71433
16689 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
16690 (compare_assert_loc): New function.
16691 (process_assert_insertions): Sort and optimize assert locations
16692 to remove duplicates and push down identical assertions on
16693 edges to their destination block.
16695 2017-01-27 Richard Biener <rguenther@suse.de>
16697 PR tree-optimization/79244
16698 * tree-vrp.c (remove_range_assertions): Forcefully propagate
16699 out SSA names even if abnormal.
16701 2017-01-27 Jakub Jelinek <jakub@redhat.com>
16703 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
16704 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
16705 instead of MPFR_RNDN.
16707 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
16710 * arm.c (arm_option_override): Don't call build_target_option_node
16711 until after doing all option overrides.
16712 (arm_valid_target_attribute_tree): Likewise.
16714 2017-01-27 Martin Liska <mliska@suse.cz>
16716 * doc/invoke.texi (-fprofile-arcs): Document profiling support
16717 for {cd}tors and C++ {cd}tors.
16719 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
16721 * config/s390/s390.md ("*setmem_long_and")
16722 ("*setmem_long_and_31z"): Use zero_extend instead of and.
16724 2017-01-26 Martin Sebor <msebor@redhat.com>
16726 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
16729 2017-01-26 Martin Sebor <msebor@redhat.com>
16731 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
16732 HAVE_DFmode before using XFmode or DFmode.
16733 (parse_directive): Avoid using the z length modifier to avoid
16734 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
16736 PR middle-end/78703
16737 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
16738 to accept adjustment as an array.
16739 (get_int_range): New function.
16740 (struct directive): Make width and prec arrays.
16741 (directive::set_width, directive::set_precision): Call get_int_range.
16742 (format_integer, format_floating): Handle width and precision ranges.
16743 (format_string, parse_directive): Same.
16745 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16748 * dwarf2out.c (generate_skeleton_bottom_up): For children with
16749 comdat_type_p set, just clone them, but keep the children in the
16753 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
16754 which have direct callers with -fvar-tracking-assignments enabled
16756 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
16757 inside of type units.
16759 2017-01-26 Martin Sebor <msebor@redhat.com>
16761 PR middle-end/78703
16762 * gimple-ssa-sprintf.c (struct result_range): Add likely and
16764 (struct format_result): Replace number_chars, number_chars_min,
16765 and number_chars_max with a single member of struct result_range.
16767 (format_result::operator+=): Adjust.
16768 (struct fmtresult): Remove bounded. Handle likely and unlikely
16770 (fmtresult::adjust_for_width_or_precision): New function.
16771 (fmtresult:type_max_digits): New function.
16772 (bytes_remaining): Handle likely and unlikely counters.
16773 (min_bytes_remaining): Remove.
16774 (format_percent): Simplify.
16775 (format_integer, format_floating): Set likely and unlikely counters.
16776 (get_string_length, format_character, format_string): Same.
16777 (format_plain, should_warn_p): New function.
16778 (maybe_warn): Call should_warn_p. Update diagnostic messages
16779 and handle those for all directives, including plain strings.
16780 (format_directive): Handle likely and unlikely counters.
16781 Remove unnecessary quoting from diagnostics. Add an informational
16783 (add_bytes): Remove.
16784 (pass_sprintf_length::compute_format_length): Simplify.
16785 (try_substitute_return_value): Handle likely and unlikely counters.
16787 2017-01-26 Carl Love <cel@us.ibm.com>
16789 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
16790 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
16792 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
16795 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
16796 endianess for subregs into account.
16797 * lra-constraints.c (lra_constraints): Do risky transformations
16798 always on the first iteration.
16799 * lra-lives.c (check_pseudos_live_through_calls): Add arg
16800 last_call_used_reg_set.
16801 (process_bb_lives): Define and use last_call_used_reg_set.
16802 * lra.c (lra): Always continue after lra_constraints on the first
16805 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
16807 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
16809 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
16811 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16813 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
16814 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
16815 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
16816 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
16817 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
16818 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
16819 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
16820 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
16821 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
16823 2017-01-26 Marek Polacek <polacek@redhat.com>
16826 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
16827 for the third operand.
16829 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16831 PR middle-end/79236
16832 * omp-low.c (struct omp_context): Add simt_stmt field.
16833 (scan_omp_for): Return omp_context *.
16834 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
16835 context to the _simt_ SIMD stmt.
16836 (lower_omp_for): For combined SIMD with sibling _simt_
16837 SIMD, make sure to use the same decls in _looptemp_
16838 clauses as in the sibling.
16840 2017-01-26 David Sherwood <david.sherwood@arm.com>
16842 PR middle-end/79212
16843 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
16846 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16849 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
16850 emit fld b; fld a; if possible.
16852 * brig-builtins.def: Update copyright years.
16853 * config/arm/arm_acle_builtins.def: Update copyright years.
16855 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16858 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
16859 constraint instead of o for the stxsd instruction.
16861 2017-01-25 Carl Love <cel@us.ibm.com>
16863 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
16864 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
16866 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
16868 * doc/invoke.texi (C++ Dialect Options): Fix typo.
16870 2017-01-25 Richard Biener <rguenther@suse.de>
16872 PR tree-optimization/69264
16873 * target.def (vector_alignment_reachable): Improve documentation.
16874 * doc/tm.texi: Regenerate.
16875 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
16877 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
16878 earlier changes with respect to TYPE_USER_ALIGN.
16879 (vector_alignment_reachable_p): Likewise. Improve dumping.
16881 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16884 * config/arm/arm.md (xordi3): Force constant operand into a register
16887 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16889 * doc/invoke.texi (-fstore-merging): Correct default optimization
16890 levels at which it is enabled.
16891 (-O): Move -fstore-merging from list to...
16894 2017-01-25 Richard Biener <rguenther@suse.de>
16897 * omp-expand.c: Include debug.h.
16898 (expand_omp_taskreg): Make sure to generate early debug before
16899 outlining anything from a function.
16900 (expand_omp_target): Likewise.
16901 (grid_expand_target_grid_body): Likewise.
16903 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
16906 * asan.c (get_translation_unit_decl): New function.
16907 (asan_add_global): Extract modules file name from globals
16908 TRANSLATION_UNIT_DECL name.
16910 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
16913 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
16914 for long calls with APCS frame and VFP.
16916 2017-01-24 David Malcolm <dmalcolm@redhat.com>
16918 * cfg.c (original_copy_tables_initialized_p): New function.
16919 * cfg.h (original_copy_tables_initialized_p): New decl.
16920 * cfgrtl.c (relink_block_chain): Guard the call to
16921 free_original_copy_tables with a call to
16922 original_copy_tables_initialized_p.
16923 * cgraph.h (symtab_node::native_rtl_p): New decl.
16924 * cgraphunit.c (symtab_node::native_rtl_p): New function.
16925 (symtab_node::needed_p): Don't assert for early assembly output
16926 for __RTL functions.
16927 (cgraph_node::finalize_function): Set "force_output" for __RTL
16929 (cgraph_node::analyze): Bail out early for __RTL functions.
16930 (analyze_functions): Update assertion to support __RTL functions.
16931 (cgraph_node::expand): Bail out early for __RTL functions.
16932 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
16934 * function.h (struct function): Update comment for field
16936 * gimple-expr.c: Include "tree-pass.h".
16937 (gimple_has_body_p): Return false for __RTL functions.
16938 * Makefile.in (OBJS): Add run-rtl-passes.o.
16939 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
16941 (gcc::pass_manager::get_clean_slate): New accessor.
16942 * passes.c: Include "insn-addr.h".
16943 (should_skip_pass_p): Add logging. Update logic for running
16944 "expand" to be compatible with both __GIMPLE and __RTL. Guard
16945 property-provider override so it is only done for gimple passes.
16947 (skip_pass): New function.
16948 (execute_one_pass): Call skip_pass when skipping passes.
16949 * read-md.c (md_reader::read_char): Support filtering
16950 the input to a subset of line numbers.
16951 (md_reader::md_reader): Initialize fields
16952 m_first_line and m_last_line.
16953 (md_reader::read_file_fragment): New function.
16954 * read-md.h (md_reader::read_file_fragment): New decl.
16955 (md_reader::m_first_line): New field.
16956 (md_reader::m_last_line): New field.
16957 * read-rtl-function.c (function_reader::create_function): Only
16958 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
16959 curr_properties. Set DECL_INITIAL to a dummy block.
16960 (read_rtl_function_body_from_file_range): New function.
16961 * read-rtl-function.h (read_rtl_function_body_from_file_range):
16963 * run-rtl-passes.c: New file.
16964 * run-rtl-passes.h: New file.
16966 2017-01-24 Jeff Law <law@redhat.com>
16968 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
16971 2017-01-24 Bin Cheng <bin.cheng@arm.com>
16973 PR tree-optimization/79159
16974 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
16975 (record_nonwrapping_iv): Improve boundary using above function if no
16976 value range information.
16978 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
16979 Martin Jambor <mjambor@suse.cz>
16981 * brig-builtins.def: New file.
16982 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
16983 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
16984 (DEF_HSAIL_SAT_BUILTIN): Likewise.
16985 (DEF_HSAIL_INTR_BUILTIN): Likewise.
16986 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
16987 * builtin-types.def (BT_INT8): New.
16988 (BT_INT16): Likewise.
16989 (BT_UINT8): Likewise.
16990 (BT_UINT16): Likewise.
16991 (BT_FN_ULONG): Likewise.
16992 (BT_FN_UINT_INT): Likewise.
16993 (BT_FN_UINT_ULONG): Likewise.
16994 (BT_FN_UINT_LONG): Likewise.
16995 (BT_FN_UINT_PTR): Likewise.
16996 (BT_FN_ULONG_PTR): Likewise.
16997 (BT_FN_INT8_FLOAT): Likewise.
16998 (BT_FN_INT16_FLOAT): Likewise.
16999 (BT_FN_UINT32_FLOAT): Likewise.
17000 (BT_FN_UINT16_FLOAT): Likewise.
17001 (BT_FN_UINT8_FLOAT): Likewise.
17002 (BT_FN_UINT64_FLOAT): Likewise.
17003 (BT_FN_UINT16_UINT32): Likewise.
17004 (BT_FN_UINT32_UINT16): Likewise.
17005 (BT_FN_UINT16_UINT16_UINT16): Likewise.
17006 (BT_FN_INT_PTR_INT): Likewise.
17007 (BT_FN_UINT_PTR_UINT): Likewise.
17008 (BT_FN_LONG_PTR_LONG): Likewise.
17009 (BT_FN_ULONG_PTR_ULONG): Likewise.
17010 (BT_FN_VOID_UINT64_UINT64): Likewise.
17011 (BT_FN_UINT8_UINT8_UINT8): Likewise.
17012 (BT_FN_INT8_INT8_INT8): Likewise.
17013 (BT_FN_INT16_INT16_INT16): Likewise.
17014 (BT_FN_INT_INT_INT): Likewise.
17015 (BT_FN_UINT_FLOAT_UINT): Likewise.
17016 (BT_FN_FLOAT_UINT_UINT): Likewise.
17017 (BT_FN_ULONG_UINT_UINT): Likewise.
17018 (BT_FN_ULONG_UINT_PTR): Likewise.
17019 (BT_FN_ULONG_ULONG_ULONG): Likewise.
17020 (BT_FN_UINT_UINT_UINT): Likewise.
17021 (BT_FN_VOID_UINT_PTR): Likewise.
17022 (BT_FN_UINT_UINT_PTR: Likewise.
17023 (BT_FN_UINT32_UINT64_PTR): Likewise.
17024 (BT_FN_INT_INT_UINT_UINT): Likewise.
17025 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
17026 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
17027 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
17028 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
17029 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
17030 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
17031 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
17032 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
17033 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
17034 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
17035 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
17036 * doc/frontends.texi: List BRIG FE.
17037 * doc/install.texi (Testing): Add BRIG tesring requirements.
17038 * doc/invoke.texi (Overall Options): Mention BRIG.
17039 * doc/standards.texi (Standards): Doucment BRIG HSA version.
17041 2017-01-24 Richard Biener <rguenther@suse.de>
17043 PR translation/79208
17044 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
17046 2017-01-24 Martin Jambor <mjambor@suse.cz>
17049 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
17050 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
17051 and known_contexts.
17053 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
17055 PR middle-end/79123
17056 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
17057 casts from signed to unsigned really don't have a range.
17059 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
17061 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
17062 GMP_RNDx for compatiblity.
17064 2017-01-24 Martin Liska <mliska@suse.cz>
17067 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
17068 that would prevent us to call alloca with -1 as argument.
17070 2017-01-24 Jakub Jelinek <jakub@redhat.com>
17072 * dwarf2out.c (output_compilation_unit_header, output_file_names):
17073 Avoid -Wformat-security warning.
17075 2017-01-23 Andrew Pinski <apinski@cavium.com>
17077 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
17080 2017-01-23 Martin Sebor <msebor@redhat.com>
17082 PR middle-end/78703
17083 * gimple-ssa-sprintf.c (warn_level): New global.
17084 (format_integer): Use it here and throughout the rest of the file.
17085 Use the same switch to compute sign as base.
17086 (maybe_warn): New function.
17087 (format_directive): Factor out warnings into maybe_warn.
17088 Add debugging output. Use warn_level.
17089 (add_bytes): Use warn_level.
17090 (pass_sprintf_length::compute_format_length): Add debugging output.
17091 (try_substitute_return_value): Same.
17092 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
17094 PR middle-end/78703
17095 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
17096 (struct fmtresult, format_integer, format_floating): Adjust.
17097 (fmtresult::fmtresult): Set max correctly in two argument ctor.
17098 (get_string_length, format_string,format_directive): Same.
17099 (pass_sprintf_length::compute_format_length): Same.
17100 (try_substitute_return_value): Simplify slightly.
17102 PR middle-end/78703
17103 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
17104 (fmtresult::operator+=): Outlined.
17105 (struct fmtresult): Add ctors.
17106 (struct conversion_spec): Rename...
17107 (struct directive): ...to this. Add and remove data members.
17108 (directive::set_width, directive::set_precision): New functions.
17109 (format_percent): Use fmtresult ctor.
17110 (get_width_and_precision): Remove.
17111 (format_integer): Make naming changes. Avoid computing width and
17113 (format_floating): Same. Adjust indentation.
17114 (format_character, format_none): New functions.
17115 (format_string): Moved character handling to format_character.
17116 (format_directive): Remove arguments, change return type.
17117 (parse_directive): New function.
17118 (pass_sprintf_length::compute_format_length): Move directive
17119 parsing to parse_directive.
17121 2017-01-23 Jakub Jelinek <jakub@redhat.com>
17123 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
17124 (assign_assembler_name_if_needed): ... this.
17125 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
17126 (assign_assembler_name_if_needed): ... this.
17127 (free_lang_data_in_cgraph): Adjust callers.
17128 * cgraphunit.c (cgraph_node::analyze): Likewise.
17129 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
17132 2017-01-23 Richard Biener <rguenther@suse.de>
17134 PR tree-optimization/79088
17135 PR tree-optimization/79188
17136 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
17137 resetting loop bounds after last path deletion. Reset loop
17138 bounds of the target loop, make code match the comments.
17139 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
17140 Make sure loops need no fixups.
17142 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
17144 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
17145 exponent support with double type for first argument.
17146 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
17147 type returned by __builtin_vec_extract_sig,
17148 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
17149 functions from "vector int" to "vector unsigned int" or from
17150 "vector long long int" to "vector unsigned long long int".
17151 Changed type returned by __builtin_vec_extract_exp,
17152 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
17153 functions from "vector int" to "vector unsigned int" or from
17154 "vector long long int" to "vector unsigned long long int".
17155 Changed return type of __builtin_vec_test_data_class,
17156 __builtin_vec_test_data_class_sp, and
17157 __builtin_vec_test_data_class_dp from "vector int" to
17158 "vector bool int" or from "vector long long int" to "vector bool
17159 long long int" and changed second argument type from "unsigned
17160 int" to "int". Added new overloaded function forms "vector float
17161 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
17162 "vector float __builtin_vec_insert_exp_sp (vector float, vector
17163 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
17164 double, vector unsigned long long int)" and "vector double
17165 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
17166 long int)". Changed return type of
17167 __builtin_scalar_test_data_class and
17168 __builtin_scalar_test_data_class_sp and
17169 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
17170 int" and changed second argument from "unsigned int" to "int".
17171 Changed type returned by __builtin_scalar_test_neg,
17172 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
17173 from "int" to "bool int". Added new overloaded function form
17174 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
17175 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
17176 exponent double-precision with floating point first argument.
17177 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
17178 documentation of scalar_test_data_class, scalar_test_neg,
17179 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
17180 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
17181 vec_test_data_class built-in functions to reflect refinements in
17182 their type signatures.
17184 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
17186 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
17188 (aarch64_elf_asm_destructor): Likewise.
17190 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
17192 PR rtl-optimization/78634
17193 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
17194 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
17195 * ifcvt.c (noce_try_cmove): Add missing cost check.
17197 PR rtl-optimization/71724
17198 * combine.c (if_then_else_cond): Look for situations where it is
17199 beneficial to undo the work of one of the recursive calls.
17201 2017-01-23 Bin Cheng <bin.cheng@arm.com>
17203 PR tree-optimization/70754
17204 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
17205 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
17206 combined stmt before it if not NULL.
17207 (combine_chains): Process refs reversely and compute dominance point
17210 2017-01-23 Martin Liska <mliska@suse.cz>
17212 PR tree-optimization/79196
17213 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
17214 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
17216 (strlen_optimize_stmt): Call the renamed function.
17218 2017-01-23 Michael Matz <matz@suse.de>
17220 PR tree-optimization/78384
17221 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
17223 2017-01-23 Richard Biener <rguenther@suse.de>
17225 PR tree-optimization/79186
17226 * tree-vrp.c (register_new_assert_for): Make sure we've seen
17227 both incoming edges before moving an assert.
17229 2017-01-23 Martin Jambor <mjambor@suse.cz>
17231 * ipa-prop.c (load_from_param_1): Removed.
17232 (load_from_unmodified_param): Bits from load_from_param_1 put back
17234 (load_from_param): Removed.
17235 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
17236 with stmt. Reverted back to use of load_from_unmodified_param.
17238 2017-01-23 Martin Jambor <mjambor@suse.cz>
17241 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
17242 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
17243 field a pointer to garbage collected vector, mark lattices and
17244 ipcp_orig_node with GTY((skip)).
17245 (ipa_get_param_count): Adjust to descriptors being a pointer.
17246 (ipa_get_param): Likewise.
17247 (ipa_get_type): Likewise.
17248 (ipa_get_param_move_cost): Likewise.
17249 (ipa_set_param_used): Likewise.
17250 (ipa_get_controlled_uses): Likewise.
17251 (ipa_set_controlled_uses): Likewise.
17252 (ipa_is_param_used): Likewise.
17253 (ipa_node_params_t): Move into garbage collector. New methods insert
17255 (ipa_node_params_sum): Annotate wth GTY(()).
17256 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
17258 (ipa_load_from_parm_agg): Adjust declaration.
17259 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
17260 * ipa-profile.c (ipa_profile): Likewise.
17261 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
17262 (ipa_populate_param_decls): Make descriptors parameter garbage
17264 (ipa_dump_param): Adjust to descriptors being a pointer.
17265 (ipa_alloc_node_params): Likewise.
17266 (ipa_initialize_node_params): Likewise.
17267 (load_from_param_1): Make descriptors parameter garbage collected.
17268 (load_from_unmodified_param): Likewise.
17269 (load_from_param): Likewise.
17270 (ipa_load_from_parm_agg): Likewise.
17271 (ipa_node_params::~ipa_node_params): Removed.
17272 (ipa_free_all_node_params): Remove call to delete operator.
17273 (ipa_node_params_t::insert): New.
17274 (ipa_node_params_t::remove): Likewise.
17275 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
17276 copy known_csts and known_contexts vectors.
17277 (ipa_read_node_info): Adjust to descriptors being a pointer.
17278 (ipcp_modif_dom_walker): Make m_descriptors field garbage
17280 (ipcp_transform_function): Make descriptors variable garbage
17283 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
17285 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
17286 * config/i386/avx512dqintrin.h: Ditto.
17287 * config/i386/avx512fintrin.h: Ditto.
17288 * config/i386/i386.c: Handle new builtins.
17289 * config/i386/i386-builtin.def: Add new builtins.
17290 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
17291 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
17293 2017-01-23 Jakub Jelinek <jakub@redhat.com>
17294 Martin Liska <mliska@suse.cz>
17296 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
17297 * asan.c (asan_expand_poison_ifn): Support stores and use
17298 appropriate ASAN report function.
17299 * internal-fn.c (expand_ASAN_POISON_USE): New function.
17300 * internal-fn.def (ASAN_POISON_USE): Declare.
17301 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
17302 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
17303 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
17304 ASAN_POISON calls w/o LHS.
17305 * tree-ssa.c (execute_update_addresses_taken): Create clobber
17306 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
17307 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
17308 * gimplify.c (asan_poison_variables): Add attribute
17309 use_after_scope_memory to variables that really needs to live
17311 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
17312 having the attribute.
17314 2017-01-23 Martin Liska <mliska@suse.cz>
17316 * asan.c (create_asan_shadow_var): New function.
17317 (asan_expand_poison_ifn): Likewise.
17318 * asan.h (asan_expand_poison_ifn): New declaration.
17319 * internal-fn.c (expand_ASAN_POISON): Likewise.
17320 * internal-fn.def (ASAN_POISON): New builtin.
17321 * sanopt.c (pass_sanopt::execute): Expand
17322 asan_expand_poison_ifn.
17323 * tree-inline.c (copy_decl_for_dup_finish): Make function
17325 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
17326 * tree-ssa.c (is_asan_mark_p): New function.
17327 (execute_update_addresses_taken): Rewrite local variables
17328 (identified just by use-after-scope as addressable) into SSA.
17330 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
17332 * doc/install.texi (Specific): opensource.apple.com uses https
17333 now. Remove trailing slash.
17335 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
17337 * README.Portability: Remove note on an Irix compatibility issue.
17339 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
17341 * gcov.c (INCLUDE_ALGORITHM): Define.
17342 (INCLUDE_VECTOR): Define.
17343 No longer include <vector> and <algorithm> directly.
17345 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
17347 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
17349 * doc/invoke.texi (Code Gen Options): Ditto.
17351 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
17354 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
17356 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
17358 rtl-optimization/79125
17359 * cprop.c (local_cprop_pass): Handle cases where we make an
17360 unconditional trap.
17362 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
17366 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
17367 read from, for big endian.
17369 2017-01-20 Jiong Wang <jiong.wang@arm.com>
17371 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
17372 register pauth builtins for LP64 only.
17374 2017-01-20 Marek Polacek <polacek@redhat.com>
17377 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
17380 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
17382 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
17384 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
17385 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
17386 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
17388 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17391 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
17392 in big-endian mode when they are not a single duplicated value.
17394 2017-01-20 Richard Biener <rguenther@suse.de>
17396 * BASE-VER: Bump to 7.0.1.
17398 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
17400 * omp-low.c (omplow_simd_context): New struct. Use it...
17401 (lower_rec_simd_input_clauses): ...here and...
17402 (lower_rec_input_clauses): ...here to hold common data. Adjust all
17403 references to idx, lane, max_vf, is_simt.
17405 2017-01-20 Graham Markall <graham.markall@embecosm.com>
17407 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
17410 2017-01-20 Martin Jambor <mjambor@suse.cz>
17412 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
17413 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
17415 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
17416 (GTFILES): Rename hsa.c to hsa-common.c.
17417 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
17418 * hsa-dump.c: Likewise.
17419 * hsa-gen.c: Likewise.
17420 * hsa-regalloc.c: Likewise.
17421 * ipa-hsa.c: Likewise.
17422 * omp-expand.c: Likewise.
17423 * omp-low.c: Likewise.
17424 * toplev.c: Likewise.
17426 2017-01-20 Marek Polacek <polacek@redhat.com>
17429 * doc/invoke.texi: Document -Wduplicated-branches.
17430 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
17431 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
17432 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
17433 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
17434 return 0 only when not OEP_LEXICOGRAPHIC.
17435 (fold_build_cleanup_point_expr): Use the expression
17436 location when building CLEANUP_POINT_EXPR.
17437 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
17438 * tree.c (add_expr): Handle error_mark_node.
17440 2017-01-20 Martin Liska <mliska@suse.cz>
17443 * tree-profile.c (init_ic_make_global_vars): Do not call
17445 (gimple_init_gcov_profiler): Likewise.
17447 2017-01-20 Martin Liska <mliska@suse.cz>
17450 * cgraph.h (maybe_create_reference): Remove argument and
17452 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
17454 * ipa-cp.c (create_specialized_node): Likewise.
17455 * symtab.c (symtab_node::maybe_create_reference): Handle
17456 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
17458 2017-01-20 Martin Liska <mliska@suse.cz>
17460 * read-rtl-function.c (function_reader::create_function): Use
17461 build_decl instread of build_decl_stat.
17463 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
17465 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
17466 * config/i386/avx512dqintrin.h: Ditto.
17467 * config/i386/avx512fintrin.h: Ditto.
17468 * config/i386/i386-builtin-types.def: Add new types.
17469 * config/i386/i386.c: Handle new types.
17470 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
17471 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
17472 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
17473 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
17474 (__builtin_ia32_kshiftridi): New.
17475 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
17477 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
17481 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
17482 define to rs6000_init_stack_protect_guard.
17483 (rs6000_init_stack_protect_guard): New function.
17485 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
17486 Yunqiang Su <yunqiang.su@imgtec.com>
17488 * config.gcc (supported_defaults): Add madd4.
17489 (with_madd4): Add validation.
17490 (all_defaults): Add madd4.
17491 * config/mips/mips.opt (mmadd4): New option.
17492 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17494 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17496 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
17497 (ISA_HAS_FUSED_MADD4): Likewise.
17498 * doc/invoke.texi (-mmadd4): Document the new option.
17499 * doc/install.texi (--with-madd4): Document the new option.
17501 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17503 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
17504 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
17505 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
17506 (aarch64_init_pauth_hint_builtins): New.
17507 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
17508 (aarch64_expand_builtin): Expand new builtins.
17510 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17512 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
17513 * combine-stack-adj.c (no_unhandled_cfa): Handle
17514 REG_CFA_TOGGLE_RA_MANGLE.
17515 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
17516 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
17517 info for return address signing.
17518 (aarch64_expand_epilogue): Likewise.
17520 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17522 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
17523 * config/aarch64/aarch64-protos.h
17524 (aarch64_return_address_signing_enabled): New declaration.
17525 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
17527 (aarch64_expand_prologue): Sign return address before it's pushed onto
17529 (aarch64_expand_epilogue): Authenticate return address fetched from
17531 (aarch64_override_options): Sanity check for ILP32 and ISA level.
17532 (aarch64_attributes): New function attributes for "sign-return-address".
17533 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
17534 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
17535 ("*do_return"): Generate combined instructions according to key index.
17536 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
17537 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
17539 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
17540 * config/aarch64/aarch64.opt (msign-return-address=): New.
17541 * doc/extend.texi (AArch64 Function Attributes): Documents
17542 "sign-return-address=".
17543 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
17545 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
17547 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
17548 overall option summary.
17550 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17552 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
17553 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
17554 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
17555 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
17557 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
17559 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
17560 -mpower9-minmax by default for -mcpu=power9.
17561 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
17562 128-bit floating point.
17564 2017-01-20 Alan Modra <amodra@gmail.com>
17566 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
17567 optimizing for size.
17569 2017-01-20 Alan Modra <amodra@gmail.com>
17572 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
17573 for strcmp and strncmp from corresponding builtin decl.
17575 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17577 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
17578 instead of i386/rtems-64.h.
17579 * config/i386/rtems-64.h: Remove.
17581 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17585 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
17587 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
17589 2017-01-19 Tamar Christina <tamar.christina@arm.com>
17591 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
17592 Change int to HOST_WIDE_INT.
17593 * config/aarch64/aarch64-protos.h
17594 (aarch64_simd_gen_const_vector_dup): Likewise.
17595 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
17597 2017-01-19 David Malcolm <dmalcolm@redhat.com>
17599 * langhooks-def.h (lhd_type_for_size): New decl.
17600 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
17601 * langhooks.c (lhd_type_for_size): New function, taken from
17604 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
17606 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
17607 define_bypass for CR latency.
17608 (power9-cracked-alu): Update bypass latency and remove power9-branch.
17609 (power9-alu2): Add define_bypass for CR latency.
17611 (power9-mul): Update insn latency.
17612 (power9-mul-compare): Update insn latency, bypass latency and remove
17615 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17617 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
17619 * config/aarch64/aarch64.md
17620 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
17621 aarch64_nopcrelative_literal_loads.
17622 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
17624 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
17626 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
17627 TARGET_LOONGSON_3A.
17628 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
17630 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
17633 * config.gcc (supported_defaults): Add lxc1-sxc1.
17634 (with_lxc1_sxc1): Add validation.
17635 (all_defaults): Add lxc1-sxc1.
17636 * config/mips/mips.opt (mlxc1-sxc1): New option.
17637 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17639 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17640 __mips_no_lxc1_sxc1.
17641 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
17642 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
17643 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
17645 2017-01-19 Richard Biener <rguenther@suse.de>
17647 PR tree-optimization/72488
17648 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
17649 sure to restore SSA info.
17650 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
17652 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
17654 PR rtl-optimization/79121
17655 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
17656 of the inner type when shifting an extended value.
17658 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
17661 * symtab.c (symtab_node::equal_address_to): Fix comparing of
17662 interposable aliases.
17664 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
17667 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
17668 Use the evmergelohi instruction.
17669 (mov_si<mode>_e500_subreg4_2_le): Likewise.
17670 (mov_sitf_e500_subreg8_2_be): Likewise.
17671 (mov_sitf_e500_subreg12_2_le): Likewise.
17672 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
17673 (mov_si<mode>_e500_subreg4_2_be): Likewise.
17674 (mov_sitf_e500_subreg8_2_le): Likewise.
17675 (mov_sitf_e500_subreg12_2_be): Likewise.
17677 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17679 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
17680 attribute from vecsimple to vecperm.
17681 (altivec_vbpermq2): Likewise.
17683 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17686 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
17688 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17689 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
17690 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
17691 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
17692 case where N arg is SIZE_MAX.
17693 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
17694 (cmpstrsi): Add pattern.
17696 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17698 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17699 __builtin_vec_revb builtins.
17700 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
17701 built-in functions to support generation of the ISA 3.0 XXBR<x>
17702 vector byte reverse instructions.
17703 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
17704 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
17705 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
17706 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
17707 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
17708 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
17709 (P9V_BUILTIN_VEC_REVB): Likewise.
17710 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
17711 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
17712 (p9_xxbrq_v16qi): Likewise.
17713 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
17714 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
17715 (p9_xxbrh_v8hi): Likewise.
17716 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
17717 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
17718 vec_revb built-in functions.
17720 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
17722 PR rtl-optimization/78952
17723 * config/i386/i386.md (any_extract): New code iterator.
17724 (*insvqi_2): Use any_extract for source operand.
17725 (*insvqi_3): Use any_shiftrt for source operand.
17727 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
17729 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
17731 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
17733 2017-01-18 Matthias Klose <doko@ubuntu.com>
17735 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
17737 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17739 * config/rs6000/altivec.h (vec_bperm): Change #define.
17740 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
17741 (altivec_vbpermq2): New define_insn.
17742 (altivec_vbpermd): Likewise.
17743 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
17744 function interface.
17745 (VBPERMD): Likewise.
17746 (VBPERM): New polymorphic function interface.
17747 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
17748 Add entries for P9V_BUILTIN_VEC_VBPERM.
17749 * doc/extend.texi: Add interfaces for vec_bperm.
17751 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17753 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
17754 first letter of error messages.
17755 (s390_resolve_overloaded_builtin): Likewise.
17756 * config/s390/s390.c (s390_expand_builtin): Likewise.
17757 (s390_invalid_arg_for_unprototyped_fn): Likewise.
17758 (s390_valid_target_attribute_inner_p): Likewise.
17759 * config/s390/s390.md ("tabort"): Likewise.
17761 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
17763 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
17764 (ISA_AVOID_DIV_HILO): New macro.
17765 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
17766 (ISA_HAS_DDIV): Likewise.
17768 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17770 * doc/invoke.texi (fabi-version): Correct number of occurrences.
17772 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17774 * doc/invoke.texi (fabi-version): Spelling fix.
17776 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17779 * doc/invoke.texi (fabi-version): Mention mangling fix for
17782 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17785 * doc/invoke.texi (fabi-version): Document discriminator mangling.
17787 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
17790 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
17791 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
17793 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
17795 (stack_protect_test): Ditto.
17796 * config/rs6000/rs6000.opt (mstack-protector-guard=,
17797 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
17799 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
17800 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
17801 -mstack-protector-guard-offset=.
17802 (RS/6000 and PowerPC Options): Ditto.
17804 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
17806 * config/i386/i386.h (MASK_CLASS_P): New define.
17807 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
17808 there are no registers from different register sets also when
17809 mask registers are used. Update function comment.
17810 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
17811 to (*k/*r) and (*k/*km) alternatives.
17813 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
17815 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
17816 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
17817 (EH_RETURN_HANDLER_RTX): New define.
17818 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
17819 Force frame pointer in EH return functions.
17820 (aarch64_expand_epilogue): Add barrier for eh_return.
17821 (aarch64_final_eh_return_addr): Remove.
17822 (aarch64_eh_return_handler_rtx): New function.
17823 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
17825 (aarch64_eh_return_handler_rtx): New prototype.
17827 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17829 * config/rs6000/altivec.h (vec_rlmi): New #define.
17830 (vec_vrlnm): Likewise.
17831 (vec_rlnm): Likewise.
17832 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
17833 (UNSPEC_VRLNM): Likewise.
17834 (VIlong): New mode iterator.
17835 (altivec_vrl<VI_char>mi): New define_insn.
17836 (altivec_vrl<VI_char>nm): Likewise.
17837 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
17839 (VRLDNM): Likewise.
17840 (RLNM): New polymorphic function entry.
17841 (VRLWMI): New monomorphic function entry.
17842 (VRLDMI): Likewise.
17843 (RLMI): New polymorphic function entry.
17844 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
17845 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
17846 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
17849 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17852 * dwarf2out.c (field_byte_offset): Restore the
17853 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
17854 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
17855 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
17858 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17861 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
17863 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17866 * dwarf2out.c (add_data_member_location_attribute): For constant
17867 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
17868 instead of DW_AT_data_member_location, DW_AT_bit_offset and
17869 DW_AT_byte_size attributes.
17871 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17873 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
17874 after forcing to constant memory when the code model is medium.
17876 2017-01-17 Julia Koval <julia.koval@intel.com>
17879 * config/i386/avx512fintrin.h
17880 (_mm512_i32gather_ps): Change __addr type to void const*.
17881 (_mm512_mask_i32gather_ps): Ditto.
17882 (_mm512_i32gather_pd): Ditto.
17883 (_mm512_mask_i32gather_pd): Ditto.
17884 (_mm512_i64gather_ps): Ditto.
17885 (_mm512_mask_i64gather_ps): Ditto.
17886 (_mm512_i64gather_pd): Ditto.
17887 (_mm512_mask_i64gather_pd): Ditto.
17888 (_mm512_i32gather_epi32): Ditto.
17889 (_mm512_mask_i32gather_epi32): Ditto.
17890 (_mm512_i32gather_epi64): Ditto.
17891 (_mm512_mask_i32gather_epi64): Ditto.
17892 (_mm512_i64gather_epi32): Ditto.
17893 (_mm512_mask_i64gather_epi32): Ditto.
17894 (_mm512_i64gather_epi64): Ditto.
17895 (_mm512_mask_i64gather_epi64): Ditto.
17896 (_mm512_i32scatter_ps): Change __addr type to void*.
17897 (_mm512_mask_i32scatter_ps): Ditto.
17898 (_mm512_i32scatter_pd): Ditto.
17899 (_mm512_mask_i32scatter_pd): Ditto.
17900 (_mm512_i64scatter_ps): Ditto.
17901 (_mm512_mask_i64scatter_ps): Ditto.
17902 (_mm512_i64scatter_pd): Ditto.
17903 (_mm512_mask_i64scatter_pd): Ditto.
17904 (_mm512_i32scatter_epi32): Ditto.
17905 (_mm512_mask_i32scatter_epi32): Ditto.
17906 (_mm512_i32scatter_epi64): Ditto.
17907 (_mm512_mask_i32scatter_epi64): Ditto.
17908 (_mm512_i64scatter_epi32): Ditto.
17909 (_mm512_mask_i64scatter_epi32): Ditto.
17910 (_mm512_i64scatter_epi64): Ditto.
17911 (_mm512_mask_i64scatter_epi64): Ditto.
17912 * config/i386/avx512pfintrin.h
17913 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
17914 (_mm512_mask_prefetch_i32gather_ps): Ditto.
17915 (_mm512_mask_prefetch_i64gather_pd): Ditto.
17916 (_mm512_mask_prefetch_i64gather_ps): Ditto.
17917 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
17918 (_mm512_prefetch_i32scatter_ps): Ditto.
17919 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
17920 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
17921 (_mm512_prefetch_i64scatter_pd): Ditto.
17922 (_mm512_prefetch_i64scatter_ps): Ditto.
17923 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
17924 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
17925 * config/i386/avx512vlintrin.h
17926 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
17927 (_mm_mmask_i32gather_ps): Ditto.
17928 (_mm256_mmask_i32gather_pd): Ditto.
17929 (_mm_mmask_i32gather_pd): Ditto.
17930 (_mm256_mmask_i64gather_ps): Ditto.
17931 (_mm_mmask_i64gather_ps): Ditto.
17932 (_mm256_mmask_i64gather_pd): Ditto.
17933 (_mm_mmask_i64gather_pd): Ditto.
17934 (_mm256_mmask_i32gather_epi32): Ditto.
17935 (_mm_mmask_i32gather_epi32): Ditto.
17936 (_mm256_mmask_i32gather_epi64): Ditto.
17937 (_mm_mmask_i32gather_epi64): Ditto.
17938 (_mm256_mmask_i64gather_epi32): Ditto.
17939 (_mm_mmask_i64gather_epi32): Ditto.
17940 (_mm256_mmask_i64gather_epi64): Ditto.
17941 (_mm_mmask_i64gather_epi64): Ditto.
17942 (_mm256_i32scatter_ps): Change __addr type to void*.
17943 (_mm256_mask_i32scatter_ps): Ditto.
17944 (_mm_i32scatter_ps): Ditto.
17945 (_mm_mask_i32scatter_ps): Ditto.
17946 (_mm256_i32scatter_pd): Ditto.
17947 (_mm256_mask_i32scatter_pd): Ditto.
17948 (_mm_i32scatter_pd): Ditto.
17949 (_mm_mask_i32scatter_pd): Ditto.
17950 (_mm256_i64scatter_ps): Ditto.
17951 (_mm256_mask_i64scatter_ps): Ditto.
17952 (_mm_i64scatter_ps): Ditto.
17953 (_mm_mask_i64scatter_ps): Ditto.
17954 (_mm256_i64scatter_pd): Ditto.
17955 (_mm256_mask_i64scatter_pd): Ditto.
17956 (_mm_i64scatter_pd): Ditto.
17957 (_mm_mask_i64scatter_pd): Ditto.
17958 (_mm256_i32scatter_epi32): Ditto.
17959 (_mm256_mask_i32scatter_epi32): Ditto.
17960 (_mm_i32scatter_epi32): Ditto.
17961 (_mm_mask_i32scatter_epi32): Ditto.
17962 (_mm256_i32scatter_epi64): Ditto.
17963 (_mm256_mask_i32scatter_epi64): Ditto.
17964 (_mm_i32scatter_epi64): Ditto.
17965 (_mm_mask_i32scatter_epi64): Ditto.
17966 (_mm256_i64scatter_epi32): Ditto.
17967 (_mm256_mask_i64scatter_epi32): Ditto.
17968 (_mm_i64scatter_epi32): Ditto.
17969 (_mm_mask_i64scatter_epi32): Ditto.
17970 (_mm256_i64scatter_epi64): Ditto.
17971 (_mm256_mask_i64scatter_epi64): Ditto.
17972 (_mm_i64scatter_epi64): Ditto.
17973 (_mm_mask_i64scatter_epi64): Ditto.
17974 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
17975 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
17976 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
17977 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
17978 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
17979 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
17980 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
17981 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
17982 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
17983 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
17984 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
17985 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
17986 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
17987 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
17988 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
17989 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
17990 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
17991 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
17992 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
17993 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
17994 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
17995 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
17996 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
17997 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
17998 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
17999 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
18000 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
18001 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
18002 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
18003 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
18004 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
18005 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
18006 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
18007 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
18008 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
18009 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
18010 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
18011 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
18012 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
18013 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
18014 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
18015 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
18016 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
18017 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
18018 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
18019 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
18020 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
18021 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
18022 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
18023 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
18024 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
18025 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
18026 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
18027 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
18028 definitions accordingly.
18030 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
18031 Kuan-Lin Chen <kuanlinchentw@gmail.com>
18034 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
18037 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
18040 * ira-conflicts.c (ira_build_conflicts): Update total conflict
18041 hard regs for inner regno.
18043 2017-01-17 Martin Liska <mliska@suse.cz>
18046 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
18047 assumption and add comment.
18049 2017-01-17 Nathan Sidwell <nathan@acm.org>
18051 * ipa-visibility.c (localize_node): New function, broken out of ...
18052 (function_and_variable_visibility): ... here. Call it.
18054 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
18056 PR middle-end/77445
18057 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
18058 correctly set frequency of oudgoing edge.
18059 (duplicate_thread_path): Fix profile updating.
18061 2017-01-17 Jakub Jelinek <jakub@redhat.com>
18064 * configure.ac: Add GCC_BASE_VER.
18065 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
18066 version from BASE-VER file.
18067 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
18068 (gcc.o): Depend on $(BASEVER).
18069 * common.opt (dumpfullversion): New option.
18070 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
18071 * doc/invoke.texi: Document -dumpfullversion.
18072 * doc/install.texi: Document --with-gcc-major-version-only.
18073 * configure: Regenerated.
18075 2017-01-17 Richard Biener <rguenther@suse.de>
18077 PR tree-optimization/71433
18078 * tree-vrp.c (register_new_assert_for): Merge same asserts
18079 on all incoming edges.
18080 (process_assert_insertions_for): Handle insertions at the
18083 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
18085 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
18086 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
18088 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
18091 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
18094 2017-01-17 Alan Modra <amodra@gmail.com>
18097 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
18098 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
18099 symbolic stack limit when pic.
18101 2017-01-16 Martin Sebor <msebor@redhat.com>
18103 PR tree-optimization/78608
18104 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
18106 2017-01-16 Jeff Law <law@redhat.com>
18109 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
18110 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
18111 for several include directories that may be relative to sysroot.
18112 * config/i386/x-mingw32 (gplus_includedir): Define.
18113 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
18114 (native_system_includedir): Likewise.
18115 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
18116 override if TARGET_SYSTEM_ROOT is defined.
18117 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
18119 PR tree-optimization/79090
18120 PR tree-optimization/33562
18121 PR tree-optimization/61912
18122 PR tree-optimization/77485
18123 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
18124 and computed trims into the dump file.
18126 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
18128 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
18130 2017-01-16 Jakub Jelinek <jakub@redhat.com>
18133 * gimplify.c (gimplify_init_constructor): If want_value and
18134 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
18138 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
18139 sequence. Formatting fixes.
18140 (doloop_optimize): Formatting fixes.
18143 * gcc.c (debug_level_greater_than_spec_func): New function.
18144 (static_spec_functions): Add debug-level-gt spec function.
18145 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
18147 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
18148 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
18149 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
18150 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
18151 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
18152 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
18154 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
18156 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
18157 QImode fixups to general and mask registers only.
18159 2017-01-16 Carl Love <cel@us.ibm.com>
18161 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
18162 for built-in functions
18163 vector signed char vec_nabs (vector signed char)
18164 vector signed short vec_nabs (vector signed short)
18165 vector signed int vec_nabs (vector signed int)
18166 vector signed long long vec_nabs (vector signed long long)
18167 vector float vec_nabs (vector float)
18168 vector double vec_nabs (vector double)
18169 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
18171 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
18172 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
18173 * doc/extend.texi: Update the documentation file for the new built-in
18176 2017-01-16 Martin Sebor <msebor@redhat.com>
18178 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
18181 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18183 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
18184 UNSPEC_VSX__XXSPLTD to require special splat handling.
18186 2017-01-16 David Malcolm <dmalcolm@redhat.com>
18189 * system.h: Poison strndup.
18191 2017-01-16 Alan Modra <amodra@gmail.com>
18194 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
18197 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
18199 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
18201 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
18203 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
18204 call recog here. Assert that INSN_CODE (insn) is non-negative.
18206 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
18209 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
18211 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
18212 in the currently scheduled RTL fragment.
18214 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
18216 PR rtl-optimization/78751
18217 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
18220 2017-01-14 Jeff Law <law@redhat.com>
18222 PR tree-optimization/79090
18223 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
18224 variable length stores.
18225 (compute_trims): Delete dead assignment to *trim_tail.
18226 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
18229 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
18231 PR rtl-optimization/78626
18232 PR rtl-optimization/78727
18233 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
18234 of a block, and split such blocks after everything else is finished.
18236 2017-01-14 Alan Modra <amodra@gmail.com>
18239 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
18240 target legitimate_combined_insn.
18241 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
18242 (rs6000_legitimate_combined_insn): New function.
18243 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
18245 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
18246 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
18247 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
18249 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
18251 * doc/frontends.texi (G++ and GCC): Remove references to Java.
18253 2017-01-13 Jeff Law <law@redhat.com>
18255 PR tree-optimization/33562
18256 PR tree-optimization/61912
18257 PR tree-optimization/77485
18258 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
18260 (delete_dead_assignment): Likewise.
18261 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
18262 statement to delete_dead_call and delete_dead_assignment.
18264 2017-01-13 David Malcolm <dmalcolm@redhat.com>
18267 * substring-locations.c (format_warning_va): Strengthen case 1 so
18268 that both endpoints of the substring must be within the format
18269 range for just the substring to be printed.
18271 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
18273 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
18274 * config/i386/i386.c (ix86_target_string): Add missing options
18275 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
18276 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
18277 flags_other and ix86_target_other to flags2_other. Display unknown
18279 (ix86_valid_target_attribute_inner_p): Add missing options and
18280 reorder options by implied ISAs, as in ix86_target_string.
18282 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
18284 * hash-table.h (hash_table::too_empty_p): New function.
18285 (hash_table::expand): Use it.
18286 (hash_table::traverse): Likewise.
18287 (hash_table::empty_slot): Use sizeof (value_type) instead of
18288 sizeof (PTR) to convert bytes to elements. Shrink the table
18289 if the current size is excessive for the current number of
18292 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
18294 * ira-costs.c (record_reg_classes): Break from the inner loop
18295 early once alt_fail is known to be true. Update outer loop
18296 handling accordingly.
18298 2017-01-13 Jeff Law <law@redhat.com>
18300 * tree-ssa-dse.c (decrement_count): New function.
18301 (increment_start_addr, maybe_trim_memstar_call): Likewise.
18302 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
18303 when we know the partially dead statement is a mem* function.
18305 PR tree-optimization/61912
18306 PR tree-optimization/77485
18307 * tree-ssa-dse.c: Include expr.h.
18308 (maybe_trim_constructor_store): New function.
18309 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
18311 PR tree-optimization/33562
18312 PR tree-optimization/61912
18313 PR tree-optimization/77485
18314 * doc/invoke.texi: Document new dse-max-object-size param.
18315 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
18316 * tree-ssa-dse.c: Include params.h.
18317 (dse_store_status): New enum.
18318 (initialize_ao_ref_for_dse): New, partially extracted from
18320 (valid_ao_ref_for_dse, normalize_ref): New.
18321 (setup_live_bytes_from_ref, compute_trims): Likewise.
18322 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
18323 (maybe_trim_partially_dead_store): Likewise.
18324 (maybe_trim_complex_store): Likewise.
18325 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
18326 Track what bytes live from the original store. Return tri-state
18327 for dead, partially dead or live.
18328 (dse_dom_walker): Add constructor, destructor and new private members.
18329 (delete_dead_call, delete_dead_assignment): New extracted from
18331 (dse_optimize_stmt): Make a member of dse_dom_walker.
18332 Use initialize_ao_ref_for_dse.
18334 PR tree-optimization/33562
18335 PR tree-optimization/61912
18336 PR tree-optimization/77485
18337 * sbitmap.h (bitmap_count_bits): Prototype.
18338 (bitmap_clear_range, bitmap_set_range): Likewise.
18339 * sbitmap.c (bitmap_clear_range): New function.
18340 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
18342 2017-01-13 Martin Liska <mliska@suse.cz>
18345 * function.c (set_cfun): Add new argument force.
18346 * function.h (set_cfun): Likewise.
18347 * ipa-inline-transform.c (inline_call): Use the function when
18348 strict alising from is dropped for function we inline to.
18350 2017-01-13 Richard Biener <rguenther@suse.de>
18352 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
18353 for dumping GIMPLE INTEGER_CSTs.
18355 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18357 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
18358 to 201112L since C++17.
18360 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
18363 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
18364 if -fsanitize=kernel-address is present.
18366 2017-01-13 Richard Biener <rguenther@suse.de>
18368 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
18369 as _Literal ( type ) number in case usual suffixes do not
18370 preserve all information.
18372 2017-01-13 Richard Biener <rguenther@suse.de>
18374 PR tree-optimization/77283
18375 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
18376 and ssa-iterators.h.
18377 (is_feasible_trace): Implement a cost model based on joiner
18380 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
18383 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
18384 char or short to __float128/_Float128 directly.
18386 2017-01-12 Martin Sebor <msebor@redhat.com>
18388 to -Wformat-overflow.
18389 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
18390 (min_bytes_remaining): Same.
18391 (get_string_length): Same.
18392 (format_string): Same.
18393 (format_directive): Same.
18395 (pass_sprintf_length::handle_gimple_call): Same.
18397 2017-01-12 Jakub Jelinek <jakub@redhat.com>
18399 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
18400 info.nowrite calls with no lhs that can't throw. Return bool
18401 whether gsi_remove has been called or not.
18402 (pass_sprintf_length::handle_gimple_call): Return bool whether
18403 try_substitute_return_value called gsi_remove. Formatting fix.
18404 (pass_sprintf_length::execute): Don't use gsi_remove if
18405 handle_gimple_call returned true.
18408 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
18409 be removed due to side-effects, don't remove following barrier nor
18410 turn the successor edge into fallthru edge.
18412 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18415 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
18416 element-reversing loads and stores as not swappable.
18418 2017-01-12 Nathan Sidwell <nathan@acm.org>
18419 Nicolai Stange <nicstange@gmail.com>
18421 * combine.c (try_combine): Don't ignore result of overlap checking
18422 loop. Combine overlap & asm check into single loop.
18424 2017-01-12 Richard Biener <rguenther@suse.de>
18426 * tree-pretty-print.c (dump_generic_node): Provide -gimple
18427 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
18429 2017-01-12 Richard Biener <rguenther@suse.de>
18431 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
18432 and TS_TARGET_OPTION directly derive from TS_BASE.
18433 * tree-core.h (tree_optimization_option): Derive from tree_base.
18434 (tree_target_option): Likewise.
18436 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
18438 * config/i386/i386.c (memory_address_length): Increase len
18439 only when rip_relative_addr_p returns false.
18441 2017-01-11 Julia Koval <julia.koval@intel.com>
18443 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
18444 (OPTION_MASK_ISA_SGX_SET): New.
18445 (ix86_handle_option): Handle OPT_msgx.
18446 * config.gcc: Added sgxintrin.h.
18447 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
18448 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
18449 * config/i386/i386.c (ix86_target_string): Add -msgx.
18451 (ix86_option_override_internal): Handle new options.
18452 (ix86_valid_target_attribute_inner_p): Add sgx.
18453 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
18454 * config/i386/i386.opt: Add msgx.
18455 * config/i386/sgxintrin.h: New file.
18456 * config/i386/x86intrin.h: Add sgxintrin.h.
18458 2017-01-11 Jakub Jelinek <jakub@redhat.com>
18461 * fold-const.c (maybe_nonzero_address): Return 1 for function
18463 (tree_single_nonzero_warnv_p): Don't handle function local objects
18467 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
18470 2017-01-11 David Malcolm <dmalcolm@redhat.com>
18473 * opts.c: Include "spellcheck.h"
18474 (struct string_fragment): New struct.
18475 (struct edit_distance_traits<const string_fragment &>): New
18477 (get_closest_sanitizer_option): New function.
18478 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
18480 2017-01-11 Jakub Jelinek <jakub@redhat.com>
18482 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
18484 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
18485 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
18486 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
18487 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
18488 for initial die_offset if dwarf_split_debug_info.
18489 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
18490 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
18492 (output_skeleton_debug_sections): Formatting fix. Use
18493 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
18494 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
18496 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
18498 * config/arm/cortex-a53.md: Add bypasses for
18499 cortex_a53_r2f_cvt.
18500 (cortex_a53_r2f): Only use for transfers.
18501 (cortex_a53_f2r): Likewise.
18502 (cortex_a53_r2f_cvt): Add reservation for conversions.
18503 (cortex_a53_f2r_cvt): Likewise.
18505 2017-01-11 Tamar Christina <tamar.christina@arm.com>
18507 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
18508 to all inlined functions, change static to extern.
18510 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
18513 * config/arm/arm.c (legitimize_pic_address): Handle reference to
18515 (arm_assemble_integer): Likewise.
18517 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
18519 * config.gcc: Use new awk script to check CPU, FPU and architecture
18520 parameters for --with-... options.
18521 * config/arm/parsecpu.awk: New file
18522 * config/arm/arm-cpus.in: New file.
18523 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
18525 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
18527 * config/arm/t-arm: Update dependency rules.
18528 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
18529 of processing .def files.
18530 * config/arm/genopt.sh: Deleted.
18531 * config/arm/gentune.sh: Deleted.
18532 * config/arm/arm-cores.def: Deleted.
18533 * config/arm/arm-arches.def: Deleted.
18534 * config/arm/arm-fpus.def: Deleted.
18535 * config/arm/arm-tune.md: Regenerated.
18536 * config/arm/arm-tables.opt: Regenerated.
18537 * config/arm/arm-cpu.h: New generated file.
18538 * config/arm/arm-cpu-data.h: New generated file.
18539 * config/arm/arm-cpu-cdata.h: New generated file.
18541 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
18544 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
18546 (input_varpool_node): Unpack dynamically_initialized bit.
18548 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
18550 PR rtl-optimization/79032
18551 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
18552 the alignment of the adjusted memory reference against that of MODE,
18553 instead of the alignment of the original memory reference.
18555 2017-01-11 Martin Jambor <mjambor@suse.cz>
18557 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
18559 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
18560 decorated functions.
18562 2017-01-11 Richard Biener <rguenther@suse.de>
18564 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
18565 set range/nonnull info for PHI results. Do not set it on
18566 stmts marked for removal.
18568 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
18570 * expr.c (store_field): In the bitfield case, fetch the return value
18571 from the registers before applying a single big-endian adjustment.
18572 Always do a final load for a BLKmode value not larger than a word.
18574 2017-01-10 David Malcolm <dmalcolm@redhat.com>
18577 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18578 that we correctly handle column numbers greater than
18579 LINE_MAP_MAX_COLUMN_NUMBER.
18581 2017-01-10 Martin Sebor <msebor@redhat.com>
18583 PR middle-end/78245
18584 * gimple-ssa-sprintf.c (get_destination_size): Call
18585 {init,fini}object_sizes.
18586 * tree-object-size.c (addr_object_size): Adjust.
18587 (pass_through_call): Adjust.
18588 (pass_object_sizes::execute): Adjust.
18589 * tree-object-size.h (fini_object_sizes): Declare.
18591 2017-01-10 Martin Sebor <msebor@redhat.com>
18593 PR tree-optimization/78775
18594 * builtins.c (get_size_range): Move...
18595 * calls.c: ...to here.
18596 (alloc_max_size): Accept zero argument.
18597 (operand_signed_p): Remove.
18598 (maybe_warn_alloc_args_overflow): Call get_size_range.
18599 * calls.h (get_size_range): Declare.
18601 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
18603 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
18604 from TI's devices.csv file as of September 2016.
18605 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
18607 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
18609 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
18610 * doc/invoke.texi: Likewise.
18611 * doc/md.texi: Likewise.
18612 * doc/objc.texi: Likewise.
18614 2017-01-10 Joshua Conner <joshconner@google.com>
18616 * config/arm/fuchsia-elf.h: New file.
18617 * config/fuchsia.h: New file.
18618 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
18619 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
18621 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
18623 2016-01-10 Richard Biener <rguenther@suse.de>
18625 PR tree-optimization/79034
18626 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18627 Propagate out degenerate PHIs in the joiner.
18629 2017-01-10 Martin Liska <mliska@suse.cz>
18631 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
18632 (sort_congruence_classes_by_decl_uid): Likewise.
18633 (sort_congruence_class_groups_by_decl_uid): Likewise.
18634 (sem_item_optimizer::merge_classes): Sort class, groups in these
18635 classes and members in the groups by DECL_UID of declarations.
18636 This would make merge operations stable.
18638 2017-01-10 Martin Liska <mliska@suse.cz>
18640 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
18641 usage of m_classes_vec.
18642 (sem_item_optimizer::~sem_item_optimizer): Likewise.
18643 (sem_item_optimizer::get_group_by_hash): Likewise.
18644 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
18645 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
18646 (sem_item_optimizer::verify_classes): Likewise.
18647 (sem_item_optimizer::process_cong_reduction): Likewise.
18648 (sem_item_optimizer::dump_cong_classes): Likewise.
18649 (sem_item_optimizer::merge_classes): Likewise.
18650 * ipa-icf.h (congruence_class_hash): Rename from
18651 congruence_class_group_hash. Remove declaration of m_classes_vec.
18653 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
18655 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
18656 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
18657 * config.gcc: Add avx512vpopcntdqintrin.h.
18658 * config/i386/avx512vpopcntdqintrin.h: New.
18659 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
18660 * config/i386/i386-builtin-types.def: Add new types.
18661 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
18662 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
18663 __builtin_ia32_vpopcountq_v8di_mask): New.
18664 * config/i386/i386-c.c (ix86_target_macros_internal): Define
18665 __AVX512VPOPCNTDQ__.
18666 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
18667 (PTA_AVX512VPOPCNTDQ): Define.
18668 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
18669 TARGET_AVX512VPOPCNTDQ_P): Define.
18670 * config/i386/i386.opt: Add mavx512vpopcntdq.
18671 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
18672 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
18674 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18676 PR middle-end/77484
18677 * predict.def (PRED_CALL): Set to 67.
18679 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
18681 * expr.c (store_field): In the bitfield case, if the value comes from
18682 a function call and is of an aggregate type returned in registers, do
18683 not modify the field mode; extract the value in all cases if the mode
18684 is BLKmode and the size is not larger than a word.
18686 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
18689 * config/i386/cpuid.h: Fix undefined behavior.
18691 2017-01-04 Jeff Law <law@redhat.com>
18693 PR tree-optimization/79007
18694 PR tree-optimization/67955
18695 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
18696 conservative for pt.null when flag_non_call_exceptions is on.
18698 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18700 PR translation/79019
18701 PR translation/79020
18702 * params.def (PARAM_INLINE_MIN_SPEEDUP,
18703 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
18704 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
18706 * config/avr/avr.opt (maccumulate-args): Likewise.
18707 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
18708 * common.opt (freport-bug): Likewise.
18709 * cif-code.def (CIF_FINAL_ERROR): Likewise.
18710 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
18711 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
18712 translatable string.
18713 * config/i386/i386.c (function_value_32): Likewise.
18714 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
18715 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
18717 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
18718 * common/config/msp430/msp430-common.c (msp430_handle_option):
18720 * symtab.c (symtab_node::verify_base): Likewise.
18721 * opts.c (set_debug_level): Likewise.
18722 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
18723 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
18724 missing whitespace to translatable strings.
18725 * config/avr/avr.md (bswapsi2): Fix typo in comment.
18726 * config/sh/superh.h: Likewise.
18727 * config/i386/xopintrin.h: Likewise.
18728 * config/i386/znver1.md: Likewise.
18729 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
18730 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
18731 * double-int.h (struct double_int): Likewise.
18732 * double-int.c (div_and_round_double): Likewise.
18733 * wide-int.cc: Likewise.
18734 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
18735 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
18736 * cfgcleanup.c (crossjumps_occured): Renamed to ...
18737 (crossjumps_occurred): ... this.
18738 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
18741 PR tree-optimization/78899
18742 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
18743 returning bool return struct loop *, NULL for failure and the new
18745 (versionable_outer_loop_p): Don't version outer loop if it has
18746 dont_vectorized bit set.
18747 (tree_if_conversion): When versioning outer loop, ensure
18748 tree_if_conversion is performed also on the inner loop of the
18749 non-vectorizable outer loop copy.
18750 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
18751 LOOP_VECTORIZED in inner loop of the scalar outer loop and
18752 prevent vectorization of it.
18753 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
18754 the outer loop vectorization of the non-scalar version is attempted
18755 before vectorization of the inner loop in scalar version. If
18756 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
18757 vectorization of its inner loop.
18758 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
18759 has 2 inner loops, rename also on edges from bb whose single pred
18760 is outer_loop->header. Fix typo in function comment.
18762 2017-01-09 Martin Sebor <msebor@redhat.com>
18765 * asan.c (asan_emit_stack_protection): Increase local buffer size
18766 to avoid snprintf truncation warning.
18768 2017-01-09 Andrew Pinski <apinski@cavium.com>
18770 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
18771 to reference thunderx2t99 for the tuning structure
18772 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
18774 (thunderx2t99_extra_costs): This.
18775 * config/aarch64/aarch64-tune.md: Regenerate.
18776 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
18777 (vulcan_addrcost_table): This.
18778 (vulcan_regmove_cost): Rename to ...
18779 (thunderx2t99_regmove_cost): This.
18780 (vulcan_vector_cost): Rename to ...
18781 (thunderx2t99_vector_cost): this.
18782 (vulcan_branch_cost): Rename to ...
18783 (thunderx2t99_branch_cost): This.
18784 (vulcan_tunings): Rename to ...
18785 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
18786 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
18788 2017-01-09 Martin Jambor <mjambor@suse.cz>
18792 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
18793 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
18794 (propagate_vr_accross_jump_function): Use the above function for all
18795 value range computations for pass-through jump functions and type
18796 converasion from explicit value range values.
18797 (ipcp_propagate_stage): Do not attempt to deduce types of formal
18798 parameters from TYPE_ARG_TYPES.
18799 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
18800 (ipa_write_node_info): Stream type of the actual argument.
18801 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
18803 2017-01-09 Martin Liska <mliska@suse.cz>
18806 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
18807 (lookup_compiler): Do not show error message with have_E.
18809 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18811 PR tree-optimization/78938
18812 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
18813 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
18814 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
18815 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
18818 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18820 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
18823 2017-01-09 Richard Biener <rguenther@suse.de>
18825 PR tree-optimization/78997
18826 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
18827 name condition properly.
18829 2017-01-09 Richard Biener <rguenther@suse.de>
18832 * dwarf2out.c (is_cxx): New overload with context.
18833 (is_naming_typedef_decl): Use it.
18835 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18837 * invoke.texi (Option Summary): Correct spacing in option lists
18838 and add line breaks to fix over-long lines.
18840 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18842 PR middle-end/17660
18844 * extend.texi (Common Variable Attributes): Add xref to GCC
18845 Internals manual to explain mode attribute keywords.
18847 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18850 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
18851 and Preprocessor Options.
18852 (Options for Linking): Document -pthread here....
18853 (RS/6000 and PowerPC Options): ...not here.
18854 (Solaris 2 Options): ...or here.
18855 * doc/cppopts.texi: Document -pthread.
18857 2017-01-08 Martin Sebor <msebor@redhat.com>
18859 PR middle-end/77708
18860 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
18861 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
18862 New member functions.
18863 (format_directive): Used them.
18865 (pass_sprintf_length::handle_gimple_call): Same.
18866 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
18867 to avoid truncation for any argument.
18868 (extract_affine_mul): Same.
18869 * tree.c (get_file_function_name): Same.
18871 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18873 PR middle-end/77484
18874 * predict.def (PRED_INDIR_CALL): Set to 86.
18876 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18878 PR preprocessor/54124
18879 * doc/cppopts.texi: Reformat -d subtable to list the full name
18880 of the options. Add cross-reference to the docs for the general
18881 compiler -d options.
18882 * doc/invoke.texi (Developer Options): Add cross-reference to the
18883 preprocessor-specific -d option documentation.
18885 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18887 PR preprocessor/13498
18888 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
18889 redudant material, and reflect new command-line options.
18890 (System Headers): Likewise.
18892 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18894 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
18895 -isystem, and -idirafter. Copy-edit.
18896 * doc/cppopts.texi: Copy-edit. Remove contradiction about
18897 default for -ftrack-macro-expansion. Delete obsolete and
18898 badly-formatted implementation details about -fdebug-cpp output.
18899 * doc/cppwarnopts.texi: Copy-edit.
18901 2017-01-07 David Malcolm <dmalcolm@redhat.com>
18904 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18905 that the transition from a max line width >= 1<<10 to narrower
18906 lines works correctly.
18908 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
18910 * doc/options.texi (PerFunction): New.
18911 * opt-functions.awk (switch_flags): Map both Optimization and
18912 PerFunction to CL_OPTIMIZATION.
18913 * opth-gen.awk: Test for PerFunction flag along with
18915 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
18916 it only when the latter is present. Skip those that don't in
18917 the hash function generator.
18918 * common.opt (fvar-tracking): Mark as PerFunction instead of
18920 (fvar-tracking-assignments): Likewise.
18921 (fvar-tracking-assignments-toggle): Likewise.
18922 (fvar-tracking-uninit): Likewise.
18924 2017-01-07 Jakub Jelinek <jakub@redhat.com>
18926 PR translation/79018
18927 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
18930 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
18933 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
18934 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
18935 TARGET_LONG_JUMP_TABLE_OFFSETS.
18936 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
18937 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
18938 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
18939 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
18940 * config/m68k/m68k.md (tablejump expander): Likewise.
18941 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
18942 TARGET_LONG_JUMP_TABLE_OFFSETS.
18943 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
18944 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
18946 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
18947 David Holsgrove <david.holsgrove@xilinx.com>
18949 * common/config/microblaze/microblaze-common.c
18950 (TARGET_EXCEPT_UNWIND_INFO): Remove.
18951 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
18953 * config/microblaze/microblaze.c (microblaze_must_save_register)
18954 (microblaze_expand_epilogue, microblaze_return_addr): Handle
18956 (microblaze_eh_return): New function.
18957 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
18958 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
18959 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
18960 * config/microblaze/microblaze.md (eh_return): New pattern.
18962 2017-01-06 Jakub Jelinek <jakub@redhat.com>
18964 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
18965 GCC_DIAGNOSTIC_STRINGIFY): Define.
18967 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
18969 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18971 * config/arm/arm.md (<mcrr>): New.
18973 * config/arm/arm.c (arm_arch5te): New.
18974 (arm_option_override): Set arm_arch5te.
18975 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
18977 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
18978 (arm_mcrr_qualifiers): ... this. New.
18979 (MRRC_QUALIFIERS): Define to...
18980 (arm_mrrc_qualifiers): ... this. New.
18981 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
18983 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
18984 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
18985 (MRRCI, mrrc, MRRC): New.
18986 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
18987 VUNSPEC_MRRC2): New.
18989 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18991 * config/arm/arm.md (<mcr>): New.
18993 * config/arm/arm.c (arm_coproc_builtin_available): Add
18994 support for mcr, mrc, mcr2 and mrc2.
18995 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
18996 (arm_mcr_qualifiers): ... this. New.
18997 (MRC_QUALIFIERS): Define to ...
18998 (arm_mrc_qualifiers): ... this. New.
18999 (MCR_QUALIFIERS): Define to ...
19000 (arm_mcr_qualifiers): ... this. New.
19001 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
19003 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
19004 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
19005 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
19006 VUNSPEC_MRC2): New.
19008 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19010 * config/arm/arm.md (*ldc): New.
19014 * config/arm/arm.c (arm_coproc_builtin_available): Add
19015 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
19016 (arm_coproc_ldc_stc_legitimate_address): New.
19017 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
19018 'qualifier_const_pointer'.
19019 (LDC_QUALIFIERS): Define to...
19020 (arm_ldc_qualifiers): ... this. New.
19021 (STC_QUALIFIERS): Define to...
19022 (arm_stc_qualifiers): ... this. New.
19023 * config/arm/arm-protos.h
19024 (arm_coproc_ldc_stc_legitimate_address): New.
19025 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
19026 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
19027 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
19028 stc2, stcl, stc2l): New.
19029 * config/arm/constraints.md (Uz): New.
19030 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
19031 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
19032 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
19033 VUNSPEC_STC2L): New.
19035 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19037 * config/arm/arm.md (<cdp>): New.
19038 * config/arm/arm.c (neon_const_bounds): Rename this ...
19039 (arm_const_bounds): ... this.
19040 (arm_coproc_builtin_available): New.
19041 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
19042 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
19043 (CDP_QUALIFIERS): Define to...
19044 (arm_cdp_qualifiers): ... this. New.
19046 (arm_expand_builtin_args): Add case for 6 arguments.
19047 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
19048 (arm_const_bounds): ... this.
19049 (arm_coproc_builtin_available): New.
19050 * config/arm/arm_acle.h (__arm_cdp): New.
19052 * config/arm/arm_acle_builtins.def (cdp): New.
19054 * config/arm/iterators.md (CDPI,CDP,cdp): New.
19055 * config/arm/neon.md: Rename all 'neon_const_bounds' to
19056 'arm_const_bounds'.
19057 * config/arm/types.md (coproc): New.
19058 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
19059 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
19060 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
19061 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
19063 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19065 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
19066 (UBINOP_QUALIFIERS): New.
19068 (acle_builtin_data): New. Change comment.
19069 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
19070 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
19071 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
19072 arm_acle_builtins.def.
19073 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
19074 (arm_init_acle_builtins): New.
19075 (CRC32_BUILTIN): Remove.
19076 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
19077 crc32cb, crc32ch and crc32cw.
19078 (arm_init_crc32_builtins): Remove.
19079 (arm_init_builtins): Use arm_init_acle_builtins rather
19080 than arm_init_crc32_builtins.
19081 (arm_expand_acle_builtin): New.
19082 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
19083 * config/arm/arm_acle_builtins.def: New.
19085 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19087 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
19088 (arm_builtin_datum): ... this.
19089 (arm_init_neon_builtin): Rename to ...
19090 (arm_init_builtin): ... this. Add a new parameters PREFIX
19091 and USE_SIG_IN_NAME.
19092 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
19093 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
19094 'arm_builtin_datum'.
19095 (arm_init_vfp_builtins): Likewise.
19096 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
19097 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
19098 (arm_expand_neon_args): Rename to ...
19099 (arm_expand_builtin_args): ... this. Rename builtin_arg
19100 enum values and differentiate between ARG_BUILTIN_MEMORY
19101 and ARG_BUILTIN_NEON_MEMORY.
19102 (arm_expand_neon_builtin_1): Rename to ...
19103 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
19104 values, arm_expand_builtin_args and add bool parameter NEON.
19105 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
19106 (arm_expand_vfp_builtin): Likewise.
19107 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
19109 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19111 PR middle-end/77484
19112 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
19113 * predict.c (tree_estimate_probability_bb): Reverse direction of
19114 polymorphic call predictor.
19116 2017-01-06 David Malcolm <dmalcolm@redhat.com>
19118 * passes.c (execute_one_pass): Split out pass-skipping logic into...
19119 (determine_pass_name_match): ...this new function and...
19120 (should_skip_pass_p): ...this new function.
19122 2017-01-06 Nathan Sidwell <nathan@acm.org>
19124 * ipa-visibility.c (function_and_variable_visibility): Reformat
19125 comments and long lines. Remove extrneous if.
19126 * symtab.c (symtab_node::make_decl_local): Fix code format.
19127 (symtab_node::set_section_for_node): Fix comment typo.
19129 2017-01-06 Martin Liska <mliska@suse.cz>
19132 * lra-constraints.c: Rename invariant to lra_invariant.
19133 * predict.c (set_even_probabilities): Initialize e to NULL.
19135 2017-01-05 Martin Sebor <msebor@redhat.com>
19137 PR tree-optimization/78910
19138 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
19139 (format_integer): Correct off-by-one error in the handling
19140 of precision with negative numbers in signed conversions..
19142 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
19144 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
19146 2017-01-05 Jakub Jelinek <jakub@redhat.com>
19148 PR tree-optimization/71016
19149 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
19150 factor_out_conditional_conversion. Formatting fix.
19151 (factor_out_conditional_conversion): Add cond_stmt argument.
19152 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
19153 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
19156 2017-01-05 David Malcolm <dmalcolm@redhat.com>
19158 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
19159 read-rtl-function.o, and selftest-rtl.o.
19160 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
19161 (selftest::aarch64_test_loading_full_dump): New function.
19162 (selftest::aarch64_run_selftests): New function.
19163 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
19164 selftest::aarch64_run_selftests.
19165 * config/i386/i386.c
19166 (selftest::ix86_test_loading_dump_fragment_1): New function.
19167 (selftest::ix86_test_loading_call_insn): New function.
19168 (selftest::ix86_test_loading_full_dump): New function.
19169 (selftest::ix86_test_loading_unspec): New function.
19170 (selftest::ix86_run_selftests): Call the new functions.
19171 * emit-rtl.c (maybe_set_max_label_num): New function.
19172 * emit-rtl.h (maybe_set_max_label_num): New decl.
19173 * function.c (instantiate_decls): Guard call to
19174 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
19175 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
19177 * gensupport.c (gen_reader::gen_reader): Pass "false"
19178 for new "compact" param of rtx_reader.
19179 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
19180 rather than an empty string for NULL strings.
19181 * read-md.c: Potentially include config.h rather than bconfig.h.
19182 Wrap include of errors.h with #ifdef GENERATOR_FILE.
19183 (have_error): New global, copied from errors.c.
19184 (md_reader::read_name): Rename to...
19185 (md_reader::read_name_1): ...this, adding "out_loc" param,
19186 and converting "missing name or number" to returning false, rather
19188 (md_reader::read_name): Reimplement in terms of read_name_1.
19189 (md_reader::read_name_or_nil): New function.
19190 (md_reader::read_string): Handle "(nil)" by returning NULL.
19191 (md_reader::md_reader): Add new param "compact".
19192 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
19193 (md_reader::read_file): New method.
19194 * read-md.h (md_reader::md_reader): Add new param "compact".
19195 (md_reader::read_file): New method.
19196 (md_reader::is_compact): New accessor.
19197 (md_reader::read_name): Convert return type from void to file_location.
19198 (md_reader::read_name_or_nil): New decl.
19199 (md_reader::read_name_1): New decl.
19200 (md_reader::m_compact): New field.
19201 (noop_reader::noop_reader): Pass "false" for new "compact" param
19203 (rtx_reader::rtx_reader): Add new "compact" param.
19204 (rtx_reader::read_rtx_operand): Make virtual and convert return
19205 type from void to rtx.
19206 (rtx_reader::read_until): New decl.
19207 (rtx_reader::handle_any_trailing_information): New virtual function.
19208 (rtx_reader::postprocess): New virtual function.
19209 (rtx_reader::finalize_string): New virtual function.
19210 (rtx_reader::m_in_call_function_usage): New field.
19211 (rtx_reader::m_reuse_rtx_by_id): New field.
19212 * read-rtl-function.c: New file.
19213 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
19214 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
19215 (selftest::verify_three_block_rtl_cfg): New decl.
19216 * read-rtl-function.h: New file.
19217 * read-rtl.c: Potentially include config.h rather than bconfig.h.
19218 For host, include function.h, memmodel.h, and emit-rtl.h.
19219 (one_time_initialization): New function.
19220 (struct compact_insn_name): New struct.
19221 (compact_insn_names): New array.
19222 (find_code): Handle insn codes in compact dumps.
19223 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
19224 (bind_subst_iter_and_attr): Likewise.
19225 (add_condition_to_string): Likewise.
19226 (add_condition_to_rtx): Likewise.
19227 (apply_attribute_uses): Likewise.
19228 (add_current_iterators): Likewise.
19229 (apply_iterators): Likewise.
19230 (initialize_iterators): Guard usage of apply_subst_iterator with
19231 #ifdef GENERATOR_FILE.
19232 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
19233 (md_reader::read_mapping): Likewise.
19234 (add_define_attr_for_define_subst): Likewise.
19235 (add_define_subst_attr): Likewise.
19236 (read_subst_mapping): Likewise.
19237 (check_code_iterator): Likewise.
19238 (rtx_reader::read_rtx): Likewise. Move one-time initialization
19240 (one_time_initialization): New function.
19241 (rtx_reader::read_until): New method.
19242 (read_flags): New function.
19243 (parse_reg_note_name): New function.
19244 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
19245 Handle reuse_rtx ids.
19246 Wrap iterator lookup within #ifdef GENERATOR_FILE.
19247 Add parsing support for RTL dumps, mirroring the special-cases in
19248 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
19249 values, and calling handle_any_trailing_information.
19250 (rtx_reader::read_rtx_operand): Convert return type from void
19251 to rtx, returning return_rtx. Handle case 'e'. Call
19252 finalize_string on XSTR and XTMPL fields.
19253 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
19254 "(nil)" values were omitted. Call the postprocess vfunc on the
19256 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
19257 class ctor. Initialize m_in_call_function_usage. Call
19258 one_time_initialization.
19259 * rtl-tests.c (selftest::test_uncond_jump): Call
19260 set_new_first_and_last_insn.
19261 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
19262 * selftest-rtl.c: New file.
19263 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
19264 (selftest::get_insn_by_uid): New decl.
19265 * selftest-run-tests.c (selftest::run_tests): Call
19266 read_rtl_function_c_tests.
19267 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
19268 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
19271 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
19273 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
19274 operands in a special way. Assert that pos+len <= mode precision.
19276 2017-01-05 Jakub Jelinek <jakub@redhat.com>
19278 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
19279 3 argument Alias with unlimited for the negative form.
19280 (fno-vect-cost-model): Removed.
19282 2017-01-05 Martin Liska <mliska@suse.cz>
19284 * hsa-gen.c (gen_hsa_divmod): New function.
19285 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
19287 2017-01-05 Martin Liska <mliska@suse.cz>
19290 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
19293 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19295 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
19296 small constant length operands.
19298 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19300 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
19301 between loop iterations.
19303 2017-01-05 Martin Liska <mliska@suse.cz>
19306 * gimplify.c (gimplify_decl_expr): Compare to
19307 asan_poisoned_variables instread of checking flags.
19308 (gimplify_target_expr): Likewise.
19309 (gimplify_expr): Likewise.
19310 (gimplify_function_tree): Conditionally initialize
19311 asan_poisoned_variables.
19313 2017-01-04 Jeff Law <law@redhat.com>
19315 PR tree-optimizatin/78812
19316 * rtl.h (contains_mem_rtx_p): Prototype.
19317 * ifcvt.c (containts_mem_rtx_p): Move from here to...
19318 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
19319 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
19320 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
19321 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
19323 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19325 * input.c (assert_char_at_range): Default-initialize actual_range.
19327 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19329 * df-scan.c (df_ref_create_structure): Make regno unsigned,
19330 to match the caller.
19332 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19334 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
19335 insns after final jump in test to emit dummy move.
19337 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19339 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
19340 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
19342 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19344 * multiple_target.c (create_dispatcher_calls): Init e_next.
19345 * tree-ssa-loop-split.c (split_loop): Init border.
19346 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
19349 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
19354 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
19355 (altivec_register_operand): Do not return true if the operand
19356 contains a SUBREG mixing SImode and SFmode.
19357 (vsx_register_operand): Likewise.
19358 (vsx_reg_sfsubreg_ok): New predicate.
19359 (vfloat_operand): Do not return true if the operand contains a
19360 SUBREG mixing SImode and SFmode.
19361 (vint_operand): Likewise.
19362 (vlogical_operand): Likewise.
19363 (gpc_reg_operand): Likewise.
19364 (int_reg_operand): Likewise.
19365 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
19366 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
19367 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
19369 (rs6000_emit_move_si_sf_subreg): New helper function.
19370 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
19371 fixup SUBREGs involving SImode and SFmode.
19372 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
19373 numbers for the new peephole2 optimization.
19374 (peephole2 for SFmode unions): New peephole2 to optimize cases in
19375 the GLIBC math library that do AND/IOR/XOR operations on single
19376 precision floating point.
19377 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
19378 target macros to say whether we need to avoid SUBREGs mixing
19380 (TARGET_ALLOW_SF_SUBREG): Likewise.
19381 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
19382 (UNSPEC_SI_FROM_SF): Likewise.
19383 (iorxor): Change spacing.
19384 (and_ior_xor): New iterator for AND, IOR, and XOR.
19385 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
19386 (movdi_from_sf_zero_ext): Likewise.
19387 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
19388 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
19389 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
19390 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
19391 (fms<mode>4): Likewise.
19392 (fnma<mode>4): Likewise.
19393 (fnms<mode>4): Likewise.
19394 (nfma<mode>4): Likewise.
19395 (nfms<mode>4): Likewise.
19397 2017-01-04 Marek Polacek <polacek@redhat.com>
19400 * doc/invoke.texi: Document -Wpointer-compare.
19402 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19404 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
19407 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
19408 descriptions for -gdwarf-5 and emit them as uleb128 instead of
19411 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
19414 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
19415 documentation of the powerpc_popcntb_ok attribute.
19416 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19417 code to issue warning messages if a requested CPU configuration is
19418 not supported by the binary (assembler and loader) toolchain.
19419 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
19420 made to define a built-in function that has been disabled.
19421 (paired_init_builtins): Add assertion to prevent ICE if attempt is
19422 made to define a built-in function that has been disabled.
19423 (altivec_init_builtins): Add comment explaining why definition
19424 of the DST built-in functions is not preceded by an assertion
19425 check. Add assertions to prevent ICE if attempts are made to
19426 define an altivec predicate or an abs* built-in function that has
19428 (htm_init_builtins): Add comment explaining why definition of the
19429 htm built-in functions is not preceded by an assertion check.
19431 2017-01-04 Jeff Law <law@redhat.com>
19433 PR tree-optimizatin/67955
19434 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
19435 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
19436 the points-to solution does not include pt_null. Use DECL_PT_UID
19439 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
19441 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
19442 Use gen_int_mode instead of gen_lopwart for const_int operands.
19444 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19446 PR tree-optimization/71563
19447 * match.pd: Simplify X << Y into X if Y is known to be 0 or
19448 out of range value - has low bits known to be zero.
19450 2017-01-04 Alan Modra <amodra@gmail.com>
19452 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
19453 * configure: Regenerate.
19454 * config.in: Regenerate.
19456 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19459 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
19460 a substring of the message, but strcmp with the whole message. Ifdef
19461 ENABLE_NLS, translate the message first using dgettext.
19463 2017-01-03 Jeff Law <law@redhat.com>
19465 PR tree-optimizatin/78856
19466 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
19467 (mark_threaded_blocks): Remove code to truncate thread paths that
19468 cross multiple loop headers. Instead invalidate the cached loop
19469 iteration information and handle case of a thread path walking
19470 into an irreducible region.
19472 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
19475 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
19476 assertions. Add support for doing the signbit if the IEEE 128-bit
19477 floating point value is in a GPR.
19478 * config/rs6000/rs6000.md (Fsignbit): Delete.
19479 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
19480 Update the length attribute if the value is in a GPR.
19481 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
19482 the sign or zero extension instruction, since the value is always 0/1.
19483 (signbit<mode>2_dm2): Delete using <Fsignbit>.
19486 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
19487 extracting SImode to a GPR register so that we can generate a
19488 store, limit the vector to be in a traditional Altivec register
19489 for the vextuwrx instruction.
19491 2017-01-03 Ian Lance Taylor <iant@google.com>
19493 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
19495 2017-01-03 Martin Sebor <msebor@redhat.com>
19497 PR tree-optimization/78696
19498 * gimple-ssa-sprintf.c (format_floating): Correct handling of
19499 precision. Use MPFR for %f for greater fidelity. Correct handling
19501 (pass_sprintf_length::compute_format_length): Set width and precision
19502 specified by asrerisk to void_node for vararg functions.
19503 (try_substitute_return_value): Adjust dump output.
19505 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
19507 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
19509 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
19511 * doc/invoke.texi (SPARC options): Document -mlra as the default.
19512 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
19513 -mlra/-mno-lra was passed to the compiler.
19515 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
19517 PR rtl-optimization/65618
19518 * emit-rtl.c (try_split): Move initialization of "before" and
19519 "after" to just before the call to emit_insn_after_setloc.
19521 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
19523 * doc/md.texi (Standard Names): Remove reference to Java frontend.
19525 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
19527 * dwarf2out.c (gen_enumeration_type_die): When
19528 -gno-strict-dwarf, add a DW_AT_encoding attribute.
19530 2017-01-03 Jakub Jelinek <jakub@redhat.com>
19532 PR tree-optimization/78965
19533 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
19534 Change first argument from const call_info & to call_info &. For %n
19535 set info.nowrite to false.
19537 PR middle-end/78901
19538 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
19539 possibly throwing calls.
19541 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
19542 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
19543 and fns handling, rather than in a separate case SSA_NAME.
19545 2017-01-02 Jeff Law <law@redhat.com>
19547 * config/darwin-driver.c (darwin_driver_init): Const-correctness
19548 fixes for first_period and second_period variables.
19550 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
19553 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
19554 (*insvqi_1): New insn pattern.
19555 (*insvqi_1_mem_rex64): Ditto.
19556 (*insvqi_2): Ditto.
19557 (*insvqi_3): Rename from *insvqi.
19559 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
19561 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
19563 * doc/cfg.texi (Edges): Remove reference to Java.
19564 (Maintaining the CFG): Ditto.
19566 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19568 PR middle-end/77674
19569 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
19570 transparent aliases.
19572 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19574 PR middle-end/77484
19575 * predict.def (PRED_CALL): Update hitrate.
19576 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
19577 * predict.c (tree_estimate_probability_bb): Split CALL predictor
19578 into direct/indirect/polymorphic variants.
19580 2017-01-01 Jakub Jelinek <jakub@redhat.com>
19582 Update copyright years.
19584 * gcc.c (process_command): Update copyright notice dates.
19585 * gcov-dump.c (print_version): Ditto.
19586 * gcov.c (print_version): Ditto.
19587 * gcov-tool.c (print_version): Ditto.
19588 * gengtype.c (create_file): Ditto.
19589 * doc/cpp.texi: Bump @copying's copyright year.
19590 * doc/cppinternals.texi: Ditto.
19591 * doc/gcc.texi: Ditto.
19592 * doc/gccint.texi: Ditto.
19593 * doc/gcov.texi: Ditto.
19594 * doc/install.texi: Ditto.
19595 * doc/invoke.texi: Ditto.
19597 Copyright (C) 2017 Free Software Foundation, Inc.
19599 Copying and distribution of this file, with or without modification,
19600 are permitted in any medium without royalty provided the copyright
19601 notice and this notice are preserved.