1 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3 * tree.h (build_var_debug_value): Remove prototype.
5 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7 * tree.c (tree_cons_stat): Remove _stat from name.
8 * tree.h (tree_cons): Remove macro.
10 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12 * tree.c (build_vl_exp_stat): Remove _stat from name.
13 * tree.h (build_vl_exp): Remove macro.
15 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17 * tree.c (build_decl_stat): Remove _stat from name.
18 * tree.h (build_decl): Remove macro.
20 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22 * gimple.c (gimple_build_with_ops_stat): Adjust.
23 (gimple_alloc_stat): Remove _stat from name.
24 * gimple.h (gimple_alloc): Remove macro.
26 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28 * tree.c (make_tree_vec_stat): Remove _stat from name.
29 (grow_tree_vec_stat): Likewise.
30 * tree.h (make_tree_vec_stat): Adjust prototype.
31 (grow_tree_vec_stat): Likewise.
32 (make_tree_vec): Remove macro.
33 (grow_tree_vec): Likewise.
35 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
37 * fold-const.c (fold_build1_stat_loc): Adjust.
38 (fold_build2_stat_loc): Likewise.
39 (fold_build3_stat_loc): Likewise.
40 * tree.c (build0_stat): Remove _stat from name.
41 (build1_stat): Likewise.
42 (build2_stat): Likewise.
43 (build3_stat): Likewise.
44 (build4_stat): Likewise.
45 (build5_stat): Likewise.
46 * tree.h (build1_loc): Remove macro, and rename _stat function
48 (build2_loc): Likewise.
49 (build3_loc): Likewise.
50 (build4_loc): Likewise.
51 (build5_loc): Likewise.
53 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
55 * tree.c (make_int_cst_stat): Remove _stat from name.
56 * tree.h (make_int_cst_stat): Adjust prototype.
57 (make_int_cst): Remove macro.
59 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
61 * tree.c (make_tre_binfo_stat): Remove _stat from name.
62 * tree.h (make_tree_binfo_stat): Adjust prototype.
63 (make_tree_binfo): Remove.
65 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
67 * tree.c (copy_node_stat): Rename to copy_node.
68 (build_distinct_type_copy): Adjust.
69 * tree.h (copy_node_stat): Adjust prototype.
70 (copy_node): Remove macro.
72 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
74 * tree.c (make_node_stat): rename to make_node.
75 (build_tree_list_stat): Adjust.
76 (build0_stat): Likewise.
77 (build2_stat): Likewise.
78 (build3_stat): Likewise.
79 (build4_stat): Likewise.
80 (build5_stat): Likewise.
81 (build_decl_stat): Likewise.
82 * tree.h (make_node_stat): Adjust prototype.
83 (make_node): remove macro.
85 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
87 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
88 (PPC_FEATURE2_SCV): Likewise.
89 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
91 2017-07-28 Tamar Christina <tamar.christina@arm.com>
93 * config/aarch64/aarch64.c
94 (aarch64_internal_mov_immediate): Add new special pattern.
95 * config/aarch64/aarch64.md (*movdi_aarch64):
96 Add reg/32bit const mov case.
98 2017-07-28 Tamar Christina <tamar.christina@arm.com>
99 Richard Sandiford <richard.sandiford@linaro.org>
101 * config/aarch64/aarch64.md (mov<mode>): Generalize.
102 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
103 Add integer and movi cases.
104 (movi-split-hf-df-sf split, fp16): New.
105 (enabled): Added TARGET_FP_F16INST.
106 * config/aarch64/iterators.md (GPF_HF): New.
107 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
109 2017-07-28 Tamar Christina <tamar.christina@arm.com>
111 * config/aarch64/aarch64.c
112 (aarch64_simd_container_mode): Add prototype.
113 (aarch64_expand_mov_immediate): Add HI support.
114 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
115 (aarch64_can_const_movi_rtx_p): New.
116 (aarch64_preferred_reload_class):
117 Remove restrictions of using FP registers for certain SIMD operations.
118 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
119 (aarch64_valid_floating_const): Add integer move validation.
120 (aarch64_simd_imm_scalar_p): Remove.
121 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
122 (aarch64_legitimate_constant_p): Expand list of supported cases.
123 * config/aarch64/aarch64-protos.h
124 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
125 (aarch64_reinterpret_float_as_int): New.
126 (aarch64_simd_imm_scalar_p): Remove.
127 * config/aarch64/constraints.md (Uvi): New.
128 (Dd): Split into Ds and new Dd.
129 * config/aarch64/aarch64.md (*movsi_aarch64):
131 (*movdi_aarch64): Add SIMD mov case.
133 2017-07-28 Bin Cheng <bin.cheng@arm.com>
135 * tree-predcom.c: (struct chain): Handle store-store chain in which
136 stores for elimination only store loop invariant values.
137 (execute_pred_commoning_chain): Ditto.
138 (prepare_initializers_chain_store_elim): Ditto.
139 (prepare_finalizers): Ditto.
140 (is_inv_store_elimination_chain): New function.
141 (initialize_root_vars_store_elim_1): New function.
143 2017-07-28 Bin Cheng <bin.cheng@arm.com>
145 * tree-predcom.c: Revise general description of the pass.
146 (enum chain_type): New enum type for store elimination.
147 (struct chain): New field supporting store elimination.
148 (struct component): Ditto.
149 (dump_chain): Dump store-stores chain.
150 (release_chain): Release resources.
151 (split_data_refs_to_components): Compute and create component
152 contains only stores for elimination.
153 (get_chain_last_ref_at): New function.
154 (make_invariant_chain): Initialization.
155 (make_rooted_chain): Specify chain type in parameter and record it.
156 (add_looparound_copies): Skip for store-stores chain.
157 (determine_roots_comp): Compute type of chain and pass it to
159 (initialize_root_vars_store_elim_2): New function.
160 (finalize_eliminated_stores): New function.
161 (remove_stmt): Handle store for elimination.
162 (execute_pred_commoning_chain): Execute predictive commoning on
164 (determine_unroll_factor): Skip unroll for store-stores chain.
165 (prepare_initializers_chain_store_elim): New function.
166 (prepare_initializers_chain): Hanlde store-store chain.
167 (prepare_finalizers_chain, prepare_finalizers): New function.
168 (tree_predictive_commoning_loop): Return integer value indicating
169 if loop is unrolled or lcssa form is corrupted.
170 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
172 2017-07-28 Bin Cheng <bin.cheng@arm.com>
174 * tree-predcom.c (initialize_root): Delete.
175 (execute_pred_commoning_chain): Initialize root vars and replace
176 reference of non-combined chain directly, rather than call above
179 2017-07-28 Bin Cheng <bin.cheng@arm.com>
181 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
182 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
184 2017-07-28 Bin Cheng <bin.cheng@arm.com>
186 * tree-predcom.c (struct chain): New field init_seq.
187 (release_chain): Release init_seq.
188 (prepare_initializers_chain): Record intialization stmts in above
190 (insert_init_seqs): New function.
191 (tree_predictive_commoning_loop): Call insert_init_seqs.
193 2017-07-28 Bin Cheng <bin.cheng@arm.com>
195 * tree-predcom.c (determine_roots_comp): Skip trivial components.
197 2017-07-28 Richard Biener <rguenther@suse.de>
199 * match.pd: Remove superfluous :c.
200 * genmatch.c (simplify::id): Add member.
201 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
203 (current_id): New global.
204 (dt_node::parent): Move from ...
205 (dt_operand::parent): ... here. Add for_id member.
206 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
207 (decision_tree::find_node): Relax order requirement when
208 merging DT_TRUE nodes to ones inbetween the current simplify
209 and the one we try to merge with. Add diagnostic whenever
210 we need to enforce pattern order by not merging.
211 (decision_tree::insert): Set current_id.
212 (decision_tree::print_node): Dump parent node and for_id.
213 (parser::last_id): Add member.
214 (parser::push_simplify): Assign unique id.
215 (parser::parser): Initialize last_id.
217 2017-07-28 Martin Liska <mliska@suse.cz>
220 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
221 gimple_build_debug_bind.
223 2017-07-28 Richard Biener <rguenther@suse.de>
225 PR tree-optimization/81502
226 * match.pd: Add pattern combining BIT_INSERT_EXPR with
228 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
230 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
231 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
232 for BIT_FIELD_REF args.
233 * fold-const.c (make_bit_field_ref): Likewise.
234 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
236 2017-07-28 Jakub Jelinek <jakub@redhat.com>
239 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
240 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
241 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
242 Or it into SANITIZER_UNDEFINED.
243 * ubsan.c: Include gimple-fold.h and varasm.h.
244 (ubsan_expand_ptr_ifn): New function.
245 (instrument_pointer_overflow): New function.
246 (maybe_instrument_pointer_overflow): New function.
247 (instrument_object_size): Formatting fix.
248 (pass_ubsan::execute): Call instrument_pointer_overflow
249 and maybe_instrument_pointer_overflow.
250 * internal-fn.c (expand_UBSAN_PTR): New function.
251 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
252 * sanitizer.def (__ubsan_handle_pointer_overflow,
253 __ubsan_handle_pointer_overflow_abort): New builtins.
254 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
255 * internal-fn.def (UBSAN_PTR): New internal function.
256 * opts.c (sanitizer_opts): Add pointer-overflow.
257 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
258 * fold-const.c (build_range_check): Compute pointer range check in
259 integral type if pointer arithmetics would be needed. Formatting
262 2017-07-28 Martin Liska <mliska@suse.cz>
265 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
266 parameters that are of a variable-length.
268 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
270 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
272 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
273 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
274 (CRT_CALL_STATIC_FUNCTION): Likewise.
275 (ASM_DEFAULT_SPEC): New define.
276 (ASM_SPEC32): Likewise.
277 (ASM_SPEC64): Likewise.
278 (ASM_SPEC_COMMON): Likewise.
279 (ASM_SPEC): Likewise.
280 (INVALID_64BIT): Likewise.
281 (LINK_OS_DEFAULT_SPEC): Likewise.
282 (LINK_OS_SPEC32): Likewise.
283 (LINK_OS_SPEC64): Likewise.
284 (POWERPC_LINUX): Likewise.
285 (PTRDIFF_TYPE): Likewise.
286 (RESTORE_FP_PREFIX): Likewise.
287 (RESTORE_FP_SUFFIX): Likewise.
288 (SAVE_FP_PREFIX): Likewise.
289 (SAVE_FP_SUFFIX): Likewise.
290 (SIZE_TYPE): Likewise.
291 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
292 (TARGET_64BIT): Likewise.
293 (TARGET_64BIT): Likewise.
294 (TARGET_AIX): Likewise.
295 (WCHAR_TYPE_SIZE): Likewise.
296 (WCHAR_TYPE): Undefine.
297 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
298 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
299 (CPP_OS_RTEMS_SPEC): Delete.
300 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
301 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
303 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
305 2017-07-28 Jakub Jelinek <jakub@redhat.com>
307 PR tree-optimization/81578
308 * tree-parloops.c (build_new_reduction): Bail out if
309 reduction_code isn't one of the standard OpenMP reductions.
310 Move the details printing after that decision.
312 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
314 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
315 related to reload_in_progress.
316 (splat_input_operand): Likewise.
317 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
319 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
321 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
322 (TARGET_INSTANTIATE_DECLS): Likewise.
323 (legitimate_indexed_address_p): Delete reload_in_progress code.
324 (rs6000_debug_legitimate_address_p): Likewise.
325 (rs6000_eliminate_indexed_memrefs): Likewise.
326 (rs6000_emit_le_vsx_store): Likewise.
327 (rs6000_emit_move_si_sf_subreg): Likewise.
328 (rs6000_emit_move): Likewise.
329 (register_to_reg_type): Likewise.
330 (rs6000_pre_atomic_barrier): Likewise.
331 (rs6000_machopic_legitimize_pic_address): Likewise.
332 (rs6000_allocate_stack_temp): Likewise.
333 (rs6000_address_for_fpconvert): Likewise.
334 (rs6000_address_for_altivec): Likewise.
335 (rs6000_secondary_memory_needed_rtx): Delete function.
336 (rs6000_check_sdmode): Likewise.
337 (rs6000_alloc_sdmode_stack_slot): Likewise.
338 (rs6000_instantiate_decls): Likewise.
339 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
340 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
341 Delete reload_in_progress.
342 (*vec_reload_and_plus_<mptrsize>): Likewise.
343 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
344 (vsx_div_v2di): Likewise.
345 (vsx_udiv_v2di): Likewise.
347 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
349 * config/rs6000/rs6000.opt (mlra): Replace with stub.
350 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
351 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
352 (rs6000_debug_reg_global): Delete print of LRA status.
353 (rs6000_option_override_internal): Delete dead LRA related code.
354 (rs6000_lra_p): Delete function.
355 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
357 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
359 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
360 * config/riscv/rtems.h: New file.
362 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
363 Sudakshina Das <sudi.das@arm.com>
365 * config/aarch64/aarch64.md
366 (define_split for and<mode>3nr_compare): Move
367 non aarch64_logical_operand to a register.
368 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
369 register immediate operand to a register.
370 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
372 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
375 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
377 2017-07-27 Richard Biener <rguenther@suse.de>
379 PR tree-optimization/81573
380 PR tree-optimization/81494
381 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
382 multi defuse cycle case.
384 2017-07-27 Richard Biener <rguenther@suse.de>
386 PR tree-optimization/81571
387 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
390 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
392 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
393 earlier and only if MASK_FPU is set. Adjust formatting.
395 2017-07-27 Martin Liska <mliska@suse.cz>
397 * opt-functions.awk: Add validation of value of Init.
398 * optc-gen.awk: Pass new argument.
400 2017-07-27 Martin Liska <mliska@suse.cz>
402 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
405 2017-07-27 Martin Liska <mliska@suse.cz>
407 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
408 BBs and edges seen by autoFDO.
410 2017-07-27 Richard Biener <rguenther@suse.de>
412 PR tree-optimization/81502
413 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
414 with incompatible but same sized type.
415 (execute_update_addresses_taken): Likewise.
417 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
419 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
420 flag_tree_loop_vectorize rather than flag_tree_vectorize.
422 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
425 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
426 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
427 Change s_operand to memory_operand.
429 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
431 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
432 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
433 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
434 Emit instructions rather than returning an expression. Handle TFmode
435 and KFmode by casting to TImode.
436 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
437 (rs6000_emit_le_vsx_store): Likewise.
438 * config/rs6000/vsx.md (VSX_TI): New iterator.
439 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
440 (*vsx_le_undo_permute_<mode>): Likewise.
441 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
442 emit the split sequence.
443 (*vsx_le_perm_store_<mode>): Likewise.
445 2017-07-27 Jakub Jelinek <jakub@redhat.com>
447 PR tree-optimization/81555
448 PR tree-optimization/81556
449 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
450 if true, force CHANGED for the recursive invocation.
451 (reassociate_bb): Remember original length of ops array, pass
452 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
454 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
455 attributes for noipa attribute. For naked attribute use
456 lookup_attribute first before lookup_attribute_spec.
457 * final.c (rest_of_handle_final): Disable IPA RA for functions with
459 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
460 for functions with noipa attribute.
461 (cgraph_externally_visible_p): Return true for functions with noipa
463 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
464 for functions with noipa attribute.
465 * doc/extend.texi: Document noipa function attribute.
466 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
467 also for functions with noipa attribute.
468 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
470 2017-07-26 Andrew Pinski <apinski@cavium.com>
472 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
473 vec_unalign_load_cost and vec_unalign_store_cost.
475 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
477 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
478 -mvsx-small-integer option.
479 (ISA_3_0_MASKS_IEEE): Likewise.
480 (OTHER_VSX_VECTOR_MASKS): Likewise.
481 (POWERPC_MASKS): Likewise.
482 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
483 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
484 code, only testing for DImode being allowed in non-VSX floating
486 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
487 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
489 (rs6000_option_override_internal): Delete -mvsx-small-integer.
490 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
491 TARGET_P8_VECTOR test.
492 (rs6000_secondary_reload_simple_move): Likewise.
493 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
494 since TARGET_P9_VECTOR was already tested.
495 (rs6000_opt_masks): Remove -mvsx-small-integer.
496 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
497 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
499 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
500 test for TARGET_VEXTRACTUB was used, and that uses
502 (p9 extract splitter): Likewise.
503 (vsx_extract_<mode>_di_p9): Likewise.
504 (vsx_extract_<mode>_store_p9): Likewise.
505 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
506 for TARGET_P9_VECTOR was used. Delete code that is now dead with
507 the elimination of TARGET_VSX_SMALL_INTEGER.
508 (vsx_extract_<mode>_p8): Likewise.
509 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
510 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
511 (vsx_set_<mode>_p9): Likewise.
512 (vsx_set_v4sf_p9): Likewise.
513 (vsx_set_v4sf_p9_zero): Likewise.
514 (vsx_insert_extract_v4sf_p9): Likewise.
515 (vsx_insert_extract_v4sf_p9_2): Likewise.
516 * config/rs6000/rs6000.md (sign extend splitter): Change
517 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
518 (floatsi<mode>2_lfiwax_mem): Likewise.
519 (floatunssi<mode>2_lfiwzx_mem): Likewise.
520 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
521 since a test for TARGET_P9_VECTOR was used.
522 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
523 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
524 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
525 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
526 TARGET_P8_VECTOR test.
527 (fix_trunc<mode>si2_stfiwx): Likewise.
528 (fix_trunc<mode>si2_internal): Likewise.
529 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
530 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
532 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
533 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
534 TARGET_P8_VECTOR test.
535 (fixuns_trunc<mode>si2_stfiwx): Likewise.
536 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
537 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
539 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
540 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
541 since a test for TARGET_P9_VECTOR was used.
542 (splitter for loading small constants): Likewise.
544 2017-07-26 Andrew Pinski <apinski@cavium.com>
546 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
549 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
552 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
553 (fp_valid_at): Likewise.
555 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
557 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
558 (qdf24xx_addrcost_table): Likewise.
559 (cortexa57_tunings): Update to use generic_branch_cost.
560 (cortexa72_tunings): Likewise.
561 (cortexa73_tunings): Likewise.
562 (qdf24xx_tunings): Likewise.
564 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
566 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
567 (thunderx2t99_branch_cost): Likewise.
568 (cortexa35_tunings): Update to use generic_branch_cost.
569 (cortexa53_tunings): Likewise.
570 (cortexa57_tunings): Likewise.
571 (cortexa72_tunings): Likewise.
572 (cortexa73_tunings): Likewise.
573 (thunderx2t99_tunings): Likewise.
575 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
577 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
578 (sparc_option_override): Honour MASK_FSMULD.
579 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
580 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
581 * config/sparc/sparc.opt (mfsmuld): New option.
582 * doc/invoke.texi (mfsmuld): Document option.
584 2017-07-26 Marek Polacek <polacek@redhat.com>
587 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
589 2017-07-26 Richard Biener <rguenther@suse.de>
591 * gimple-match-head.c (do_valueize): Return OP if valueize
593 (get_def): New helper to get at the def stmt of a SSA name
595 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
596 do_valueize to get at the def stmt.
597 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
599 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
602 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
604 2017-07-26 Martin Liska <mliska@suse.cz>
607 * function.c (expand_function_start): Make expansion of
608 nonlocal_goto_save_area after parm_birth_insn.
610 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
612 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
613 from all CPU target flags enable members.
615 2017-07-26 Richard Biener <rguenther@suse.de>
617 * genmatch.c (dt_simplify::gen): Make iterator vars const.
618 (decision_tree::gen): Make 'type' const.
619 (write_predicate): Likewise.
621 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
623 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
624 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
625 (rs6000_option_override_internal): Likewise.
626 (rs6000_expand_vector_set): Likewise.
627 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
628 (TARGET_UPPER_REGS_SF): Likewise.
629 (TARGET_UPPER_REGS_DI): Likewise.
630 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
631 (TARGET_DIRECT_MOVE_64BIT): Likewise.
632 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
633 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
634 (Splitters for DI constants in Altivec registers): Likewise.
635 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
636 (vsx_set_v4sf_p9): Likewise.
637 (vsx_set_v4sf_p9_zero): Likewise.
638 (vsx_insert_extract_v4sf_p9): Likewise.
639 (vsx_insert_extract_v4sf_p9_2): Likewise.
641 2017-07-25 Carl Love <cel@us.ibm.com>
643 * doc/extend.texi: Update the built-in documentation file for the
644 existing built-in functions
645 vector signed char vec_cnttz (vector signed char);
646 vector unsigned char vec_cnttz (vector unsigned char);
647 vector signed short vec_cnttz (vector signed short);
648 vector unsigned short vec_cnttz (vector unsigned short);
649 vector signed int vec_cnttz (vector signed int);
650 vector unsigned int vec_cnttz (vector unsigned int);
651 vector signed long long vec_cnttz (vector signed long long);
652 vector unsigned long long vec_cnttz (vector unsigned long long);
654 2017-07-25 Andrew Pinski <apinski@cavium.com>
656 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
657 accesses where the use is for the first operand of a BIT_INSERT.
659 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
662 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
663 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
665 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
667 * config/i386/gstabs.h: Delete.
668 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
670 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
672 * config/i386/i386.c (ix86_decompose_address): Do not check for
673 register RTX when looking at index_reg or base_reg.
674 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
676 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
678 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
679 to update EH info here.
681 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
683 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
685 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
687 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
689 2017-07-25 Torsten Duwe <duwe@suse.de>
691 * common.opt: Introduce -fpatchable-function-entry
692 command line option, and its variables function_entry_patch_area_size
693 and function_entry_patch_area_start.
694 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
695 including a two-value parser.
696 * target.def (print_patchable_function_entry): New target hook.
697 * targhooks.h (default_print_patchable_function_entry): New function.
698 * targhooks.c (default_print_patchable_function_entry): Likewise.
699 * toplev.c (process_options): Switch off IPA-RA if
700 patchable function entries are being generated.
701 * varasm.c (assemble_start_function): Look at the
702 patchable-function-entry command line switch and current
703 function attributes and maybe generate NOP instructions by
704 calling the print_patchable_function_entry hook.
705 * doc/extend.texi: Document patchable_function_entry attribute.
706 * doc/invoke.texi: Document -fpatchable_function_entry
708 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
710 * doc/tm.texi: Re-generate.
712 2017-07-25 Jakub Jelinek <jakub@redhat.com>
715 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
716 TARGET_AVX512DQ rather than TARGET_AVX512BW.
718 2017-07-25 Tamar Christina <tamar.christina@arm.com>
720 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
722 2017-07-25 Richard Biener <rguenther@suse.de>
724 PR tree-optimization/81455
725 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
726 not walk in cycles when looking for guards.
728 2017-07-25 Richard Biener <rguenther@suse.de>
730 PR tree-optimization/81529
731 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
732 when optimizing backedge uses.
734 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
736 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
738 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
739 to dl_section_ref. On AIX, append an expression to subtract
740 the size of the section length to dl_section_ref.
742 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
744 * configure.ac: If any of the config.* scripts fail, exit 1.
745 * configure: Regenerate.
747 2017-07-25 Richard Biener <rguenther@suse.de>
750 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
751 of immediate uses, be more verbose on errors.
753 2017-07-25 Richard Biener <rguenther@suse.de>
755 PR tree-optimization/81510
756 * tree-vect-loop.c (vect_is_simple_reduction): When the
757 reduction stmt is not inside the loop bail out.
759 2017-07-25 Richard Biener <rguenther@suse.de>
761 PR tree-optimization/81303
762 * tree-vect-loop-manip.c (vect_loop_versioning): Build
763 profitability check against LOOP_VINFO_NITERSM1.
765 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
767 * domwalk.c (cmp_bb_postorder): Simplify.
768 (sort_bbs_postorder): New function. Use it...
769 (dom_walker::walk): ...here to optimize common cases.
771 2017-07-25 Martin Liska <mliska@suse.cz>
774 * ipa-visibility.c (function_and_variable_visibility): Make the
775 redirection just on target that supports aliasing.
776 Fix GNU coding style.
778 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
781 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
783 * config/arm/rtems.h: Likewise.
784 * config/bfin/rtems.h: Likewise.
785 * config/i386/rtemself.h: Likewise.
786 * config/lm32/rtems.h: Likewise.
787 * config/m32c/rtems.h: Likewise.
788 * config/m68k/rtemself.h: Likewise.
789 * config/microblaze/rtems.h: Likewise.
790 * config/mips/rtems.h: Likewise.
791 * config/moxie/rtems.h: Likewise.
792 * config/nios2/rtems.h: Likewise.
793 * config/powerpcspe/rtems.h: Likewise.
794 * config/rs6000/rtems.h: Likewise.
795 * config/rtems.h: Likewise.
796 * config/sh/rtems.h: Likewise.
797 * config/sh/rtemself.h: Likewise.
798 * config/sparc/rtemself.h: Likewise.
800 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
803 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
804 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
805 * tree-ssa-structalias.c (alias_get_name): Same.
807 2017-07-25 Bin Cheng <bin.cheng@arm.com>
810 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
811 instructions if no du chain is found.
813 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
815 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
817 2017-07-25 Richard Biener <rguenther@suse.de>
820 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
823 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
825 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
827 (ISA_2_7_MASKS_SERVER): Likewise.
828 (ISA_3_0_MASKS_IEEE): Likewise.
829 (OTHER_P8_VECTOR_MASKS): Likewise.
830 (OTHER_VSX_VECTOR_MASKS): Likewise.
831 (POWERPC_MASKS): Likewise.
832 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
833 duplicate list of options.
834 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
835 explicit -mupper-regs options.
836 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
837 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
838 alias for -mupper-regs-df.
839 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
840 (rs6000_init_hard_regno_mode_ok): Likewise.
841 (rs6000_option_override_internal): Likewise.
842 (rs6000_opt_masks): Likewise.
843 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
844 options in terms of whether -mvsx or -mpower8-vector was used.
845 (TARGET_UPPER_REGS_DI): Likewise.
846 (TARGET_UPPER_REGS_SF): Likewise.
847 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
848 -mupper-regs-* options.
850 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
852 * passes.c (emergency_dump_function): Print some empty lines and a
853 header before the RTL dump.
855 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
857 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
859 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
862 * config/aarch64/aarch64.c (aarch64_classify_symbol):
863 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
865 2017-07-24 Carl Love <cel@us.ibm.com>
867 * config/rs6000/rs6000-c.c: Add support for built-in functions
868 vector float vec_extract_fp32_from_shorth (vector unsigned short);
869 vector float vec_extract_fp32_from_shortl (vector unsigned short);
870 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
871 vec_extract_fp_from_shortl): Add defines for the two builtins.
872 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
873 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
875 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
876 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
877 * doc/extend.texi: Update the built-in documentation file for the
878 new built-in function.
880 2017-07-24 Jakub Jelinek <jakub@redhat.com>
883 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
885 * doc/generic.texi: Likewise.
886 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
887 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
889 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
891 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
892 (aarch64_mls_elt_merge<mode>): Likewise.
894 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
896 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
899 2017-07-23 Michael Collison <michael.collison@arm.com>
901 * config/arm/arm.c (arm_option_override): Deprecate
902 use of -mstructure-size-boundary.
903 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
904 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
906 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
909 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
910 Reduce cost estimate for direct moves.
912 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
915 * config/i386/i386.c (ix86_option_override_internal): Disable
916 BMI, BMI2 and TBM instructions for -m16.
918 2017-07-21 Carl Love <cel@us.ibm.com>
920 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
921 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
922 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
923 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
924 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
925 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
926 VMULOSW): New enum "unspec" values.
927 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
928 altivec_vmulosw): New patterns.
929 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
930 VMULOSW): Add definitions.
932 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
934 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
936 * config/aarch64/aarch64-options-extensions.def (rdma); New.
937 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
938 (AARCH64_FL_V8_1): Renumber.
939 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
940 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
941 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
942 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
943 rdma to feature modifiers list.
945 2017-07-21 Yury Gribov <tetra2005@gmail.com>
948 * ipa-visibility (function_and_variable_visibility): Convert
949 recursive PLT call to direct call if appropriate.
951 2017-07-21 Andrew Pinski <apinski@cavium.com>
953 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
954 operand 1 to see if the types precision matches.
955 * fold-const.c (operand_equal_p): Likewise.
957 2017-07-21 Richard Biener <rguenther@suse.de>
959 PR tree-optimization/81303
960 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
961 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
962 (vect_peeling_hash_get_lowest_cost): Adjust.
963 (vect_enhance_data_refs_alignment): Likewise. Use
964 vect_get_peeling_costs_all_drs to compute the penalty for no
965 peeling to match up costs.
967 2017-07-21 Richard Biener <rguenther@suse.de>
969 PR tree-optimization/81500
970 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
971 we didn't identify a reduction path.
973 2017-07-21 Tom de Vries <tom@codesourcery.com>
974 Cesar Philippidis <cesar@codesourcery.com>
976 PR gcov-profile/81442
977 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
980 2017-07-21 Tom de Vries <tom@codesourcery.com>
983 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
985 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
986 nvptx_override_options_after_change.
988 2017-07-21 Ulrich Drepper <drepper@redhat.com>
990 * dwarf2out.c (output_file_names): Avoid double testing for
993 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
995 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
997 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
999 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
1001 (try_crossjump_to_edge): Do not punt on partitioned functions.
1003 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
1005 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
1006 Put all BBs reachable only via paths crossing cold region to cold
1008 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
1010 2016-07-21 Richard Biener <rguenther@suse.de>
1012 PR tree-optimization/81303
1013 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
1014 into account prologue and epilogue iterations when raising
1015 min_profitable_iters to sth at least covering one vector iteration.
1017 2017-07-21 Tamar Christina <tamar.christina@arm.com>
1019 * config/arm/arm.c (arm_test_cpu_arch_dat):
1022 2017-07-20 Nathan Sidwell <nathan@acm.org>
1024 Remove TYPE_METHODS.
1025 * tree.h (TYPE_METHODS): Delete.
1026 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
1027 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
1028 (dbxout_type_methods): Scan TYPE_FIELDS.
1029 (dbxout_type): Don't check TYPE_METHODS here.
1030 * function.c (use_register_for_decl): Always ignore register for
1031 class types when not optimizing.
1032 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
1033 * tree.c (free_lang_data_in_type): Stitch out member functions and
1034 templates from TYPE_FIELDS.
1035 (build_distinct_type_copy, verify_type_variant,
1036 verify_type): Member fns are on TYPE_FIELDS.
1037 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
1038 * tree-pretty-print.c (dump_generic_node): Likewise.
1040 2017-07-20 Jakub Jelinek <jakub@redhat.com>
1043 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
1044 V2TImode and V4TImode.
1045 (ix86_expand_vector_extract): Likewise.
1046 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
1047 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
1048 (ssescalarmode): Handle V4TImode and V2TImode.
1049 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
1050 (*vec_extractv2ti, *vec_extractv4ti): New insns.
1051 (VEXTRACTI128_MODE): New mode iterator.
1052 (splitter for *vec_extractv?ti first element): New.
1053 (VEC_INIT_MODE): New mode iterator.
1054 (vec_init<mode>): Consolidate 3 expanders into one using
1055 VEC_INIT_MODE mode iterator.
1057 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
1059 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
1060 non_spilled_static_chain_regno_p.
1062 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
1064 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
1066 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
1068 * bb-reorder.c (connect_traces): Allow copying of blocks within
1071 2017-07-20 Richard Biener <rguenther@suse.de>
1073 * gimple.h (gimple_phi_result): Add gphi * overload.
1074 (gimple_phi_result_ptr): Likewise.
1075 (gimple_phi_arg): Likewise. Adjust index assert to only
1076 allow actual argument accesses rather than all slots available
1078 (gimple_phi_arg_def): Add gphi * overload.
1079 * tree-phinodes.c (make_phi_node): Initialize only actual
1081 (resize_phi_node): Clear memory not covered by old node,
1082 do not initialize excess argument slots.
1083 (reserve_phi_args_for_new_edge): Initialize new argument slot
1086 2017-07-20 Bin Cheng <bin.cheng@arm.com>
1088 PR tree-optimization/81388
1090 2016-07-21 Bin Cheng <bin.cheng@arm.com>
1092 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
1093 by removing computation of may_be_zero.
1095 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1096 Tom de Vries <tom@codesourcery.com>
1099 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
1100 when gimple level profile disagrees with what RTL expander did.
1102 2017-07-20 Richard Biener <rguenther@suse.de>
1104 PR tree-optimization/61171
1105 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
1106 (vect_analyze_stmt): Add slp instance parameter.
1107 (vectorizable_reduction): Likewise.
1108 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
1109 (vect_is_simple_reduction): Deal with chains not detected
1110 as SLP reduction chain, specifically not properly associated
1111 chains containing a mix of plus/minus.
1112 (get_reduction_op): Remove.
1113 (get_initial_defs_for_reduction): Simplify, pass in whether
1114 this is a reduction chain, pass in the SLP node for the PHIs.
1115 (vect_create_epilog_for_reduction): Get the SLP instance as
1117 (vectorizable_reduction): Get the SLP instance as arg.
1118 During analysis remember the SLP node with the PHIs in the
1119 instance. Simplify getting at the vectorized reduction PHIs.
1120 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
1121 through SLP instance.
1122 (vect_slp_analyze_operations): Likewise.
1123 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
1124 (vect_transform_stmt): Likewise.
1126 2017-07-20 Tom de Vries <tom@codesourcery.com>
1128 PR tree-optimization/81489
1129 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
1130 read of phi arg location to before loop that modifies phi.
1132 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
1134 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
1137 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1140 * except.c (execute): Fix ordering issue.
1142 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
1144 PR rtl-optimization/81423
1145 * combine.c (make_compound_operation_int): Don't try to optimize
1146 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
1148 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
1150 PR rtl-optimization/81423
1151 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
1152 with a constant that is -1 in the truncated to mode.
1154 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1156 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
1157 (determine_unlikely_bbs): ... here.
1158 * predict.h (propagate_unlikely_bbs_forward): Declare.
1159 * cfgexpand.c (pass_expand::execute): Use it.
1160 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
1162 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
1163 propagate_unlikely_bbs_forward.
1165 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1168 * except.c (maybe_add_nop_after_section_switch): New function.
1171 2017-07-19 Tom de Vries <tom@codesourcery.com>
1173 * gimple.h (gimple_phi_set_arg): Make assert more strict.
1175 2017-07-19 Tom de Vries <tom@codesourcery.com>
1177 * gimple.h (gimple_phi_arg): Make assert more strict.
1179 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
1181 * config.gcc (powerpc*-*-*): Add mmintrin.h.
1182 * config/rs6000/mmintrin.h: New file.
1183 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
1185 2017-07-19 Jakub Jelinek <jakub@redhat.com>
1187 PR tree-optimization/81346
1188 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
1190 2017-07-19 Tom de Vries <tom@codesourcery.com>
1192 * config/nvptx/nvptx.md (VECIM): Add V2DI.
1194 2017-07-19 Tom de Vries <tom@codesourcery.com>
1196 * config/nvptx/nvptx-modes.def: Add V2DImode.
1197 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
1198 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
1199 (nvptx_output_mov_insn): Handle lack of mov.b128.
1200 (nvptx_print_operand): Handle 'H' and 'L' codes.
1201 (nvptx_vector_mode_supported): Allow V2DImode.
1202 (nvptx_preferred_simd_mode): New function.
1203 (nvptx_data_alignment): New function.
1204 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
1205 nvptx_preferred_simd_mode.
1206 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
1208 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
1210 2017-07-19 Tom de Vries <tom@codesourcery.com>
1212 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
1213 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
1214 (nvptx_vector_mode_supported): New function. Allow V2SImode.
1215 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
1216 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
1217 (mov<VECIM>_insn): New define_insn.
1218 (define_expand "mov<VECIM>): New define_expand.
1220 2017-07-19 Tom de Vries <tom@codesourcery.com>
1222 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
1224 2017-07-19 Jakub Jelinek <jakub@redhat.com>
1226 PR tree-optimization/81346
1227 * fold-const.h (fold_div_compare, range_check_type): Declare.
1228 * fold-const.c (range_check_type): New function.
1229 (build_range_check): Use range_check_type.
1230 (fold_div_compare): No longer static, rewritten into
1231 a match.pd helper function.
1232 (fold_comparison): Don't call fold_div_compare here.
1233 * match.pd (X / C1 op C2): New optimization using fold_div_compare
1236 2017-07-19 Nathan Sidwell <nathan@acm.org>
1238 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
1239 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
1240 * tree.c (find_decls_types_r, verify_type): Use
1241 TYPE_{MIN,MAX}_VALUE_RAW.
1242 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
1243 (hash_tree): Likewise.
1244 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
1246 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
1249 2017-07-18 Tom de Vries <tom@codesourcery.com>
1252 * omp-expand.c (expand_omp_for_static_chunk): Handle
1253 equal-argument loop exit phi.
1255 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
1258 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
1259 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
1260 operand 2 predicate.
1261 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
1262 operand 2 predicate.
1263 (ror,rol -> rorx splitters): Use const_int_operand as
1264 operand 2 predicate.
1266 2017-06-18 Richard Biener <rguenther@suse.de>
1268 PR tree-optimization/81410
1269 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
1270 the gap in the ! slp_perm SLP case after each group.
1272 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1275 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
1278 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1281 * predict.c (set_even_probabilities): Cleanup; do not affect
1282 probabilities that are already known.
1283 (combine_predictions_for_bb): Call even when count is set.
1285 2017-07-18 Nathan Sidwell <nathan@acm.org>
1287 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
1290 2017-07-18 Bin Cheng <bin.cheng@arm.com>
1293 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
1294 optimization for loop niter analysis.
1296 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
1299 * config/avr/avr.c (avr_optimize_casesi): Don't use
1300 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
1302 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1304 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
1305 body_cost_vec from _vect_peel_extended_info.
1306 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
1307 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
1310 2017-07-18 Bin Cheng <bin.cheng@arm.com>
1312 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
1314 2017-07-18 Richard Biener <rguenther@suse.de>
1316 PR tree-optimization/80620
1317 PR tree-optimization/81403
1318 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
1319 info when re-using a VN table entry.
1321 2017-07-18 Richard Biener <rguenther@suse.de>
1323 PR tree-optimization/81418
1324 * tree-vect-loop.c (vectorizable_reduction): Properly compute
1325 vectype_in. Verify that with lane-reducing reduction operations
1326 we have a single def-use cycle.
1328 2017-07-17 Carl Love <cel@us.ibm.com>
1330 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
1332 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1333 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
1334 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
1335 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
1336 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
1337 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
1338 VMULOSW): New enum "unspec" values.
1339 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
1340 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
1341 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
1342 altivec_vmulosw): New patterns.
1343 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
1344 VMULOSW): Add definitions.
1345 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
1347 * config/alpha/alpha.c: Include predict.h.
1349 2017-07-17 Yury Gribov <tetra2005@gmail.com>
1351 * tree-vrp.c (compare_assert_loc): Fix comparison function
1352 to return predictable results.
1354 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1356 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
1359 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
1360 * doc/invoke.texi (mexpand-adddi): Update text.
1362 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1364 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
1365 that also clobbers the CC register. The old expand code is moved
1367 (*arc_clzsi2): ... here.
1368 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
1369 the CC register. The old expand code is moved to ...
1370 (arc_ctzsi2): ... here.
1372 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1374 * config/arc/arc.opt (mindexed-loads): Use initial value
1375 TARGET_INDEXED_LOADS_DEFAULT.
1376 (mauto-modify-reg): Use initial value
1377 TARGET_AUTO_MODIFY_REG_DEFAULT.
1378 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
1379 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
1380 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
1381 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
1383 2017-07-17 Martin Liska <mliska@suse.cz>
1386 * opts.c (finish_options): Do not allow -fgnu-tm
1387 w/ -fsanitize={kernel-,}address. Say sorry.
1389 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1392 * tree-loop-distribution.c (classify_partition): Only assert on
1393 numer of iterations.
1394 (merge_dep_scc_partitions): Delete prameter. Update function call.
1395 (distribute_loop): Remove code handling loop with unknown niters.
1396 (pass_loop_distribution::execute): Skip loop with unknown niters.
1398 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1401 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
1402 function sort_partitions_by_post_order.
1404 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1406 PR tree-optimization/81374
1407 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
1408 the max index of basic blocks, rather than number of basic blocks.
1410 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1412 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
1414 (arc_legitimate_pic_operand_p): Likewise.
1415 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
1417 (arc_needs_pcl_p): Likewise.
1418 (arc_legitimate_pc_offset_p): Likewise.
1419 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
1420 function is also used in constrains.md.
1421 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
1422 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
1423 PLUS. Only return true/false in known cases, otherwise assert.
1424 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
1425 is already called in arc_legitimate_constant_p.
1426 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
1428 (LEGITIMATE_PIC_OPERAND_P): Use
1429 arc_raw_symbolic_reference_mentioned_p function.
1430 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
1435 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1436 Andrew Burgess <andrew.burgess@embecosm.com>
1438 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
1439 (arc_return_address_register): New function.
1440 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
1441 (arc_handle_fndecl_attribute): Add naked attribute.
1442 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
1443 (TARGET_WARN_FUNC_RETURN): Likewise.
1444 (arc_allocate_stack_slots_for_args): New function.
1445 (arc_warn_func_return): Likewise.
1446 (machine_function): Change type fn_type.
1447 (arc_compute_function_type): Consider new naked function type,
1448 change function return type.
1449 (arc_must_save_register): Adapt to handle new
1450 arc_compute_function_type's return type.
1451 (arc_expand_prologue): Likewise.
1452 (arc_expand_epilogue): Likewise.
1453 (arc_return_address_regs): Delete.
1454 (arc_return_address_register): New function.
1455 (arc_epilogue_uses): Use above function.
1456 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
1457 (arc_function_type): Change encoding, add naked type.
1458 (ARC_INTERRUPT_P): Change to handle the new encoding.
1459 (ARC_FAST_INTERRUPT_P): Likewise.
1460 (ARC_NORMAL_P): Define.
1461 (ARC_NAKED_P): Likewise.
1462 (arc_compute_function_type): Delete prototype.
1463 * config/arc/arc.md (in_ret_delay_slot): Use
1464 arc_return_address_register function.
1465 (simple_return): Likewise.
1466 (p_return_i): Likewise.
1468 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1470 PR tree-optimization/81428
1471 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
1472 can't be built for those types.
1474 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
1476 Remove stuff dead since r239246.
1478 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
1479 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
1480 (avr_inform_devices): Remove dead stuff.
1482 2017-07-17 Tamar Christina <tamar.christina@arm.com>
1484 * config/arm/arm_neon.h: Fix softp typo.
1486 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1488 PR tree-optimization/81365
1489 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
1490 aggregate moves onto bb predecessor edges, make sure there are no
1491 loads that could alias the lhs in between the start of bb and the
1494 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
1497 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
1498 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
1499 [LSHIFTRT, outer_code = TRUNCATE]: Same.
1501 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1503 PR tree-optimization/81396
1504 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
1505 (init_symbolic_number): Initialize it to 1.
1506 (perform_symbolic_merge): Add n_ops from both operands into the new
1508 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
1509 without base_addr as useless if they need more than one operation.
1510 (bswap_replace): Handle !bswap case for NULL base_addr.
1512 2017-07-17 Tom de Vries <tom@codesourcery.com>
1515 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
1518 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
1520 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
1521 conditional builtin define __FIX_LEON3FT_B2BST.
1523 2017-07-17 Daniel Cederman <cederman@gaisler.com>
1525 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
1526 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
1529 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1531 PR rtl-optimization/81424
1532 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
1533 to remove potential trapping from operands if -fnon-call-exceptions.
1535 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1537 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
1538 profile_proability for scalling.
1539 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
1541 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1543 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
1545 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1547 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
1548 fixpoint arithmetics.
1550 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1552 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
1553 fixpoint arithmetics.
1555 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1557 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
1558 fixpoint arithmetics.
1560 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1562 * profile-count.h (profile_probability::from_reg_br_prob_note,
1563 profile_probability::to_reg_br_prob_note): New functions.
1564 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
1565 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
1566 * predict.c (probability_reliable_p): Update.
1567 (edge_probability_reliable_p): Update.
1568 (br_prob_note_reliable_p): Update.
1569 (invert_br_probabilities): Update.
1570 (add_reg_br_prob_note): New function.
1571 (combine_predictions_for_insn): Update.
1572 * asan.c (asan_clear_shadow): Update.
1573 * cfgbuild.c (compute_outgoing_frequencies): Update.
1574 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
1575 (update_br_prob_note): Update.
1576 (rtl_verify_edges): Update.
1577 (purge_dead_edges): Update.
1578 (fixup_reorder_chain): Update.
1579 * emit-rtl.c (try_split): Update.
1580 * ifcvt.c (cond_exec_process_insns): Update.
1581 (cond_exec_process_if_block): Update.
1582 (dead_or_predicable): Update.
1583 * internal-fn.c (expand_addsub_overflow): Update.
1584 (expand_neg_overflow): Update.
1585 (expand_mul_overflow): Update.
1586 * loop-doloop.c (doloop_modify): Update.
1587 * loop-unroll.c (compare_and_jump_seq): Update.
1588 * optabs.c (emit_cmp_and_jump_insn_1): Update.
1589 * predict.h: Update.
1590 * reorg.c (mostly_true_jump): Update.
1592 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
1593 * config/alpha/alpha.c (emit_unlikely_jump): Update.
1594 * config/arc/arc.c: (emit_unlikely_jump): Update.
1595 * config/arm/arm.c: (emit_unlikely_jump): Update.
1596 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
1597 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
1598 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
1599 (ix86_print_operand): Update.
1600 (ix86_split_fp_branch): Update.
1601 (predict_jump): Update.
1602 * config/ia64/ia64.c (ia64_print_operand): Update.
1603 * config/mmix/mmix.c (mmix_print_operand): Update.
1604 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
1605 (rs6000_expand_split_stack_prologue): Update.
1606 * config/rs6000/rs6000.c: Update.
1607 * config/s390/s390.c (s390_expand_vec_strlen): Update.
1608 (s390_expand_vec_movstr): Update.
1609 (s390_expand_cs_tdsi): Update.
1610 (s390_expand_split_stack_prologue): Update.
1611 * config/sh/sh.c (sh_print_operand): Update.
1612 (expand_cbranchsi4): Update.
1613 (expand_cbranchdi4): Update.
1614 * config/sparc/sparc.c (output_v9branch): Update.
1615 * config/spu/spu.c (get_branch_target): Update.
1616 (ea_load_store_inline): Update.
1617 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
1618 * config/tilepro/tilepro.c: Update.
1620 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1622 * gimplify.c (mostly_copy_tree_r): Revert latest change.
1623 (gimplify_save_expr): Likewise.
1625 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1627 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
1629 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1631 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
1633 * timevar.def (TV_IPA_FNSUMMARY): Define.
1635 2017-07-16 Daniel Cederman <cederman@gaisler.com>
1637 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
1638 to back store errata sensitive sequence from being generated.
1639 (sqrtdf2_fix): Likewise.
1641 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1643 * tree-ssa-threadupdate.c (compute_path_counts,
1644 update_joiner_offpath_counts): Use profile_probability.
1646 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
1649 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1651 * config/arm/arm-c.c (arm_cpu_builtins): Define
1652 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1654 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
1656 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1657 array entries to represent __ieee128 versions of the
1658 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
1659 scalar_extract_sig, and scalar_insert_exp built-in functions.
1660 (altivec_resolve_overloaded_builtin): Add special case handling
1661 for the __builtin_scalar_insert_exp function, as represented by
1662 the P9V_BUILTIN_VEC_VSIEDP constant.
1663 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
1664 exponent support for __ieee128 argument.
1665 (VSESQP): Add scalar extract signature support for __ieee128
1667 (VSTDCNQP): Add scalar test negative support for __ieee128
1669 (VSIEQP): Add scalar insert exponent support for __int128 argument
1670 with __ieee128 result.
1671 (VSIEQPF): Add scalar insert exponent support for __ieee128
1672 argument with __ieee128 result.
1673 (VSTDCQP): Add scalar test data class support for __ieee128
1675 (VSTDCNQP): Add overload support for scalar test negative with
1677 (VSTDCQP): Add overload support for scalar test data class
1679 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
1681 (UNSPEC_VSX_SIEXPQP): New constant.
1682 (xsxexpqp): New insn for VSX scalar extract exponent quad
1684 (xsxsigqp): New insn for VSX scalar extract significand quad
1686 (xsiexpqpf): New insn for VSX scalar insert exponent quad
1687 precision with floating point argument.
1688 (xststdcqp): New expand for VSX scalar test data class quad
1690 (xststdcnegqp): New expand for VSX scalar test negative quad
1692 (xststdcqp): New insn to match expansions for VSX scalar test data
1693 class quad precision and VSX scalar test negative quad precision.
1694 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
1695 special case operand checking to enforce that second operand of
1696 VSX scalar test data class with quad precision argument is a 7-bit
1698 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
1699 prototypes and descriptions of __ieee128 versions of
1700 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
1701 scalar_test_data_class, and scalar_test_neg built-in functions.
1703 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1705 PR tree-optimization/81162
1706 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
1707 replace a negate with an add.
1709 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
1711 * doc/invoke.texi (arm/-mcpu): Document +crypto.
1713 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1715 * config/arm/arm-c.c (arm_cpu_builtins): Define
1716 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1718 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1720 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
1721 (armv8-r): Set ARM Cortex-R52 as default CPU.
1722 * config/arm/arm-tables.opt: Regenerate.
1723 * config/arm/arm-tune.md: Regenerate.
1724 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
1726 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
1727 extension for -mcpu=cortex-r52.
1729 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1731 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
1732 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
1733 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
1734 (fp-armv8): Define it as FP_ARMv8 only.
1735 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
1736 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
1738 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
1739 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
1740 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
1741 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
1742 than TARGET_FPU_ARMV8.
1743 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
1744 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
1745 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
1747 * config/arm/neon.md (neon_vrint): Likewise.
1748 (neon_vcvt): Likewise.
1749 (neon_<fmaxmin_op><mode>): Likewise.
1750 (<fmaxmin><mode>3): Likewise.
1751 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
1752 * config/arm/predicates.md (arm_cond_move_operator): Check against
1753 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
1755 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
1757 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
1760 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1762 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
1763 loop in comment with memset.
1765 2017-07-14 Martin Liska <mliska@suse.cz>
1767 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
1768 * dwarf2out.c (is_java): Remove the function.
1769 (output_pubname): Remove usage of the function.
1770 (lower_bound_default): Remove usage of DW_LANG_Java.
1771 (gen_compile_unit_die): Likewise.
1772 * gcc.c: Remove compiler defaults for .java and .zip files.
1773 * gimple-expr.c (remove_suffix): Change as there's no longer
1774 extension than 4-letter one.
1775 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
1776 (gimplify_save_expr): Likewise.
1777 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
1778 as it's possible even for other languages than Java.
1779 * langhooks.h (struct lang_hooks): Remove Java from a comment.
1780 * lto-opts.c (lto_write_options): Remove reference to Java.
1781 * opts.c (strip_off_ending): Update file extension handling.
1782 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
1783 * tree-eh.c (lower_resx): Likewise.
1784 * tree.c (free_lang_data_in_type): Remove dead code.
1785 (find_decls_types_r): Likewise.
1786 (build_common_builtin_nodes): Remove Java from a comment.
1787 (verify_type): Remove dead code.
1788 * varasm.c (assemble_external): Remove Java from a comment.
1790 2017-07-14 Martin Liska <mliska@suse.cz>
1792 * opts.c (finish_options): Add quotes.
1793 (common_handle_option): Likewise.
1795 2017-07-14 Martin Liska <mliska@suse.cz>
1797 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
1798 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
1800 * dwarf2out.c (lower_bound_default): Do not handle
1802 (gen_compile_unit_die): Likewise.
1803 * gcc.c: Remove default extension binding for GNU Pascal.
1804 * stmt.c: Remove Pascal language from a comment.
1805 * xcoffout.c: Likewise.
1807 2017-07-13 David Malcolm <dmalcolm@redhat.com>
1810 * diagnostic-show-locus.c (fixit_cmp): New function.
1811 (layout::layout): Sort m_fixit_hints.
1812 (column_range::column_range): Assert that the values are valid.
1813 (struct char_span): New struct.
1814 (correction::overwrite): New method.
1815 (struct source_line): New struct.
1816 (line_corrections::add_hint): Add assertions. Reimplement memcpy
1817 calls in terms of classes source_line and char_span, and
1818 correction::overwrite.
1819 (selftest::test_overlapped_fixit_printing_2): New function.
1820 (selftest::diagnostic_show_locus_c_tests): Call it.
1822 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
1824 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
1825 early if there is no lhs.
1827 2017-07-13 Martin Liska <mliska@suse.cz>
1829 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
1830 (gen_reference_type_die): Likewise.
1831 * stor-layout.c: Remove Pascal-related comment.
1833 2017-07-13 Martin Liska <mliska@suse.cz>
1835 * opts.c (finish_options): Add quotes to error messages.
1836 (parse_sanitizer_options): Likewise.
1838 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
1840 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
1842 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
1844 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
1846 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
1848 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
1850 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
1852 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1855 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
1856 provides the hardware capability bits, define the macro
1857 __BUILTIN_CPU_SUPPORTS__.
1858 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
1859 if GLIBC does not provide the hardware capability bits. Add a
1860 gcc_unreachable call if the built-in cpu function is neither
1861 __builtin_cpu_is nor __builtin_cpu_supports.
1862 (rs6000_get_function_versions_dispatcher): Change the warning
1863 that an old GLIBC is used which does not export the capability
1864 bits to be an error.
1865 * doc/extend.texi (target_clones attribute): Document the
1866 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
1867 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
1868 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
1869 the macros defined by GCC if the newer GLIBC is available.
1871 2017-07-12 Jeff Law <law@redhat.com>
1873 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
1874 remaining includes slightly.
1875 * config/riscv/riscv-builtins.c: Include profile-count.h.
1877 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1880 * config/avr/avr.c (avr_set_current_function): In diagnostic
1881 messages: Quote keywords and (parts of) identifiers.
1882 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
1885 2017-07-12 Carl Love <cel@us.ibm.com>
1887 * config/rs6000/rs6000-c.c: Add support for built-in functions
1888 vector bool char vec_revb (vector bool char);
1889 vector bool short vec_revb (vector short char);
1890 vector bool int vec_revb (vector bool int);
1891 vector bool long long vec_revb (vector bool long long);
1892 * doc/extend.texi: Update the built-in documentation file for the
1893 new built-in functions.
1895 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1897 * config/s390/s390.md: Remove movcc splitter.
1899 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1901 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
1902 load/store on condition.
1904 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1907 * config/avr/avr.c (avr_encode_section_info)
1908 [progmem && !TREE_READONLY]: Error if progmem object needs
1911 2017-07-11 Michael Collison <michael.collison@arm.com>
1913 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
1916 2017-07-11 Carl Love <cel@us.ibm.com>
1918 * config/rs6000/rs6000-c.c: Add support for builtins
1919 vector unsigned int vec_parity_lsbb (vector signed int);
1920 vector unsigned int vec_parity_lsbb (vector unsigned int);
1921 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
1922 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
1923 vector unsigned long long vec_parity_lsbb (vector signed long long);
1924 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
1925 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
1926 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
1927 * doc/extend.texi: Update the built-in documentation file for the
1928 new built-in functions.
1930 2017-07-11 David Malcolm <dmalcolm@redhat.com>
1932 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
1933 (layout::m_primary_loc): New field.
1934 (layout::layout): Initialize new field. Move location filtering
1935 logic from here to...
1936 (layout::maybe_add_location_range): ...this new method. Add
1937 support for filtering to just the lines already specified by other
1939 (layout::will_show_line_p): New method.
1940 (gcc_rich_location::add_location_if_nearby): New method.
1941 (selftest::test_add_location_if_nearby): New test function.
1942 (selftest::diagnostic_show_locus_c_tests): Call it.
1943 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
1946 2017-07-11 Tom de Vries <tom@codesourcery.com>
1948 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
1949 (bb_first_real_insn): New function.
1950 (nvptx_single): Add extra initialization of broadcasted condition
1953 2017-07-11 Nathan Sidwell <nathan@acm.org>
1955 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
1957 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
1959 * doc/extend.texi (AVR Function Attributes): Remove weblink to
1960 Binutils doc as TEXI will mess them up.
1961 * doc/invoke.texi (AVR Options): Same here.
1963 2017-07-11 Daniel Cederman <cederman@gaisler.com>
1965 * config/sparc/sparc.opt (mfix-ut700): New option.
1966 (mfix-gr712rc): Likewise.
1967 (sparc_fix_b2bst): New variable.
1968 * doc/invoke.texi (SPARC options): Document them.
1969 (ARM options): Fix warnings.
1970 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
1971 instructions to prevent sequences that can trigger the store-store
1972 errata for certain LEON3FT processors.
1973 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
1974 (sparc_option_override): Set sparc_fix_b2bst appropriately.
1975 * config/sparc/sparc.md (fix_b2bst): New attribute.
1976 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
1978 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
1981 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
1983 (*rsqrtsf2_sse): Ditto.
1985 (div<mode>3): Macroize insn from divdf3 and divsf3
1986 using MODEF mode iterator.
1988 2017-07-10 Martin Sebor <msebor@redhat.com>
1990 PR tree-optimization/80397
1991 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
1992 instead of testing for equality to INTEGER_TYPE.
1994 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
1996 * config.gcc: Remove uclibc from arc target spec.
1998 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
2000 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
2002 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
2005 * lto-wrapper.c (remove_option): New function.
2006 (merge_and_complain): Merge PIC/PIE options more realistically.
2008 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
2010 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
2014 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
2015 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
2016 * config.in: Regenerate.
2017 * configure: Regenerate.
2018 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
2019 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
2020 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
2021 (TARGET_GASISR_PROLOGUES): ...target mask.
2022 * common/config/avr/avr-common.c
2023 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
2024 Set -mgas-isr-prologues.
2025 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
2026 INSERT_PASS_BEFORE for it.
2027 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
2028 * config/avr/avr.c (avr_option_override)
2029 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
2030 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
2031 (avr_attribute_table) <no_gccisr>: Add new function attribute.
2032 (avr_set_current_function) <is_no_gccisr>: Init machine field.
2033 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
2035 (make_avr_pass_pre_proep): New function.
2036 (emit_push_sfr) <treg>: Add argument to function and use it
2038 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
2039 and set machine->gasisr.yes.
2040 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
2041 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
2042 __gcc_isr.n_pushed to .L__stack_usage.
2043 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
2044 (avr_asm_final_postscan_insn): ...this new static function.
2045 * config/avr/avr.h (machine_function)
2046 <is_no_gccisr, use_L__stack_usage>: New fields.
2047 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
2048 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
2049 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
2050 (gasisr, *gasisr): New expander and insn.
2051 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
2052 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
2053 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
2055 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
2057 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
2060 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
2062 Move jump-tables out of .text again.
2065 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
2066 (ASM_OUTPUT_ADDR_VEC): New function.
2067 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
2068 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
2069 INSN_ADDRESSes as asm comment.
2070 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
2071 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
2072 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
2073 * config/avr/avr.md (*tablejump): Adjust comment.
2074 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
2075 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
2077 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
2078 (avr_output_addr_vec): New proto.
2079 (avr_log_t) <insn_addresses>: New field.
2081 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
2084 * config/i386/i386.c (ix86_function_arg_advance): Set
2085 outgoing_args_on_stack to true if there are outgoing arguments
2087 (ix86_function_arg): Likewise.
2088 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
2089 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
2090 * config/i386/i386.h (machine_function): Add
2091 outgoing_args_on_stack.
2093 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
2095 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
2097 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
2099 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
2101 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
2102 (REAL_H): Remove $(MACHMODE_H).
2103 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
2105 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
2106 $(MACHMODE_H) and double-int.h.
2107 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
2109 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
2112 2017-07-07 Andrew Pinski <apinski@cavium.com>
2114 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
2115 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
2117 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
2119 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
2120 Add warning if GCC was not configured to link against a GLIBC that
2121 exports the hardware capability bits.
2122 (make_resolver_func): Make resolver function private and not a
2123 COMDAT function. Create the name with clone_function_name instead
2124 of make_unique_name.
2127 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
2128 correct operand in doing the split.
2130 2017-07-07 Carl Love <cel@us.ibm.com>
2132 * config/rs6000/rs6000-c: Add support for built-in function
2133 vector unsigned short vec_pack_to_short_fp32 (vector float,
2135 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
2136 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
2137 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
2138 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
2139 (convert_4f32_8i16): Add define_expand.
2140 * doc/extend.texi: Update the built-in documentation file for the
2141 new built-in function.
2143 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2145 * config/sparc/m8.md: New file.
2146 * config/sparc/sparc.md: Include m8.md.
2148 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2150 * config/sparc/sparc.opt: New option -mvis4b.
2151 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
2152 (sparc_option_override): Handle VIS4B.
2153 (enum sparc_builtins): Define
2154 SPARC_BUILTIN_DICTUNPACK{8,16,32},
2155 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
2156 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
2157 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
2158 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
2159 (check_constant_argument): New function.
2160 (sparc_vis_init_builtins): Define builtins
2161 __builtin_vis_dictunpack{8,16,32},
2162 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
2163 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
2164 __builtin_vis_fpcmpde{8,16,32}shl and
2165 __builtin_vis_fpcmpur{8,16,32}shl.
2166 (sparc_expand_builtin): Check that the constant operands to
2167 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
2168 constant and in range.
2169 * config/sparc/sparc-c.c (sparc_target_macros): Handle
2171 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
2172 (SPARC_IMM5_P): Likewise.
2173 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
2174 (enabled): Handle vis4b.
2175 (UNSPEC_DICTUNPACK): New unspec.
2176 (UNSPEC_FPCMPSHL): Likewise.
2177 (UNSPEC_FPUCMPSHL): Likewise.
2178 (UNSPEC_FPCMPDESHL): Likewise.
2179 (UNSPEC_FPCMPURSHL): Likewise.
2180 (cpu_feature): New CPU feature `vis4b'.
2181 (dictunpack{8,16,32}): New insns.
2182 (FPCSMODE): New mode iterator.
2183 (fpcscond): New code iterator.
2184 (fpcsucond): Likewise.
2185 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
2186 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
2187 (fpcmpde{8,16,32}{si,di}shl): Likewise.
2188 (fpcmpur{8,16,32}{si,di}shl): Likewise.
2189 * config/sparc/constraints.md: Define constraints `q' for unsigned
2190 2-bit integer constants and `t' for unsigned 5-bit integer
2192 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
2194 (imm5_operand_dictunpack16): Likewise.
2195 (imm5_operand_dictunpack32): Likewise.
2196 (imm2_operand): Likewise.
2197 * doc/invoke.texi (SPARC Options): Document -mvis4b.
2198 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
2199 ditunpack* and fpcmp*shl builtins.
2201 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2203 * config.gcc: Handle m8 in --with-{cpu,tune} options.
2204 * config.in: Add HAVE_AS_SPARC6 define.
2205 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
2207 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
2209 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
2210 (CPP_CPU_SPEC): Handle m8.
2211 (ASM_CPU_SPEC): Likewise.
2212 * config/sparc/sparc-opts.h (enum processor_type): Add
2214 * config/sparc/sparc.c (m8_costs): New struct.
2215 (sparc_option_override): Handle TARGET_CPU_m8.
2216 (sparc32_initialize_trampoline): Likewise.
2217 (sparc64_initialize_trampoline): Likewise.
2218 (sparc_issue_rate): Likewise.
2219 (sparc_register_move_cost): Likewise.
2220 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
2221 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
2222 (ASM_CPU64_DEFAULT_SPEC): Likewise.
2223 (CPP_CPU_SPEC): Handle M8.
2224 (ASM_CPU_SPEC): Likewise.
2225 (AS_M8_FLAG): Define.
2226 * config/sparc/sparc.md: Add m8 to the cpu attribute.
2227 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
2228 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
2230 * configure: Regenerate.
2231 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
2234 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2236 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
2238 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
2239 ("*movdi_insn_sp32"): Do not set v3pipe.
2240 ("*movsi_insn"): Likewise.
2241 ("*movdi_insn_sp64"): Likewise.
2242 ("*movsf_insn"): Likewise.
2243 ("*movdf_insn_sp32"): Likewise.
2244 ("*movdf_insn_sp64"): Likewise.
2245 ("*zero_extendsidi2_insn_sp64"): Likewise.
2246 ("*sign_extendsidi2_insn"): Likewise.
2247 ("*mov<VM32:mode>_insn"): Likewise.
2248 ("*mov<VM64:mode>_insn_sp64"): Likewise.
2249 ("*mov<VM64:mode>_insn_sp32"): Likewise.
2250 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
2251 ("<vlop:code><VL:mode>3"): Likewise.
2252 ("*not_<vlop:code><VL:mode>3"): Likewise.
2253 ("*nand<VL:mode>_vis"): Likewise.
2254 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
2255 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
2256 ("one_cmpl<VL:mode>2"): Likewise.
2257 ("faligndata<VM64:mode>_vis"): Likewise.
2258 ("alignaddrsi_vis"): Likewise.
2259 ("alignaddrdi_vis"): Likweise.
2260 ("alignaddrlsi_vis"): Likewise.
2261 ("alignaddrldi_vis"): Likewise.
2262 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
2263 ("bmaskdi_vis"): Likewise.
2264 ("bmasksi_vis"): Likewise.
2265 ("bshuffle<VM64:mode>_vis"): Likewise.
2266 ("cmask8<P:mode>_vis"): Likewise.
2267 ("cmask16<P:mode>_vis"): Likewise.
2268 ("cmask32<P:mode>_vis"): Likewise.
2269 ("pdistn<P:mode>_vis"): Likewise.
2270 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
2272 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2274 * config/sparc/sparc.md ("subtype"): New insn attribute.
2275 ("*wrgsr_sp64"): Set insn subtype.
2276 ("*rdgsr_sp64"): Likewise.
2277 ("alignaddrsi_vis"): Likewise.
2278 ("alignaddrdi_vis"): Likewise.
2279 ("alignaddrlsi_vis"): Likewise.
2280 ("alignaddrldi_vis"): Likewise.
2281 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
2282 ("fexpand_vis"): Likewise.
2283 ("fpmerge_vis"): Likewise.
2284 ("faligndata<VM64:mode>_vis"): Likewise.
2285 ("bshuffle<VM64:mode>_vis"): Likewise.
2286 ("cmask8<P:mode>_vis"): Likewise.
2287 ("cmask16<P:mode>_vis"): Likewise.
2288 ("cmask32<P:mode>_vis"): Likewise.
2289 ("fchksm16_vis"): Likewise.
2290 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
2291 ("fmean16_vis"): Likewise.
2292 ("fp<plusminus_insn>64_vis"): Likewise.
2293 ("<plusminus_insn>v8qi3"): Likewise.
2294 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
2295 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
2296 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
2297 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
2298 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
2299 ("*movqi_insn"): Likewise.
2300 ("*movhi_insn"): Likewise.
2301 ("*movsi_insn"): Likewise.
2302 ("movsi_pic_gotdata_op"): Likewise.
2303 ("*movdi_insn_sp32"): Likewise.
2304 ("*movdi_insn_sp64"): Likewise.
2305 ("movdi_pic_gotdata_op"): Likewise.
2306 ("*movsf_insn"): Likewise.
2307 ("*movdf_insn_sp32"): Likewise.
2308 ("*movdf_insn_sp64"): Likewise.
2309 ("*zero_extendhisi2_insn"): Likewise.
2310 ("*zero_extendqihi2_insn"): Likewise.
2311 ("*zero_extendqisi2_insn"): Likewise.
2312 ("*zero_extendqidi2_insn"): Likewise.
2313 ("*zero_extendhidi2_insn"): Likewise.
2314 ("*zero_extendsidi2_insn_sp64"): Likewise.
2315 ("ldfsr"): Likewise.
2316 ("prefetch_64"): Likewise.
2317 ("prefetch_32"): Likewise.
2318 ("tie_ld32"): Likewise.
2319 ("tie_ld64"): Likewise.
2320 ("*tldo_ldub_sp32"): Likewise.
2321 ("*tldo_ldub1_sp32"): Likewise.
2322 ("*tldo_ldub2_sp32"): Likewise.
2323 ("*tldo_ldub_sp64"): Likewise.
2324 ("*tldo_ldub1_sp64"): Likewise.
2325 ("*tldo_ldub2_sp64"): Likewise.
2326 ("*tldo_ldub3_sp64"): Likewise.
2327 ("*tldo_lduh_sp32"): Likewise.
2328 ("*tldo_lduh1_sp32"): Likewise.
2329 ("*tldo_lduh_sp64"): Likewise.
2330 ("*tldo_lduh1_sp64"): Likewise.
2331 ("*tldo_lduh2_sp64"): Likewise.
2332 ("*tldo_lduw_sp32"): Likewise.
2333 ("*tldo_lduw_sp64"): Likewise.
2334 ("*tldo_lduw1_sp64"): Likewise.
2335 ("*tldo_ldx_sp64"): Likewise.
2336 ("*mov<VM32:mode>_insn"): Likewise.
2337 ("*mov<VM64:mode>_insn_sp64"): Likewise.
2338 ("*mov<VM64:mode>_insn_sp32"): Likewise.
2340 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2342 * config/sparc/sparc.md ("type"): New insn type viscmp.
2343 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
2345 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
2346 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
2347 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
2348 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
2350 ("n7_vis_logical_11cycle"): Likewise.
2351 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
2352 * config/sparc/niagara2.md ("niag3_vis": Likewise.
2353 * config/sparc/niagara.md ("niag_vis"): Likewise.
2354 * config/sparc/ultra3.md ("us3_fga"): Likewise.
2355 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
2357 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2359 * config/sparc/sparc.md: New instruction type `bmask'.
2360 (bmaskdi_vis): Use the `bmask' type.
2361 (bmasksi_vis): Likewise.
2362 * config/sparc/ultra3.md (us3_array): Likewise.
2363 * config/sparc/niagara7.md (n7_array): Likewise.
2364 * config/sparc/niagara4.md (n4_array): Likewise.
2365 * config/sparc/niagara2.md (niag2_vis): Likewise.
2366 (niag3_vis): Likewise.
2367 * config/sparc/niagara.md (niag_vis): Likewise.
2369 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2371 * ipa-comdats.c: Remove optimize check from gate.
2372 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
2373 for functions not optimized.
2374 (ipa_fn_summary_read): Skip optimize check.
2375 (ipa_fn_summary_write): Likewise.
2376 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
2378 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
2380 (can_inline_edge_p): Check flag_pcc_struct_return for match.
2381 (check_callers): Give up on caller which is not optimized.
2382 (inline_small_functions): Likewise.
2383 (ipa_inline): Do not give up when not optimizing.
2384 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
2385 away unoptimizes cdtors.
2386 (whole_program_function_and_variable_visibility): Do
2387 ipa_discover_readonly_nonaddressable_vars in LTO mode.
2388 * ipa.c (process_references): Do not check optimize.
2389 (symbol_table::remove_unreachable_nodes): Update optimize check.
2390 (set_writeonly_bit): Update optimize check.
2391 (pass_ipa_cdtor_merge::gate): Do not check optimize.
2392 (pass_ipa_single_use::gate): Remove.
2394 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2396 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
2397 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
2398 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
2399 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
2400 permute_load, permute_store, adjust_extract, adjust_splat,
2401 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
2402 replace_swap_with_copy, dump_swap_insn_table,
2403 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
2404 recombine_lvx_pattern, recombine_stvx_pattern,
2405 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
2406 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
2407 to file rs6000-p8swap.c.
2408 * config/rs6000/rs6000-p8swap.c: New file.
2409 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
2410 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
2411 and rs6000*-*-* targets.
2413 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2415 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
2417 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2419 * lto-wrapper.c (merge_and_complain): Do not merge
2420 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
2421 fsigned_zeros, ftrapping_math, fwrapv.
2422 (append_compiler_options): Do not track these options.
2423 (append_linker_options): Likewie
2425 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2427 * cgraphunit.c (cgraph_node::finalize_function): When
2428 !flag_toplevel_reorde set no_reorder flag.
2429 (varpool_node::finalize_decl): Likewise.
2430 (symbol_table::compile): Drop no toplevel reorder path.
2432 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2434 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
2435 edges; zero probability is not better than uninitialized.
2437 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
2439 * asan.h (asan_sanitize_allocas_p): Declare.
2440 * asan.c (asan_sanitize_allocas_p): New function.
2441 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
2442 (handle_builtin_alloca): Likewise.
2443 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
2444 if !asan_sanitize_allocas_p.
2445 * params.def (asan-instrument-allocas): Add new option.
2446 * params.h (ASAN_PROTECT_ALLOCAS): Define.
2447 * opts.c (common_handle_option): Disable allocas sanitization for
2450 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
2452 * asan.c: Include gimple-fold.h.
2453 (get_last_alloca_addr): New function.
2454 (handle_builtin_stackrestore): Likewise.
2455 (handle_builtin_alloca): Likewise.
2456 (asan_emit_allocas_unpoison): Likewise.
2457 (get_mem_refs_of_builtin_call): Add new parameter, remove const
2458 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
2459 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
2460 (instrument_builtin_call): Pass gimple iterator to
2461 get_mem_refs_of_builtin_call.
2462 (last_alloca_addr): New global.
2463 * asan.h (asan_emit_allocas_unpoison): Declare.
2464 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
2465 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
2466 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
2467 if function calls alloca.
2468 * gimple-fold.c (replace_call_with_value): Remove static keyword.
2469 * gimple-fold.h (replace_call_with_value): Declare.
2470 * internal-fn.c: Include asan.h.
2471 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
2472 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
2474 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2476 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
2477 (C_SELFTEST_FLAGS): New.
2478 (CPP_SELFTEST_FLAGS): New.
2479 (SELFTEST_DEPS): New, from deps of s-selftest.
2480 (C_SELFTEST_DEPS): New, from deps of s-selftest.
2481 (CPP_SELFTEST_DEPS): New.
2482 (selftest): Add dependency on s-selftest-c++.
2483 (s-selftest): Rename to...
2484 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
2485 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
2486 than SELFTEST_FLAGS.
2487 (selftest-gdb): Rename to...
2488 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
2490 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
2491 (selftest-valgrind): Rename to...
2492 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
2494 (selftest-valgrind): Reintroduce as an alias for
2495 selftest-c-valgrind.
2496 (s-selftest-c++): New.
2497 (selftest-c++-gdb): New.
2498 (selftest-c++-valgrind): New.
2500 2017-07-06 Olivier Hainque <hainque@adacore.com>
2502 * gcc.c (process_command): When deciding if undefined variables
2503 should be ignored when processing specs, accept "gcc -v" as well.
2505 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2507 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
2508 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
2510 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
2512 * config/arm/arm-cpus.in (armv8-r): Add new entry.
2513 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
2514 * config/arm/arm-tables.opt: Regenerate.
2515 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
2517 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
2519 2017-07-06 Carl Love <cel@us.ibm.com>
2521 * ChangeLog: Clean up from mid air collision
2523 2017-07-06 Carl Love <cel@us.ibm.com>
2525 * config/rs6000/rs6000-c.c: Add support for built-in functions
2526 vector signed int vec_subc (vector signed int, vector signed int);
2527 vector signed __int128 vec_subc (vector signed __int128,
2528 vector signed __int128);
2529 vector unsigned __int128 vec_subc (vector unsigned __int128,
2530 vector unsigned __int128);
2531 vector signed int vec_sube (vector signed int, vector signed int,
2533 vector unsigned int vec_sube (vector unsigned int,
2534 vector unsigned int,
2535 vector unsigned int);
2536 vector signed __int128 vec_sube (vector signed __int128,
2537 vector signed __int128,
2538 vector signed__int128);
2539 vector unsigned __int128 vec_sube (vector unsigned __int128,
2540 vector unsigned __int128,
2541 vector unsigned __int128);
2542 vector signed int vec_subec (vector signed int, vector signed int,
2544 vector unsigned int vec_subec (vector unsigned int,
2545 vector unsigned int,
2546 vector unsigned int);
2547 vector signed __int128 vec_subec (vector signed __int128,
2548 vector signed __int128,
2549 vector signed__int128);
2550 vector unsigned __int128 vec_subec (vector unsigned __int128,
2551 vector unsigned __int128,
2552 vector unsigned __int128);
2553 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
2554 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
2555 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
2556 BU_ALTIVEC_OVERLOAD_X definitions.
2557 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
2558 * doc/extend.texi: Update the built-in documentation file for the new
2561 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2564 * diagnostic-show-locus.c (layout::layout): Use start and finish
2565 spelling location for the start and finish of each range.
2566 * genmatch.c (linemap_client_expand_location_to_spelling_point):
2567 Add unused aspect param.
2568 * input.c (expand_location_1): Add "aspect" param, and use it
2569 to access the correct part of the location.
2570 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
2572 (expand_location_to_spelling_point): Likewise.
2573 (linemap_client_expand_location_to_spelling_point): Add "aspect"
2574 param, and pass it to expand_location_1.
2576 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
2578 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
2579 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
2580 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
2581 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
2582 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
2583 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
2584 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
2585 _mm_maskz_getmant_ss): New intrinsics.
2586 (__builtin_ia32_getexpss128_mask): Changed to ...
2587 __builtin_ia32_getexpss128_round ... this.
2588 (__builtin_ia32_getexpsd128_mask): Changed to ...
2589 __builtin_ia32_getexpsd128_round ... this.
2590 * config/i386/i386-builtin-types.def
2591 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
2592 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
2593 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
2594 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
2595 __builtin_ia32_getmantss_mask_round): New builtins.
2596 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
2597 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
2598 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
2599 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
2600 * config/i386/sse.md
2601 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
2602 avx512f_sgetexp<mode><mask_scalar_name>
2603 <round_saeonly_scalar_name> ... this.
2604 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
2605 %0, %1, %2<round_saeonly_op3>}): Changed to ...
2606 vgetexp<ssescalarmodesuffix>
2607 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2608 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
2609 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
2610 avx512f_vgetmant<mode><mask_scalar_name>
2611 <round_saeonly_scalar_name> ... this.
2612 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
2613 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
2614 vgetmant<ssescalarmodesuffix>
2615 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
2616 %0<mask_scalar_operand4>, %1, %2
2617 <round_saeonly_scalar_mask_op4>, %3} ... this.
2618 * config/i386/subst.md (mask_scalar_operand4,
2619 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
2620 round_saeonly_scalar_nimm_predicate): New subst attributes.
2622 2017-07-06 Julia Koval <julia.koval@intel.com>
2624 * config/i386/i386.c (ix86_erase_embedded_rounding):
2625 Remove code for old rounding pattern.
2627 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
2629 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
2631 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
2633 * doc/sourcebuild.texi (Test Directives, Variants of
2634 dg-require-support): Add documentation for dg-require-stack-check.
2636 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
2638 * config/i386/subst.md (mask_scalar, round_scalar,
2639 round_saeonly_scalar): New meta-templates.
2640 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
2641 round_scalar_mask_operand3, round_scalar_mask_op3,
2642 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
2643 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
2644 round_saeonly_scalar_constraint,
2645 round_saeonly_scalar_prefix): New subst attribute.
2646 * config/i386/sse.md
2647 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
2648 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
2649 <round_scalar_name> ... this.
2650 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
2651 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
2652 <round_scalar_name> ... this.
2653 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
2654 <sse>_vm<code><mode>3<mask_scalar_name>
2655 <round_saeonly_scalar_name> ... this.
2656 (v<plusminus_mnemonic><ssescalarmodesuffix>
2657 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2658 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2659 v<plusminus_mnemonic><ssescalarmodesuffix>
2660 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2661 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2662 (v<multdiv_mnemonic><ssescalarmodesuffix>
2663 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2664 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2665 v<multdiv_mnemonic><ssescalarmodesuffix>
2666 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2667 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2668 (v<maxmin_float><ssescalarmodesuffix>
2669 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
2670 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
2671 v<maxmin_float><ssescalarmodesuffix>
2672 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2673 %0<mask_scalar_operand3>, %1, %<iptr>2
2674 <round_saeonly_scalar_mask_op3>} ... this.
2676 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
2678 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
2679 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
2681 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
2682 Alan Hayward <alan.hayward@arm.com>
2683 David Sherwood <david.sherwood@arm.com>
2685 * combine.c (simplify_if_then_else): Remove "enum" before
2687 * compare-elim.c (can_eliminate_compare): Likewise.
2688 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
2690 (aarch64_lookup_simd_builtin_type): Likewise.
2691 (aarch64_simd_builtin_type): Likewise.
2692 (aarch64_init_simd_builtin_types): Likewise.
2693 (aarch64_simd_expand_args): Likewise.
2694 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
2696 (aarch64_reverse_mask): Likewise.
2697 (aarch64_simd_emit_reg_reg_move): Likewise.
2698 (aarch64_gen_adjusted_ldpstp): Likewise.
2699 (aarch64_ccmp_mode_to_code): Likewise.
2700 (aarch64_operands_ok_for_ldpstp): Likewise.
2701 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2702 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
2704 (aarch64_min_divisions_for_recip_mul): Likewise.
2705 (aarch64_reassociation_width): Likewise.
2706 (aarch64_get_condition_code_1): Likewise.
2707 (aarch64_simd_emit_reg_reg_move): Likewise.
2708 (aarch64_simd_attr_length_rglist): Likewise.
2709 (aarch64_reverse_mask): Likewise.
2710 (aarch64_operands_ok_for_ldpstp): Likewise.
2711 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2712 (aarch64_gen_adjusted_ldpstp): Likewise.
2713 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
2715 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
2716 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
2717 (arm_lookup_simd_builtin_type): Likewise.
2718 (arm_simd_builtin_type): Likewise.
2719 (arm_init_simd_builtin_types): Likewise.
2720 (arm_expand_builtin_args): Likewise.
2721 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
2722 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
2723 (ft32_setup_incoming_varargs): Likewise.
2724 (ft32_function_arg): Likewise.
2725 (ft32_function_arg_advance): Likewise.
2726 (ft32_pass_by_reference): Likewise.
2727 (ft32_arg_partial_bytes): Likewise.
2728 (ft32_valid_pointer_mode): Likewise.
2729 (ft32_addr_space_pointer_mode): Likewise.
2730 (ft32_addr_space_legitimate_address_p): Likewise.
2731 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
2733 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
2734 (ix86_emit_outlined_ms2sysv_restore): Likewise.
2735 (iamcu_alignment): Likewise.
2736 (canonicalize_vector_int_perm): Likewise.
2737 (ix86_noce_conversion_profitable_p): Likewise.
2738 (ix86_mpx_bound_mode): Likewise.
2739 (ix86_operands_ok_for_move_multiple): Likewise.
2740 * config/microblaze/microblaze-protos.h
2741 (microblaze_expand_conditional_branch_reg): Likewise.
2742 * config/microblaze/microblaze.c
2743 (microblaze_expand_conditional_branch_reg): Likewise.
2744 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
2746 (rs6000_reassociation_width): Likewise.
2747 (rs6000_invalid_binary_op): Likewise.
2748 (fusion_p9_p): Likewise.
2749 (emit_fusion_p9_load): Likewise.
2750 (emit_fusion_p9_store): Likewise.
2751 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
2753 (riscv_hard_regno_mode_ok_p): Likewise.
2754 (riscv_address_insns): Likewise.
2755 (riscv_split_symbol): Likewise.
2756 (riscv_legitimize_move): Likewise.
2757 (riscv_function_value): Likewise.
2758 (riscv_hard_regno_nregs): Likewise.
2759 (riscv_expand_builtin): Likewise.
2760 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
2761 (riscv_build_integer): Likewise.
2762 (riscv_split_integer): Likewise.
2763 (riscv_legitimate_constant_p): Likewise.
2764 (riscv_cannot_force_const_mem): Likewise.
2765 (riscv_regno_mode_ok_for_base_p): Likewise.
2766 (riscv_valid_base_register_p): Likewise.
2767 (riscv_valid_offset_p): Likewise.
2768 (riscv_valid_lo_sum_p): Likewise.
2769 (riscv_classify_address): Likewise.
2770 (riscv_legitimate_address_p): Likewise.
2771 (riscv_address_insns): Likewise.
2772 (riscv_load_store_insns): Likewise.
2773 (riscv_force_binary): Likewise.
2774 (riscv_split_symbol): Likewise.
2775 (riscv_force_address): Likewise.
2776 (riscv_legitimize_address): Likewise.
2777 (riscv_move_integer): Likewise.
2778 (riscv_legitimize_const_move): Likewise.
2779 (riscv_legitimize_move): Likewise.
2780 (riscv_address_cost): Likewise.
2781 (riscv_subword): Likewise.
2782 (riscv_output_move): Likewise.
2783 (riscv_canonicalize_int_order_test): Likewise.
2784 (riscv_emit_int_order_test): Likewise.
2785 (riscv_function_arg_boundary): Likewise.
2786 (riscv_pass_mode_in_fpr_p): Likewise.
2787 (riscv_pass_fpr_single): Likewise.
2788 (riscv_pass_fpr_pair): Likewise.
2789 (riscv_get_arg_info): Likewise.
2790 (riscv_function_arg): Likewise.
2791 (riscv_function_arg_advance): Likewise.
2792 (riscv_arg_partial_bytes): Likewise.
2793 (riscv_function_value): Likewise.
2794 (riscv_pass_by_reference): Likewise.
2795 (riscv_setup_incoming_varargs): Likewise.
2796 (riscv_print_operand): Likewise.
2797 (riscv_elf_select_rtx_section): Likewise.
2798 (riscv_save_restore_reg): Likewise.
2799 (riscv_for_each_saved_reg): Likewise.
2800 (riscv_register_move_cost): Likewise.
2801 (riscv_hard_regno_mode_ok_p): Likewise.
2802 (riscv_hard_regno_nregs): Likewise.
2803 (riscv_class_max_nregs): Likewise.
2804 (riscv_memory_move_cost): Likewise.
2805 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
2806 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
2807 (rl78_addr_space_address_mode): Likewise.
2808 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2810 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2811 (rs6000_reassociation_width): Likewise.
2812 (rs6000_invalid_binary_op): Likewise.
2813 (fusion_p9_p): Likewise.
2814 (emit_fusion_p9_load): Likewise.
2815 (emit_fusion_p9_store): Likewise.
2816 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
2817 (ok_for_simple_move_operands): Likewise.
2818 (ok_for_simple_move_strict_operands): Likewise.
2819 (ok_for_simple_arith_logic_operands): Likewise.
2820 (visium_legitimize_reload_address): Likewise.
2821 (visium_select_cc_mode): Likewise.
2822 (output_cbranch): Likewise.
2823 (visium_split_double_move): Likewise.
2824 (visium_expand_copysign): Likewise.
2825 (visium_expand_int_cstore): Likewise.
2826 (visium_expand_fp_cstore): Likewise.
2827 * config/visium/visium.c (visium_pass_by_reference): Likewise.
2828 (visium_function_arg): Likewise.
2829 (visium_function_arg_advance): Likewise.
2830 (visium_libcall_value): Likewise.
2831 (visium_setup_incoming_varargs): Likewise.
2832 (visium_legitimate_constant_p): Likewise.
2833 (visium_legitimate_address_p): Likewise.
2834 (visium_legitimize_address): Likewise.
2835 (visium_secondary_reload): Likewise.
2836 (visium_register_move_cost): Likewise.
2837 (visium_memory_move_cost): Likewise.
2838 (prepare_move_operands): Likewise.
2839 (ok_for_simple_move_operands): Likewise.
2840 (ok_for_simple_move_strict_operands): Likewise.
2841 (ok_for_simple_arith_logic_operands): Likewise.
2842 (visium_function_value_1): Likewise.
2843 (rtx_ok_for_offset_p): Likewise.
2844 (visium_legitimize_reload_address): Likewise.
2845 (visium_split_double_move): Likewise.
2846 (visium_expand_copysign): Likewise.
2847 (visium_expand_int_cstore): Likewise.
2848 (visium_expand_fp_cstore): Likewise.
2849 (visium_split_cstore): Likewise.
2850 (visium_select_cc_mode): Likewise.
2851 (visium_split_cbranch): Likewise.
2852 (output_cbranch): Likewise.
2853 (visium_print_operand_address): Likewise.
2854 * expmed.c (flip_storage_order): Likewise.
2855 * expmed.h (emit_cstore): Likewise.
2856 (flip_storage_order): Likewise.
2857 * genrecog.c (validate_pattern): Likewise.
2858 * hsa-gen.c (gen_hsa_addr): Likewise.
2859 * internal-fn.c (expand_arith_overflow): Likewise.
2860 * ira-color.c (allocno_copy_cost_saving): Likewise.
2861 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2862 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
2863 (process_invariant_for_inheritance): Likewise.
2864 * lra-eliminations.c (move_plus_up): Likewise.
2865 * omp-low.c (lower_oacc_reductions): Likewise.
2866 * simplify-rtx.c (simplify_subreg): Likewise.
2867 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
2868 (TARGET_CHKP_BOUND_MODE): Likewise..
2869 * targhooks.c (default_chkp_bound_mode): Likewise.
2870 (default_setup_incoming_vararg_bounds): Likewise.
2871 * targhooks.h (default_chkp_bound_mode): Likewise.
2872 (default_setup_incoming_vararg_bounds): Likewise.
2873 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
2874 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
2875 (have_whole_vector_shift): Likewise.
2876 * tree-vect-stmts.c (vectorizable_load): Likewise.
2877 * doc/tm.texi: Regenerate.
2879 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2881 Graceful degrade if Binutils PR21472 is not available.
2884 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
2885 .rodata in flash test fails.
2886 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
2887 * confgure: Regenerate.
2888 * config.in: Regenerate.
2889 * config/avr/avr.c (avr_asm_named_section)
2890 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
2891 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
2892 (avr_asm_init_sections): Same.
2894 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2896 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
2897 (fma<VH:mode>4_intrinsic): Likewise.
2898 (*fmsub<VCVTF:mode>4): Likewise.
2899 (*fmsub<VH:mode>4_intrinsic): Likewise.
2901 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2904 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
2905 Don't depend on "optimize > 0".
2906 (out_movhi_r_mr, out_movqi_mr_r): Same.
2907 (out_movhi_mr_r, out_movqi_r_mr): Same.
2908 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
2909 io_address_operand on "optimize > 0".
2911 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2913 * tree-loop-distribution.c: Add general explanantion on the pass.
2914 (generate_loops_for_partition): Mark distributed loop.
2915 (pg_add_dependence_edges): New parameter. Handle alias data
2916 dependence specially and record it in the parameter if asked.
2917 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
2918 (init_partition_graph_vertices, add_partition_graph_edge): New.
2919 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
2920 (free_partition_graph_vdata, build_partition_graph): New.
2921 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
2922 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
2923 (data_ref_segment_size, latch_dominated_by_data_ref): New.
2924 (compute_alias_check_pairs, version_loop_by_alias_check): New.
2925 (version_for_distribution_p, finalize_partitions): New.
2926 (distribute_loop): Handle alias data dependence specially. Factor
2927 out loop fusion code as functions and call these functions.
2929 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2931 * tree-loop-distribution.c (classify_partition): New parameter and
2932 better handle reduction statement.
2933 (rdg_build_partitions): Revise comment.
2934 (distribute_loop): Compute statements in all partitions and pass it
2935 to classify_partition.
2937 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2939 * tree-loop-distribution.c (enum partition_type): New.
2940 (struct partition): New field type.
2941 (partition_merge_into): Add parameter. Update partition type.
2942 (data_dep_in_cycle_p, update_type_for_merge): New functions.
2943 (build_rdg_partition_for_vertex): Compute partition type.
2944 (rdg_build_partitions): Dump partition type.
2945 (distribute_loop): Update calls to partition_merge_into.
2947 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2949 * tree-loop-distribution.c (struct ddr_hasher): New.
2950 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
2952 (classify_partition): Call get_data_dependence.
2953 (pg_add_dependence_edges): Ditto.
2954 (distribute_loop): Release data dependence hash table.
2956 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2958 * tree-loop-distribution.c (ref_base_address): Delete.
2959 (similar_memory_accesses): Rename ...
2960 (share_memory_accesses): ... to this. Check if partitions access
2961 the same memory reference.
2962 (distribute_loop): Call share_memory_accesses.
2964 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2966 * tree-loop-distribution.c (struct partition): New field recording
2968 (partition_alloc, partition_free): Init and release data refs.
2969 (partition_merge_into): Merge data refs.
2970 (build_rdg_partition_for_vertex): Collect data refs for partition.
2971 (pg_add_dependence_edges): Change parameters from vector to bitmap.
2973 (distribute_loop): Remve data refs from vertice data of partition
2976 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2978 * tree-loop-distribution.c (params.h): Include header file.
2979 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
2980 (datarefs_vec): New global var.
2981 (create_rdg_vertices): Use datarefs_vec directly.
2982 (free_rdg): Don't free data references.
2983 (build_rdg): Update use. Don't free data references.
2984 (distribute_loop): Compute global variable for data references.
2985 Bail out if there are too many data references.
2987 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2989 * tree-loop-distribution.c (loop_nest): New global var.
2990 (build_rdg): Use loop directly, rather than loop nest.
2991 (pg_add_dependence_edges): Remove loop nest parameter. Use global
2993 (distribute_loop): Compute global variable loop nest. Update use.
2995 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2997 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
2998 (partition_merge_into): New parameter. Dump reason for fusion.
2999 (distribute_loop): Update use of partition_merge_into.
3001 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3003 * tree-loop-distribution.c (bb_top_order_index): New.
3004 (bb_top_order_index_size, bb_top_order_cmp): New.
3005 (stmts_from_loop): Use topological order.
3006 (pass_loop_distribution::execute): Compute and release topological
3007 order for basic blocks.
3009 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3011 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
3014 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3016 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
3017 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
3018 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
3019 * internal-fn.def (LOOP_DIST_ALIAS): New.
3020 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
3021 (fold_loop_internal_call): ... this.
3022 (vect_loop_dist_alias_call): New function.
3023 (set_uid_loop_bbs): Call fold_loop_internal_call.
3024 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
3027 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
3030 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
3031 Require dead FLAGS_REG at the beginning of a peephole.
3033 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
3036 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
3037 arguments in the call to __builtin_ia32_sbb_u32.
3038 (_subborrow_u64): Swap _X and _Y arguments in the call to
3039 __builtin_ia32_sbb_u64.
3041 2017-07-04 Jakub Jelinek <jakub@redhat.com>
3044 * tree-vrp.c (compare_assert_loc): Turn into a function template
3045 with stable template parameter. Only test if a->e is NULL,
3046 !a->e == !b->e has been verified already. Use e == NULL or
3047 e != NULL instead of e or ! e tests. If stable is true, don't use
3048 iterative_hash_expr, on the other side allow a or b or both NULL
3049 and sort the NULLs last.
3050 (process_assert_insertions): Sort using compare_assert_loc<false>
3051 instead of compare_assert_loc, later sort using
3052 compare_assert_loc<true> before calling process_assert_insertions_for
3053 in a loop. Use break instead of continue once seen NULL pointer.
3055 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
3057 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
3058 Cortex-R7 and Cortex-R8 processors.
3060 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
3062 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
3063 uninitialized while src is not.
3065 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
3067 * common/config/arm/arm-common.c: Adjust include path for
3069 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
3070 (arm-cpu.h): Create in build directory. Adjust dependency rules.
3071 (arm-cpu-data.h): Likewise.
3072 (arm-cpu-cdata.h): Likewise.
3073 * config/arm/arm-cpu.h: Delete.
3074 * config/arm/arm-cpu-cdata.h: Delete.
3075 * config/arm/arm-cpu-data.h: Delete.
3077 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
3079 * config/arm/arm-cpus.in (cortex-a55): New.
3080 (cortex-a75): Likewise.
3081 (cortex-a75.cortex-a55): Likewise.
3082 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
3084 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
3085 * config/arm/arm-cpu-cdata.h: Regenerate.
3086 * config/arm/arm-cpu-data.h: Regenerate.
3087 * config/arm/arm-cpu.h: Regenerate.
3088 * config/arm/arm-tables.opt: Regenerate.
3089 * config/arm/arm-tune.md: Regenerate.
3091 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
3093 * haifa-sched.c (sched_create_recovery_edges): Update profile.
3095 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
3097 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
3100 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
3102 PR tree-optimization/81292
3103 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
3104 full_string_p, also call adjust_related_strinfos if the adjustment
3105 is simple, otherwise invalidate related strinfos.
3107 2017-07-04 Martin Liska <mliska@suse.cz>
3110 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
3111 newly created variable as DECL_IGNORED_P.
3113 2017-07-04 Martin Liska <mliska@suse.cz>
3116 * ipa-inline.c (inline_small_functions):
3117 Use xstrdup_for_dump.
3119 2017-07-04 Tom de Vries <tom@codesourcery.com>
3121 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
3123 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
3126 * config/darwin.c (darwin_function_switched_text_sections):
3129 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
3131 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
3133 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
3135 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
3137 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3139 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
3140 min_profitable_iters, and th as inclusive lower bounds.
3141 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
3142 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
3143 for min_profitable_iters and min_profitable_estimate.
3144 (vect_transform_loop): Treat th as an inclusive lower bound.
3145 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
3147 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
3150 * config/darwin.c (darwin_function_switched_text_sections):
3151 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
3152 in two pieces, and suppress the use of buf.
3154 2017-07-03 Nathan Sidwell <nathan@acm.org>
3156 * hash-table.h (hash_table_mod1): Fix indentation.
3158 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3161 * predict.c (force_edge_cold): Be more careful about propagation
3163 * profile-count.h (profile_probability::guessed,
3164 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
3166 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
3168 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
3170 * doc/invoke.texi (rcpc architecture extension): Document it.
3172 2017-07-03 Richard Biener <rguenther@suse.de>
3174 PR tree-optimization/60510
3175 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
3176 the scalar reduction PHI and use it.
3177 (vectorizable_reduction): Properly guard the single_defuse_cycle
3178 path for non-SLP reduction chains where we cannot use it.
3179 Rework reduc_def/index and vector type deduction. Rework
3180 vector operand gathering during reduction op code-gen.
3181 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
3182 chains dissolve the chain and leave it to non-SLP reduction
3185 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3187 * tree-data-ref.h (dr_alignment): Declare.
3188 * tree-data-ref.c (dr_alignment): New function.
3189 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
3190 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
3192 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
3194 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3196 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
3197 and base_misalignment fields.
3198 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
3199 * tree-data-ref.c: Include builtins.h.
3200 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
3201 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
3202 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
3203 * tree-vect-data-refs.c: Include tree-cfg.h.
3204 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
3205 fields instead of calculating an alignment here.
3206 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
3207 innermost_loop_behavior fields.
3209 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3211 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
3213 (DR_STEP_ALIGNMENT): New macro.
3214 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
3215 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
3216 (create_data_ref): Print it.
3217 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
3218 to tell whether the step preserves vector (mis)alignment.
3219 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
3220 Move the check for an integer step and generalise to all INTEGER_CST.
3221 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
3222 Print the outer step alignment.
3224 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3226 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
3227 with offset_alignment.
3228 (DR_ALIGNED_TO): Delete.
3229 (DR_OFFSET_ALIGNMENT): New macro.
3230 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
3231 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
3232 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
3233 (create_data_ref): Likewise.
3234 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
3235 (vect_analyze_data_refs): Likewise.
3236 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
3237 creating dummy innermost behavior.
3239 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3241 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
3242 with a "innermost_loop_behavior *" and refeence tree.
3243 * tree-data-ref.c (dr_analyze_innermost): Likewise.
3244 (create_data_ref): Update call accordingly.
3245 * tree-predcom.c (find_looparound_phi): Likewise.
3247 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3249 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
3250 fields with dr_wrt_vec_loop.
3251 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
3252 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
3253 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
3254 (vect_dr_behavior): New function.
3255 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
3256 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
3257 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
3258 track whether the step preserves the misalignment.
3259 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
3260 Use vect_dr_behavior.
3261 (vect_setup_realignment): Update call accordingly.
3262 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
3263 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
3264 call to vect_create_addr_base_for_vector_ref.
3265 (vect_create_cond_for_align_checks): Likewise.
3266 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
3267 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
3268 (vect_recog_mask_conversion_pattern): Likewise.
3269 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
3270 (new_stmt_vec_info): Remove redundant zeroing.
3272 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
3274 * common/config/arm/arm-common.c (arm_be8_option): New function.
3275 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
3276 (ISA_ARMv6): Add isa_bit_be8.
3277 * config/arm/arm.h (arm_be8_option): Add prototype.
3278 (BE8_SPEC_FUNCTION): New define.
3279 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
3280 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
3281 (mlittle-endian): Similarly.
3282 (mbe8, mbe32): New options.
3283 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
3284 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
3286 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3288 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
3290 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3292 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
3293 (cleanup_tree_cfg_bb): Use it.
3294 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
3296 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
3298 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3301 * loop-doloop.c (add_test): Update profile.
3303 2017-07-03 Martin Liska <mliska@suse.cz>
3306 * sanopt.c (rewrite_usage_of_param): New function.
3307 (sanitize_rewrite_addressable_params): Likewise.
3308 (pass_sanopt::execute): Call rewrite_usage_of_param.
3310 2017-07-03 Richard Biener <rguenther@suse.de>
3312 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
3313 back to using VIEW_CONVERT_EXPR.
3315 2017-07-03 Martin Liska <mliska@suse.cz>
3318 * doc/extend.texi: Document when a resolver function is
3319 generated for target_clones.
3321 2017-07-03 Martin Liska <mliska@suse.cz>
3323 * asan.c (asan_emit_stack_protection): Unpoison just red zones
3324 and shadow memory of auto variables which are subject of
3325 use-after-scope sanitization.
3326 (asan_expand_mark_ifn): Add do set only when is_poison.
3328 2016-07-03 Richard Biener <rguenther@suse.de>
3330 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
3332 (vect_force_simple_reduction): Record reduction def -> phi mapping.
3333 (vectorizable_reduction): Perform reduction PHI creation when
3334 visiting a reduction PHI and adjust and simplify code generation
3335 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
3336 (vect_transform_loop): Visit reduction PHIs.
3337 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
3338 defs into the SLP tree.
3339 (vect_build_slp_tree): Reduction defs terminate the recursion.
3340 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
3342 (vect_get_vec_defs_for_stmt_copy): Export.
3343 (vect_get_vec_defs): Likewise.
3344 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
3346 (vect_get_vec_defs_for_stmt_copy): Declare.
3347 (vect_get_vec_defs): Likewise.
3349 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3351 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
3352 parameter with a "loop" parameter and use it instead of the
3353 loop containing DR_STMT. Don't check simple_iv when doing
3354 BB analysis. Describe the two analysis modes in the comment.
3356 2017-07-03 Tom de Vries <tom@codesourcery.com>
3358 PR tree-optimization/69468
3359 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
3360 (find_same_succ_bb): Handle ignore_edge_flags.
3362 2017-07-03 Tom de Vries <tom@codesourcery.com>
3364 PR tree-optimization/81192
3365 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
3367 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
3369 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
3371 2017-07-03 Tom de Vries <tom@codesourcery.com>
3373 PR tree-optimization/81192
3374 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
3375 BB_SAME_SUCC (bb) == NULL.
3377 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3379 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
3382 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3384 * dumpfile.c: Include profile-count.h
3385 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
3387 (insert_cond_bb): Update profile.
3388 * tree-cfg.h (insert_cond_bb): Update prototype.
3389 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
3390 * tree-dump.c: Do not include tree-cfg.
3392 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3394 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
3396 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3398 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
3401 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3403 * tree-complex.c (expand_complex_div_wide): update profile.
3405 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3406 Alan Hayward <alan.hayward@arm.com>
3407 David Sherwood <david.sherwood@arm.com>
3409 * Makefile.in (MACHMODE_H): Remove insn-modes.h
3410 (CORETYPES_H): New define.
3411 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
3412 (insn-modes-inline.h, s-modes-inline-h): New rules.
3413 (generated_files): Add insn-modes-inline.h.
3414 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
3415 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
3416 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
3417 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
3418 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
3419 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
3420 (build/gencodes.o, build/genconditions.o): Likewise.
3421 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
3422 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
3423 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
3424 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
3425 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
3426 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
3427 * coretypes.h: Include everything up to real.h for generators.
3428 Include insn-modes.h first. Include wide-int-print.h after
3429 wide-int.h. Include insn-modes-inline.h and then machmode.h.
3430 * machmode.h: Don't include insn-modes.h here.
3431 * function-tests.c: Remove includes of signop.h, machmode.h,
3432 double-int.h and wide-int.h.
3434 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
3436 * optc-save-gen.awk: Likewise.
3437 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
3438 * godump.c: Remove include of wide-int-print.h.
3439 * pretty-print.h: Likewise.
3440 * wide-int-print.cc: Likewise.
3441 * wide-int.cc: Likewise.
3442 * hash-map-tests.c: Remove include of signop.h.
3443 * hash-set-tests.c: Likewise.
3444 * rtl-tests.c: Likewise.
3445 * mkconfig.sh: Remove include of machmode.h.
3446 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
3448 (emit_insn_modes_inline_h): ...this new function. Emit the code
3449 into an insn-modes-inline.h header file, adding appropriate
3450 include guards and end comments.
3451 (emit_insn_modes_c_header): Remove include of machmode.h.
3452 (emit_min_insn_modes_c_header): Include coretypes.h rather than
3454 (main): Handle -i flag and call emit_insn_modes_inline_h when
3457 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3459 * tree-ssa-strlen.c (strinfo): Rename the length field to
3460 nonzero_chars. Add a full_string_p field.
3461 (compare_nonzero_chars, zero_length_string_p): New functions.
3462 (get_addr_stridx): Add an offset_out parameter.
3463 Use compare_nonzero_chars.
3464 (get_stridx): Update accordingly. Use compare_nonzero_chars.
3465 (new_strinfo): Update after above changes to strinfo.
3466 (set_endptr_and_length): Set full_string_p.
3467 (get_string_length): Update after above changes to strinfo.
3468 (unshare_strinfo): Update call to new_strinfo.
3469 (maybe_invalidate): Likewise.
3470 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
3471 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
3472 as a uhwi instead of an shwi. Update after above changes to
3473 strinfo and new_strinfo.
3474 (zero_length_string): Assert that chainsi contains full strings.
3475 Use zero_length_string_p. Update call to new_strinfo.
3476 (adjust_related_strinfos): Update after above changes to strinfo.
3477 Copy full_string_p from origsi.
3478 (adjust_last_stmt): Use zero_length_string_p.
3479 (handle_builtin_strlen): Update after above changes to strinfo and
3480 new_strinfo. Install the lhs as the string length if the previous
3481 entry didn't describe a full string.
3482 (handle_builtin_strchr): Update after above changes to strinfo
3484 (handle_builtin_strcpy): Likewise.
3485 (handle_builtin_strcat): Likewise.
3486 (handle_builtin_malloc): Likewise.
3487 (handle_pointer_plus): Likewise.
3488 (handle_builtin_memcpy): Likewise. Track nonzero characters
3489 that aren't necessarily followed by a nul terminator.
3490 (handle_char_store): Likewise.
3492 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3494 PR tree-optimization/80769
3495 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
3496 for malloc and calloc. Document the new invariant that all related
3497 strinfos have delayed lengths or none do.
3498 (verify_related_strinfos): Move earlier in file.
3499 (set_endptr_and_length): New function, split out from...
3500 (get_string_length): ...here. Also set the lengths of related
3502 (zero_length_string): Assert that chainsi has known (rather than
3504 (adjust_related_strinfos): Likewise.
3506 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3508 PR tree-optimization/81136
3509 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
3510 assert that two references with the same misalignment have the same
3511 compile-time misalignment if those compile-time misalignments
3514 2017-07-01 Andi Kleen <ak@linux.intel.com>
3516 * print-tree.c (print_node): Print all attributes.
3518 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3520 * cfg.c (scale_bbs_frequencies): New function.
3521 * cfg.h (scale_bbs_frequencies): Declare it.
3522 * cfgloopanal.c (single_likely_exit): Cleanup.
3523 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
3525 (scale_loop_profile): Likewise.
3526 (loop_version): Likewise.
3527 (create_empty_loop_on_edge): Update.
3528 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
3529 scale_loop_frequencies, scale_loop_profile, loopify,
3530 loop_version): Update prototypes.
3531 * modulo-sched.c (sms_schedule): Update.
3532 * predict.c (unlikely_executed_edge_p): Also check probability.
3533 (probably_never_executed_edge_p): Fix typo.
3534 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3535 * tree-parloops.c (gen_parallel_loop): Update.
3536 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
3537 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3538 * tree-ssa-loop-split.c (split_loop): Update.
3539 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3540 * tree-vect-loop-manip.c (vect_do_peeling): Update.
3541 (vect_loop_versioning): Update.
3542 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3544 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3546 * trans-mem.c (split_bb_make_tm_edge): Update profile.
3548 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3550 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
3551 to keep profile consistent.
3553 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3555 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
3556 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
3557 * profile-count.h (max_safe_multiplier): Make unsigned.
3558 (profile_count::guessed_zero): New.
3560 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3562 * bb-reorder.c (fix_up_crossing_landing_pad,
3563 fix_crossing_conditional_branches): Use make_single_succ_edge
3564 to keep profile consistent.
3566 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3568 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
3571 2017-07-01 Jakub Jelinek <jakub@redhat.com>
3574 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
3575 the right scopes, make sure cond_jump isn't preserved between multiple
3576 iterations. Search for fallthru edge whenever there are 3+ edges and
3577 use find_fallthru_edge for it.
3579 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3581 Patch by Alexander Monakov <amonakov@ispras.ru>
3582 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
3583 probabilities consistently.
3585 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3587 * pa.c (pa_expand_compare_and_swap_loop): Update call of
3588 emit_cmp_and_jump_insns.
3590 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3593 * tree-inline.c (expand_call_inline): Combine profile statuses.
3595 2017-06-30 Andrew Pinski <apinski@cavium.com>
3597 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
3598 fold_stmt returned true.
3600 2017-06-30 Nathan Sidwell <nathan@acm.org>
3602 * ggc.h (empty_string): Delete.
3603 * cfgexpand.c (expand_asm_stmt): Use plain "".
3604 * optabs.c (expand_asm_memory_barrier): Likewise.
3605 * stringpool.c (empty_string): Delete.
3606 (digit_vector, digit_string): Delete.
3607 (ggc_alloc_string): Use plain "", don't optimize single digit
3608 strings. Use ggc_alloc_atomic.
3610 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
3612 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
3613 comparison set and one other set, use the cost of the non-comparison
3616 2017-06-30 Nathan Sidwell <nathan@acm.org>
3618 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
3621 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
3623 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
3624 loops. Remove now unneeded calls to gimple_switch_set_label() that
3625 just set removed labels to NULL_TREE.
3627 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
3629 * tree-ssanames.c (set_range_info_raw): Abstract from ...
3630 (set_range_info): ...here. Only call set_range_info_raw if domain
3632 (set_nonzero_bits): Call set_range_info_raw.
3633 * tree-ssanames.h (set_range_info_raw): New.
3635 2017-06-30 Jakub Jelinek <jakub@redhat.com>
3638 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
3639 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
3640 of nonimmediate_operand and <store_mask_constraint> instead of m
3641 for the input operand. For V8FI iterator, always split if input
3642 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
3643 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
3644 <store_mask_predicate> instead of register_operand and
3645 <store_mask_constraint> instead of v for the input operand. Make
3646 sure both operands aren't MEMs for if not <mask_applied>.
3648 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
3650 * lto-wrapper.c (copy_file) Close both file descriptors before
3653 2017-06-30 Martin Liska <mliska@suse.cz>
3656 * multiple_target.c (create_dispatcher_calls): Make ifunc
3657 also for function that don't have calls or are not referenced.
3659 2017-06-30 Richard Biener <rguenther@suse.de>
3661 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
3662 analyze the first scalar stmt. Move vector type computation
3663 for the BB case here from ...
3664 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
3665 live operation processing in the SLP case properly.
3667 2017-06-30 Richard Biener <rguenther@suse.de>
3669 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
3671 2017-06-30 Martin Liska <mliska@suse.cz>
3674 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
3675 before BUILT_IN_UNWIND_RESUME when ASAN is used.
3677 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
3679 * doc/invoke.texi (AArch64): Add missing options and remove redundant
3682 2017-06-30 Richard Biener <rguenther@suse.de>
3684 PR tree-optimization/81249
3685 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
3686 condition reduction result to original scalar type.
3688 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3690 * profile-count.h (enum profile_quality): Fix typos and whitespace
3693 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3695 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
3696 type for branch probabilities.
3698 2017-06-29 Julian Brown <julian@codesourcery.com>
3699 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3701 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
3702 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
3703 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
3704 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
3706 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3708 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
3709 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
3710 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
3711 CC usage from generic code to here.
3712 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
3713 CC usage into the target macros.
3715 2017-06-29 Maya Rashish <coypu@sdf.org>
3717 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
3720 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3722 * arm/arm-builtins.c: Include profile-count.h
3723 * except.c (sjlj_emit_function_enter): Use
3724 profile_probability::unlikely.
3726 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3728 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
3729 and tocrel_offset be pointer args rather than implicitly using
3731 (legitimate_constant_pool_address_p, rs6000_emit_move,
3732 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
3733 tocrel_offset and use in toc_relative_expr_p call.
3734 (print_operand, print_operand_address): Use static tocrel_base_oac
3735 and tocrel_offset_oac.
3736 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
3739 2017-06-29 Maya Rashish <coypu@sdf.org>
3741 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
3743 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
3745 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
3746 objects, take into account only the alignment of 'op0' and 'mode1' if
3749 2017-06-29 Steve Ellcey <sellcey@cavium.com>
3751 * ccmp.c (ccmp_tree_comparison_p): New function.
3752 (ccmp_candidate_p): Update to use above function.
3753 (get_compare_parts): New function.
3754 (expand_ccmp_next): Update to use new functions.
3755 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
3757 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
3758 take mode as argument.
3759 * ccmp.h (expand_ccmp_expr): Add mode as argument.
3760 * expr.c (expand_expr_real_1): Pass mode as argument.
3762 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
3764 * combine.c (combine_instructions): Print insns to dump_file, together
3767 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3769 * asan.c (asan_emit_stack_protection): Update.
3770 (create_cond_insert_point): Update.
3771 * auto-profile.c (afdo_propagate_circuit): Update.
3772 * basic-block.h (struct edge_def): Turn probability to
3773 profile_probability.
3774 (EDGE_FREQUENCY): Update.
3775 * bb-reorder.c (find_traces_1_round): Update.
3776 (better_edge_p): Update.
3777 (sanitize_hot_paths): Update.
3778 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
3779 (make_single_succ_edge): Update.
3780 (check_bb_profile): Update.
3781 (dump_edge_info): Update.
3782 (update_bb_profile_for_threading): Update.
3783 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
3784 probabilitycount to 0.
3785 * cfgbuild.c (compute_outgoing_frequencies): Update.
3786 * cfgcleanup.c (try_forward_edges): Update.
3787 (outgoing_edges_match): Update.
3788 (try_crossjump_to_edge): Update.
3789 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
3790 (expand_gimple_tailcall): Update.
3791 (construct_init_block): Use make_single_succ_edge.
3792 (construct_exit_block): Use make_single_succ_edge.
3793 * cfghooks.c (verify_flow_info): Update.
3794 (redirect_edge_succ_nodup): Update.
3795 (split_edge): Update.
3796 (account_profile_record): Update.
3797 * cfgloopanal.c (single_likely_exit): Update.
3798 * cfgloopmanip.c (scale_loop_profile): Update.
3799 (set_zero_probability): Remove.
3800 (duplicate_loop_to_header_edge): Update.
3801 * cfgloopmanip.h (loop_version): Update prototype.
3802 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3803 (force_nonfallthru_and_redirect): Update.
3804 (update_br_prob_note): Update.
3805 (rtl_verify_edges): Update.
3806 (purge_dead_edges): Update.
3807 (rtl_lv_add_condition_to_bb): Update.
3808 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
3809 * cgraphunit.c (init_lowered_empty_function): Update.
3810 (cgraph_node::expand_thunk): Update.
3811 * cilk-common.c: Include profile-count.h
3812 * dojump.c (inv): Remove.
3813 (jumpifnot): Update.
3814 (jumpifnot_1): Update.
3815 (do_jump_1): Update.
3817 (do_jump_by_parts_greater_rtx): Update.
3818 (do_compare_rtx_and_jump): Update.
3819 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
3820 do_jump_1. do_compare_rtx_and_jump): Update prototype.
3821 * dwarf2cfi.c: Include profile-count.h
3822 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
3823 (sjlj_emit_dispatch_table): Likewise.
3824 * explow.c: Include profile-count.h
3825 * expmed.c (emit_store_flag_force): Update.
3826 (do_cmp_and_jump): Update.
3827 * expr.c (compare_by_pieces_d::generate): Update.
3828 (compare_by_pieces_d::finish_mode): Update.
3829 (emit_block_move_via_loop): Update.
3830 (store_expr_with_bounds): Update.
3831 (store_constructor): Update.
3832 (expand_expr_real_2): Update.
3833 (expand_expr_real_1): Update.
3834 * expr.h (try_casesi, try_tablejump): Update prototypes.
3835 * gimple-pretty-print.c (dump_probability): Update.
3836 (dump_profile): New.
3837 (dump_gimple_label): Update.
3838 (dump_gimple_bb_header): Update.
3839 * graph.c (draw_cfg_node_succ_edges): Update.
3840 * hsa-gen.c (convert_switch_statements): Update.
3841 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
3842 (find_if_case_1): Update.
3843 (find_if_case_2): Update.
3844 * internal-fn.c (expand_arith_overflow_result_store): Update.
3845 (expand_addsub_overflow): Update.
3846 (expand_neg_overflow): Update.
3847 (expand_mul_overflow): Update.
3848 (expand_vector_ubsan_overflow): Update.
3849 * ipa-cp.c (good_cloning_opportunity_p): Update.
3850 * ipa-split.c (split_function): Use make_single_succ_edge.
3851 * ipa-utils.c (ipa_merge_profiles): Update.
3852 * loop-doloop.c (add_test): Update.
3853 (doloop_modify): Update.
3854 * loop-unroll.c (compare_and_jump_seq): Update.
3855 (unroll_loop_runtime_iterations): Update.
3856 * lra-constraints.c (lra_inheritance): Update.
3857 * lto-streamer-in.c (input_cfg): Update.
3858 * lto-streamer-out.c (output_cfg): Update.
3859 * mcf.c (adjust_cfg_counts): Update.
3860 * modulo-sched.c (sms_schedule): Update.
3861 * omp-expand.c (expand_omp_for_init_counts): Update.
3862 (extract_omp_for_update_vars): Update.
3863 (expand_omp_ordered_sink): Update.
3864 (expand_omp_for_ordered_loops): Update.
3865 (expand_omp_for_generic): Update.
3866 (expand_omp_for_static_nochunk): Update.
3867 (expand_omp_for_static_chunk): Update.
3868 (expand_cilk_for): Update.
3869 (expand_omp_simd): Update.
3870 (expand_omp_taskloop_for_outer): Update.
3871 (expand_omp_taskloop_for_inner): Update.
3872 * omp-simd-clone.c (simd_clone_adjust): Update.
3873 * optabs.c (expand_doubleword_shift): Update.
3874 (expand_abs): Update.
3875 (emit_cmp_and_jump_insn_1): Update.
3876 (expand_compare_and_swap_loop): Update.
3877 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
3878 * predict.c (predictable_edge_p): Update.
3879 (edge_probability_reliable_p): Update.
3880 (set_even_probabilities): Update.
3881 (combine_predictions_for_insn): Update.
3882 (combine_predictions_for_bb): Update.
3883 (propagate_freq): Update.
3884 (estimate_bb_frequencies): Update.
3885 (force_edge_cold): Update.
3886 * profile-count.c (profile_count::dump): Add missing space into dump.
3887 (profile_count::debug): Add newline.
3888 (profile_count::differs_from_p): Explicitly convert to unsigned.
3889 (profile_count::stream_in): Update.
3890 (profile_probability::dump): New member function.
3891 (profile_probability::debug): New member function.
3892 (profile_probability::differs_from_p): New member function.
3893 (profile_probability::differs_lot_from_p): New member function.
3894 (profile_probability::stream_in): New member function.
3895 (profile_probability::stream_out): New member function.
3896 * profile-count.h (profile_count_quality): Rename to ...
3897 (profile_quality): ... this one.
3898 (profile_probability): New.
3899 (profile_count): Update.
3900 * profile.c (compute_branch_probabilities): Update.
3901 * recog.c (peep2_attempt): Update.
3902 * sched-ebb.c (schedule_ebbs): Update.
3903 * sched-rgn.c (find_single_block_region): Update.
3904 (compute_dom_prob_ps): Update.
3905 (schedule_region): Update.
3906 * sel-sched-ir.c (compute_succs_info): Update.
3907 * stmt.c (struct case_node): Update.
3908 (do_jump_if_equal): Update.
3909 (get_outgoing_edge_probs): Update.
3910 (conditional_probability): Update.
3911 (emit_case_dispatch_table): Update.
3912 (expand_case): Update.
3913 (expand_sjlj_dispatch_table): Update.
3914 (emit_case_nodes): Update.
3915 * targhooks.c: Update.
3916 * tracer.c (better_p): Update.
3917 (find_best_successor): Update.
3918 * trans-mem.c (expand_transaction): Update.
3919 * tree-call-cdce.c: Update.
3920 * tree-cfg.c (gimple_split_edge): Upate.
3921 (move_sese_region_to_fn): Upate.
3922 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
3923 * tree-eh.c (lower_resx): Upate.
3924 (cleanup_empty_eh_move_lp): Upate.
3925 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3926 * tree-inline.c (copy_edges_for_bb): Update.
3927 (copy_cfg_body): Update.
3928 * tree-parloops.c (gen_parallel_loop): Update.
3929 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
3930 (gimple_gen_time_profiler): Update.
3931 * tree-ssa-dce.c (remove_dead_stmt): Update.
3932 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3933 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3934 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3935 (unloop_loops): Update.
3936 (try_peel_loop): Update.
3937 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3938 * tree-ssa-loop-split.c (connect_loops): Update.
3939 (split_loop): Update.
3940 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3941 (hoist_guard): Update.
3942 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3943 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3944 (value_replacement): Update.
3945 * tree-ssa-reassoc.c (branch_fixup): Update.
3946 * tree-ssa-tail-merge.c (replace_block_by): Update.
3947 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3948 (create_edge_and_update_destination_phis): Update.
3949 (compute_path_counts): Update.
3950 (recompute_probabilities): Update.
3951 (update_joiner_offpath_counts): Update.
3952 (freqs_to_counts_path): Update.
3953 (duplicate_thread_path): Update.
3954 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3955 (struct switch_conv_info): Update.
3956 (gen_inbound_check): Update.
3957 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3958 (vect_do_peeling): Update.
3959 (vect_loop_versioning): Update.
3960 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3961 (optimize_mask_stores): Update.
3962 * ubsan.c (ubsan_expand_null_ifn): Update.
3963 * value-prof.c (gimple_divmod_fixed_value): Update.
3964 (gimple_divmod_fixed_value_transform): Update.
3965 (gimple_mod_pow2): Update.
3966 (gimple_mod_pow2_value_transform): Update.
3967 (gimple_mod_subtract): Update.
3968 (gimple_mod_subtract_transform): Update.
3969 (gimple_ic): Update.
3970 (gimple_stringop_fixed_value): Update.
3971 (gimple_stringops_transform): Update.
3972 * value-prof.h: Update.
3974 2017-06-29 Carl Love <cel@us.ibm.com>
3976 * config/rs6000/rs6000-c.c: Add support for built-in functions
3977 vector signed int vec_signed (vector float);
3978 vector signed long long vec_signed (vector double);
3979 vector signed int vec_signed2 (vector double, vector double);
3980 vector signed int vec_signede (vector double);
3981 vector signed int vec_signedo (vector double);
3982 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
3983 instruction generator.
3984 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3985 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
3986 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
3988 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
3989 vunsignede_v2df): Add define_expands.
3990 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
3991 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
3992 VEC_UNSIGNEDO): Add definitions.
3993 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3994 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
3995 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
3996 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
3997 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
3998 * config/rs6000/altivec.h (vec_signed, vec_signed2,
3999 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
4000 vec_unsignede, vec_unsignedo): Add builtin defines.
4001 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
4003 * doc/extend.texi: Update the built-in documentation file for the
4004 new built-in functions.
4006 2017-06-29 Richard Biener <rguenther@suse.de>
4008 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
4009 reduction chains to LOOP_VINFO_REDUCTIONS.
4010 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
4011 SLP reductions after processing reduction chains.
4013 2017-06-29 Nathan Sidwell <nathan@acm.org>
4015 * builtins.c (fold_builtin_FUNCTION): Use
4016 lang_hooks.decl_printable_name.
4018 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
4021 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
4022 with only one label.
4023 * stmt.c (expand_case): Assert NCASES is greater than one.
4025 2017-06-29 Richard Biener <rguenther@suse.de>
4027 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
4029 (group_case_labels): Likewise.
4030 (find_taken_edge): Push sanity checking on val to workers...
4031 (find_taken_edge_cond_expr): ... here
4032 (find_taken_edge_switch_expr): ... and here, handle cases
4033 with just a default label.
4034 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
4035 (group_case_labels): Likewise.
4036 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
4037 group_case_labels does anything cleanup the CFG again.
4039 2017-06-29 Bin Cheng <bin.cheng@arm.com>
4041 PR tree-optimization/81196
4042 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
4043 exit condition comparing two IVs.
4045 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
4047 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
4048 profile to the dummy entry at the end of the list of architectures.
4049 * config/arm/arm-cpu-cdata.h: Regenerated.
4051 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4052 Michael Collison <michael.collison@arm.com>
4055 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
4057 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
4058 (*aarch64_reg_<mode>3_minus_mask): New pattern.
4059 (*aarch64_<optab>_reg_di3_mask2): New pattern.
4060 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
4061 of shift when the shift amount is masked with constant equal to
4062 the size of the mode.
4063 * config/aarch64/predicates.md (subreg_lowpart_operator): New
4066 2017-06-29 Martin Liska <mliska@suse.cz>
4068 * config/i386/i386.opt: Change range from [1,5] to [0,5].
4070 2017-06-29 Yury Gribov <tetra2005@gmail.com>
4073 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
4076 (edge_growth_cache_entry::edge_growth_cache_entry): New
4078 (reset_edge_growth_cache): Update to use constructor.
4080 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4082 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
4083 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
4084 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
4086 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
4088 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
4089 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
4091 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
4093 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
4094 (*-linux-uclibc*): Add t-uclibc tmake_file.
4095 * config/t-musl: New.
4096 * config/t-uclibc: New.
4098 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
4100 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
4102 (gen_comm_data): Emit architectural setting of arch_prof.
4103 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
4105 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
4106 (armv8-m.base, armv8-m.main): Likewise.
4107 * arm-protos.h (arm_build_target): Add profile field.
4108 (arch_option): Likewise.
4109 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
4111 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
4112 arm_active_target.profile.
4114 2017-06-28 Richard Biener <rguenther@suse.de>
4117 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
4118 TYPE_OVERFLOW_WRAPS.
4119 * match.pd (negate_expr_p): Likewise.
4120 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
4121 fold_build2, not fold_binary.
4123 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
4125 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
4126 Convert memory address to Pmode.
4127 (aarch64_print_operand): Assert MEM operands are always Pmode.
4129 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
4132 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
4133 Remove redundant if.
4134 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
4135 * config/arm/aarch-common-protos.h
4136 (aarch_forward_to_shift_is_not_shifted_re): Remove.
4137 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
4139 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
4142 * multiple_target.c (create_dispatcher_calls): Set the default
4143 clone to be static, not public.
4145 2017-06-28 Richard Biener <rguenther@suse.de>
4147 * tree-vect-loop.c (vectorizable_reduction): Move special
4148 cond reduction IV var creation ...
4149 (vect_create_epilog_for_reduction): ... here. Remove induction_index
4150 parameter. Use STMT_VINFO_VECTYPE.
4151 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
4154 2017-06-28 Martin Liska <mliska@suse.cz>
4157 * ipa-visibility.c (non_local_p): Handle visibility.
4159 2017-06-28 Martin Liska <mliska@suse.cz>
4162 * common.opt: Add IntegerRange to various options.
4163 * opt-functions.awk (integer_range_info): New function.
4164 * optc-gen.awk: Add integer_range_info to cl_options struct.
4165 * opts-common.c (decode_cmdline_option): Handle
4166 CL_ERR_INT_RANGE_ARG.
4167 (cmdline_handle_error): Likewise.
4168 * opts.c (print_filtered_help): Show valid interval in
4169 when --help is provided.
4170 * opts.h (struct cl_option): Add range_min and range_max fields.
4171 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
4173 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
4175 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
4176 (x * C EQ/NE y * C): New transformation.
4178 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
4180 * genmultilib (combination_space): Accept '+' in option names.
4182 2017-06-28 Martin Liska <mliska@suse.cz>
4185 * asan.c (instrument_derefs): Bail out inner references
4186 that are hard register variables.
4188 2017-06-28 Jakub Jelinek <jakub@redhat.com>
4191 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
4192 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
4194 2017-06-28 Richard Biener <rguenther@suse.de>
4196 * tree-vectorizer.h (vect_get_vec_defs): Remove.
4197 (vect_get_slp_defs): Adjust.
4198 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
4200 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
4202 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4203 get_initial_defs_for_reduction instead of vect_get_vec_defs.
4204 (vectorizable_reduction): Adjust.
4205 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
4207 (vect_get_slp_defs): Likewise.
4208 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
4209 (vectorizable_bswap): Adjust.
4210 (vectorizable_call): Likewise.
4211 (vectorizable_conversion): Likewise.
4212 (vectorizable_assignment): Likewise.
4213 (vectorizable_shift): Likewise.
4214 (vectorizable_operation): Likewise.
4215 (vectorizable_store): Likewise.
4216 (vectorizable_condition): Likewise.
4217 (vectorizable_comparison): Likewise.
4219 2017-06-28 Michael Collison <michael.collison@arm.com>
4222 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
4224 (arm_gen_movmemqi): Removed unused variable 'i'.
4225 Convert 'for' loop into 'while' loop.
4226 (arm_expand_prologue): Remove last unnecessary set of insn.
4227 (thumb_pop): Remove unused variable 'pushed_words'.
4228 (thumb_exit): Remove last unnecessary set of regs_to_pop.
4230 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4232 * config/s390/predicates.md: Use s390_rel_address_ok_p.
4233 * config/s390/s390-protos.h: Add prototype of
4234 s390_rel_address_ok_p.
4235 * config/s390/s390.c (s390_got_symbol): New function.
4236 (s390_rel_address_ok_p): New function.
4237 (legitimize_pic_address): Use s390_rel_address_ok_p.
4238 (s390_load_got): Use s390_got_symbol.
4239 (s390_option_override): Issue error if
4240 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
4241 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
4243 * config/s390/s390.opt: New option mpic-data-is-text-relative.
4245 2017-06-27 Andrew Pinski <apinski@cavium.com>
4247 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
4248 (X * copysign (1.0, X)): New pattern.
4249 (X * copysign (1.0, -X)): New pattern.
4250 (copysign (-1.0, CST)): New pattern.
4252 2017-06-27 Joseph Myers <joseph@codesourcery.com>
4254 * genmultilib (combination_space): Remove variable.
4255 Validate reuse rules against regular expression for any sequence
4256 of multilib options in any order.
4258 2017-06-27 Michael Collison <michael.collison@arm.com>
4260 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
4261 call aarch64_split_simd_combine.
4262 * (aarch64_combine_internal<mode>): Delete pattern.
4263 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
4264 Allow register and subreg operands.
4266 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4268 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
4269 specific need, just fallback on defaults.
4270 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
4272 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4273 Olivier Hainque <hainque@adacore.com>
4275 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
4277 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
4278 targets. Pick a default if no particular attempt applied.
4279 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
4282 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4284 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
4285 (x86_64-wrs-vxworks7): Likewise.
4287 2017-06-27 Marek Polacek <polacek@redhat.com>
4290 * ubsan.c (instrument_null): Check get_base_address's result for null.
4292 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
4294 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
4296 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
4298 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
4299 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
4300 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
4301 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
4303 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
4304 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
4305 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
4306 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
4307 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
4308 BUILT_IN_FEUPDATEENV): New builtins.
4309 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
4310 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
4311 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
4312 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
4314 (builtin_structptr_types): Adjust size.
4315 * tree.c (builtin_structptr_types): Add four entries.
4317 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4318 Olivier Hainque <hainque@adacore.com>
4320 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
4321 (TLS_SYM): New local macro, forcing reference to __tls__ on
4322 link command lines for VxWorks 7 RTPs, triggering initialization
4324 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
4325 OS features TLS support, true for RTPs on VxWorks 7.
4326 * config/vxworks.c (vxworks_override_options): Setup emutls
4329 2017-06-27 Jakub Jelinek <jakub@redhat.com>
4331 * predict.c (test_prediction_value_range): Use -1U instead of -1
4332 to avoid narrowing conversion warning.
4333 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
4334 to avoid narrowing conversion warning.
4335 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
4337 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
4339 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4341 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
4342 64bit configurations.
4343 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
4344 (SIZE_TYPE): Likewise.
4345 * config/vxworks.c (vxworks_emutls_var_fields): Use
4346 long_unsigned_type_node instead of unsigned_type_node as the offset
4347 field type, which is "pointer" mode in emutls.c.
4349 2017-06-27 Jakub Jelinek <jakub@redhat.com>
4352 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
4355 * gimple-fold.c (replace_call_with_call_and_fold): Handle
4356 gimple_vuse copying separately from gimple_vdef copying.
4358 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4360 * value-prof.c (free_hist): Remove call to memset and the enclosing if
4363 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
4364 Olivier Hainque <hainque@adacore.com>
4366 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
4367 for all vxworks7 targets.
4368 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
4369 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
4370 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
4371 variations for VX6/VX7 and 32/64bits later on in ...
4372 (VXWORKS_LIB_SPEC): Leverage new macros.
4373 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
4374 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
4376 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
4378 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
4379 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
4381 2017-06-26 Carl Love <cel@us.ibm.com>
4383 * config/rs6000/rs6000-c.c: Add support for built-in functions
4384 vector bool char vec_reve (vector bool char);
4385 vector signed char vec_reve (vector signed char);
4386 vector unsigned char vec_reve (vector unsigned char);
4387 vector bool int vec_reve (vector bool int);
4388 vector signed int vec_reve (vector signed int);
4389 vector unsigned int vec_reve (vector unsigned int);
4390 vector bool long long vec_reve (vector bool long long);
4391 vector signed long long vec_reve (vector signed long long);
4392 vector unsigned long long vec_reve (vector unsigned long long);
4393 vector bool short vec_reve (vector bool short);
4394 vector signed short vec_reve (vector signed short);
4395 vector double vec_reve (vector double);
4396 vector float vec_reve (vector float);
4397 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
4398 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
4399 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
4400 (altivec_vreve): New pattern.
4401 * config/rs6000/altivec.h (vec_reve): New define.
4402 * doc/extend.texi (vec_rev): Update the built-in documentation file
4403 for the new built-in functions.
4405 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4407 PR tree-optimization/71815
4408 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
4410 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
4412 (slsr_process_phi): Likewise.
4413 (replace_uncond_cands_and_profitable_phis): Don't replace a
4414 multiply candidate with a stride of 1 (copy or cast).
4415 (phi_incr_cost): Call uses_consumed_by_stmt rather than
4417 (lowest_cost_path): Likewise.
4418 (total_savings): Likewise.
4420 2017-06-26 Richard Biener <rguenther@suse.de>
4423 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
4424 Use def_builtin_pure for all gather builtins.
4426 2017-06-26 Richard Biener <rguenther@suse.de>
4428 PR tree-optimization/81203
4429 * tree-tailcall.c (find_tail_calls): Do not move stmts into
4432 2017-06-26 Marek Polacek <polacek@redhat.com>
4435 * doc/invoke.texi: Document -Wmultistatement-macros.
4437 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
4439 * doc/sourcebuild.texi (ARM-specific attributes): Document new
4440 arm_neon_ok_no_float_abi effective target.
4442 2017-06-26 Richard Biener <rguenther@suse.de>
4444 PR tree-optimization/80928
4445 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
4446 (copy_bbs): Set BB_DUPLICATED flag early.
4447 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
4449 (execute_on_shrinking_pred): Likewise.
4450 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
4451 BB_DUPLICATED blocks.
4452 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
4453 iterate over all PHIs considering removal of *gsi.
4455 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
4457 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
4460 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4462 * config/rs6000/rs6000-string.c: (expand_block_clear,
4463 do_load_for_compare, select_block_compare_mode,
4464 compute_current_alignment, expand_block_compare,
4465 expand_strncmp_align_check, expand_strn_compare,
4466 expand_block_move, rs6000_output_load_multiple)
4467 Move functions related to string/block move/compare
4469 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
4470 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
4471 for this function which is now used in two files.
4472 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
4473 * config.gcc: Add rs6000-string.o to extra_objs for
4474 targets powerpc*-*-* and rs6000*-*-*.
4476 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4479 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
4480 32-bit, since indexed is not valid for DImode.
4481 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
4482 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
4483 (define_peephole2 for Altivec d-form load): Add 32-bit support.
4484 (define_peephole2 for Altivec d-form store): Likewise.
4487 * multiple_target.c (create_dispatcher_calls): Only create the
4488 dispatcher call if the function is the default clone of a
4491 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
4494 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
4495 a call, force the call to not be a tail call.
4497 2017-06-23 Jeff Law <law@redhat.com>
4499 * doc/contrib.texi: Add entry for Steven Pemberton's work on
4502 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
4504 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
4505 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
4506 handling for early expansion of vector shifts (sl,sr,sra,rl).
4507 (builtin_function_type): Add vector shift right instructions
4508 to the unsigned argument list.
4510 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
4512 rtl-optimizatoin/79286
4513 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
4514 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
4515 trap. PIC register plus a const unspec without offset can never trap.
4517 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
4519 * tree.h (builtin_structptr_type): New type.
4520 (builtin_structptr_types): Declare new array.
4521 * tree.c (builtin_structptr_types): New array.
4522 (free_lang_data, build_common_tree_nodes): Use it.
4524 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
4527 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
4530 2017-06-22 Matt Turner <mattst88@gmail.com>
4532 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
4533 Lake models to skylake case. Assume skylake for unknown
4534 models with clflushopt.
4536 2017-06-22 Jeff Law <law@redhat.com>
4538 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
4539 frame sizes that do not satisfy aarch64_uimm12_shift.
4541 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
4543 * profile-count.h (apply_probability,
4544 apply_scale, probability_in): Fix checks for zero.
4546 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4548 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
4549 * doc/cppdiropts.texi (-I @var{dir}): Document it.
4551 2016-06-22 Richard Biener <rguenther@suse.de>
4553 * tree-vect-loop.c (vect_model_reduction_cost): Handle
4554 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
4555 REDUC_MAX_EXPR support.
4556 (vectorizable_reduction): Likewise.
4557 (vect_create_epilog_for_reduction): Likewise.
4559 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
4561 * match.pd (A / (1 << B) -> A >> B): New.
4562 * generic-match-head.c: Include optabs-tree.h.
4563 * gimple-match-head.c: Likewise.
4564 * optabs-tree.h (target_supports_op_p): New.
4565 * optabs-tree.c (target_supports_op_p): New.
4567 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4569 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
4570 $gcc_cv_ld --help output.
4571 (gcc_cv_ld_demangle): Likewise.
4572 (gcc_cv_ld_eh_frame_hdr): Likewise.
4573 (gcc_cv_ld_pie): Likewise.
4574 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
4575 (gcc_cv_ld_buildid): Likewise.
4576 (gcc_cv_ld_sysroot): Likewise.
4577 (ld_bndplt_support): Likewise.
4578 (ld_pushpopstate_support): Likewise.
4579 * configure: Regenerate.
4580 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
4582 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4585 * config/i386/sse.md (round<mode>2): Renumber match_dup and
4586 operands indexes to avoid gap between operands and match_dups.
4588 2017-06-21 Andrew Pinski <apinski@cavium.com>
4590 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
4591 Increment Arith_shift and Arith_shift_reg by 1.
4592 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
4594 * config/aarch64/aarch64.c (thunderx_tunings): Enable
4595 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
4596 (aarch64_strip_extend): Add new argument and test for it.
4597 (aarch64_cheap_mult_shift_p): New function.
4598 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
4599 add a cost if it is true.
4600 Update calls to aarch64_strip_extend.
4601 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
4603 2017-06-21 Andrew Pinski <apinski@cavium.com>
4605 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
4607 (thunderxt88): Likewise.
4608 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
4609 (thunderx_prefetch_tune): New variable.
4610 (thunderx2t99_prefetch_tune): Update for the correct values.
4611 (thunderxt88_tunings): New variable.
4612 (thunderx_tunings): Use thunderx_prefetch_tune instead of
4613 generic_prefetch_tune.
4614 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
4616 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4618 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
4619 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
4620 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
4621 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
4622 (aarch64_atomic_cas<mode>, GPI): Likewise.
4624 2017-06-21 Martin Liska <mliska@suse.cz>
4626 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
4627 statements on cold and hot labels.
4628 * predict.c (tree_estimate_probability_bb): Remove the
4629 prediction from this place.
4631 2017-06-21 Martin Liska <mliska@suse.cz>
4633 PR tree-optimization/79489
4634 * gimplify.c (maybe_add_early_return_predict_stmt): New
4636 (gimplify_return_expr): Call the function.
4637 * predict.c (tree_estimate_probability_bb): Remove handling
4639 * predict.def: Update comment about early return predictor.
4640 * gimple-predict.h (is_gimple_predict): New function.
4641 * predict.def: Change default value of early return to 66.
4642 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
4644 * passes.def: Put pass_strip_predict_hints to the beginning of
4647 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
4649 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
4650 FUNCTION_DECL declarations.
4651 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
4653 (dwaf2out_decl): Likewise.
4654 * godump.c (go_early_global_decl): Skip call to the real debug hook
4655 for FUNCTION_DECL declarations.
4656 * passes.c (rest_of_decl_compilation): Skip call to the
4657 early_global_decl debug hook for FUNCTION_DECL declarations, unless
4658 -fdump-go-spec is passed.
4660 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4662 * config/i386/i386.c (struct builtin_isa): New field pure_p.
4663 Reorder for compactness.
4664 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
4665 (def_builtin_pure, def_builtin_pure2): New functions.
4666 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
4668 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4670 * match.pd (nop_convert): New predicate.
4671 ((A +- CST1) +- CST2): Allow some NOP conversions.
4673 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4676 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
4677 with ctors/dtors if GOVD_SHARED is set.
4679 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4681 * config/aarch64/aarch64.md (movti_aarch64):
4682 Emit mov rather than orr.
4683 (movtf_aarch64): Likewise.
4684 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
4685 Emit mov rather than orr.
4687 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4689 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
4690 Swap alternatives, make integer dup more expensive.
4692 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4694 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
4695 Return true for non-tls symbols.
4697 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
4699 * config/aarch64/aarch64-cores.def (cortex-a55): New.
4700 (cortex-a75): Likewise.
4701 (cortex-a75.cortex-a55): Likewise.
4702 * config/aarch64/aarch64-tune.md: Regenerate.
4703 * doc/invoke.texi (-mtune): Document new values for -mtune.
4705 2017-06-21 Tom de Vries <tom@codesourcery.com>
4707 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
4709 (Effective-Target Keywords, Other attributes): Suggest using
4710 dg-add-options stack_size feature to get stack limit in stack_size
4711 effective target documentation.
4713 2017-06-21 Julian Brown <julian@codesourcery.com>
4714 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4716 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
4717 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
4718 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
4720 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
4721 attribute type list for neon_multiply.
4722 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
4723 attribute type list for neon_multiply.
4724 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
4725 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
4726 attribute type list for neon_multiply.
4727 * config/arm/types.md (crypto_pmull): Add.
4728 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
4729 attribute type list.
4731 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
4733 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
4736 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4738 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
4739 to make sure not to dereference a NULL cost_classes_ptr pointer.
4741 2017-06-20 Carl Love <cel@us.ibm.com>
4743 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4744 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4745 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4746 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4747 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4748 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4749 VMULOSW): New enum "unspec" values.
4750 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
4751 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
4752 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4753 altivec_vmulosw): New patterns.
4754 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4755 VMULOSW): Add definitions.
4757 2017-06-20 Julia Koval <julia.koval@intel.com>
4759 * config/i386/i386.c: Fix rounding expand for new pattern.
4760 * config/i386/subst.md: Fix pattern (parallel -> unspec).
4762 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4764 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
4765 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
4767 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4769 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
4772 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4774 * config/aarch64/aarch64-cores.def: Rearrange to sort by
4775 architecture, then by implementer ID.
4776 * config/aarch64/aarch64-tune.md: Regenerate.
4778 2017-06-20 Richard Biener <rguenther@suse.de>
4781 * fold-const.c (split_tree): Fold to type before negating.
4783 2017-06-20 David Malcolm <dmalcolm@redhat.com>
4785 * diagnostic-show-locus.c
4786 (selftest::test_fixit_deletion_affecting_newline): New function.
4787 (selftest::diagnostic_show_locus_c_tests): Call it.
4789 2017-06-20 Andreas Schwab <schwab@suse.de>
4792 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
4795 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
4797 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
4798 __ARM_FEATURE_COPROC according to support.
4800 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4802 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
4803 Rewritten to avoid overflow for > 32-bit pointers.
4806 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
4807 by removing enum keyword.
4808 (ubsan_type_descriptor): Likewise. Formatting fix.
4811 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
4812 splitter): Require TARGET_SSE2 in the condition.
4814 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
4817 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
4818 for doing vector set of SFmode on ISA 3.0.
4819 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
4820 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
4822 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
4823 SFmode value into a V4SF variable that was extracted from another
4824 V4SF variable without converting the element to double precision
4825 and back to single precision vector format.
4826 (vsx_insert_extract_v4sf_p9_2): Likewise.
4828 2017-06-19 Jakub Jelinek <jakub@redhat.com>
4830 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
4831 in UWHI to avoid undefined overflow.
4834 * ubsan.h (enum ubsan_encode_value_phase): New.
4835 (ubsan_encode_value): Change second argument to
4836 enum ubsan_encode_value_phase with default value of
4837 UBSAN_ENCODE_VALUE_GENERIC.
4838 * ubsan.c (ubsan_encode_value): Change second argument to
4839 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
4840 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
4841 create_tmp_var_raw instead of create_tmp_var and use a
4843 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
4844 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
4845 ubsan_encode_value callers.
4848 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
4849 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
4850 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
4852 2017-06-19 Richard Biener <rguenther@suse.de>
4855 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
4856 estimates if we changed anything.
4858 2017-06-19 Richard Biener <rguenther@suse.de>
4860 PR tree-optimization/80887
4861 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
4862 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
4863 simplified lookups, then reset mprts_hook.
4864 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
4866 (try_to_simplify): Likewise.
4868 2017-06-19 Martin Liska <mliska@suse.cz>
4871 * gimplify.c (gimplify_switch_expr):
4872 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
4874 2017-06-19 Martin Liska <mliska@suse.cz>
4876 * doc/install.texi: Document that PGO runs in 4 stages.
4878 2017-06-19 Martin Liska <mliska@suse.cz>
4881 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
4882 to dispatcher function name.
4883 * multiple_target.c (replace_function_decl): New function.
4884 (create_dispatcher_calls): Redirect both edges and references.
4886 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
4888 * profile-count.c (profile_count::dump): Dump quality.
4889 (profile_count::differs_from_p): Update for unsigned val.
4890 * profile-count.h (profile_count_quality): New enum.
4891 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
4893 2017-06-19 Richard Biener <rguenther@suse.de>
4895 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
4896 struct function as arg.
4897 (estimate_numbers_of_iterations): Export overload with loop arg.
4898 (free_numbers_of_iterations_estimates_loop): Use an overload of
4899 free_numbers_of_iterations_estimates instead.
4900 * tree-cfg.c (remove_bb): Adjust.
4901 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
4902 * tree-parloops.c (gen_parallel_loop): Likewise.
4903 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
4905 (tree_unroll_loops_completely): Likewise.
4906 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
4907 Use an overload instead and export.
4908 (estimated_loop_iterations): Adjust.
4909 (max_loop_iterations): Likewise.
4910 (likely_max_loop_iterations): Likewise.
4911 (estimate_numbers_of_iterations): Take struct function as arg
4913 (loop_exits_before_overflow): Adjust.
4914 (free_numbers_of_iterations_estimates_loop): Use an overload.
4915 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
4916 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
4918 2017-06-19 Richard Biener <rguenther@suse.de>
4921 * ipa-prop.c (find_constructor_constant_at_offset): Handle
4922 RANGE_EXPR conservatively.
4924 2017-06-16 Carl Love <cel@us.ibm.com>
4926 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4927 definitions for vec_float, vec_float2, vec_floato,
4928 vec_floate built-ins.
4929 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
4930 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
4932 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
4933 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
4934 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
4935 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
4936 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
4937 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
4938 vec_floato): Add builtin defines.
4939 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
4940 Update the built-in documentation file for the new built-in
4943 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4945 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
4946 (mthumb): Mark as the negative of -marm.
4948 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4950 * doc/invoke.texi (ARM Options, -mcpu): Document supported
4952 (ARM Options, -mtune): Document that this accepts the same
4953 extension options as -mcpu.
4954 (ARM Options, -mfpu): Document addition of -mfpu=auto.
4956 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4958 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
4959 permitted extensions.
4961 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4963 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
4964 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
4965 (armv8-m.main): Add option +nodsp.
4966 * config/arm/arm-cpu-cdata.h: Regenerated.
4968 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4970 * config/arm/t-fuchsia: New file.
4971 * config.gcc (arm*-*-fuchsia*): Use it.
4973 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4975 * config/arm/t-symbian: Rewrite for new option infrastructure.
4977 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4979 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
4980 (MULTILIB_REQUIRED): Likewise.
4982 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4984 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
4985 (MULTILIB_RESUE): Likewise.
4986 (MULTILIB_MATCHES): Likewise.
4987 (MULTLIB_REQUIRED): Likewise.
4989 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4991 * config/arm/t-rtems: Rewrite for new option framework.
4993 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4995 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
4996 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
4997 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
4998 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
4999 * config/arm/t-multilib: ... here.
5000 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
5001 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
5002 armv7-a and armv8*-a when A-profile libraries have not been built.
5003 * config/arm/t-rmprofile: Rewrite.
5005 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5007 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
5008 with a backslash. Remove the backslash after substituting unescaped
5010 * doc/fragments.texi (MULTILIB_REUSE): Document it.
5012 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5014 * config.gcc: (arm*-*-*): When building a-profile libraries, force
5015 the driver to pass through the default setting of -mfloat-abi.
5016 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
5018 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
5019 (all_feat_combs): New rule.
5020 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
5022 * config/arm/t-aprofile: Rewrite.
5024 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5026 * config/arm/arm.h (FPUTYPE_AUTO): Define.
5027 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
5028 fpu is not specified by the user/command-line.
5029 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
5030 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
5031 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
5032 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
5033 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
5034 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
5036 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5038 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
5039 * config/arm/t-arm-elf: Rewritten.
5041 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5043 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
5044 have some floating-point instructions.
5045 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
5046 (TARGET_MAYBE_HARD_FLOAT): New macro.
5047 * config/arm/arm-builtins.c (arm_init_builtins): Use
5048 TARGET_MAYBE_HARD_FLOAT.
5049 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
5051 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5053 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
5054 (configargs.h): Include it.
5055 (arm_print_hint_for_fpu_option): New function.
5056 (arm_parse_fpu_option): New function.
5057 (candidate_extension): New class.
5058 (arm_canon_for_multilib): New function.
5059 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
5060 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
5061 (ARCH_CANONICAL_SPECS): New macro.
5062 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
5064 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5066 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
5067 are set after handling multilib fragments. Set target_cpu_default2
5070 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5072 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
5074 (arm*-*-*): Set target_cpu_default2 to a quoted string.
5075 * config/arm/parsecpu.awk (check_cpu): Validate any extension
5077 (check_arch): Likewise.
5078 * config/arm/arm.c (arm_configure_build_target): Handle
5079 TARGET_CPU_DEFAULT being a string constant. Scan any feature
5080 options in the default.
5082 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5084 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
5085 when an option is an alias of another.
5086 * config/arm/parsecpu.awk (optalias): New parser token.
5087 (gen_comm_data): Mark non-alias options as such. Emit entries
5088 for extension aliases.
5089 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
5090 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
5091 (armv6kz, armv6zk, armv6t2): Likewise.
5092 (armv7): Make vfpv3-d16 an alias.
5093 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
5095 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
5096 Sort in canonical order.
5097 (armv8-a): Sort in canonical order.
5098 (armv8.1-a, armv8.2-a): Likewise.
5099 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
5101 (cortex-a9): Sort in canonical order.
5102 * config/arm/arm.c (selftests.h): Include it.
5103 (arm_test_cpu_arch_data): New function.
5104 (arm_run_self_tests): New function.
5105 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
5106 (targetm): Move declaration to the end of the file.
5107 * arm-cpu-cdata.h: Regenerated.
5109 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5111 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
5112 call to target_mode_check describing the type of option passed.
5113 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
5114 (arm_target_thumb_only): Use arm_parse_arch_option_name or
5115 arm_parse_cpu_option_name to match parameters against list of
5117 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
5118 arm_arch_core_flags data structure.
5119 * config/arm/arm-cpu_cdata.h: Regenerated.
5121 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5123 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
5125 (arm_print_hint_for_cpu_option): Likewise.
5126 (arm_print_hint_for_arch_option): Likewise.
5127 (arm_parse_cpu_option_name): Likewise.
5128 (arm_parse_arch_option_name): Likewise.
5129 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
5130 of entries in the all_fpus list.
5131 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
5132 (arm_parse_cpu_option_name): Declare.
5133 (arm_parse_arch_option_name): Declare.
5134 (arm_parse_option_features): Declare.
5135 (arm_intialize_isa): Declare.
5136 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
5138 (gen_comm_data): ... here. Make definitions non-static.
5139 * config/arm/arm-cpu-data.h: Regenerated.
5140 * config/arm/arm-cpu-cdata.h: Regenerated.
5142 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5144 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
5145 (cpu_arch_extension): New structure.
5146 (cpu_arch_option, arch_option, cpu_option): New structures.
5147 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
5149 (gen_data): Generate new format data tables.
5150 * config/arm/arm.c (cpu_tune): New structure.
5151 (cpu_option, processors): Delete.
5152 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
5153 (arm_print_hint_for_cpu_option): ... this and ...
5154 (arm_print_hint_for_arch_option): ... this.
5155 (arm_parse_arch_cpu_name): Delete. Replace with ...
5156 (arm_parse_cpu_option_name): ... this and ...
5157 (arm_parse_arch_option_name): ... this.
5158 (arm_unrecognized_feature): Change type of target parameter to
5160 (arm_parse_arch_cpu_features): Delete. Replace with ...
5161 (arm_parse_option_features): ... this.
5162 (arm_configure_build_target): Rework to use new configuration data
5164 (arm_print_tune_info): Rework for new configuration data tables.
5165 * config/arm/arm-cpu-data.h: Regenerated.
5166 * config/arm/arm-cpu.h: Regenerated.
5168 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5170 * Makefile.in (OBJS): Move sbitmap.o from here ...
5171 (OBJS-libcommon): ... to here.
5173 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5175 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
5176 (ISA_ALL_CRYPTO): New macro.
5177 (ISA_ALL_SIMD): New macro
5178 (ISA_ALL_FP): New macro.
5179 * config/arm/arm.c (fpu_bitlist): Update initializer.
5180 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
5182 (arm9e): Add fpu. Add option for nofp
5183 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
5184 (arm926ej-s, arm1026ej-s): Likewise.
5185 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
5186 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
5187 neon-fp16, neon-vfpv4, nofp and nosimd.
5188 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
5189 (cortex-a8): Add fpu. Add option for nofp.
5190 (cortex-a9): Add fpu. Add options for nosimd and nofp.
5191 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
5192 (cortex-r4f): Add fpu.
5193 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
5194 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
5196 (cortex-r8): Likewise.
5197 (cortex-m4): Add fpu. Add option for nofp.
5198 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
5199 (cortex-a17.cortex-a7): Likewise.
5200 (cortex-a32): Add fpu. Add options for crypto and nofp.
5201 (cortex-a35, cortex-a53): Likewise.
5202 (cortex-a57): Add fpu. Add option for crypto.
5203 (cortex-a72, cortex-a73): Likewise.
5204 (exynos-m1): Likewise.
5205 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
5206 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
5207 (cortex-m33): Add fpu. Add option for nofp.
5208 * config/arm/arm-cpu-cdata.h: Regenerated
5209 * config/arm/arm-cpu-data.h: Regenerated.
5211 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5213 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
5214 (armv5te, armv5tej): Likewise.
5215 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
5216 (armv7): Add options fp and vfpv3-d16.
5217 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
5218 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
5220 (armv7ve): Likewise.
5221 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
5222 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
5223 (armv8-a): Add nocrypto option.
5224 (armv8.1-a, armv8.2-a): Likewise.
5225 (armv8-m.main): add options fp, fp.dp and nofp.
5227 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5229 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
5231 (armv8-a+crc): Delete.
5232 (armv8.1-a): Add options simd, crypto and nofp.
5233 (armv8.2-a): Add options fp16, simd, crypto and nofp.
5234 (armv8.2-a+fp16): Delete.
5235 (armv8-m.main): Add option dsp.
5236 (armv8-m.main+dsp): Delete.
5237 (cortex-a8): Add fpu. Add nofp option.
5238 (cortex-a9): Add fpu. Add nofp and nosimd options.
5239 * config/arm/parsecpu.awk (gen_data): Generate option tables and
5240 link to main cpu and architecture data structures.
5241 (gen_comm_data): Only put isa attributes from the main architecture
5243 (option): New statement for architecture and CPU entries.
5244 * arm.c (struct cpu_option): New structure.
5245 (struct processors): Add entry for options.
5246 (arm_unrecognized_feature): New function.
5247 (arm_parse_arch_cpu_name): Ignore any characters after the first
5249 (arm_parse_arch_cpu_feature): New function.
5250 (arm_configure_build_target): Separate out any CPU and architecture
5251 features and parse separately. Don't error out if -mfpu=auto is
5252 used with only an architecture string.
5253 (arm_print_asm_arch_directives): New function.
5254 (arm_file_start): Call it.
5255 * config/arm/arm-cpu-cdata.h: Regenerated.
5256 * config/arm/arm-cpu-data.h: Likewise.
5257 * config/arm/arm-tables.opt: Likewise.
5259 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5261 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
5262 assembler when it is not -mfpu=auto.
5264 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5266 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
5267 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
5268 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
5269 (ASM_CPU_SPEC): Rewrite.
5270 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
5271 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
5272 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
5273 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
5274 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
5275 copied string is NUL-terminated. Also strip any characters prefixed
5277 (arm_rewrite_selected_arch): New function.
5278 (arm_rewrite_march): New function.
5280 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5282 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
5283 (x_arm_cpu_string, x_arm_tune_string): Likewise.
5284 (march, mcpu, mtune): Convert to string-based options.
5285 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
5286 (arm_parse_arch_cpu_name): New function.
5287 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
5288 identify selected architecture or CPU.
5289 (arm_option_save): New function.
5290 (TARGET_OPTION_SAVE): Redefine.
5291 (arm_option_restore): Restore string options.
5292 (arm_option_print): Print string options.
5294 2017-06-16 Martin Sebor <msebor@redhat.com>
5296 PR tree-optimization/80933
5297 PR tree-optimization/80934
5298 * builtins.c (fold_builtin_3): Do not handle bcmp here.
5299 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
5300 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
5301 (gimple_fold_builtin): Call them.
5303 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5305 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
5306 as unlikely; update profile.
5308 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5310 * predict.c (force_edge_cold): Handle declaring edges impossible
5313 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5315 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
5317 (try_unroll_loop_completely): Fix reporting.
5319 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5321 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
5323 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
5326 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
5327 if given a non-constant argument for an intrinsic which requires a
5330 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5332 * profile.c (compare_freqs): New function.
5333 (branch_prob): Sort edge list.
5334 (find_spanning_tree): Assume that the list is priority sorted.
5336 2017-06-16 Richard Biener <rguenther@suse.de>
5338 PR tree-optimization/81090
5339 * passes.def (pass_record_bounds): Remove.
5340 * tree-pass.h (make_pass_record_bounds): Likewise.
5341 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
5342 make_pass_record_bounds): Likewise.
5343 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
5344 not free niter estimates at the beginning but at the end.
5345 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
5347 2017-06-16 Richard Biener <rguenther@suse.de>
5349 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
5350 initializer to workaround ICE in host GCC 4.8.
5352 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5354 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
5356 (clone_inlined_nodes): Update.
5358 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5360 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
5361 prefetch settings, and enable prefetching by default at -O3.
5363 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5365 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5366 Set flag_prefetch_loop_arrays according to tuning data.
5368 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5370 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
5372 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
5373 [Unrelated to main purpose of the patch] Place the pointer field last
5374 to enable type checking errors when tune structure are wrongly merged.
5375 * config/aarch64/aarch64.c (generic_prefetch_tune,)
5376 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
5377 (thunderx2t99_prefetch_tune): New tune constants.
5378 (tune_params *_tunings): Update all tunings (no functional change).
5379 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
5380 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
5381 from tunings structures.
5383 2017-06-16 Jakub Jelinek <jakub@redhat.com>
5386 * ubsan.c (instrument_null): Add T argument, use it instead
5387 of computing it based on IS_LHS.
5388 (instrument_object_size): Likewise.
5389 (pass_ubsan::execute): Adjust instrument_null and
5390 instrument_object_size callers to pass gimple_get_lhs or
5391 gimple_assign_rhs1 result to it. Use instrument_null instead of
5392 calling get_base_address and instrument_mem_ref. Handle
5393 aggregate call arguments for object-size sanitization.
5395 2017-06-16 Yury Gribov <tetra2005@gmail.com>
5397 PR tree-optimization/81089
5398 * tree-vrp.c (is_masked_range_test): Validate operands of
5401 2017-06-15 Martin Sebor <msebor@redhat.com>
5404 * dumpfile.c (dump_register): Avoid calling memset to initialize
5405 a class with a default ctor.
5406 * gcc.c (struct compiler): Remove const qualification.
5407 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
5408 * hash-table.h: Ditto.
5409 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
5411 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
5412 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
5414 * params.h (struct param_info): Make struct members non-const.
5415 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
5416 with default initialization.
5417 * vec.h (vec_copy_construct, vec_default_construct): New helper
5419 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
5420 with vec_copy_construct.
5421 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
5422 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
5423 * doc/invoke.texi (-Wclass-memaccess): Document.
5425 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5427 * emit-rtl.h (is_leaf): Update comment about local
5430 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5433 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
5434 for a variable to have a section before checking if the section has a
5436 Set section to.persistent if persistent attribute is set.
5437 Warn if .persistent attribute is used on an automatic variable.
5439 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
5441 PR rtl-optimization/80474
5442 * reorg.c (update_block): Do not ignore instructions in a delay slot.
5444 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
5446 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
5449 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
5451 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
5452 (casesi): Emit bounds checking as RTL.
5453 (casesi_internal_mips16_<mode>): Remove bounds checking.
5455 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
5457 * config/xtensa/xtensa.c (xtensa_option_override): Append
5458 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
5459 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
5460 xtensa_doloop_hooks): Define unconditionally.
5461 (xtensa_reorg_loops): Only call reorg_loops in the presence of
5463 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
5464 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
5465 for it in xtensa_option_override.
5466 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
5467 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
5469 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
5471 * doc/cppopts.texi: Document '-' special value to -MF.
5473 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
5475 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
5476 (cortex_a53_fconst): Likewise.
5477 (cortex_a53_fpmul): Likewise.
5478 (cortex_a53_f_load_64): Likewise.
5479 (cortex_a53_f_load_many): Likewise.
5480 (cortex_a53_advsimd_alu): Likewise.
5481 (cortex_a53_advsimd_alu_q): Likewise.
5482 (cortex_a53_advsimd_mul): Likewise.
5483 (cortex_a53_advsimd_mul_q): Likewise.
5484 (fpmac bypass): Add new bypass for fpmac-fpmac case.
5485 Add missing fmul, r2f_cvt and fconst cases.
5487 2017-06-14 Richard Biener <rguenther@suse.de>
5490 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
5492 (fold_binary_loc): When associating do not treat pre-existing
5493 TREE_OVERFLOW on literal constants as a reason to allow
5494 TREE_OVERFLOW on associated literal constants.
5496 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5498 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
5499 (MASK_FEATURES): New macro.
5500 * config/sparc/sparc.c (sparc_option_override): Remove the special
5501 handling of -mfpu and generalize it to all MASK_FEATURES switches.
5503 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5505 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
5506 a division of 0 if non-call exceptions are enabled.
5508 2017-06-14 Andrew Pinski <apinski@cavium.com>
5509 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5512 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
5513 Improve vector initialization code gen for only variable case.
5515 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5517 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
5519 2017-06-14 Richard Biener <rguenther@suse.de>
5521 PR tree-optimization/81083
5522 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
5525 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5527 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
5528 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
5529 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
5530 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
5531 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
5532 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
5534 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5536 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
5537 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
5539 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5541 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
5543 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5545 * config/rs6000/t-rtems: Don't handle SPE.
5547 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5549 * config/rs6000/t-linux: Don't handle SPE.
5551 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5553 * config/rs6000/eabispe.h: Delete file.
5555 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5557 * config/rs6000/t-spe: Delete file.
5559 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5561 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
5562 (rs6000_legitimate_offset_address_p): Return false for anything in
5563 V2SImode or V2SFmode.
5565 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5567 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
5568 except V2SF and V2SI. Rearrange the vector modes, and add comments.
5569 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
5571 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
5572 (rs6000_legitimate_offset_address_p): Ditto.
5573 (rs6000_emit_move): Ditto.
5574 (rs6000_init_builtins): Remove V4HI_type_node.
5576 2017-06-13 Martin Liska <mliska@suse.cz>
5579 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
5580 (gate_asan): Likewise.
5581 * asan.h (asan_no_sanitize_address_p): Remove the function.
5582 (sanitize_flags_p): New function.
5583 * builtins.def: Fix coding style.
5584 * common.opt: Use renamed enum value.
5585 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
5586 * doc/extend.texi: Document no_sanitize attribute.
5587 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
5588 to SANITIZE_UNDEFINED_NONDEFAULT.
5589 * gcc.c (sanitize_spec_function): Use the renamed enum value.
5590 * gimple-fold.c (optimize_atomic_compare_exchange_p):
5591 Use sanitize_flags_p.
5592 * gimplify.c (gimplify_function_tree): Likewise.
5593 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
5594 * opts.c (parse_no_sanitize_attribute): New function.
5595 (common_handle_option): Use renamed enum value.
5596 * opts.h (parse_no_sanitize_attribute): Declare.
5597 * tree.c (sanitize_flags_p): New function.
5598 * tree.h: Declared here.
5599 * tsan.c: Use sanitize_flags_p.
5600 * ubsan.c (ubsan_expand_null_ifn): Likewise.
5601 (instrument_mem_ref): Likewise.
5602 (instrument_bool_enum_load): Likewise.
5603 (do_ubsan_in_current_function): Remove the function.
5604 (pass_ubsan::execute): Use sanitize_flags_p.
5605 * ubsan.h: Remove do_ubsan_in_current_function
5606 * tree-cfg.c (print_no_sanitize_attr_value): New function.
5607 (dump_function_to_file): Use it here.
5609 2017-06-13 Martin Jambor <mjambor@suse.cz>
5611 PR tree-optimization/80803
5612 PR tree-optimization/81063
5613 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
5614 (propagate_subaccesses_across_link): Enqueue subtree whenever
5615 necessary instead of relying on the caller.
5617 2017-06-13 Martin Jambor <mjambor@suse.cz>
5619 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
5620 that have a first_link.
5621 (sort_and_splice_var_accesses): Do not check first_link before
5623 (subtree_mark_written_and_enqueue): Likewise.
5624 (propagate_all_subaccesses): Likewise and do not stop at first
5625 parent with a first_link.
5627 2017-06-13 Martin Jambor <mjambor@suse.cz>
5629 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
5632 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5634 * match.pd: New pattern.
5636 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5638 * tree-vrp.c (is_masked_range_test): New function.
5639 (register_edge_assert_for): Determine ranges for
5642 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5644 PR tree-optimization/67328
5645 * fold-const.c (maskable_range_p): New function.
5646 (build_range_check): Generate bittests if possible.
5648 2017-06-13 Martin Liska <mliska@suse.cz>
5650 * gimple-pretty-print.c (dump_probability): Add new argument.
5651 (dump_edge_probability): Dump both probability and count.
5652 (dump_gimple_label): Likewise.
5653 (dump_gimple_bb_header): Likewise.
5655 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
5658 * config/avr/avr-devices.c: Fix indentation.
5659 * config/avr/gen-avr-mmcu-specs.c: Dito.
5661 2017-06-13 Richard Biener <rguenther@suse.de>
5663 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
5664 instead get vector type from stmt_info.
5665 (vectorizable_reduction): Adjust. Remove dead code.
5667 2017-06-13 Richard Biener <rguenther@suse.de>
5670 * fold-const.c (extract_muldiv_1): Remove bogus distribution
5671 case of C * (x * C2 + C3).
5672 (fold_addr_of_array_ref_difference): Properly fold index difference.
5674 2017-06-12 David S. Miller <davem@davemloft.net>
5677 * config/sparc/sparc.md (return expander): Emit frame blockage if
5678 function uses alloca.
5680 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
5682 * combine.c (make_field_assignment): Check len rather than the mode
5683 precision when calling force_to_mode.
5685 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
5687 Support multilibs and devices that see flash in RAM address range.
5690 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
5691 (avr_mcu_t) <flash_pm_offset>: New field.
5692 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
5693 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
5694 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
5695 (AVR_TINY_PM_OFFSET): Remove macro.
5696 * config/avr/avr.opt (-mshort-calls): New option.
5697 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
5698 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
5699 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
5700 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
5701 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
5702 instead of avr_arch->have_jmp_call.
5703 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
5704 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
5705 avr_arch->flash_pm_offset to define.
5706 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
5707 new field flash_pm_offset. Add entry for avrxmega3.
5708 (avr_texinfo): Add entry for avrxmega3.
5709 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
5710 attiny212, attiny214,
5711 attiny412, attiny414, attiny416, attiny417,
5712 attiny814, attiny816, attiny817,
5713 attiny1614, attiny1616, attiny1617,
5714 attiny3214, attiny3216, attiny3217.
5715 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
5716 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
5717 (avr_print_operand_address) [AVR_TINY]: Same.
5718 (avr_asm_init_sections) <readonly_data_section>: Only patch
5719 callback if avr_arch->flash_pm_offset = 0.
5720 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
5721 for rodata if avr_arch->flash_pm_offset != 0.
5722 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
5723 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
5724 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
5725 (m_options): Append opt_rcall.
5726 (m_dirnames): Append dir_rcall.
5727 * config/avr/t-multilib: Regenerate.
5729 * configure.ac [target=avr]: Check whether avrxmega3 default
5730 linker description file works as needed.
5731 * configure: Regenerate.
5732 * doc/avr-mmcu.texi: Regenerate.
5733 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
5734 <__AVR_ARCH__>: Document avrxmega3 and 103.
5735 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
5736 <__AVR_SHORT_CALLS__>: Document it.
5737 <__AVR_PM_BASE_ADDRESS__>: Document it.
5738 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
5739 (AVR Variable Attributes) <progmem>: Document this is
5740 not needed for avrxmega3.
5741 (AVR Named Address Spaces) <__flash>: Dito.
5743 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
5745 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
5747 2017-06-12 Doug Rupp <rupp@adacore.com>
5749 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
5750 Append vxworks-stdint.h to the tm_file list.
5751 * config/vxworks-stdint.h: New file.
5753 2017-06-12 Martin Liska <mliska@suse.cz>
5755 PR tree-optimization/81041
5756 * tree-profile.c (gimple_gen_ic_func_profiler):
5757 Create an extra BB in profile-generate
5758 (gimple_gen_time_profiler): Likewise.
5760 2017-06-12 Jakub Jelinek <jakub@redhat.com>
5762 PR tree-optimization/81003
5763 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
5764 (update_range_test): Use it instead of force_gimple_operand_gsi.
5766 2017-06-12 Richard Biener <rguenther@suse.de>
5768 PR tree-optimization/81053
5769 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
5770 with backedge value not defined in loop. Simplify def stmt
5773 2017-06-11 Tom de Vries <tom@codesourcery.com>
5776 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
5778 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
5779 nvptx_cannot_force_const_mem.
5781 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5783 * opts.c (finish_options): Move test for flag_split_stack after
5784 it has been initialized.
5786 2017-06-11 Jason Merrill <jason@redhat.com>
5788 * tree.h (id_equal): New.
5789 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
5790 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
5791 instead of strcmp of IDENTIFIER_POINTER.
5793 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5795 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
5796 (mark_all_inlined_calls_cdtor): Fix formating.
5797 (inline_transform): Rescale profile before inlining.
5799 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5801 * cgraph.h (cgraph_edge::clone): Update prototype.
5802 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
5803 (cgraph_node::create_clone): Update.
5804 (cgraph_node::create_version_clone): Update.
5805 * tree-inline.c (copy_bb): Update.
5806 (expand_call_inline): Update.
5808 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5810 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
5811 factored out from ...
5812 (rs6000_emit_prologue): ... here.
5814 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5816 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
5817 factored out from ...
5818 (rs6000_emit_prologue): ... here.
5820 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5822 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
5824 (handle_missing_profiles): Fix computation of tp_first_run.
5825 (counts_to_freqs): Do not touch freqs when count is 0.
5827 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5829 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
5832 2017-06-10 Tom de Vries <tom@codesourcery.com>
5834 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
5835 attributes): Document signal effective target.
5837 2017-06-10 Tom de Vries <tom@codesourcery.com>
5839 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
5840 Document effective target stack_size.
5842 2017-06-09 David Malcolm <dmalcolm@redhat.com>
5844 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
5845 to the edit_context if they can be auto-applied.
5847 2017-06-9 Ian Lance Taylor <iant@golang.org>
5849 * opts.c (finish_options): If -fsplit-stack, disable implicit
5850 -forder-blocks-and-partition.
5851 * doc/invoke.texi (Optimize Options): Document that when using
5852 -fsplit-stack -forder-blocks-and-partition is not implicitly
5855 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5857 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
5858 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
5859 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
5860 * builtins.def (abort, trap, unreachable): Declare cold.
5861 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
5862 * tree-core.h (ECF_COLD): New.
5863 * tree.c (set_call_expr_flags): Handle ECF_COLD.
5864 (build_common_builtin_nodes): Mark unreachable and abort as cold.
5866 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5868 * predict.c (unlikely_executed_stmt_p): Cleanup.
5870 2017-06-09 Richard Biener <rguenther@suse.de>
5872 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
5873 model if the ref is always written to.
5875 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5877 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
5879 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5881 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
5884 2017-06-09 Tom de Vries <tom@codesourcery.com>
5887 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
5888 "target cannot support label values" when encountering LABEL_REF.
5890 2017-06-09 Martin Liska <mliska@suse.cz>
5892 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
5893 (gimple_gen_ic_func_profiler): Emit direct comparison
5894 of __gcov_indirect_call_callee with NULL.
5895 (gimple_gen_time_profiler): Change probability from
5896 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
5898 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5900 * profile.c (edge_gcov_counts): Turn to pointer.
5901 (compute_branch_probabilities, compute_branch_probabilities): Update.
5902 (branch_prob): Do not clear edge_gcov_count.
5903 * profile.h (edge_gcov_counts): Turn to pointer.
5904 (edge_gcov_count): Update.
5906 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5908 * gimple.h (gimple_check_failed): Mark cold.
5910 2017-06-09 Richard Biener <rguenther@suse.de>
5912 PR tree-optimization/66623
5913 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
5914 refactor check_reduction into two parts, properly computing
5915 whether we have to check reduction validity for outer loop
5918 2017-06-09 Richard Biener <rguenther@suse.de>
5920 PR tree-optimization/79483
5921 * graphite-scop-detection.c (order): New global.
5922 (get_order): Compute bb to order mapping that satisfies code
5923 generation constraints.
5924 (cmp_pbbs): New helper.
5925 (build_scops): Start domwalk at entry block, sort generated
5928 2017-06-09 Richard Biener <rguenther@suse.de>
5931 * ipa-polymorphic-call.c
5932 (ipa_polymorphic_call_context::restrict_to_inner_class):
5933 Skip FIELD_DECLs with error_mark_node type.
5934 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
5937 2017-06-09 Martin Liska <mliska@suse.cz>
5939 * predict.c (struct branch_predictor): New struct.
5940 (test_prediction_value_range): New test.
5941 (predict_c_tests): New function.
5942 * selftest-run-tests.c (selftest::run_tests): Run the function.
5943 * selftest.h: Declare new tests.
5945 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
5948 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
5949 gen_add3_insn did not fail.
5950 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
5951 r0, construct that number in a temporary reg and add that reg to r0.
5952 If asked to put the result in r0 as well, fail.
5954 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
5956 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5957 for early expansion of vec_eqv.
5959 2017-06-08 Jakub Jelinek <jakub@redhat.com>
5962 * ubsan.c (instrument_null): Avoid pointless code temporary.
5963 (pass_ubsan::execute): Instrument aggregate arguments of calls.
5965 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
5969 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
5972 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
5975 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5977 * predict.c (unlikely_executed_edge_p): Move ahead.
5978 (probably_never_executed_edge_p): Use it.
5980 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5983 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
5984 gimple_call_builtin_p call.
5986 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5988 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
5989 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
5990 rtl_check_failed_type2, rtl_check_failed_code1,
5991 rtl_check_failed_code2, rtl_check_failed_code_mode,
5992 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
5993 rtvec_check_failed_bounds, rtl_check_failed_flag,
5994 _fatal_insn_not_found, _fatal_insn): Likewise.
5995 * tree.h (tree_contains_struct_check_failed,
5996 tree_check_failed, tree_not_check_failed,
5997 tree_class_check_failed, tree_range_check_failed,
5998 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
5999 tree_vec_elt_check_failed, phi_node_elt_check_failed,
6000 tree_operand_check_failed, omp_clause_check_failed,
6001 omp_clause_operand_check_failed, omp_clause_range_check_failed):
6004 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6006 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
6007 flag_branch_probabilities.
6008 * ipa-inline.c (edge_badness): Likewise.
6009 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
6010 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
6011 * predict.c (maybe_hot_frequency_p): Likewise.
6012 (probably_never_executed): Likewise.
6013 * sched-ebb.c (schedule_ebbs): Likewise.
6014 * sched-rgn.c (find_single_block_region): Likewise.
6015 * tracer.c (tail_duplicate): Likewise.
6017 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6019 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
6020 longer requires x_flag_profile_use.
6022 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6024 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
6025 instead of flag_reorder_blocks_and_partition.
6026 * dbxout.c (dbxout_function_end): Likewise.
6027 * dwarf2out.c (gen_subprogram_die): Likewise.
6028 * haifa-sched.c (sched_create_recovery_edges): Likewise.
6029 * hw-doloop.c (reorg_loops): Likewise.
6030 * varasm.c (assemble_start_function,
6031 assemble_end_function): Likewise.
6032 (decide_function_section): Do not check for
6033 flag_reorder_blocks_and_partition.
6035 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
6037 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
6039 (chkp_get_hard_register_fake_addr_expr): Ditto.
6040 (chkp_build_addr_expr): Add check for hard reg case.
6041 (chkp_parse_array_and_component_ref): Ditto.
6042 (chkp_find_bounds_1): Ditto.
6043 (chkp_process_stmt): Don't generate bounds store for
6046 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6048 * predict.c (maybe_hot_bb_p): Do not check profile status.
6049 (maybe_hot_edge_p): Likewise.
6050 (probably_never_executed): Check for zero counts even if profile
6052 (unlikely_executed_edge_p): New function.
6053 (unlikely_executed_stmt_p): New function.
6054 (unlikely_executed_bb_p): New function.
6055 (set_even_probabilities): Use unlikely predicates.
6056 (combine_predictions_for_bb): Likewise.
6057 (predict_paths_for_bb): Likewise.
6058 (predict_paths_leading_to_edge): Likewise.
6059 (determine_unlikely_bbs): New function.
6060 (estimate_bb_frequencies): Use it.
6061 (compute_function_frequency): Use zero counts even if profile is
6063 * profile-count.h: Fix typo.
6065 2017-08-08 Julia Koval <julia.koval@intel.com>
6067 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
6068 _mm512_mask_cvtsepi16_storeu_epi8,
6069 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
6070 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
6071 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
6072 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
6073 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
6074 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
6075 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
6076 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
6077 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
6078 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
6079 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
6080 __builtin_ia32_pmovuswb256mem_mask,
6081 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
6082 __builtin_ia32_pmovwb512mem_mask): New builtins.
6084 2017-08-08 Julia Koval <julia.koval@intel.com>
6086 PR target/73350,80862
6087 * config/i386/subst.md (round): Fix round pattern.
6088 * config/i386/i386.c (ix86_erase_embedded_rounding):
6089 Fix erasing rounding for the fixed pattern.
6091 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6093 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
6095 2017-06-08 Martin Liska <mliska@suse.cz>
6097 PR gcov-profile/80911
6098 * gcov.c (block_info::block_info): New constructor.
6100 2017-06-07 Carl Love <cel@us.ibm.com>
6102 * config/rs6000/rs6000-c: The return type of the following
6103 built-in functions was implemented as int not long long. Fix sign
6104 of return value for the unsigned version of vec_mulo and vec_mule.
6105 vector unsigned long long vec_bperm (vector unsigned long long,
6106 vector unsigned char)
6107 vector signed long long vec_mule (vector signed int,
6109 vector unsigned long long vec_mule (vector unsigned int,
6110 vector unsigned int)
6111 vector signed long long vec_mulo (vector signed int,
6113 vector unsigned long long vec_mulo (vector unsigned int,
6114 vector unsigned int)
6115 * doc/extend.texi: Fix the documentation for the built-in
6118 2017-06-07 Carl Love <cel@us.ibm.com>
6121 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
6124 2017-06-07 Carl Love <cel@us.ibm.com>
6126 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
6127 support, Generate doublehv for signed int/float for BE case only.
6129 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
6131 * doc/invoke.texi (mcx16): Rewrite.
6133 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6135 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
6136 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
6137 *mov<mode>_softfloat, and an anonymous splitter): Use
6138 nonimmediate_operand instead of rs6000_nonimmediate_operand.
6140 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6142 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
6144 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
6145 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
6146 (rs6000_debug_reg_global): Adjust.
6147 (rs6000_init_hard_regno_mode_ok): Adjust.
6148 (rs6000_dbx_register_number): Adjust.
6149 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
6150 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
6151 Remove SPE_ACC and SPEFSCR.
6152 (REG_ALLOC_ORDER): Ditto.
6153 (FRAME_POINTER_REGNUM): Change to 111.
6154 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
6155 (REG_CLASS_NAMES): Ditto.
6156 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
6157 (REGISTER_NAMES): Ditto.
6158 (ADDITIONAL_REG_NAMES): Ditto.
6159 (rs6000_reg_names): Ditto.
6160 * config/rs6000/rs6000.md: Renumber some register number
6163 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6165 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
6167 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
6168 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
6170 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
6171 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
6172 Delete the SPE high registers.
6173 (REG_ALLOC_ORDER): Ditto.
6174 (enum reg_class): Remove SPE_HIGH_REGS.
6175 (REG_CLASS_NAMES): Ditto.
6176 (REG_CLASS_CONTENTS): Delete the SPE high registers.
6177 (REGISTER_NAMES): Ditto.
6178 (rs6000_reg_names): Ditto.
6179 * doc/tm.texi.in: Remove SPE as example.
6180 * doc/tm.texi: Regenerate.
6182 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6184 * config/rs6000/8540.md (ppc8540_brinc): Delete.
6185 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
6186 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
6187 * config/rs6000/rs6000.md (type): Remove "brinc".
6189 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6191 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
6192 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
6193 * config/rs6000/linuxspe.h: Delete file.
6194 * config/rs6000/rs6000.md: Don't include spe.md.
6195 * config/rs6000/spe.h: Delete file.
6196 * config/rs6000/spe.md: Delete file.
6197 * config/rs6000/t-rs6000: Remove spe.md.
6199 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6201 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
6202 (reg_or_none500mem_operand): Delete.
6203 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
6204 instead of reg_or_none500mem_operand.
6206 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6208 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
6209 handling of SPE flags.
6210 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
6212 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6214 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
6216 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
6217 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
6218 paired_divv2sf3): Similar.
6219 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
6220 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
6221 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
6223 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
6224 Rename the paired_* instruction patterns.
6225 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
6227 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
6228 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
6229 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
6230 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
6231 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
6233 (struct machine_function): Delete field spe_insn_chain_scanned_p.
6234 (spe_func_has_64bit_regs_p): Delete.
6235 (spe_expand_predicate_builtin): Delete.
6236 (spe_expand_evsel_builtin): Delete.
6237 (TARGET_DWARF_REGISTER_SPAN): Do not define.
6238 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
6239 (invalid_e500_subreg): Delete.
6240 (rs6000_legitimize_address): Always force_reg op2 as well, for
6241 paired single memory accesses.
6242 (rs6000_member_type_forces_blk): Delete.
6243 (rs6000_spe_function_arg): Delete.
6244 (rs6000_expand_unop_builtin): Delete SPE handling.
6245 (rs6000_expand_binop_builtin): Ditto.
6246 (spe_expand_stv_builtin): Delete.
6247 (bdesc_2arg_spe): Delete.
6248 (spe_expand_builtin): Delete.
6249 (spe_expand_predicate_builtin): Delete.
6250 (spe_expand_evsel_builtin): Delete.
6251 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
6252 (spe_init_builtins): Delete.
6253 (spe_func_has_64bit_regs_p): Delete.
6254 (savres_routine_name): Delete "info" parameter. Adjust callers.
6255 (rs6000_emit_stack_reset): Ditto.
6256 (rs6000_dwarf_register_span): Delete.
6257 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
6258 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
6259 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
6261 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
6263 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
6264 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
6265 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
6266 mulv2sf3, divv2sf3): Delete expanders.
6268 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6270 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
6272 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6274 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
6275 * config/rs6000/rs6000.c: Ditto.
6277 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6279 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
6280 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
6281 comparison_operator.
6283 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6285 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
6286 * config/rs6000/rs6000.opt: Ditto.
6287 * config/rs6000/t-rtems: Ditto.
6289 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6291 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
6292 TARGET_E500_SINGLE by 0, simplify.
6293 * config/rs6000/rs6000.c: Ditto.
6294 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
6295 (spe_build_register_parallel): Delete.
6296 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
6297 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
6298 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
6299 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
6300 (E500_CONVERT): Delete.
6301 * config/rs6000/spe.md: Remove many patterns and all define_constants.
6303 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6305 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
6306 * config/rs6000/dfp.md: Ditto.
6307 (negdd2, *negdd2_fpr): Merge.
6308 (absdd2, *absdd2_fpr): Merge.
6309 (negtd2, *negtd2_fpr): Merge.
6310 (abstd2, *abstd2_fpr): Merge.
6311 * config/rs6000/e500.h: Delete file.
6312 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
6313 TARGET_FPRS by 1 and simplify.
6314 * config/rs6000/rs6000-c.c: Ditto.
6315 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
6317 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
6319 * config/rs6000/rs6000.md: Ditto.
6320 (floatdidf2, *floatdidf2_fpr): Merge.
6321 (move_from_CR_gt_bit): Delete.
6322 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
6323 (E500_CR_IOR_COMPARE): Delete.
6324 (All patterns that require !TARGET_FPRS): Delete.
6325 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
6327 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6329 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
6331 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6333 * graphds.c (add_edge): Intitialize edge's attached data.
6334 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
6335 pointer parameter. Call pointed function on each edge during
6336 graph traversing. Skip traversing the edge when the function
6338 (graphds_dfs, graphds_scc): Ditto.
6339 (for_each_edge): New parameter. Pass the new parameter to callback
6341 * graphds.h (skip_edge_callback): New function pointer type.
6342 (graphds_dfs, graphds_scc): New function pointer parameter.
6343 (graphds_edge_callback, for_each_edge): New parameter.
6345 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6347 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
6348 out code checking if runtime alias check is possible to below ...
6349 Call the new function.
6350 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
6351 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
6353 2017-06-07 Marek Polacek <polacek@redhat.com>
6356 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
6357 TYPE_OVERFLOW_WRAPS check.
6359 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6361 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
6362 if versioning is required.
6363 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
6364 peeling with the check for versioning.
6366 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6368 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
6369 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
6370 Set true to new parameter if new ssa variable is defined.
6371 (vect_gen_vector_loop_niters): Refactor. Set range information
6372 for the new vector loop bound variable.
6373 (vect_do_peeling): Ditto.
6375 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6377 * tree-affine.c (ssa.h): Include header file.
6378 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
6379 has wrapping overflow behavior.
6381 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6383 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
6385 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6387 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
6388 (tree_to_aff_combination): ... here.
6390 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6392 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
6393 reg_pressure model function.
6394 (ivopts_global_cost_for_size): Delete.
6395 (determine_set_costs, iv_ca_recount_cost): Call new model function
6396 ivopts_estimate_reg_pressure.
6398 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6400 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
6401 expensive than udiv. Remove floating point cases from mod.
6403 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6405 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
6408 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6410 * config/aarch64/aarch64.md
6411 (copysignsf3): Fix mask generation.
6413 2017-06-07 Jakub Jelinek <jakub@redhat.com>
6415 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
6417 (class dump_manager): Add register_dumps method.
6418 * dumpfile.c: Include langhooks.h.
6419 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
6420 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
6421 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
6422 (dump_manager::dump_register): Start with 512 entries instead of 32.
6423 (dump_manager::register_dumps): New method.
6424 * toplev.c (general_init): Instead of invoking register_dumps
6425 langhook, invoke register_dumps method on the dump manager.
6426 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
6429 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
6431 * doc/md.texi: Clarify the restrictions on a define_insn condition.
6432 Say that # requires an associated define_split to exist, and that
6433 the define_split must be suitable for use after register allocation.
6435 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6437 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
6438 (compute_outgoing_frequencies): Also initialize zero counts.
6439 (find_many_sub_basic_blocks): Do not produce uninitialized profile
6440 around loops; preserve more of profile when nothing changes.
6442 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
6444 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
6446 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
6447 * config/arm/arm-cpu-cdata.h: Regenerate.
6448 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
6449 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
6450 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
6451 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
6452 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
6454 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
6456 * config/arm/t-rmprofile: Likewise.
6457 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
6459 2017-06-06 David S. Miller <davem@davemloft.net>
6462 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
6463 blockage if function uses alloca.
6465 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
6467 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
6468 New "uid" fields to hold pretty-print IDs of group and ref.
6469 Memory references are now identified as <group_id>:<ref_id>
6470 instead of using [random] addresses.
6471 (dump_mem_details): Simplify, no functional change.
6472 (dump_mem_ref): Simplify and make output more concise.
6473 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
6474 (find_or_create_group): Initialize group uid.
6475 (record_ref): Initialize ref uid. Improve debug output.
6476 (prune_group_by_reuse, should_issue_prefetch_p,)
6477 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
6478 (mark_nontemporal_store, determine_loop_nest_reuse):
6479 Improve debug output.
6481 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
6483 * dbgcnt.def (prefetch): New debug counter.
6484 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
6485 (schedule_prefetches): Stop issueing prefetches if debug counter
6488 2017-06-06 Tom de Vries <tom@codesourcery.com>
6490 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
6491 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
6493 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6495 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
6496 Use aarch64_reg_or_zero predicate for operand 4.
6497 (aarch64_compare_and_swap<mode> define_insn_and_split):
6498 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
6499 (aarch64_store_exclusive<mode>): Likewise for operand 2.
6501 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
6503 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
6504 (arm_compute_save_core_reg_mask): This.
6505 (thumb1_compute_save_reg_mask): Rename into ...
6506 (thumb1_compute_save_core_reg_mask): This.
6507 (arm_compute_save_reg0_reg12_mask): Adapt comment.
6508 (arm_compute_frame_layout): Likewise.
6510 2017-06-06 Richard Biener <rguenther@suse.de>
6512 PR tree-optimization/80974
6513 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
6514 keep or clear leaders SSA info.
6516 2017-06-06 Tom de Vries <tom@codesourcery.com>
6518 * config/nvptx/nvptx.c (split_mode_p): New function.
6519 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
6521 2017-06-06 Tom de Vries <tom@codesourcery.com>
6523 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
6525 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6528 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
6531 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6533 * shrink-wrap.c (handle_simple_exit): Update profile.
6534 (try_shrink_wrapping): Upate profile.
6536 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6538 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
6539 (tree_guess_outgoing_edge_probabilities): New.
6540 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
6541 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
6543 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6545 * ipa-split.c (split_function): Initialize return bb profile.
6547 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6549 * profile.c (compute_branch_probabilities): Also initialize
6552 2017-06-06 Richard Biener <rguenther@suse.de>
6554 PR tree-optimization/80928
6555 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
6556 (vect_analyze_loop_operations): Properly guard analysis for
6558 (vect_transform_loop): Likewise.
6559 (vect_analyze_loop_2): Also reset SLP type on PHIs.
6560 (vect_model_induction_cost): Do not cost for pure SLP.
6561 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
6562 of induction in inner loop vectorization.
6563 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
6564 (vect_get_and_check_slp_defs): Handle vect_induction_def.
6565 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
6567 (vect_analyze_slp_cost_1): Cost induction.
6568 (vect_detect_hybrid_slp_stmts): Handle PHIs.
6569 (vect_get_slp_vect_defs): Likewise.
6570 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
6571 (vect_transform_stmt): Handle SLP reductions.
6572 * tree-vectorizer.h (vectorizable_induction): Adjust.
6574 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6576 * config/rs6000/rs6000.c (make_resolver_func): Update
6577 init_lowered_empty_function call.
6579 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
6581 * doc/invoke.texi: Document the -fprofile-abs-path option.
6582 * common.opt (fprofile-abs-path): New option.
6583 * gcov-io.h (gcov_write_filename): Declare.
6584 * gcov-io.c (gcov_write_filename): New function.
6585 * coverage.c (coverage_begin_function): Use gcov_write_filename.
6586 * profile.c (output_location): Likewise.
6588 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6590 * shring-wrap.c: Revert accidental commit.
6592 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
6594 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
6596 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6598 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
6600 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
6601 profile in callgraph edge.
6602 * profile-count.h (apply_probability): If THIS is 0, then result is 0
6603 (apply_scale): Likewise.
6604 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
6605 Also scale profile when inlining function with zero profile.
6606 (initialize_cfun): Update exit block profile even when it is zero.
6607 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
6608 when profile is read.
6610 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6612 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
6613 (CLONE_*): New constants to define the processors we can generate
6614 code for with the target_clone attribute.
6615 (rs6000_clone_map): New array to identify which clone processors
6616 the current program is running on.
6617 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
6618 target_clone attribute.
6619 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
6620 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
6621 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
6622 (cpu_expand_builtin): Add support for target_clone attribute.
6623 (rs6000_valid_attribute_p): Allow "default" attribute.
6624 (get_decl_name): New debug function to simplify printing the
6625 current function name in debugging statements.
6626 (rs6000_clone_priority): New functions to support the target_clone
6627 attribute, and be able to generate code to switch between ISA 2.05
6628 through ISA 3.0 (power6 through power9).
6629 (rs6000_compare_version_priority): Likewise.
6630 (rs6000_get_function_versions_dispatcher): Likewise.
6631 (make_resolver_func): Likewise.
6632 (add_condition_to_bb): Likewise.
6633 (dispatch_function_versions): Likewise.
6634 (rs6000_generate_version_dispatcher_body): Likewise.
6635 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
6636 (fusion_gpr_load_p): Fix a spacing issue.
6637 * doc/extend.texi (Common Function Attributes): Document that the
6638 PowerPC supports the target_clone attribute.
6640 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
6642 * config/arm/arm.h: explain F symbol found in description of ARM
6643 register allocation in its legend.
6645 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6647 * config/mips/frame-header-opt.c: Include profile-count.h.
6648 * config/riscv/riscv.c: Include profile-count.h
6650 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6652 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
6654 (sm_set_flag_if_changed): Add bbs field.
6655 (execute_sm_if_changed_flag_set): Pass BBS.
6656 (execute_sm): Update.
6658 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6660 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6663 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6665 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
6666 (peephole2): New peephole2 to emit the above.
6667 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
6669 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6671 * config/aarch64/aarch64.c (define_peephole2 above
6672 *sub_<shift>_<mode>): New peephole.
6674 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
6676 * config/i386/i386.c (make_resolver_func): Update.
6677 * Makefile.in: Add profile-count.h and profile-count.o
6678 * auto-profile.c (afdo_indirect_call): Update to new API.
6679 (afdo_set_bb_count): Update.
6680 (afdo_propagate_edge): Update.
6681 (afdo_propagate_circuit): Update.
6682 (afdo_calculate_branch_prob): Update.
6683 (afdo_annotate_cfg): Update.
6684 * basic-block.h: Include profile-count.h
6685 (struct edge_def): Turn count to profile_count.
6686 (struct basic_block_def): Likewie.
6687 (REG_BR_PROB_BASE): Move to profile-count.h
6688 (RDIV): Move to profile-count.h
6689 * bb-reorder.c (max_entry_count): Turn to profile_count.
6690 (find_traces): Update.
6691 (rotate_loop):Update.
6692 (connect_traces):Update.
6693 (sanitize_hot_paths):Update.
6694 * bt-load.c (migrate_btr_defs): Update.
6695 * cfg.c (RDIV): Remove.
6696 (init_flow): Use alloc_block.
6697 (alloc_block): Uninitialize count.
6698 (unchecked_make_edge): Uninitialize count.
6699 (check_bb_profile): Update.
6700 (dump_edge_info): Update.
6701 (dump_bb_info): Update.
6702 (update_bb_profile_for_threading): Update.
6703 (scale_bbs_frequencies_int): Update.
6704 (scale_bbs_frequencies_gcov_type): Update.
6705 (scale_bbs_frequencies_profile_count): New.
6706 * cfg.h (update_bb_profile_for_threading): Update.
6707 (scale_bbs_frequencies_profile_count): Declare.
6708 * cfgbuild.c (compute_outgoing_frequencies): Update.
6709 (find_many_sub_basic_blocks): Update.
6710 * cfgcleanup.c (try_forward_edges): Update.
6711 (try_crossjump_to_edge): Update.
6712 * cfgexpand.c (expand_gimple_tailcall): Update.
6713 (construct_exit_block): Update.
6714 * cfghooks.c (verify_flow_info): Update.
6715 (dump_bb_for_graph): Update.
6716 (split_edge): Update.
6717 (make_forwarder_block): Update.
6718 (duplicate_block): Update.
6719 (account_profile_record): Update.
6720 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
6721 (get_estimated_loop_iterations): Update.
6722 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
6723 (single_likely_exit): Update.
6724 * cfgloopmanip.c (scale_loop_profile): Update.
6726 (set_zero_probability): Update.
6727 (lv_adjust_loop_entry_edge): Update.
6728 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
6729 (purge_dead_edges): Update.
6730 (rtl_account_profile_record): Update.
6731 * cgraph.c (cgraph_node::create): Uninitialize count.
6732 (symbol_table::create_edge): Uninitialize count.
6733 (cgraph_update_edges_for_call_stmt_node): Update.
6734 (cgraph_edge::dump_edge_flags): Update.
6735 (cgraph_node::dump): Update.
6736 (cgraph_edge::maybe_hot_p): Update.
6737 * cgraph.h: Include profile-count.h
6738 (create_clone), create_edge, create_indirect_edge): Update.
6739 (cgraph_node): Turn count to profile_count.
6740 (cgraph_edge0: Likewise.
6741 (make_speculative, clone): Update.
6742 (create_edge): Update.
6743 (init_lowered_empty_function): Update.
6744 * cgraphclones.c (cgraph_edge::clone): Update.
6745 (duplicate_thunk_for_node): Update.
6746 (cgraph_node::create_clone): Update.
6747 * cgraphunit.c (cgraph_node::analyze): Update.
6748 (cgraph_node::expand_thunk): Update.
6749 * final.c (dump_basic_block_info): Update.
6750 * gimple-streamer-in.c (input_bb): Update.
6751 * gimple-streamer-out.c (output_bb): Update.
6752 * graphite.c (print_global_statistics): Update.
6753 (print_graphite_scop_statistics): Update.
6754 * hsa-brig.c: Include basic-block.h.
6755 * hsa-dump.c: Include basic-block.h.
6756 * hsa-gen.c (T sum_slice): Update.
6757 (convert_switch_statements):Update.
6758 * hsa-regalloc.c: Include basic-block.h.
6759 * ipa-chkp.c (chkp_produce_thunks): Update.
6760 * ipa-cp.c (struct caller_statistics): Update.
6761 (init_caller_stats): Update.
6762 (gather_caller_stats): Update.
6763 (ipcp_cloning_candidate_p): Update.
6764 (good_cloning_opportunity_p): Update.
6765 (get_info_about_necessary_edges): Update.
6766 (dump_profile_updates): Update.
6767 (update_profiling_info): Update.
6768 (update_specialized_profile): Update.
6769 (perhaps_add_new_callers): Update.
6770 (decide_about_value): Update.
6771 (ipa_cp_c_finalize): Update.
6772 * ipa-devirt.c (struct odr_type_warn_count): Update.
6773 (struct decl_warn_count): Update.
6774 (struct final_warning_record): Update.
6775 (possible_polymorphic_call_targets): Update.
6776 (ipa_devirt): Update.
6777 * ipa-fnsummary.c (redirect_to_unreachable): Update.
6778 * ipa-icf.c (sem_function::merge): Update.
6779 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
6780 * ipa-inline.c (compute_uninlined_call_time): Update.
6781 (compute_inlined_call_time): Update.
6782 (want_inline_small_function_p): Update.
6783 (want_inline_self_recursive_call_p): Update.
6784 (edge_badness): Update.
6785 (lookup_recursive_calls): Update.
6786 (recursive_inlining): Update.
6787 (inline_small_functions): Update.
6788 (dump_overall_stats): Update.
6789 (dump_inline_stats): Update.
6790 * ipa-profile.c (ipa_profile_generate_summary): Update.
6791 (ipa_propagate_frequency): Update.
6792 (ipa_profile): Update.
6793 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
6794 * ipa-utils.c (ipa_merge_profiles): Update.
6795 * loop-doloop.c (doloop_modify): Update.
6796 * loop-unroll.c (report_unroll): Update.
6797 (unroll_loop_runtime_iterations): Update.
6798 * lto-cgraph.c (lto_output_edge): Update.
6799 (lto_output_node): Update.
6800 (input_node): Update.
6801 (input_edge): Update.
6802 (merge_profile_summaries): Update.
6803 * lto-streamer-in.c (input_cfg): Update.
6804 * lto-streamer-out.c (output_cfg): Update.
6805 * mcf.c (create_fixup_graph): Update.
6806 (adjust_cfg_counts): Update.
6807 (sum_edge_counts): Update.
6808 * modulo-sched.c (sms_schedule): Update.
6809 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
6810 * predict.c (maybe_hot_count_p): Update.
6811 (probably_never_executed): Update.
6812 (dump_prediction): Update.
6813 (combine_predictions_for_bb): Update.
6814 (propagate_freq): Update.
6815 (handle_missing_profiles): Update.
6816 (counts_to_freqs): Update.
6817 (rebuild_frequencies): Update.
6818 (force_edge_cold): Update.
6819 * predict.h: Include profile-count.h
6820 (maybe_hot_count_p, counts_to_freqs): UPdate.
6821 * print-rtl-function.c: Do not include cfg.h
6822 * print-rtl.c: Include basic-block.h
6823 * profile-count.c: New file.
6824 * profile-count.h: New file.
6825 * profile.c (is_edge_inconsistent): Update.
6826 (correct_negative_edge_counts): Update.
6827 (is_inconsistent): Update.
6828 (set_bb_counts): Update.
6829 (read_profile_edge_counts): Update.
6830 (compute_frequency_overlap): Update.
6831 (compute_branch_probabilities): Update; Initialize and deinitialize
6833 (branch_prob): Update.
6834 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
6835 (edge_gcov_count): New.
6836 (bb_gcov_count): New.
6837 * shrink-wrap.c (try_shrink_wrapping): Update.
6838 * tracer.c (better_p): Update.
6839 * trans-mem.c (expand_transaction): Update.
6840 (ipa_tm_insert_irr_call): Update.
6841 (ipa_tm_insert_gettmclone_call): Update.
6842 * tree-call-cdce.c: Update.
6843 * tree-cfg.c (gimple_duplicate_sese_region): Update.
6844 (gimple_duplicate_sese_tail): Update.
6845 (gimple_account_profile_record): Update.
6846 (execute_fixup_cfg): Update.
6847 * tree-inline.c (copy_bb): Update.
6848 (copy_edges_for_bb): Update.
6849 (initialize_cfun): Update.
6850 (freqs_to_counts): Update.
6851 (copy_cfg_body): Update.
6852 (expand_call_inline): Update.
6853 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
6854 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
6855 (try_unroll_loop_completely): Update.
6856 (try_peel_loop): Update.
6857 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6858 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
6859 * tree-ssa-loop-split.c (connect_loops): Update.
6860 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
6861 * tree-ssa-reassoc.c (branch_fixup): Update.
6862 * tree-ssa-tail-merge.c (replace_block_by): Update.
6863 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
6864 (compute_path_counts): Update.
6865 (update_profile): Update.
6866 (recompute_probabilities): Update.
6867 (update_joiner_offpath_counts): Update.
6868 (estimated_freqs_path): Update.
6869 (freqs_to_counts_path): Update.
6870 (clear_counts_path): Update.
6871 (ssa_fix_duplicate_block_edges): Update.
6872 (duplicate_thread_path): Update.
6873 * tree-switch-conversion.c (case_bit_test_cmp): Update.
6874 (struct switch_conv_info): Update.
6875 * tree-tailcall.c (decrease_profile): Update.
6876 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
6877 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6878 * value-prof.c (check_counter): Update.
6879 (gimple_divmod_fixed_value): Update.
6880 (gimple_mod_pow2): Update.
6881 (gimple_mod_subtract): Update.
6882 (gimple_ic_transform): Update.
6883 (gimple_stringop_fixed_value): Update.
6884 * value-prof.h (gimple_ic): Update.
6886 2017-06-02 Carl Love <cel@us.ibm.com>
6888 * config/rs6000/rs6000-c: Add support for built-in functions
6889 vector double vec_doublee (vector signed int);
6890 vector double vec_doublee (vector unsigned int);
6891 vector double vec_doublee (vector float);
6892 vector double vec_doubleh (vector signed int);
6893 vector double vec_doubleh (vector unsigned int);
6894 vector double vec_doubleh (vector float);
6895 vector double vec_doublel (vector signed int);
6896 vector double vec_doublel (vector unsigned int);
6897 vector double vec_doublel (vector float);
6898 vector double vec_doubleo (vector signed int);
6899 vector double vec_doubleo (vector unsigned int);
6900 vector double vec_doubleo (vector float);.
6901 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
6902 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
6904 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
6905 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
6906 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
6908 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
6909 vec_doublel, vec_doubleh.
6910 * doc/extend.texi: Update the built-in documentation file for the
6911 new built-in functions.
6913 2017-06-02 David Malcolm <dmalcolm@redhat.com>
6916 * ipa-inline-analysis.c (free_growth_caches): Set
6917 edge_removal_hook_holder to NULL after removing it.
6919 2017-06-02 Sudakshina Das <sudi.das@arm.com>
6921 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
6922 comparision with zero.
6924 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6925 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6926 for early expansion of vec_min and vec_max builtins.
6927 (builtin_function_type): Add min/max unsigned variants to those
6928 identified as having unsigned arguments.
6930 2017-06-02 Olivier Hainque <hainque@adacore.com>
6932 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
6934 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6936 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
6937 Use VALL_F16 iterator rather than VALL.
6939 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6941 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
6942 Emit CBNZ inside loop when doing a strong exchange and comparing
6943 against zero. Generate the CC flags after the loop.
6945 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
6947 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
6948 (dl_section_ref): New.
6949 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
6950 On AIX, append an expression to subtract the size of the
6951 section length to dl_section_ref.
6953 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6955 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6956 for early expansion of vector absolute builtins.
6958 2017-06-02 Richard Biener <rguenther@suse.de>
6960 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
6961 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
6963 2017-06-02 Richard Biener <rguenther@suse.de>
6965 PR tree-optimization/80948
6966 * tree-tailcall.c (find_tail_calls): Track stmts to move in
6969 2017-06-02 Richard Biener <rguenther@suse.de>
6971 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
6973 * tree-vect-stmts.c (process_use): Do not mark backedge defs
6974 for inductions as relevant.
6976 2017-06-02 Richard Biener <rguenther@suse.de>
6978 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
6979 (vectorizable_induction): ... this. Remove dead code.
6981 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
6983 * builtins. (expand_builtin_alloca): Remove second parameter and
6984 infer its value from the first parameter instead.
6985 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
6987 2017-06-02 Jakub Jelinek <jakub@redhat.com>
6989 PR rtl-optimization/80903
6990 * loop-doloop.c (add_test): Unshare sequence.
6992 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
6994 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
6996 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
6998 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
7000 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
7001 xlogue_layout::get_instance, logue_layout::xlogue_layout,
7002 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
7003 (xlogue_layout::get_stub_rtx): Make static.
7004 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
7005 (xlogue_layout::compute_stub_managed_regs): Rename to...
7006 (xlogue_layout::count_stub_managed_regs): ...this.
7007 (xlogue_layout::is_stub_managed_reg): New function.
7008 (xlogue_layout::m_stub_names): Rename to...
7009 (xlogue_layout::s_stub_names): ...this, make static.
7010 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
7011 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
7012 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
7013 xlogue_layout::s_stub_names): Instantiate statics.
7014 (stub_managed_regs): Remove.
7015 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
7016 (disable_call_ms2sysv_xlogues): Rename to...
7017 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
7018 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
7020 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
7021 change after reload_completed.
7022 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
7024 (ix86_expand_prologue): Likewise.
7025 (ix86_expand_epilogue): Likewise.
7026 (ix86_expand_split_stack_prologue): Likewise.
7027 (ix86_compute_frame_layout): Remove frame parameter ...
7028 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
7029 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
7031 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
7032 (ix86_frame): Move from here ...
7033 * config/i386/i386.h (ix86_frame): ... to here.
7034 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
7035 complete ix86_frame data structure instead. Remove some_ld_name.
7037 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
7039 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
7040 symbols that hold a DECL_VALUE_EXPR.
7042 2017-06-01 Martin Jambor <mjambor@suse.cz>
7044 PR tree-optimization/80898
7045 * tree-sra.c (process_subtree_disqualification): Removed.
7046 (disqualify_candidate): Do not acll
7047 process_subtree_disqualification.
7048 (subtree_mark_written_and_enqueue): New function.
7049 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
7050 RHS has been disqualified and re-queue LHS if necessary. Apart
7051 from that, ignore disqualified RHS.
7053 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7055 * config/s390/s390.c (s390_emit_epilogue): Disable early return
7056 address fetch for z10 or later.
7058 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7060 * config/arc/arc.md (tst_movb): Add guard when splitting.
7062 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7064 * config/arc/arc.c (arc_can_eliminate): Test against
7065 arc_frame_pointer_needed.
7067 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7069 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
7070 to prevent store reordering.
7071 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
7072 (type): Add block type.
7073 (stack_tie): Define special instruction to be used in
7076 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7078 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
7079 constraint. It is not valid for the pattern.
7080 (noncommutative_binary_comparison): Likewise.
7082 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7084 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
7087 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7089 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
7090 be used by the reg-alloc.
7092 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7094 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
7095 reg-alloc when having mul64 or mul32x16 instructions.
7096 (mulsidi3): Likewise.
7097 (umulsidi3): Likewise.
7098 (mulsi32x16): New pattern.
7099 (mulsi64): Likewise.
7100 (mulsidi64): Likewise.
7101 (umulsidi64): Likewise.
7102 (MUL32x16_REG): Define.
7103 (mul64_600): Use MUL32x16_REG.
7104 (mac64_600): Likewise.
7105 (umul64_600): Likewise.
7106 (umac64_600): Likewise.
7108 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7110 * config/arc/arc.md (mulsi3_700): Make it commutative.
7112 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
7114 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
7116 (*sign_extendsidi2_insn): Likewise for movstosw.
7118 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
7120 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
7121 the type of the input discriminant value. Convert the
7122 discriminant value of signedness vary.
7124 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
7126 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
7127 Add to -Wall section.
7129 2017-06-01 Richard Biener <rguenther@suse.de>
7132 * fold-const.c (fold_plusminus_mult_expr): If the factored
7133 factor may be zero use a wrapping type for the inner operation.
7134 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
7135 and handle moved defs.
7136 (process_assignment): Properly guard the unary op case. Return a
7137 tri-state indicating that moving the stmt before the call may allow
7138 to continue. Pass through to_move.
7139 (find_tail_calls): Handle moving unrelated defs before
7142 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
7145 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
7146 splitter result in the canonical way.
7148 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
7150 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
7151 also for 32bit target. Update insn attributes.
7152 (zero-extendsidi2 splitter): Allow all registers for operand 1.
7154 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
7156 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
7157 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
7158 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
7159 (_mm_maskz_min_ss): New intrinsics.
7161 2017-05-31 Martin Liska <mliska@suse.cz>
7163 * tree-vect-loop.c (vect_create_epilog_for_reduction):
7164 Change comment style to one we normally use.
7165 (vectorizable_reduction): Likewise.
7166 (vectorizable_induction): Likewise.
7167 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7168 (vectorizable_call): Likewise.
7169 (vectorizable_simd_clone_call): Likewise.
7170 (vectorizable_conversion): Likewise.
7171 (vectorizable_assignment): Likewise.
7172 (vectorizable_shift): Likewise.
7173 (vectorizable_operation): Likewise.
7174 (vectorizable_store): Likewise.
7175 (vectorizable_load): Likewise.
7176 * tree-vectorizer.h: Likewise.
7178 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
7180 * passes.c (emergency_dump_function): New.
7181 * tree-pass.h (emergency_dump_function): Declare.
7182 * plugin.c (plugins_internal_error_function): Remove.
7183 * plugin.h (plugins_internal_error_function): Remove declaration.
7184 * toplev.c (internal_error_function): New static function. Use it...
7185 (general_init): ...here.
7187 2017-05-31 Graham Markall <graham.markall@embecosm.com>
7189 * config/arc/arc.c (arc_print_operand): Handle constant operands.
7190 (arc_rtx_costs): Add costs for new patterns.
7191 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
7192 * config/arc/predicates.md: Add _1_2_3_operand predicate.
7194 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
7196 * tree-ssa-strlen.c (get_next_strinfo): New function.
7197 (get_stridx_plus_constant): Use it.
7198 (zero_length_string): Likewise.
7199 (adjust_related_strinfos): Likewise.
7200 (adjust_last_stmt): Likewise.
7202 2017-05-31 Richard Biener <rguenther@suse.de>
7205 * config/i386/i386.c (ix86_expand_builtin): Remove assert
7206 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
7208 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
7210 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
7211 loop_vinfo argument and use of dependence distance vectors.
7212 Check instead whether the two references differ only in their
7213 initial value and assume that they have the same alignment if the
7214 difference is a multiple of the vector alignment.
7215 (vect_analyze_data_refs_alignment): Update call accordingly.
7217 2017-05-31 Martin Liska <mliska@suse.cz>
7220 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
7222 2017-05-31 Bin Cheng <bin.cheng@arm.com>
7224 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
7225 (create_intersect_range_checks): Move from ...
7226 * tree-data-ref.c (create_intersect_range_checks_index)
7227 (create_intersect_range_checks): ... to here.
7228 (create_runtime_alias_checks): New function factored from ...
7229 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
7230 here. Call above function.
7231 * tree-data-ref.h (create_runtime_alias_checks): New function.
7233 2017-05-31 Bin Cheng <bin.cheng@arm.com>
7235 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
7236 segment length for dr_b and compute it in wide_int.
7238 2017-05-31 Richard Biener <rguenther@suse.de>
7240 PR tree-optimization/80906
7241 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
7242 and pass through iv_map.
7243 (copy_bb_and_scalar_dependences): Adjust.
7244 (translate_pending_phi_nodes): Likewise.
7245 (copy_loop_close_phi_args): Handle code-generating IVs instead
7248 2017-05-30 David Malcolm <dmalcolm@redhat.com>
7250 * diagnostic-color.c (color_dict): Add "type-diff".
7251 (parse_gcc_colors): Update comment.
7252 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
7253 -fdiagnostics-show-template-tree and -fno-elide-type.
7254 (GCC_COLORS): Add type-diff to example.
7256 (-fdiagnostics-show-template-tree): New.
7257 (-fno-elide-type): New.
7258 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
7259 the pp_format_decoder callback. Call any m_format_postprocessor's
7261 (pretty_printer::pretty_printer): Initialize
7262 m_format_postprocessor.
7263 (pretty_printer::~pretty_printer): Delete any
7264 m_format_postprocessor.
7265 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
7266 (class format_postprocessor): New class.
7267 (struct pretty_printer::format_decoder): Document the new parameters.
7268 (struct pretty_printer::m_format_postprocessor): New field.
7269 * tree-diagnostic.c (default_tree_printer): Update for new
7270 bool and const char ** params.
7271 * tree-diagnostic.h (default_tree_printer): Likewise.
7273 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
7275 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
7276 (lwa_operand): Delete rs6000_gen_cell_microcode test.
7277 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
7278 rs6000_gen_cell_microcode code.
7279 (rs6000_final_prescan_insn): Delete.
7280 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
7281 "warn-cell-microcode" entries.
7282 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
7283 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
7284 throughout. Change cc_reg_not_micro_cr0_operand to
7285 cc_reg_not_cr0_operand throughout.
7286 (*extendhi<mode>2_noload): Delete.
7287 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
7288 (mwarn-cell-microcode): Delete.
7289 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
7290 -mgen-cell-microcode and -mwarn-cell-microcode.
7292 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
7295 * config/i386/constraints.md (Yd): New constraint.
7297 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
7298 and (?Yd, r) alternatives. Update insn attributes.
7299 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
7300 and (?*Yd, r) alternatives. Update insn attributes.
7301 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
7303 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
7305 * gimplify.c (gimplify_modify_expr): Don't create a
7306 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
7309 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
7311 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
7313 2017-05-30 Richard Biener <rguenther@suse.de>
7315 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
7316 and reduc_def fields.
7317 (STMT_VINFO_REDUC_TYPE): New define.
7318 (STMT_VINFO_REDUC_DEF): Likewise.
7319 (vect_force_simple_reduction): Adjust prototype.
7320 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
7321 (vect_is_simple_reduction): Remove check_reduction argument.
7322 (vect_force_simple_reduction): Adjust and set
7323 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
7324 (vectorizable_reduction): Do not re-do reduction analysis
7325 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
7326 * tree-parloops.c (gather_scalar_reductions): Adjust.
7328 2017-05-30 Richard Biener <rguenther@suse.de>
7331 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
7334 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
7336 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
7337 Introduce unknown_misalignment parameter and remove vf.
7338 (vect_peeling_hash_get_lowest_cost):
7339 Pass unknown_misalignment parameter.
7340 (vect_enhance_data_refs_alignment):
7341 Fix unsupportable data ref treatment.
7343 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7345 * tree-vect-data-refs.c (vect_get_data_access_cost):
7346 Workaround for SLP handling.
7347 (vect_enhance_data_refs_alignment):
7348 Compute costs for doing no peeling at all, compare to the best
7349 peeling costs so far and avoid peeling if cheaper.
7351 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7353 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
7354 Return peeling info and set costs to zero for unlimited cost
7356 (vect_enhance_data_refs_alignment): Also inspect all datarefs
7357 with unknown misalignment. Compute and costs for unknown
7358 misalignment, compare them to the costs for known misalignment
7359 and choose the cheapest for peeling.
7361 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7363 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
7364 (vect_get_peeling_costs_all_drs): Create function.
7365 (vect_peeling_hash_get_lowest_cost):
7366 Use vect_get_peeling_costs_all_drs.
7367 (vect_peeling_supportable): Create function.
7368 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
7370 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7372 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
7373 DR_HAS_NEGATIVE_STEP.
7374 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
7375 (vect_enhance_data_refs_alignment): Use.
7376 (vect_duplicate_ssa_name_ptr_info): Use.
7377 * tree-vectorizer.h (dr_misalignment): Use.
7378 (known_alignment_for_access_p): Use.
7380 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7383 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
7385 (has_section_name): New function.
7387 2017-05-30 Martin Liska <mliska@suse.cz>
7390 * auto-profile.c (get_function_decl_from_block): Fix
7393 2017-05-30 Richard Biener <rguenther@suse.de>
7396 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
7398 2017-05-30 Martin Liska <mliska@suse.cz>
7400 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
7401 * dumpfile.h (struct dump_file_info): Remove ctors.
7403 2017-05-30 Martin Liska <mliska@suse.cz>
7405 * predict.def: Fix GNU coding style.
7407 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
7409 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
7410 Mark 'to' argument with ATTRIBUTE_UNUSED.
7412 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
7414 * config/xtensa/xtensa.c (xtensa_emit_call): Use
7415 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
7416 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
7419 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
7421 * doc/install.texi (Options specification): Restore entry of
7422 --enable-sjlj-exceptions.
7424 2017-05-27 Michael Eager <eager@eagercon.com>
7427 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
7429 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
7431 * config/microblaze/microblaze.h
7432 (FIXED_REGISTERS): Update in macro.
7433 (CALL_USED_REGISTERS): Update in macro.
7435 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
7437 * doc/install.texi: Add links to macOS binary distributions.
7439 2017-05-27 Jakub Jelinek <jakub@redhat.com>
7443 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7445 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
7447 2017-05-26 Martin Liska <mliska@suse.cz>
7449 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
7451 2017-05-26 Martin Liska <mliska@suse.cz>
7453 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
7454 always leading ';; '.
7455 (dump_bb_info): Likewise.
7456 (brief_dump_cfg): Likewise.
7457 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
7458 * dumpfile.c: Remove usage of TDF_VERBOSE.
7459 * dumpfile.h (enum dump_kind): Likewise.
7460 (dump_gimple_bb_header): Do not use TDF_COMMENT.
7461 * print-tree.c (debug_verbose): Remove.
7462 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
7463 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
7464 * tree-diagnostic.c (default_tree_printer): Replace
7465 TDF_DIAGNOSTIC with TDF_SLIM.
7467 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7469 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
7470 in parameter loop, rather than loop_vinfo.
7471 (create_intersect_range_checks): Ditto.
7472 (vect_create_cond_for_alias_checks): Update call to above functions.
7474 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7476 PR tree-optimization/80815
7477 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
7478 for merging runtime alias checks. Handle negative DR_STEPs.
7480 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7482 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
7484 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
7485 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
7486 out code pruning runtime alias checks.
7487 * tree-data-ref.c (prune_runtime_alias_test_list): New function
7488 factored out from above.
7489 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
7491 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
7493 (prune_runtime_alias_test_list): New decalaration.
7495 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7497 * tree-vect-data-refs.c (compare_tree): Rename and move ...
7498 * tree-data-ref.c (data_ref_compare_tree): ... to here.
7499 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
7500 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
7501 (operator==, comp_dr_with_seg_len_pair): Ditto.
7502 (vect_prune_runtime_alias_test_list): Ditto.
7504 2017-05-26 Martin Liska <mliska@suse.cz>
7507 * params.def: Bound partial-inlining-entry-probability param.
7509 2017-05-26 Marek Polacek <polacek@redhat.com>
7512 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
7515 2017-05-26 Richard Biener <rguenther@suse.de>
7517 PR tree-optimization/80842
7518 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
7521 2017-05-26 Richard Biener <rguenther@suse.de>
7523 PR tree-optimization/80844
7524 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
7526 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
7528 * doc/md.texi (Machine Constraints): Update x86 family
7529 machine constraints section to match 'config/i386/constraints.md'.
7531 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
7533 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
7535 2017-05-25 Nathan Sidwell <nathan@acm.org>
7537 * doc/invoke.texi (--enable-languages): Update documentation.
7539 2017-05-25 Martin Liska <mliska@suse.cz>
7541 * dumpfile.c: Add TDF_FOLDING.
7542 * dumpfile.h (enum dump_kind): Likewise.
7543 * genmatch.c (dt_simplify::gen_1): Use it.
7545 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7547 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
7549 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7551 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
7552 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
7554 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7556 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
7557 * match.pd (X == C): Rewrite it here.
7558 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
7559 with_certain_nonzero_bits2): New predicates.
7560 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
7562 2017-05-24 Nathan Sidwell <nathan@acm.org>
7564 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
7567 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
7570 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
7572 * config/powerpcspe: New port. Files are copied from the rs6000
7573 port, with "rs6000" in filenames replaced by "powerpcspe".
7575 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
7577 PR rtl-optimization/80754
7578 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
7580 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
7582 * config/sparc/sparc.md (length): Return the correct value for -mflat
7583 sibcalls to match output_sibcall.
7585 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
7589 * config/rs6000/rs6000.c (struct machine_function): Add new field
7591 (rs6000_get_separate_components): Init that field, use it.
7592 (rs6000_components_for_bb): Use the field.
7594 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
7596 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
7598 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
7601 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
7603 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7606 * config/s390/s390.c (s390_check_qrst_address): Check incoming
7607 address against address_operand predicate.
7608 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
7610 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
7612 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
7613 parameters passed indirectly.
7615 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
7617 * config/i386/i386.md (*movdi_internal): Remove SSE4
7618 alternative 18 (?r, *v). Update insn attributes.
7619 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
7620 Update insn attributes.
7621 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
7622 Update insn attributes.
7623 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
7624 alternative 1 (r, v). Remove isa attribute.
7625 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
7626 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
7627 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
7629 2017-05-23 Tom de Vries <tom@codesourcery.com>
7631 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
7634 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7636 * cgraphunit.c (symbol_table::process_new_functions): Update.
7637 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
7638 (inline_generate_summary): Rename to ...
7639 (ipa_fn_summary_generate): ... this one.
7640 (inline_read_summary): Rename to ...
7641 (ipa_fn_summary_read): ... this one.
7642 (inline_write_summary): Rename to ...
7643 (ipa_fn_summary_write): ... this one.
7644 (inline_free_summary): Rename to ...
7645 (ipa_free_fn_summary): ... this one.
7646 (pass_data_local_fn_summary, pass_local_fn_summary,
7647 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
7648 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
7649 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
7650 make_pass_ipa_fn_summary): New.
7651 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
7652 inline_write_summary, inline_free_summary): Remove.
7653 (ipa_free_fn_summary) : New.
7654 * ipa-inline.c (ipa_inline): Update.
7655 (pass_ipa_inline): Do not generate summaries.
7656 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
7658 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
7659 and add pass_ipa_fn_summary.
7660 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
7662 (make_pass_inline_parameters): Remove.
7664 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
7666 * omp-low.c (struct omp_context): Remove "default_kind" member.
7669 * omp-offload.c (execute_oacc_device_lower): Remove the
7670 parallelism dimensions function attributes for unparallelized
7671 OpenACC kernels constructs.
7673 2017-05-23 Martin Liska <mliska@suse.cz>
7675 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
7677 (cgraph_edge::make_speculative): Likewise.
7678 (cgraph_edge::resolve_speculation): Likewise.
7679 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
7680 (cgraph_node::dump): Likewise.
7681 * cgraph.h: Likewise.
7682 * cgraphunit.c (analyze_functions): Likewise.
7683 (symbol_table::compile): Likewise.
7684 * ipa-cp.c (print_all_lattices): Likewise.
7685 (determine_versionability): Likewise.
7686 (initialize_node_lattices): Likewise.
7687 (ipcp_verify_propagated_values): Likewise.
7688 (estimate_local_effects): Likewise.
7689 (update_profiling_info): Likewise.
7690 (create_specialized_node): Likewise.
7691 (perhaps_add_new_callers): Likewise.
7692 (decide_about_value): Likewise.
7693 (decide_whether_version_node): Likewise.
7694 (identify_dead_nodes): Likewise.
7695 (ipcp_store_bits_results): Likewise.
7696 * ipa-devirt.c (dump_targets): Likewise.
7697 (ipa_devirt): Likewise.
7698 * ipa-icf.c (sem_item::dump): Likewise.
7699 (sem_function::equals): Likewise.
7700 (sem_variable::equals): Likewise.
7701 (sem_item_optimizer::read_section): Likewise.
7702 (sem_item_optimizer::execute): Likewise.
7703 (congruence_class::dump): Likewise.
7704 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
7705 (dump_inline_summary): Likewise.
7706 (estimate_node_size_and_time): Likewise.
7707 (inline_analyze_function): Likewise.
7708 * ipa-inline-transform.c (inline_call): Likewise.
7709 * ipa-inline.c (report_inline_failed_reason): Likewise.
7710 (want_early_inline_function_p): Likewise.
7711 (edge_badness): Likewise.
7712 (update_edge_key): Likewise.
7713 (inline_small_functions): Likewise.
7714 * ipa-profile.c (ipa_profile): Likewise.
7715 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
7716 (ipa_make_edge_direct_to_target): Likewise.
7717 (remove_described_reference): Likewise.
7718 (ipa_impossible_devirt_target): Likewise.
7719 (propagate_controlled_uses): Likewise.
7720 (ipa_print_node_params): Likewise.
7721 (ipcp_transform_function): Likewise.
7722 * ipa-pure-const.c (pure_const_read_summary): Likewise.
7723 (propagate_pure_const): Likewise.
7724 * ipa-reference.c (generate_summary): Likewise.
7725 (read_write_all_from_decl): Likewise.
7726 (propagate): Likewise.
7727 (ipa_reference_read_optimization_summary): Likewise.
7728 * ipa-utils.c (ipa_merge_profiles): Likewise.
7729 * ipa.c (walk_polymorphic_call_targets): Likewise.
7730 (symbol_table::remove_unreachable_nodes): Likewise.
7731 (ipa_single_use): Likewise.
7732 * passes.c (execute_todo): Likewise.
7733 * predict.c (drop_profile): Likewise.
7734 * symtab.c (symtab_node::get_dump_name): New function.
7735 (symtab_node::dump_name): Likewise.
7736 (symtab_node::dump_asm_name): Likewise.
7737 (symtab_node::dump_references): Likewise.
7738 (symtab_node::dump_referring): Likewise.
7739 (symtab_node::dump_base): Likewise.
7740 (symtab_node::debug_symtab): Likewise.
7741 * tree-sra.c (convert_callers_for_node): Likewise.
7742 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
7743 * value-prof.c (init_node_map): Likewise.
7745 2017-05-23 Martin Liska <mliska@suse.cz>
7747 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
7748 and symtab_node::debug_symtab to symbol_table::debug.
7749 * cgraphunit.c (analyze_functions): Use the renamed function.
7750 (symbol_table::compile): Likewise.
7751 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
7752 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
7753 * passes.c (execute_todo): Likewise.
7754 * symtab.c (symbol_table::dump): New function.
7755 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
7757 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7759 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
7760 that nonconst implies exec.
7762 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7764 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7765 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7766 (inline_edge_summary_vec): Turn into ...
7767 (ipa_call_summaries): ... this one.
7768 (redirect_to_unreachable, edge_set_predicate,
7769 evaluate_properties_for_edge, inline_summary_alloc,
7770 reset_ipa_call_summary, reset_inline_summary,
7771 inline_summary_t::duplicate): Update.
7772 (inline_edge_duplication_hook): Turn to ...
7773 (ipa_call_summary_t::duplicate): ... this one.
7774 (inline_edge_removal_hook): Turn to ...
7775 (ipa_call_summary_t::remove): ... this one.
7776 (dump_inline_edge_summary): Turn to ...
7777 (dump_ipa_call_summary): ... this one.
7778 (estimate_function_body_sizes): Update.
7779 (inline_update_callee_summaries): Update.
7780 (remap_edge_change_prob): Update.
7781 (remap_edge_summaries): Update.
7782 (inline_merge_summary): Update.
7783 (do_estimate_edge_time): Update.
7784 (inline_generate_summary): Update.
7785 (inline_read_section): Update.
7786 (inline_read_summary): Update.
7787 (inline_free_summary): Update.
7788 * ipa-inline.c (can_inline_edge_p): Update.
7789 (compute_inlined_call_time): Update.
7790 (want_inline_small_function_p): Update.
7791 (edge_badness): Update.
7792 (early_inliner): Update.
7793 * ipa-inline.h (inline_edge_summary): Turn to ...
7794 (ipa_call_summary): ... this one.
7795 (ipa_call_summary_t): New class.
7796 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
7797 (ipa_call_summaries): New.
7798 (inline_edge_summary): Remove.
7799 (estimate_edge_growth): Update.
7800 * ipa-profile.c (ipa_propagate_frequency_1): Update.
7801 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7802 * ipa-split.c (execute_split_functions): Update.
7803 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
7805 2017-05-23 Tom de Vries <tom@codesourcery.com>
7807 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7808 attributes): Document rdrand effective target.
7810 2017-05-23 Tom de Vries <tom@codesourcery.com>
7812 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7813 attributes): Sort alphabetically.
7815 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
7817 * config/avr/genmultilib.awk: Use gsub instead of gensub.
7819 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
7822 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
7823 V2DF/V2DI splat into two separate patterns, one that handles
7824 registers, and the other that only handles memory. Drop support
7825 for splatting from a GPR on ISA 2.07 and then splitting the
7826 splat into direct move and splat.
7827 (vsx_splat_<mode>_reg): Likewise.
7828 (vsx_splat_<mode>_mem): Likewise.
7830 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
7832 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
7834 2017-05-22 Jakub Jelinek <jakub@redhat.com>
7837 * omp-low.c (finish_taskreg_remap): New function.
7838 (finish_taskreg_scan): If unit size of ctx->record_type
7839 is non-constant, unshare the size expression and replace
7840 decls in it with possible outer var refs.
7843 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
7844 GOVD_SHARED rather than GOVD_PRIVATE with it.
7845 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
7846 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
7849 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
7850 as last argument to build_outer_var_ref for pointer bases of array
7853 2017-05-19 Martin Sebor <msebor@redhat.com>
7855 * print-tree.c (print_node): Print DECL_READ_P flag.
7857 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7859 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
7860 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
7861 * cgraph.c: Likewise.
7862 * cgraphunit.c: Likewise.
7863 * gengtype.c: Likewise.
7864 * ipa-cp.c: Likewise.
7865 * ipa-devirt.c: Likewise.
7866 * ipa-icf.c: Likewise.
7867 * ipa-predicate.c: Likewise.
7868 * ipa-profile.c: Likewise.
7869 * ipa-prop.c: Likewise.
7870 * ipa-split.c: Likewise.
7872 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
7873 edge_predicate_pool, dump_inline_hints,
7874 inline_summary::account_size_time, redirect_to_unreachable,
7875 edge_set_predicate, set_hint_predicate,
7876 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
7877 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
7878 inline_summary_t::remove, remap_hint_predicate_after_duplication,
7879 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
7880 ipa_call_summary_t::remove, initialize_growth_caches,
7881 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
7882 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
7883 mark_modified, unmodified_parm_1, unmodified_parm,
7884 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
7885 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
7886 compute_bb_predicates, will_be_nonconstant_expr_predicate,
7887 will_be_nonconstant_predicate, record_modified_bb_info,
7888 get_minimal_bb, record_modified, param_change_prob,
7889 phi_result_unknown_predicate, predicate_for_phi_result,
7890 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
7891 estimate_function_body_sizes, compute_inline_parameters,
7892 compute_inline_parameters_for_curren, pass_data_inline_parameters,
7893 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
7894 inline_update_callee_summaries, remap_edge_change_prob,
7895 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
7896 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
7897 inline_analyze_function, inline_summary_t::insert,
7898 inline_generate_summary, read_ipa_call_summary, inline_read_section,
7899 inline_read_summary, write_ipa_call_summary, inline_write_summary,
7900 inline_free_summary): Move to ipa-fnsummary.h
7901 (predicate_t): Remove.
7902 * ipa-fnsummary.c: New file.
7903 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
7904 (enum inline_hints_vals, inline_hints, agg_position_info,
7905 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
7906 inline_summaries, ipa_call_summary, ipa_call_summary_t,
7907 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
7908 dump_inline_summary, dump_inline_hints, inline_generate_summary,
7909 inline_read_summary, inline_write_summary, inline_free_summary,
7910 inline_analyze_function, initialize_inline_failed,
7911 inline_merge_summary, inline_update_overall_summary,
7912 compute_inline_parameters): Move to ipa-fnsummary.h
7913 * ipa-fnsummary.h: New file.
7914 * ipa-inline-transform.h: Include ipa-inline.h.
7915 * ipa-inline.c: LIkewise.
7917 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7919 * ipa-inline.c (edge_badness): Use inlined_time instead of
7920 inline_summaries->get.
7922 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7924 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
7926 2017-05-22 Nathan Sidwell <nathan@acm.org>
7928 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
7929 (fdump-lang): Document 'raw' option.
7930 * dumpfile.h (TDI_tu): Delete.
7931 * dumpfile.c (dump_files): Remove translation-unit.
7932 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
7934 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
7936 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
7937 command option from $(AWK) call.
7938 * config/avr/genmultilib.awk: Simplify and rewrite so that it
7939 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
7940 [FORMAT]: Remove handling of variable.
7941 * config/avr/t-multilib: Regenerate.
7943 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7945 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
7947 (dump_inline_summary): Do not print self_time.
7948 (estimate_function_body_sizes): Do not set self_time.
7949 (compute_inline_parameters): Likewise.
7950 (inline_read_section, inline_write_summary): Do not stream self_time.
7951 * ipa-inline.h (inline_summary): Drop self_time.
7953 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7955 * ipa-inline-analysis.c (account_size_time): Rename to ...
7956 (inline_summary::account_size_time): ... this one.
7957 (reset_ipa_call_summary): Turn to ...
7958 (ipa_call_summary::reset): ... this one.
7959 (reset_inline_summary): Turn to ...
7960 (inline_summary::reset): ... this one.
7961 (inline_summary_t::remove): Update.
7962 (inline_summary_t::duplicate): Update.
7963 (ipa_call_summary_t::remove): Update.
7964 (dump_inline_summary): Update.
7965 (estimate_function_body_sizes): Update.
7966 (compute_inline_parameters): Update.
7967 (estimate_node_size_and_time): Update.
7968 (inline_merge_summary): Update.
7969 (inline_update_overall_summary): Update.
7970 (inline_read_section): Update.
7971 (inline_write_summary): Update.
7972 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
7973 add account_size_time and reset member functions.
7974 (ipa_call_summary): Add reset function.
7975 * ipa-predicate.h (predicate::operator &): Constify.
7977 2017-05-22 Richard Biener <rguenther@suse.de>
7979 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
7981 2017-05-19 Jason Merrill <jason@redhat.com>
7983 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
7985 2017-05-19 Marek Polacek <polacek@redhat.com>
7988 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
7989 TYPE_OVERFLOW_WRAPS checks.
7991 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
7993 * tree-core.h (enum omp_clause_default_kind): Add
7994 "OMP_CLAUSE_DEFAULT_PRESENT".
7995 * tree-pretty-print.c (dump_omp_clause): Handle it.
7996 * gimplify.c (enum gimplify_omp_var_data): Add
7997 "GOVD_MAP_FORCE_PRESENT".
7998 (gimplify_adjust_omp_clauses_1): Map it to
7999 "GOMP_MAP_FORCE_PRESENT".
8000 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
8002 * gimplify.c (oacc_default_clause): Clarify.
8004 2017-05-19 Nathan Sidwell <nathan@acm.org>
8006 LANG_HOOK_REGISTER_DUMPS
8007 * toplev.c (general_init): Call register dump lang hook.
8008 * doc/invoke.texi: Document -fdump-lang option family.
8009 * dumpfile.c (dump_files): Remove class dump here.
8010 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
8011 * dumpfile.h (tree_dump_index): Remove TDI_class.
8012 * langhooks-def.h (lhd_register_dumps): Declare.
8013 (LANG_HOOKS_REGISTER_DUMPS): Define.
8014 (LANG_HOOKS_INITIALIZER): Add it.
8015 * langhooks.c (lhd_register_dumps): Define.
8016 * langhooks.h (struct lang_hooks): Add register_dumps.
8018 2017-05-19 Nathan Sidwell <nathan@acm.org>
8020 * context.h (context::set_passes): New.
8021 * context.c (context::context): Do not create pass manager.
8022 * toplev.c (general_init): Create pass manager here.
8024 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
8026 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
8027 use this splitter if two add or or instructions would also work for
8028 the constant we want to generate.
8030 2017-05-19 Richard Biener <rguenther@suse.de>
8033 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
8034 predicate evaluation.
8036 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
8038 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
8040 * ipa-inline.c (want_inline_small_function_p): Do not cast to
8043 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
8045 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
8046 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
8047 (inline_edge_summary_vec): Turn into ...
8048 (ipa_call_summaries): ... this one.
8049 (redirect_to_unreachable, edge_set_predicate,
8050 evaluate_properties_for_edge, inline_summary_alloc,
8051 reset_ipa_call_summary, reset_inline_summary,
8052 inline_summary_t::duplicate): Update.
8053 (inline_edge_duplication_hook): Turn to ...
8054 (ipa_call_summary_t::duplicate): ... this one.
8055 (inline_edge_removal_hook): Turn to ...
8056 (ipa_call_summary_t::remove): ... this one.
8057 (dump_inline_edge_summary): Turn to ...
8058 (dump_ipa_call_summary): ... this one.
8059 (estimate_function_body_sizes): Update.
8060 (inline_update_callee_summaries): Update.
8061 (remap_edge_change_prob): Update.
8062 (remap_edge_summaries): Update.
8063 (inline_merge_summary): Update.
8064 (do_estimate_edge_time): Update.
8065 (inline_generate_summary): Update.
8066 (inline_read_section): Update.
8067 (inline_read_summary): Update.
8068 (inline_free_summary): Update.
8069 * ipa-inline.c (can_inline_edge_p): Update.
8070 (compute_inlined_call_time): Update.
8071 (want_inline_small_function_p): Update.
8072 (edge_badness): Update.
8073 (early_inliner): Update.
8074 * ipa-inline.h (inline_edge_summary): Turn to ...
8075 (ipa_call_summary): ... this one.
8076 (ipa_call_summary_t): New class.
8077 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
8078 (ipa_call_summaries): New.
8079 (inline_edge_summary): Remove.
8080 (estimate_edge_growth): Update.
8081 * ipa-profile.c (ipa_propagate_frequency_1): Update.
8082 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
8083 * ipa-split.c (execute_split_functions): Update.
8084 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
8086 2017-05-19 Richard Biener <rguenther@suse.de>
8089 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
8091 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
8093 * config/rs6000/rs6000.c (struct machine_function): Add field
8094 fpr_is_wrapped_separately.
8095 (rs6000_get_separate_components): Use 64 components. Handle the
8097 (rs6000_components_for_bb): Handle the FPR components.
8098 (rs6000_emit_prologue_components): Handle the FPR components.
8099 (rs6000_emit_epilogue_components): Handle the FPR components.
8100 (rs6000_set_handled_components): Handle the FPR components.
8101 (rs6000_emit_prologue): Don't output prologue code for those FPRs
8102 that are already separately shrink-wrapped.
8103 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
8104 that are already separately shrink-wrapped.
8106 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
8109 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
8112 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
8113 (define_peephole2 for Altivec d-form load): Add peepholes to catch
8114 cases where the register allocator uses a move and an offsettable
8115 memory operation to/from a FPR register on ISA 2.06/2.07.
8116 (define_peephole2 for Altivec d-form store): Likewise.
8118 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
8121 * config/i386/mmx.md (*mov<mode>_internal): Enable
8122 alternatives 11, 12, 13 and 14 also for 32bit targets.
8123 Remove alternatives 15, 16, 17 and 18.
8124 * config/i386/sse.md (vec_concatv2di): Change
8125 alternative (!x, *y) to (x, ?!*Yn).
8127 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
8129 * dumpfile.h (enum dump_kind): Remove stray comma.
8131 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
8133 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
8134 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
8135 predicate::num_conditions
8136 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
8137 (CHANGED): turn into predicate::changed.
8138 (agg_position_info): Move to ipa-predicate.h
8139 (add_condition, predicate::add_clause, predicate::operator &=,
8140 predicate::or_with, predicate::evaluate, predicate::probability,
8141 dump_condition, dump_clause, predicate::dump,
8142 predicate::remap_after_duplication, predicate::remap_after_inlining,
8143 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
8144 (evaluate_conditions_for_known_args): Update.
8145 (set_cond_stmt_execution_predicate): Update.
8146 * ipa-inline.h: Include ipa-predicate.h
8147 (condition, inline_param_summary, conditions, agg_position_info,
8148 predicate): Move to ipa-predicate.h
8149 * ipa-predicate.c: New file.
8150 * ipa-predicate.h: New file.
8152 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
8154 * final.c (leaf_function_p): Check we are not in a sequence.
8156 2017-05-18 Martin Liska <mliska@suse.cz>
8158 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
8159 * dumpfile.c (dump_register): Use new enum dump_kind.
8160 (get_dump_file_name): Likewise.
8161 (dump_enable_all): Likewise.
8162 (dump_switch_p_1): Likewise.
8163 (enable_rtl_dump_file): Remove usage of TDF_RTL.
8164 * dumpfile.h (enum dump_kind): New enum type.
8165 (struct dump_file_info): Create constructor and
8166 format fields and comments.
8167 * passes.c (pass_manager::register_one_dump_file):
8169 * statistics.c (statistics_early_init): Likewise.
8170 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
8171 TDF_TREE with TDF_SLIM.
8172 (gather_memory_references_ref): Likewise.
8174 2017-05-18 Martin Liska <mliska@suse.cz>
8176 * vec.h (struct vnull): Use it.
8178 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
8180 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
8181 (true_predicate, false_predicate, true_predicate_p,
8182 false_predicate_p): Remove.
8183 (single_cond_predicate, not_inlined_predicate): Turn to member function
8185 (add_condition): Update.
8186 (add_clause): Turn to...
8187 (predicate::add_clause): ... this one; update; allow passing NULL
8189 (and_predicates): Turn to ...
8190 (predicate::operator &=): ... this one.
8191 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
8192 (or_predicates): Turn to ...
8193 (predicate::or_with): ... this one.
8194 (evaluate_predicate): Turn to ...
8195 (predicate::evaluate): ... this one.
8196 (predicate_probability): Turn to ...
8197 (predicate::probability): ... this one.
8198 (dump_condition): Update.
8199 (dump_predicate): Turn to ...
8200 (predicate::dump): ... this one.
8201 (account_size_time): Update.
8202 (edge_set_predicate): Update.
8203 (set_hint_predicate): UPdate.
8204 (evaluate_conditions_for_known_args): Update.
8205 (evaluate_properties_for_edge): Update.
8206 (remap_predicate_after_duplication): Turn to...
8207 (predicate::remap_after_duplication): ... this one.
8208 (remap_hint_predicate_after_duplication): Update.
8209 (inline_summary_t::duplicate): UPdate.
8210 (dump_inline_edge_summary): Update.
8211 (dump_inline_summary): Update.
8212 (set_cond_stmt_execution_predicate): Update.
8213 (set_switch_stmt_execution_predicate): Update.
8214 (compute_bb_predicates): Update.
8215 (will_be_nonconstant_expr_predicate): Update.
8216 (will_be_nonconstant_predicate): Update.
8217 (phi_result_unknown_predicate): Update.
8218 (predicate_for_phi_result): Update.
8219 (array_index_predicate): Update.
8220 (estimate_function_body_sizes): Update.
8221 (estimate_node_size_and_time): Update.
8222 (estimate_ipcp_clone_size_and_time): Update.
8223 (remap_predicate): Rename to ...
8224 (predicate::remap_after_inlining): ... this one.
8225 (remap_hint_predicate): Update.
8226 (inline_merge_summary): Update.
8227 (inline_update_overall_summary): Update.
8228 (estimate_size_after_inlining): Update.
8229 (read_predicate): Rename to ...
8230 (predicate::stream_in): ... this one.
8231 (read_inline_edge_summary): Update.
8232 (write_predicate): Rename to ...
8233 (predicate::stream_out): ... this one.
8234 (write_inline_edge_summary): Update.
8235 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
8236 (clause_t): Turn to uint32_t
8237 (predicate): Turn to class; implement constructor and operators
8239 (size_time_entry): Update.
8240 (inline_summary): Update.
8241 (inline_edge_summary): Update.
8243 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
8245 * fold-const.c (fold_binary_loc): Move transformation...
8246 * match.pd (C - X CMP X): ... here.
8248 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
8250 * config/sparc/sparc.c (sparc_option_override): Set function
8251 alignment for -mcpu=niagara7 to 64 to match the I$ line.
8252 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
8254 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
8256 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
8258 2017-05-18 Marek Polacek <polacek@redhat.com>
8261 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
8262 (pass_ubsan::execute): Call gimple_assign_single_p instead of
8263 gimple_assign_load_p.
8265 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
8268 * real.c (do_compare): Give decimal_do_compare preference over
8269 comparing just the signs.
8271 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
8273 * doc/md.texi (Canonicalization of Instructions): Describe the
8274 canonical form of instructions that inherently set a condition
8277 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
8280 * tree-cfg.c: Move deletion of unreachable case statements to after
8281 the merging of consecutive case labels.
8283 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
8285 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
8286 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
8287 restoring of callee-saved registers.
8289 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
8291 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
8292 * config/visium/visium.c (single_set_and_flags): Likewise.
8293 * config/visium/visium.md (Substitutions): Likewise.
8295 2017-05-17 Martin Liska <mliska@suse.cz>
8297 * cfg.c: Introduce dump_flags_t type and
8298 use it instead of int type.
8300 * cfghooks.c: Likewise.
8301 * cfghooks.h (struct cfg_hooks): Likewise.
8302 * cfgrtl.c: Likewise.
8303 * cfgrtl.h: Likewise.
8304 * cgraph.c (cgraph_node::get_body): Likewise.
8305 * coretypes.h: Likewise.
8306 * domwalk.c: Likewise.
8307 * domwalk.h: Likewise.
8308 * dumpfile.c (struct dump_option_value_info): Likewise.
8309 (dump_enable_all): Likewise.
8310 (dump_switch_p_1): Likewise.
8311 (opt_info_switch_p): Likewise.
8312 * dumpfile.h (enum tree_dump_index): Likewise.
8313 (struct dump_file_info): Likewise.
8314 * genemit.c: Likewise.
8315 * generic-match-head.c: Likewise.
8316 * gengtype.c (open_base_files): Likewise.
8317 * gimple-pretty-print.c: Likewise.
8318 * gimple-pretty-print.h: Likewise.
8319 * graph.c (print_graph_cfg): Likewise.
8320 * graphite-scop-detection.c (dot_all_sese): Likewise.
8321 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
8322 * loop-unroll.c (report_unroll): Likewise.
8323 * passes.c (pass_manager::register_one_dump_file): Likewise.
8324 * print-tree.c: Likewise.
8325 * statistics.c: Likewise.
8326 * tree-cfg.c: Likewise.
8327 * tree-cfg.h: Likewise.
8328 * tree-dfa.c: Likewise.
8329 * tree-dfa.h: Likewise.
8330 * tree-dump.c (dump_function): Likewise.
8331 * tree-dump.h (struct dump_info): Likewise.
8332 * tree-pretty-print.c: Likewise.
8333 * tree-pretty-print.h: Likewise.
8334 * tree-ssa-live.c: Likewise.
8335 * tree-ssa-live.h: Likewise.
8336 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
8337 * tree-vect-loop.c: Likewise.
8338 * tree-vect-slp.c: Likewise.
8340 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
8341 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8343 PR tree-optimization/80457
8344 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
8345 of all arguments to a statement as scalar_to_vec operations.
8346 (vectorizable_call): Adjust call to vect_model_simple_cost for
8348 (vectorizable_conversion): Likewise.
8349 (vectorizable_assignment): Likewise.
8350 (vectorizable_shift): Likewise.
8351 (vectorizable_operation): Likewise.
8352 (vectorizable_comparison): Likewise.
8353 (vect_is_simple_cond): Record the def types for operands.
8354 (vectorizable_condition): Likewise, call vect_model_simple_cost.
8355 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
8356 for statement argument count.
8358 2017-05-16 Carl Love <cel@us.ibm.com>
8360 * config/rs6000/rs6000-c: Add support for built-in functions
8361 vector unsigned long long vec_bperm (vector unsigned long long,
8362 vector unsigned char)
8363 vector signed long long vec_mule (vector signed int,
8365 vector unsigned long long vec_mule (vector unsigned int,
8366 vector unsigned int)
8367 vector signed long long vec_mulo (vector signed int,
8369 vector unsigned long long vec_mulo (vector unsigned int,
8370 vector unsigned int)
8371 vector signed char vec_sldw (vector signed char,
8374 vector unsigned char vec_sldw (vector unsigned char,
8375 vector unsigned char,
8377 vector signed short vec_sldw (vector signed short,
8378 vector signed short,
8380 vector unsigned short vec_sldw (vector unsigned short,
8381 vector unsigned short,
8383 vector signed int vec_sldw (vector signed int,
8386 vector unsigned int vec_sldw (vector unsigned int,
8387 vector unsigned int,
8389 vector signed long long vec_sldw (vector signed long long,
8390 vector signed long long,
8392 vector unsigned long long vec_sldw (vector unsigned long long,
8393 vector unsigned long long,
8395 * config/rs6000/rs6000-c: Add support for built-in functions
8396 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
8397 * config/rs6000/altivec.h: Add defintion for vec_sldw.
8398 * doc/extend.texi: Update the built-in documentation for the
8399 new built-in functions.
8401 2017-05-16 Marek Polacek <polacek@redhat.com>
8405 * tree.c (save_expr): Don't fold the expression.
8407 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
8409 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
8410 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
8411 and (?*y,m). Update insn attributes.
8413 2017-05-16 Martin Liska <mliska@suse.cz>
8415 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
8416 flags argument of print_gimple_stmt, print_gimple_expr,
8417 print_generic_stmt and print_generic_expr.
8418 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
8419 * coretypes.h: Likewise.
8420 * except.c (dump_eh_tree): Likewise.
8421 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8422 * gimple-pretty-print.h: Likewise.
8423 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
8424 (backprop::push_to_worklist): Likewise.
8425 (backprop::pop_from_worklist): Likewise.
8426 (backprop::process_use): Likewise.
8427 (backprop::intersect_uses): Likewise.
8428 (note_replacement): Likewise.
8429 * gimple-ssa-store-merging.c
8430 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
8431 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
8432 (pass_store_merging::execute): Likewise.
8433 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
8434 (ssa_base_cand_dump_callback): Likewise.
8435 (dump_incr_vec): Likewise.
8436 (replace_refs): Likewise.
8437 (replace_mult_candidate): Likewise.
8438 (create_add_on_incoming_edge): Likewise.
8439 (create_phi_basis): Likewise.
8440 (insert_initializers): Likewise.
8441 (all_phi_incrs_profitable): Likewise.
8442 (introduce_cast_before_cand): Likewise.
8443 (replace_one_candidate): Likewise.
8444 * gimplify.c (gimplify_expr): Likewise.
8445 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
8446 (set_rename): Likewise.
8447 (rename_uses): Likewise.
8448 (copy_loop_phi_nodes): Likewise.
8449 (add_close_phis_to_merge_points): Likewise.
8450 (copy_loop_close_phi_args): Likewise.
8451 (copy_cond_phi_args): Likewise.
8452 (graphite_copy_stmts_from_block): Likewise.
8453 (translate_pending_phi_nodes): Likewise.
8454 * graphite-poly.c (print_pdr): Likewise.
8455 (dump_gbb_cases): Likewise.
8456 (dump_gbb_conditions): Likewise.
8457 (print_scop_params): Likewise.
8458 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
8459 (build_cross_bb_scalars_use): Likewise.
8460 (gather_bbs::before_dom_children): Likewise.
8461 * hsa-dump.c (dump_hsa_immed): Likewise.
8462 * ipa-cp.c (print_ipcp_constant_value): Likewise.
8463 (get_replacement_map): Likewise.
8464 * ipa-inline-analysis.c (dump_condition): Likewise.
8465 (estimate_function_body_sizes): Likewise.
8466 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
8467 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
8468 * ipa-prop.c (ipa_dump_param): Likewise.
8469 (ipa_print_node_jump_functions_for_edge): Likewise.
8470 (ipa_modify_call_arguments): Likewise.
8471 (ipa_modify_expr): Likewise.
8472 (ipa_dump_param_adjustments): Likewise.
8473 (ipa_dump_agg_replacement_values): Likewise.
8474 (ipcp_modif_dom_walker::before_dom_children): Likewise.
8475 * ipa-pure-const.c (check_stmt): Likewise.
8476 (pass_nothrow::execute): Likewise.
8477 * ipa-split.c (execute_split_functions): Likewise.
8478 * omp-offload.c (dump_oacc_loop_part): Likewise.
8479 (dump_oacc_loop): Likewise.
8480 * trans-mem.c (tm_log_emit): Likewise.
8481 (tm_memopt_accumulate_memops): Likewise.
8482 (dump_tm_memopt_set): Likewise.
8483 (dump_tm_memopt_transform): Likewise.
8484 * tree-cfg.c (gimple_verify_flow_info): Likewise.
8485 (print_loop): Likewise.
8486 * tree-chkp-opt.c (chkp_print_addr): Likewise.
8487 (chkp_gather_checks_info): Likewise.
8488 (chkp_get_check_result): Likewise.
8489 (chkp_remove_check_if_pass): Likewise.
8490 (chkp_use_outer_bounds_if_possible): Likewise.
8491 (chkp_reduce_bounds_lifetime): Likewise.
8492 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
8493 (chkp_mark_completed_bounds): Likewise.
8494 (chkp_register_incomplete_bounds): Likewise.
8495 (chkp_mark_invalid_bounds): Likewise.
8496 (chkp_maybe_copy_and_register_bounds): Likewise.
8497 (chkp_build_returned_bound): Likewise.
8498 (chkp_get_bound_for_parm): Likewise.
8499 (chkp_build_bndldx): Likewise.
8500 (chkp_get_bounds_by_definition): Likewise.
8501 (chkp_generate_extern_var_bounds): Likewise.
8502 (chkp_get_bounds_for_decl_addr): Likewise.
8503 * tree-chrec.c (chrec_apply): Likewise.
8504 * tree-data-ref.c (dump_data_reference): Likewise.
8505 (dump_subscript): Likewise.
8506 (dump_data_dependence_relation): Likewise.
8507 (analyze_overlapping_iterations): Likewise.
8508 * tree-inline.c (expand_call_inline): Likewise.
8509 (tree_function_versioning): Likewise.
8510 * tree-into-ssa.c (dump_defs_stack): Likewise.
8511 (dump_currdefs): Likewise.
8512 (dump_names_replaced_by): Likewise.
8513 (dump_update_ssa): Likewise.
8514 (update_ssa): Likewise.
8515 * tree-object-size.c (pass_object_sizes::execute): Likewise.
8516 * tree-parloops.c (build_new_reduction): Likewise.
8517 (try_create_reduction_list): Likewise.
8518 (ref_conflicts_with_region): Likewise.
8519 (oacc_entry_exit_ok_1): Likewise.
8520 (oacc_entry_exit_single_gang): Likewise.
8521 * tree-pretty-print.h: Likewise.
8522 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
8523 (get_scalar_evolution): Likewise.
8524 (add_to_evolution): Likewise.
8525 (get_loop_exit_condition): Likewise.
8526 (analyze_evolution_in_loop): Likewise.
8527 (analyze_initial_condition): Likewise.
8528 (analyze_scalar_evolution): Likewise.
8529 (instantiate_scev): Likewise.
8530 (number_of_latch_executions): Likewise.
8531 (gather_chrec_stats): Likewise.
8532 (final_value_replacement_loop): Likewise.
8533 (scev_const_prop): Likewise.
8534 * tree-sra.c (dump_access): Likewise.
8535 (disqualify_candidate): Likewise.
8536 (create_access): Likewise.
8538 (maybe_add_sra_candidate): Likewise.
8539 (create_access_replacement): Likewise.
8540 (analyze_access_subtree): Likewise.
8541 (analyze_all_variable_accesses): Likewise.
8542 (sra_modify_assign): Likewise.
8543 (initialize_constant_pool_replacements): Likewise.
8544 (find_param_candidates): Likewise.
8545 (decide_one_param_reduction): Likewise.
8546 (replace_removed_params_ssa_names): Likewise.
8547 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
8548 * tree-ssa-copy.c (dump_copy_of): Likewise.
8549 (copy_prop_visit_cond_stmt): Likewise.
8550 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
8551 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
8552 (record_equivalences_from_stmt): Likewise.
8553 * tree-ssa-dse.c (compute_trims): Likewise.
8554 (delete_dead_call): Likewise.
8555 (delete_dead_assignment): Likewise.
8556 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
8557 (forward_propagate_into_cond): Likewise.
8558 (pass_forwprop::execute): Likewise.
8559 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8560 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
8562 (move_computations_worker): Likewise.
8563 (execute_sm): Likewise.
8564 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
8565 (remove_exits_and_undefined_stmts): Likewise.
8566 (remove_redundant_iv_tests): Likewise.
8567 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
8568 (adjust_iv_update_pos): Likewise.
8569 * tree-ssa-math-opts.c (bswap_replace): Likewise.
8570 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
8571 (value_replacement): Likewise.
8572 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
8573 * tree-ssa-pre.c (print_pre_expr): Likewise.
8574 (get_representative_for): Likewise.
8575 (create_expression_by_pieces): Likewise.
8576 (insert_into_preds_of_block): Likewise.
8577 (eliminate_insert): Likewise.
8578 (eliminate_dom_walker::before_dom_children): Likewise.
8579 (eliminate): Likewise.
8580 (remove_dead_inserted_code): Likewise.
8581 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
8582 * tree-ssa-reassoc.c (get_rank): Likewise.
8583 (eliminate_duplicate_pair): Likewise.
8584 (eliminate_plus_minus_pair): Likewise.
8585 (eliminate_not_pairs): Likewise.
8586 (undistribute_ops_list): Likewise.
8587 (eliminate_redundant_comparison): Likewise.
8588 (update_range_test): Likewise.
8589 (optimize_range_tests_var_bound): Likewise.
8590 (optimize_vec_cond_expr): Likewise.
8591 (rewrite_expr_tree): Likewise.
8592 (rewrite_expr_tree_parallel): Likewise.
8593 (linearize_expr): Likewise.
8594 (break_up_subtract): Likewise.
8595 (linearize_expr_tree): Likewise.
8596 (attempt_builtin_powi): Likewise.
8597 (attempt_builtin_copysign): Likewise.
8598 (transform_stmt_to_copy): Likewise.
8599 (transform_stmt_to_multiply): Likewise.
8600 (dump_ops_vector): Likewise.
8601 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
8602 (print_scc): Likewise.
8603 (set_ssa_val_to): Likewise.
8604 (visit_reference_op_store): Likewise.
8605 (visit_use): Likewise.
8606 (sccvn_dom_walker::before_dom_children): Likewise.
8607 (run_scc_vn): Likewise.
8608 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
8610 (expr_hash_elt::print): Likewise.
8611 (const_and_copies::pop_to_marker): Likewise.
8612 (const_and_copies::record_const_or_copy_raw): Likewise.
8613 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
8614 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
8615 (dump_predicates): Likewise.
8616 (find_uninit_use): Likewise.
8617 (warn_uninitialized_phi): Likewise.
8618 (pass_late_warn_uninitialized::execute): Likewise.
8619 * tree-ssa.c (verify_vssa): Likewise.
8620 (verify_ssa): Likewise.
8621 (maybe_optimize_var): Likewise.
8622 * tree-vrp.c (dump_value_range): Likewise.
8623 (dump_all_value_ranges): Likewise.
8624 (dump_asserts_for): Likewise.
8625 (register_edge_assert_for_2): Likewise.
8626 (vrp_visit_cond_stmt): Likewise.
8627 (vrp_visit_switch_stmt): Likewise.
8628 (vrp_visit_stmt): Likewise.
8629 (vrp_visit_phi_node): Likewise.
8630 (simplify_cond_using_ranges_1): Likewise.
8631 (fold_predicate_in): Likewise.
8632 (evrp_dom_walker::before_dom_children): Likewise.
8633 (evrp_dom_walker::push_value_range): Likewise.
8634 (evrp_dom_walker::pop_value_range): Likewise.
8635 (execute_early_vrp): Likewise.
8637 2017-05-16 Richard Biener <rguenther@suse.de>
8639 * dwarf2out.c (loc_list_from_tree_1): Do not create
8640 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
8642 2017-05-16 Richard Biener <rguenther@suse.de>
8644 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
8646 (note_variable_value_in_expr): If we resolved the decl ref
8647 do not push to the stack.
8649 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
8651 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
8652 operations in fast-math mode.
8653 (vaddq_f16): Likewise.
8654 (vmul_f16): Likewise.
8655 (vmulq_f16): Likewise.
8656 (vsub_f16): Likewise.
8657 (vsubq_f16): Likewise.
8658 * config/arm/neon.md (add<mode>3): New.
8660 (fma:<VH:mode>3): New. Also remove outdated comment.
8663 2017-05-16 Martin Liska <mliska@suse.cz>
8667 * ipa-devirt.c (warn_types_mismatch): Fix typo.
8668 (odr_types_equivalent_p): Likewise.
8670 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
8672 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
8674 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
8677 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
8678 non-interunit SSE move alternatives with '?'.
8679 (zero-extendsidi peephole2): New peephole to skip intermediate
8680 general register in SSE zero-extend sequence.
8682 2017-05-15 Jeff Law <law@redhat.com>
8684 * reorg.c (relax_delay_slots): Create a new variable to hold
8685 the temporary target rather than clobbering TARGET_LABEL.
8687 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
8688 missing argument to extract_bit_field call.
8689 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
8691 2017-05-15 Martin Liska <mliska@suse.cz>
8694 * gcc.c (process_command): Do not allow empty argument of -o option.
8696 2017-05-15 Renlin Li <renlin.li@arm.com>
8698 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
8699 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
8700 * config/aarch64/constraints.md (Usf): Add long call check.
8701 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
8702 (call_value): Likewise.
8703 (sibcall): Likewise.
8704 (sibcall_value): Likewise.
8706 (call_value_insn): New.
8707 (sibcall_insn): Update rtx pattern.
8708 (sibcall_value_insn): Likewise.
8709 (call_internal): Remove.
8710 (call_value_internal): Likewise.
8711 (sibcall_internal): Likewise.
8712 (sibcall_value_internal): Likewise.
8713 (call_reg): Likewise.
8714 (call_symbol): Likewise.
8715 (call_value_reg): Likewise.
8716 (call_value_symbol): Likewise.
8718 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
8721 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
8723 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
8725 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
8726 compatible with CCGOCmode and with CCZmode.
8728 2017-05-14 Martin Sebor <msebor@redhat.com>
8731 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
8732 (gimple_fold_builtin_snprintf): Same.
8733 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
8734 (gimple_fold_builtin_snprintf): Same.
8735 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
8737 (is_call_safe): New function.
8738 (try_substitute_return_value): Call it.
8739 (try_simplify_call): New function.
8740 (pass_sprintf_length::handle_gimple_call): Call it.
8742 2017-05-14 Martin Sebor <msebor@redhat.com>
8745 * builtins.c (expand_builtin_stpncpy): Simplify.
8747 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
8749 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
8750 * config/i386/i386.h
8751 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
8752 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
8753 (struct machine_function): Add new members call_ms2sysv,
8754 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
8755 (struct machine_frame_state): New fields sp_realigned and
8756 sp_realigned_offset.
8757 * config/i386/i386.c
8758 (enum xlogue_stub): New enum.
8759 (enum xlogue_stub_sets): New enum.
8760 (class xlogue_layout): New class.
8761 (struct ix86_frame): New fields stack_realign_allocate_offset,
8762 stack_realign_offset and outlined_save_offset. Modify comments to
8763 detail stack layout when using out-of-line stubs.
8764 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
8765 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
8766 -mcall-ms2sysv-xlogues.
8767 (stub_managed_regs): New static variable.
8768 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
8769 registers managed by out-of-line stub.
8770 (disable_call_ms2sysv_xlogues): New function.
8771 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
8772 m->call_ms2sysv when appropriate and compute frame layout for
8774 (sp_valid_at, fp_valid_at): New inline functions.
8775 (choose_basereg): New function.
8776 (choose_baseaddr): Add align parameter, use choose_basereg and modify
8778 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
8779 Use align parameter of choose_baseaddr to generated aligned SSE movs
8781 (pro_epilogue_adjust_stack): Modify to track
8782 machine_frame_state::sp_realigned.
8783 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
8784 (ix86_nsaved_sseregs): Likewise.
8785 (ix86_emit_save_regs): Likewise.
8786 (ix86_emit_save_regs_using_mov): Likewise.
8787 (ix86_emit_save_sse_regs_using_mov): Likewise.
8788 (get_scratch_register_on_entry): Likewise.
8789 (gen_frame_set): New function.
8790 (gen_frame_load): Likewise.
8791 (gen_frame_store): Likewise.
8792 (emit_outlined_ms2sysv_save): Likewise.
8793 (emit_outlined_ms2sysv_restore): Likewise.
8794 (ix86_expand_prologue): Modify stack re-alignment code and call
8795 emit_outlined_ms2sysv_save when appropriate.
8796 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
8797 parameter rtx_insn *insn, which allows the function to be used to only
8799 (ix86_expand_epilogue): Modify validity checks of frame and stack
8800 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
8801 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
8802 * config/i386/predicates.md
8803 (save_multiple): New predicate.
8804 (restore_multiple): Likewise.
8805 * config/i386/sse.md
8806 (save_multiple<mode>): New pattern.
8807 (save_multiple_realign<mode>): Likewise.
8808 (restore_multiple<mode>): Likewise.
8809 (restore_multiple_and_return<mode>): Likewise.
8810 (restore_multiple_leave_return<mode>): Likewise.
8811 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
8813 2017-05-14 Julia Koval <julia.koval@intel.com>
8815 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
8816 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
8817 (__builtin_ia32_xsetbv): New builtins.
8818 * config/i386/i386.c (ix86_expand_special_args_builtin):
8820 (ix86_expand_builtin): Special expand for new intrinsics.
8821 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
8822 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
8823 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
8825 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8827 * cfganal.c (inverted_post_order_compute): Change argument type
8829 * cfganal.h (inverted_post_order_compute): Adjust prototype.
8830 * df-core.c (rest_of_handle_df_initialize): Adjust.
8831 (rest_of_handle_df_finish): Likewise.
8832 (df_analyze_1): Likewise.
8833 (df_analyze): Likewise.
8834 (loop_inverted_post_order_compute): Change argument to be a vec *.
8835 (df_analyze_loop): Adjust.
8836 (df_get_n_blocks): Likewise.
8837 (df_get_postorder): Likewise.
8838 * df.h (struct df_d): Change field to be a vec.
8839 * lcm.c (compute_laterin): Adjust.
8840 (compute_available): Likewise.
8841 * lra-lives.c (lra_create_live_ranges_1): Likewise.
8842 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
8843 * tree-ssa-pre.c (compute_antic): Likewise.
8845 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8847 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
8848 (depth_first_search::depth_first_search): Change structure init
8849 function to this constructor.
8850 (depth_first_search::add_bb): Rename function to this member.
8851 (depth_first_search::execute): Likewise.
8852 (flow_dfs_compute_reverse_finish): Adjust.
8854 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8856 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
8857 (longest_simple_path): Likewise.
8858 * shrink-wrap.c (spread_components): Likewise.
8859 (disqualify_problematic_components): Likewise.
8860 (emit_common_heads_for_components): Likewise.
8861 (emit_common_tails_for_components): Likewise.
8862 (insert_prologue_epilogue_for_components): Likewise.
8864 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8866 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
8869 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8871 * df-core.c (df_set_blocks): Start using auto_bitmap.
8872 (df_compact_blocks): Likewise.
8873 * df-problems.c (df_rd_confluence_n): Likewise.
8874 * df-scan.c (df_insn_rescan_all): Likewise.
8875 (df_process_deferred_rescans): Likewise.
8876 (df_update_entry_block_defs): Likewise.
8877 (df_update_exit_block_uses): Likewise.
8878 (df_entry_block_bitmap_verify): Likewise.
8879 (df_exit_block_bitmap_verify): Likewise.
8880 (df_scan_verify): Likewise.
8881 * lra-constraints.c (lra_constraints): Likewise.
8882 (undo_optional_reloads): Likewise.
8883 (lra_undo_inheritance): Likewise.
8884 * lra-remat.c (calculate_gen_cands): Likewise.
8885 (do_remat): Likewise.
8886 * lra-spills.c (assign_spill_hard_regs): Likewise.
8887 (spill_pseudos): Likewise.
8888 * tree-ssa-pre.c (bitmap_set_and): Likewise.
8889 (bitmap_set_subtract_values): Likewise.
8891 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8893 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
8894 management with auto_bitmap.
8895 (fix_inter_tick): Likewise.
8896 (fix_recovery_deps): Likewise.
8897 * ira.c (add_store_equivs): Likewise.
8898 (find_moveable_pseudos): Likewise.
8899 (split_live_ranges_for_shrink_wrap): Likewise.
8900 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
8901 (rtx_reuse_manager::seen_def_p): Likewise.
8902 (rtx_reuse_manager::set_seen_def): Likewise.
8903 * print-rtl.h (class rtx_reuse_manager): Likewise.
8905 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8907 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
8909 (migrate_btr_def): Likewise.
8910 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
8911 * df-core.c (loop_post_order_compute): Likewise.
8912 (loop_inverted_post_order_compute): Likewise.
8913 * hsa-common.h: Likewise.
8914 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
8915 * init-regs.c (initialize_uninitialized_regs): Likewise.
8916 * ipa-inline.c (resolve_noninline_speculation): Likewise.
8917 (inline_small_functions): Likewise.
8918 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
8919 * ira.c (combine_and_move_insns): Likewise.
8920 (build_insn_chain): Likewise.
8921 * loop-invariant.c (find_invariants): Likewise.
8922 * lower-subreg.c (propagate_pseudo_copies): Likewise.
8923 * predict.c (tree_predict_by_opcode): Likewise.
8924 (predict_paths_leading_to): Likewise.
8925 (predict_paths_leading_to_edge): Likewise.
8926 (estimate_loops_at_level): Likewise.
8927 (estimate_loops): Likewise.
8928 * shrink-wrap.c (try_shrink_wrapping): Likewise.
8929 (spread_components): Likewise.
8930 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
8931 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
8932 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
8933 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
8934 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
8935 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
8936 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
8937 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
8938 (mark_threaded_blocks): Likewise.
8939 (thread_through_all_blocks): Likewise.
8940 * tree-ssa.c (verify_ssa): Likewise.
8941 (execute_update_addresses_taken): Likewise.
8942 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
8944 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8946 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
8948 (post_order_compute): Likewise.
8949 (inverted_post_order_compute): Likewise.
8950 (pre_and_rev_post_order_compute_fn): Likewise.
8952 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8954 * genrecog.c (int_set::int_set): Explicitly construct our
8955 auto_vec base class.
8956 * vec.h (auto_vec::auto_vec): New constructor.
8958 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8960 * bitmap.h (class auto_bitmap): New constructor taking
8961 bitmap_obstack * argument.
8963 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8965 * bitmap.h (class auto_bitmap): Change type of m_bits to
8966 bitmap_head, and adjust ctor / dtor and member operators.
8968 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
8970 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
8971 when returned register mode doesn't match original mode.
8973 2017-05-12 Jeff Law <law@redhat.com>
8974 Jakub Jelinek <jakub@redhat.com>
8976 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
8977 we look for cc setter after the compare-elim changes.
8978 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
8979 within the vector to match what compare-elim now expects.
8980 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
8981 (xorsi3_flags, one_cmplsi2_flags): Likewise.
8983 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
8984 after the compare-elim changes.
8985 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
8986 the vector to match what compare-elim now expects.
8987 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
8988 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
8989 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
8990 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
8991 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
8993 * config/visium/visium.c (single_set_and_flags): Fix where
8994 we look for cc setter after the compare-elim changes.
8995 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
8996 with the vector to match what compare-elim now expects.
8997 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
8998 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
8999 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
9000 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
9001 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
9002 (neg<mode>2_insn_set_overflow): Likewise.
9004 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
9007 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
9008 maybe_expand_insn call, set ops[0].target. If still set after call,
9009 set alt_rtl. Add extra arg to recursive calls.
9010 (extract_bit_field): Add alt_rtl argument. Pass to
9012 * expmed.h (extract_bit_field): Fix prototype.
9013 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
9014 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
9015 to extract_bit_field_calls.
9016 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
9017 Pass alt_rtl to extract_bit_field calls.
9018 * calls.c (store_unaligned_arguments_into_psuedos)
9019 load_register_parameters): Pass extra NULL to extract_bit_field calls.
9020 * optabs.c (maybe_legitimize_operand): Clear op->target when call
9022 * optabs.h (struct expand_operand): Add target bitfield.
9024 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
9026 * compare-elim.c (try_eliminate_compare): Canonicalize
9027 operation with embedded compare to
9028 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
9029 (set (reg) (operation)].
9031 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
9033 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
9036 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
9037 cost of adding a carry flag for ADC instruction.
9038 [case MINUS]: Ignore the cost of subtracting a carry flag
9039 for SBB instruction.
9041 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
9043 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
9045 * config/rs6000/bmiintrin.h: New file.
9046 * config/rs6000/bmi2intrin.h: New file.
9047 * config/rs6000/x86intrin.h: New file.
9049 2017-05-12 Jeff Law <law@redhat.com>
9051 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
9054 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
9057 * tree-cfg.c: Remove cfg edges of unreachable case statements.
9059 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9061 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
9062 early expansion of vector divide builtins.
9063 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
9064 builtins identified as having unsigned arguments.
9066 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9068 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
9069 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
9070 expansion of vector logical operations (and, andc, or, xor,
9073 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9075 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
9076 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
9078 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9080 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
9081 early GIMPLE expansion of vector multiplies.
9083 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9085 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
9086 TARGET_HAVE_MOVT conditional.
9087 (movt splitter): Likewise.
9089 2017-05-12 Richard Biener <rguenther@suse.de>
9091 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
9092 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9093 Fold all stmts not inplace.
9095 2017-05-12 Richard Biener <rguenther@suse.de>
9097 PR tree-optimization/80713
9098 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
9099 inserted_exprs bit for not removed stmts.
9101 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
9104 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
9105 parallelized" attribute for parallelized OpenACC kernels.
9106 * omp-offload.c (execute_oacc_device_lower): Use it.
9108 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
9109 Set "oacc kernels" attribute.
9110 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
9111 parameter. Adjust all users.
9112 (oacc_fn_attrib_kernels_p): Remove function.
9113 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
9114 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
9115 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
9116 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
9117 assert "oacc kernels" attribute is set.
9119 2017-05-11 Carl Love <cel@us.ibm.com>
9121 * config/rs6000/rs6000-c: Add support for built-in functions
9122 vector unsigned char vec_popcnt (vector signed char)
9123 vector unsigned char vec_popcnt (vector unsigned char)
9124 vector unsigned short vec_popcnt (vector signed short)
9125 vector unsigned short vec_popcnt (vector unsigned short)
9126 vector unsigned int vec_popcnt (vector signed int)
9127 vector unsigned int vec_popcnt (vector unsigned int)
9128 vector unsigned long long vec_popcnt (vector signed long long)
9129 vector unsigned long long vec_popcnt (vector unsigned long long)
9130 vector signed long long vec_slo (vector signed long long,
9132 vector signed long long vec_slo (vector signed long long,
9133 vector unsigned char)
9134 vector unsigned long long vec_slo (vector unsigned long long,
9136 vector unsigned long long vec_slo (vector unsigned long long,
9137 vector unsigned char)
9138 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
9139 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
9140 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
9141 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
9142 * doc/extend.texi: Update the built-in documentation file for the
9143 new built-in functions.
9145 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
9147 * attribs.h (sorted_attr_string): Move machine independent
9148 functions for target clone support from the i386 port to common
9149 code. Rename ix86_function_versions to common_function_versions.
9150 Rename make_name to make_unique_name.
9151 (common_function_versions): Likewise.
9152 (make_unique_name): Likewise.
9153 (make_dispatcher_decl): Likewise.
9154 (is_function_default_version): Likewise.
9155 * attribs.c (attr_strcmp): Likewise.
9156 (sorted_attr_string): Likewise.
9157 (common_function_versions): Likewise.
9158 (make_unique_name): Likewise.
9159 (make_dispatcher_decl): Likewise.
9160 (is_function_default_version): Likewise.
9161 * config/i386/i386.c (attr_strcmp): Likewise.
9162 (sorted_attr_string): Likewise.
9163 (ix86_function_versions): Likewise.
9164 (make_name): Likewise.
9165 (make_dispatcher_decl): Likewise.
9166 (is_function_default_version): Likewise.
9167 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
9169 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9172 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
9173 Account for direct move costs for vec_construct of integer
9176 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
9179 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
9180 (UNSPEC_STX_ATOMIC): Ditto.
9181 (loaddi_via_sse): New insn.
9182 (storedi_via_sse): Ditto.
9183 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
9184 Update corresponding peephole2 patterns.
9185 (atomic_storedi_fpu): Ditto.
9187 2017-05-11 Julia Koval <julia.koval@intel.com>
9189 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
9190 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
9192 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
9193 (__builtin_ia32_rsqrt14ss_mask): New builtins.
9194 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
9196 2017-05-11 Nathan Sidwell <nathan@acm.org>
9198 * graphite-poly.c: Include dumpfile.h.
9200 * dumpfle.h (dump_function): Declare here ...
9201 * tree-dump.h (dump_function): ... not here.
9202 * dumpfile.c: #include tree-cfg.h.
9203 (dump_function): Move here from ...
9204 * tree-dump.c (dump_function): ... here.
9205 * gimplify.c: #include splay-tree.h, not tree-dump.h.
9206 * graphite-poly.c: Don't include tree-dump.h.
9207 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
9208 * print-tree.c: Likewise.
9209 * stor-layout.c: Likewise.
9210 * tree-nested.c: Likewise.
9212 * dumpfile.c (dump_start): Use TDF_FLAGS.
9213 (dump_enable_all): Fix TDF_KIND check thinko.
9215 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
9217 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9218 array entries to represent two legal parameterizations of the
9219 overloaded __builtin_cmpb function, as represented by the
9220 P6_OV_BUILTIN_CMPB constant.
9221 (altivec_resolve_overloaded_builtin): Add special case handling
9222 for the __builtin_cmpb function, as represented by the
9223 P6_OV_BUILTIN_CMPB constant.
9224 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
9225 (BU_P6_64BIT_2): New macro.
9226 (BU_P6_OVERLOAD_2): New macro
9227 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
9228 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
9229 (CMPB): Add overload support to represent both 32-bit and 64-bit
9230 compare-bytes function.
9231 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
9232 support for TARGET_CMPB.
9233 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
9234 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
9235 documentation of the __builtin_cmpb overloaded built-in function.
9237 2017-05-11 Richard Biener <rguenther@suse.de>
9239 PR tree-optimization/80705
9240 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
9241 bases are not vectorizable.
9243 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9245 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
9246 when counting register pressure.
9248 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9250 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
9251 (iv_ca_more_deps): Renamed to ...
9252 (iv_ca_compare_deps): ... this.
9253 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
9255 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9257 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
9259 (determine_group_iv_costs): ... here.
9260 (find_inv_vars_cb): Record inv var if it's not recorded before.
9262 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9264 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
9265 (get_shiftadd_cost): Ditto.
9267 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9269 * tree-ssa-address.c: Include header file.
9270 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
9272 (add_to_parts): Refactor.
9273 (addr_to_parts): New parameter. Update use of move_hint_to_base.
9274 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
9277 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9279 PR tree-optimization/53090
9280 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
9282 (extract_cond_operands): Detect condition with IV on both sides
9283 and return COMP_IV_EXPR_2.
9284 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
9285 (rewrite_use_compare): Simplify by removing call to function
9286 extract_cond_operands.
9288 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9290 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
9291 (extract_cond_operands): Detect condition comparing against non-
9292 invariant bound and return appropriate enum value.
9293 (find_interesting_uses_cond): Update use of extract_cond_operands.
9294 Handle its return value accordingly.
9295 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
9297 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9299 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
9300 nonlinear iv_use computation in loop invariant sensitive way.
9302 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9304 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
9305 (find_iv_candidates): Call relate_compare_use_with_all_cands.
9307 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9309 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
9310 (dump_cand): Support iv_cand.inv_exprs.
9311 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
9313 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
9316 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9318 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
9320 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
9321 as local function. Include necessary header files.
9322 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
9324 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9326 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
9328 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9330 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
9331 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
9332 RSHIFT_EXPR and BIT_NOT_EXPR.
9334 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9336 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
9337 (adjust_setup_cost): New parameter supporting round up adjustment.
9338 (struct address_cost_data): Delete.
9339 (force_expr_to_var_cost): Don't bound cost with spill_cost.
9340 (split_address_cost, ptr_difference_cost): Delete.
9341 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
9342 (struct ainc_cost_data): New struct.
9343 (get_address_cost_ainc): New function.
9344 (get_address_cost, get_computation_cost): Reimplement.
9345 (determine_group_iv_cost_address): Record inv_expr for all uses of
9347 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
9348 (iv_ca_has_deps): Reimplemented to ...
9349 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
9351 (iv_ca_extend): Call iv_ca_more_deps.
9353 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9355 * tree-ssa-address.c (struct mem_address): Move to header file.
9356 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
9357 * tree-ssa-address.h (struct mem_address): Move from C file.
9358 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
9360 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9362 * tree-affine.h (aff_combination_type): New interface.
9363 (aff_combination_zero_p): Remove static.
9364 (aff_combination_const_p): New interface.
9365 (aff_combination_singleton_var_p): New interfaces.
9367 2017-05-11 Richard Biener <rguenther@suse.de>
9369 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9370 Skip unreachable blocks and destinations.
9371 (eliminate): Move stmt removal and fixup ...
9372 (fini_eliminate): ... here. Skip inserted exprs.
9373 (pass_pre::execute): Move fini_pre after fini_eliminate.
9374 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
9375 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
9376 PRE to get rid of dead code that has invalid SSA form and
9377 split critical edges again.
9379 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9381 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
9383 2017-05-11 Richard Biener <rguenther@suse.de>
9385 * passes.c (execute_function_todo): Verify loops if they are
9386 said to be up-to-date.
9387 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
9388 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
9390 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
9393 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
9394 handle calling assemble_external ourself.
9397 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
9398 modes with zero size. Enhance comment.
9400 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9402 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
9403 built-ins for vec_xl and vec_xst with short and char pointer
9406 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
9408 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
9409 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
9410 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
9411 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
9412 (_mm_maskz_min_round_ss): New intrinsics.
9413 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
9414 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
9415 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
9416 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
9417 (__builtin_ia32_minss_mask_round): New builtins.
9418 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9419 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9420 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
9422 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
9423 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
9425 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
9428 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
9430 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
9431 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
9432 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
9433 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
9434 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
9435 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
9436 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
9437 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9438 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
9439 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
9440 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
9441 (__builtin_ia32_mulss_mask_round): New builtins.
9442 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9443 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9444 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
9446 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
9447 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
9449 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
9452 2017-05-10 Julia Koval <julia.koval@intel.com>
9454 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
9455 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
9456 (_mm256_setr_m128i): New intrinsics.
9458 2017-05-10 Julia Koval <julia.koval@intel.com>
9460 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
9461 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
9462 (_mm_maskz_rcp14_ss): New intrinsics.
9463 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
9464 (__builtin_ia32_rcp14ss_mask): New builtins.
9465 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
9467 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
9469 PR tree-optimization/51513
9470 * tree-cfg.c (gimple_seq_unreachable_p): New function.
9471 (assert_unreachable_fallthru_edge_p): Use it.
9472 (group_case_labels_stmt): Likewise.
9473 * tree-cfg.h: Prototype it.
9474 * stmt.c: Include cfghooks.h and tree-cfg.h.
9475 (emit_case_dispatch_table) <gap_label>: New local variable.
9476 Use it to fill dispatch table gaps.
9477 Test for default_label before updating probabilities.
9478 (expand_case) <default_label>: Remove unneeded initialization.
9479 Test for unreachable default case statement and remove its edge.
9480 Set default_label accordingly.
9481 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
9483 2017-05-10 Carl Love <cel@us.ibm.com>
9485 * config/rs6000/rs6000-c: Add support for built-in functions
9486 vector signed char vec_neg (vector signed char)
9487 vector signed short int vec_neg (vector short int)
9488 vector signed int vec_neg (vector signed int)
9489 vector signed long long vec_neg (vector signed long long)
9490 vector float vec_neg (vector float)
9491 vector double vec_neg (vector double)
9492 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
9494 * config/rs6000/altivec.h: Add define for vec_neg
9495 * doc/extend.texi: Update the built-in documentation for the
9496 new built-in functions.
9498 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9500 PR tree-optimization/77644
9501 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
9503 2017-05-10 Nathan Sidwell <nathan@acm.org>
9505 * dumpfile.h (TDI_lang_all): New.
9506 (TDF_KIND): New. Renumber others
9507 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
9509 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
9511 (get_dump_file_name): Adjust suffix generation.
9512 (dump_enable_all): Use TDF_KIND.
9513 * doc/invoke.texi (-fdump-lang-all): Document.
9515 * dumpfile.h: Tabify.
9517 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
9520 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
9521 Move member access before delete.
9523 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
9525 * tree-inline.c (expand_call_inline): Split block at stmt
9528 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9531 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
9532 are now unused after splitting mov{sf,sd}_hardfloat.
9533 (f32_lr2): Likewise.
9535 (f32_lm2): Likewise.
9537 (f32_li2): Likewise.
9540 (f32_sr2): Likewise.
9542 (f32_sm2): Likewise.
9544 (f32_si2): Likewise.
9547 (f32_vsx): Likewise.
9549 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
9550 For movsf, order stores so the VSX stores occur before the GPR
9551 store which encourages the register allocator to use a traditional
9552 FPR instead of a GPR. For movsd, order the stores so that the GPR
9553 store comes before the VSX stores to allow the power6 to work.
9554 This is due to the power6 not having a 32-bit integer store
9555 instruction from a FPR.
9556 (movsf_hardfloat): Likewise.
9557 (movsd_hardfloat): Likewise.
9559 2017-05-09 Martin Sebor <msebor@redhat.com>
9561 PR translation/80280
9562 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
9565 PR translation/80280
9566 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
9567 data member added in r247778.
9568 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
9570 2017-05-09 Nathan Sidwell <nathan@acm.org>
9572 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
9574 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
9577 2017-05-09 Marek Polacek <polacek@redhat.com>
9579 * doc/invoke.texi: Fix typo.
9581 2017-05-09 Richard Biener <rguenther@suse.de>
9583 * tree-vrp.c (vrp_val_is_max): Adjust comment.
9584 (vrp_val_is_min): Likewise.
9585 (set_value_range_to_value): Likewise.
9586 (set_value_range_to_nonnegative): Likewise.
9587 (gimple_assign_nonzero_p): Likewise.
9588 (gimple_stmt_nonzero_p): Likewise.
9589 (vrp_int_const_binop): Likewise. Remove unreachable case.
9590 (adjust_range_with_scev): Adjust comments.
9591 (compare_range_with_value): Likewise.
9592 (extract_range_from_phi_node): Likewise.
9593 (test_for_singularity): Likewise.
9595 2017-05-09 Richard Biener <rguenther@suse.de>
9597 * tree-vrp.c (get_single_symbol): Add assert that we don't
9598 get overflowed constants as invariant part.
9599 (compare_values_warnv): Add comment before the TREE_NO_WARNING
9600 checks. Use wi::cmp instead of recursing for integer constants.
9601 (compare_values): Just ignore whether we assumed undefined
9602 overflow instead of failing the compare.
9603 (extract_range_for_var_from_comparison_expr): Add comment before the
9604 TREE_NO_WARNING sets.
9605 (test_for_singularity): Likewise.
9606 (extract_range_from_comparison): Do not disable optimization
9607 when we assumed undefined overflow.
9608 (extract_range_basic): Remove init of unused var.
9610 2017-05-09 Richard Biener <rguenther@suse.de>
9612 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
9613 (extract_range_from_multiplicative_op_1): Adjust.
9614 (extract_range_from_binary_expr_1): Use int_const_binop.
9616 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
9619 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
9620 rs6000_store_data_bypass_p in seven define_bypass directives and
9621 in several comments.
9622 * config/rs6000/rs6000-protos.h: Add prototype for
9623 rs6000_store_data_bypass_p function.
9624 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
9625 function implements slightly different (rs6000-specific) semantics
9626 than store_data_bypass_p, returning false rather than aborting
9627 with assertion error when arguments do not satisfy the
9628 requirements of store data bypass.
9629 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
9630 rs6000_store_data_bypass_p.
9632 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
9634 * config/xtensa/xtensa-protos.h
9635 (xtensa_initial_elimination_offset): New declaration.
9636 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
9637 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
9638 macro definition, add case for FRAME_POINTER_REGNUM when
9639 FRAME_GROWS_DOWNWARD.
9640 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
9641 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
9642 xtensa_initial_elimination_offset.
9644 2017-05-08 Nathan Sidwell <nathan@acm.org>
9646 * doc/invoke.texi: Alphabetize -fdump options.
9648 2017-05-08 Martin Sebor <msebor@redhat.com>
9650 PR translation/80280
9651 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
9653 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
9655 * target.def (compute_frame_layout): New optional target hook.
9656 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
9657 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
9658 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
9660 * reload1.c (verify_initial_elim_offsets): Likewise.
9661 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
9662 (use_simple_return_p): Call arm_compute_frame_layout if needed.
9663 (arm_get_frame_offsets): Split up into this ...
9664 (arm_compute_frame_layout): ... and this function.
9666 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
9668 * config/aarch64/constraints.md (Usa): New constraint.
9669 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
9671 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
9673 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
9674 with_multilib_list after it has been checked.
9676 2017-05-08 Richard Biener <rguenther@suse.de>
9678 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
9679 (bitmap_set_subtract_values): Likewise.
9681 2017-05-08 Richard Biener <rguenther@suse.de>
9683 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
9684 (gimple_assign_nonzero): ... this and remove strict_overflow_p
9686 (gimple_stmt_nonzero_warnv_p): Rename to ...
9687 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
9689 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
9690 (extract_range_basic): Adjust, do not disable propagation on
9691 strict overflow sensitive simplification.
9692 (vrp_visit_cond_stmt): Likewise.
9694 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
9696 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
9697 body size unconditionally.
9699 2017-05-07 Jeff Law <law@redhat.com>
9702 2017-05-06 Jeff Law <law@redhat.com>
9703 PR tree-optimization/78496
9704 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9707 PR tree-optimization/78496
9708 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9709 (simplify_stmt_using_ranges): Call it.
9710 (vrp_dom_walker::before_dom_children): Extract equivalences
9711 from an ASSERT_EXPR with an equality comparison against a
9714 2017-05-06 Jeff Law <law@redhat.com>
9716 PR tree-optimization/78496
9717 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9720 PR tree-optimization/78496
9721 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9722 (simplify_stmt_using_ranges): Call it.
9723 (vrp_dom_walker::before_dom_children): Extract equivalences
9724 from an ASSERT_EXPR with an equality comparison against a
9727 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9729 * lra-constraints.c (lra_copy_reg_equiv): New function.
9730 (split_reg): Use it to copy equivalence information from the
9731 original register to the spill register.
9733 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9735 PR rtl-optimization/75964
9736 * simplify-rtx.c (simplify_const_relational_operation): Remove
9737 invalid handling of comparisons of integer ABS.
9739 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
9741 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
9743 (init_regs): Remove declaration.
9744 (function_arg_advance_32): Initialize error_p as boolean variable.
9746 2017-05-05 Nathan Sidwell <nathan@acm.org>
9748 * store-motion.c (remove_reachable_equiv_notes): Reformat long
9749 lines. Use for (;;).
9751 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9753 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
9754 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
9755 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
9756 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
9757 VF=2 that require versioning.
9759 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9761 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
9764 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9766 * diagnostic.h (diagnostic_override_option_index): Convert from
9767 macro to inline function.
9769 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9771 * diagnostic.c (last_module_changed_p): New function.
9772 (set_last_module): New function.
9773 (diagnostic_report_current_module): Convert macro usage to
9774 the above functions.
9775 * diagnostic.h (diagnostic_context::last_module): Strengthen
9776 from const line_map * to const line_map_ordinary *.
9777 (diagnostic_last_module_changed): Delete macro.
9778 (diagnostic_set_last_module): Delete macro.
9780 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9782 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
9783 with diagnostic_report_diagnostic.
9784 (diagnostic_n_impl_richloc): Likewise.
9785 * diagnostic.h (report_diagnostic): Delete macro.
9786 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
9787 with diagnostic_report_diagnostic.
9788 * substring-locations.c (format_warning_va): Likewise.
9790 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9792 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
9793 save/restor of format_spec. Move option-printing code to...
9794 (print_option_information): ...this new function, and
9795 reimplement by simply printing to the pretty_printer,
9796 rather than appending to the format string.
9798 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9800 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
9801 handling logic into...
9802 (update_effective_level_from_pragmas): ...this new function.
9804 2017-05-04 Andrew Waterman <andrew@sifive.com>
9806 * config/riscv/riscv.opt (mstrict-align): New option.
9807 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
9808 (SLOW_UNALIGNED_ACCESS): Define.
9809 (riscv_slow_unaligned_access): Declare.
9810 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
9812 (riscv_slow_unaligned_access): New variable.
9813 (rocket_tune_info): Set slow_unaligned_access to true.
9814 (optimize_size_tune_info): Set slow_unaligned_access to false.
9815 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
9816 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
9817 (riscv_option_override): Set riscv_slow_unaligned_access.
9818 * doc/invoke.texi: Add -mstrict-align to RISC-V.
9820 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
9822 * config/riscv/riscv.md: Unify indentation.
9824 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
9829 * config/rs6000/rs6000.md (u code attribute): Add FIX and
9831 (extendsi<mode>2): Add support for doing sign extension via
9832 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
9833 don't have ISA 3.0 instructions.
9834 (extendsi<mode>2 splitter): Likewise.
9835 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
9836 generate the normal insns since SImode can now go in vector
9837 registers. Disallow the special UNSPECs needed for previous
9838 machines to hide SImode being used. Add new insns
9839 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
9840 (fix_trunc<mode>si2_stfiwx): Likewise.
9841 (fix_trunc<mode>si2_internal): Likewise.
9842 (fixuns_trunc<mode>si2): Likewise.
9843 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9844 (fctiw<u>z_<mode>_smallint): Likewise.
9845 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
9846 of floating point to 32-bit integer from doing a direct move to
9847 the GPR registers to do a store.
9848 (fctiwz_<mode>): Break long line.
9850 2017-05-05 Bin Cheng <bin.cheng@arm.com>
9852 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
9853 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
9854 (addr_list, addr_offset_valid_p): New.
9855 (split_address_groups): Check offset validity with above function.
9856 (gt-tree-ssa-loop-ivopts.h): Include header file.
9858 2017-05-05 Nathan Sidwell <nathan@acm.org>
9860 * config.gcc (arm*-*-*): Add missing 'fi'.
9862 2017-05-05 Steve Ellcey <sellcey@cavium.com>
9864 * doc/invoke.texi (-fopt-info): Explicitly say order of options
9865 included in -fopt-info does not matter.
9866 * doc/optinfo.texi (-fopt-info): Fix description of default
9867 behavour. Explicitly say order of options included in -fopt-info
9870 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
9872 * config.gcc: Allow combinations of aprofile and rmprofile values for
9873 --with-multilib-list.
9874 * config/arm/t-multilib: New file.
9875 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
9876 variables. Remove setting of ISA and floating-point ABI in
9877 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
9878 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
9879 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
9880 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
9882 * config/arm/t-rmprofile: Likewise except for the matches changes.
9883 * doc/install.texi (--with-multilib-list): Document the combination of
9884 aprofile and rmprofile values and warn about pitfalls in doing that.
9886 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9888 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
9889 (movdi_aarch64): Likewise.
9891 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9893 PR tree-optimization/80632
9894 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
9896 (build_arrays): Initialize it for virtual phis.
9897 (fix_phi_nodes): Use it for virtual phis.
9899 PR tree-optimization/80558
9900 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
9901 [x, y] op z into [x op, y op z] for op & or | if conditions
9904 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
9905 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9908 * config/arm/arm.md (use_literal_pool): Remove.
9909 (64-bit immediate split): No longer takes cost into consideration
9910 if arm_disable_literal_pool is enabled.
9911 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
9912 used when arm_disable_literal_pool is enabled.
9913 (arm_max_const_double_inline_cost): Remove use of
9914 arm_disable_literal_pool.
9915 (push_minipool_fix): Add assert.
9916 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
9917 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
9918 (no_literal_pool_sf_immediate): New.
9920 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9922 PR tree-optimization/80613
9923 * tree-ssa-dce.c (propagate_necessity): Remove cases for
9924 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
9926 2017-05-05 Richard Biener <rguenther@suse.de>
9928 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
9930 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
9932 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
9933 of this flag from insn conditions due to removal from r247495.
9935 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9937 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
9939 (arm_early_store_addr_dep_ptr): Likewise.
9940 * config/arm/aarch-common-protos.h
9941 (arm_early_load_addr_dep_ptr): Add prototype.
9942 (arm_early_store_addr_dep_ptr): Likewise.
9943 * config/arm/cortex-a53.md: Add new bypasses.
9945 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9947 * tree.c (next_type_uid): Change type to unsigned.
9948 (type_hash_canon): Decrement back next_type_uid if
9949 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
9950 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
9953 2017-05-04 Martin Sebor <msebor@redhat.com>
9955 * builtins.c: Fix a trivial typo in a comment.
9958 * builtins.c (check_sizes): Adjust to handle reading past the end.
9959 Avoid printing excessive upper bound of ranges. Use %E to print
9960 tree nodes instead of converting them to %wu.
9961 (expand_builtin_memchr): New function.
9962 (compute_dest_size): Rename...
9963 (compute_objsize): ...to this.
9964 (expand_builtin_memcpy): Adjust.
9965 (expand_builtin_mempcpy): Adjust.
9966 (expand_builtin_strcat): Adjust.
9967 (expand_builtin_strcpy): Adjust.
9968 (check_strncat_sizes): Adjust.
9969 (expand_builtin_strncat): Adjust.
9970 (expand_builtin_strncpy): Adjust and simplify.
9971 (expand_builtin_memset): Adjust.
9972 (expand_builtin_bzero): Adjust.
9973 (expand_builtin_memcmp): Adjust.
9974 (expand_builtin): Handle memcmp.
9975 (maybe_emit_chk_warning): Check strncat just once.
9977 2017-05-04 Martin Sebor <msebor@redhat.com>
9979 PR preprocessor/79214
9982 * builtins.c (check_sizes): Add inlining context and issue
9983 warnings even when -Wno-system-headers is set.
9984 (check_strncat_sizes): Same.
9985 (expand_builtin_strncat): Same.
9986 (expand_builtin_memmove): New function.
9987 (expand_builtin_stpncpy): Same.
9988 (expand_builtin): Handle memmove and stpncpy.
9990 2017-05-04 Bin Cheng <bin.cheng@arm.com>
9992 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
9993 which is not used any more.
9995 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9997 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
9999 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
10001 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
10002 (cortexa53_tunings): Likewise.
10003 (cortexa57_tunings): Likewise.
10004 (cortexa72_tunings): Likewise.
10005 (cortexa73_tunings): Likewise.
10007 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
10009 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
10010 Set loop alignment to 8.
10012 2017-05-04 Martin Sebor <msebor@redhat.com>
10014 PR translation/80280
10015 * builtins.c (expand_builtin_object_size): Add missing quoting to
10016 %D and like directives.
10017 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
10018 (hsa_type_for_tree_type): Same.
10019 (verify_function_arguments): Same.
10020 * symtab.c (symbol_table::change_decl_assembler_name): Same.
10021 * varasm.c (get_section): Same.
10024 2017-05-04 Martin Sebor <msebor@redhat.com>
10026 PR translation/80280
10027 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
10029 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
10031 * config/aarch64/aarch64.c (generic_addrcost_table):
10032 Change HI/TI mode setting.
10034 2017-05-04 Martin Jambor <mjambor@suse.cz>
10036 PR tree-optimization/80622
10037 * tree-sra.c (comes_initialized_p): New function.
10038 (build_accesses_from_assign): Only set write lazily when
10039 comes_initialized_p is false.
10040 (analyze_access_subtree): Use comes_initialized_p.
10041 (propagate_subaccesses_across_link): Assert !comes_initialized_p
10042 instead of testing for PARM_DECL.
10044 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10046 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
10047 constraint on operand 0 to allow more general addressing modes.
10048 Adjust output template.
10049 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
10051 * config/aarch64/aarch64-protos.h
10052 (aarch64_address_valid_for_prefetch_p): Declare prototype.
10053 * config/aarch64/constraints.md (Dp): New address constraint.
10054 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
10057 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
10059 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
10060 update use of estimate_ipcp_clone_size_and_time.
10061 (estimate_local_effects): Update use of
10062 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
10063 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
10064 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
10065 Return nonspecialized time.
10067 2017-05-04 Richard Biener <rguenther@suse.de>
10069 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
10070 for the last VUSE which def dominates the PHI. Directly call
10072 (get_continuation_for_phi_1): Remove.
10074 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
10076 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
10077 to explain the use of truncating division. Cap the number of
10078 iterations to the maximum given by nb_iterations_upper_bound,
10081 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
10083 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
10084 * configure: Regenerate.
10085 * config.in: Regenerate.
10086 * config/i386/driver-mingw32.c: new file.
10087 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
10088 * config.host: Link driver-mingw32.o on MinGW host.
10089 * doc/install.texi: Document new --enable-mingw-wildcard configure
10092 2017-05-04 Marek Polacek <polacek@redhat.com>
10094 PR tree-optimization/80612
10095 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
10097 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10098 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
10100 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
10101 (movt splitter): Likewise.
10102 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
10103 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
10104 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
10105 block for Thumb-1 with MOVT.
10106 (thumb2_legitimate_address_p): Move code block ...
10107 (can_avoid_literal_pool_for_label_p): ... into this new function.
10108 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
10110 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
10111 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
10112 "M-profile targets with the MOVT instruction".
10114 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10116 * config/arm/arm-builtins.c (arm_init_builtins): Rename
10117 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
10118 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
10120 2017-05-04 Martin Liska <mliska@suse.cz>
10122 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
10123 variable cond_code.
10125 2017-05-04 Richard Biener <rguenther@suse.de>
10127 * tree.c (array_at_struct_end_p): Handle arrays at struct
10128 end with flexarrays more conservatively. Refactor and treat
10129 arrays of arrays or aggregates more strict. Fix
10130 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
10131 * tree.c (array_at_struct_end_p): Adjust prototype.
10132 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
10133 * gimple-fold.c (get_range_strlen): Likewise.
10134 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
10136 2017-05-04 Richard Biener <rguenther@suse.de>
10138 PR tree-optimization/31130
10139 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
10141 (supports_overflow_infinity): Likewise.
10142 (is_negative_overflow_infinity): Likewise.
10143 (is_positive_overflow_infinity): Likewise.
10144 (is_overflow_infinity): Likewise.
10145 (stmt_overflow_infinity): Likewise.
10146 (overflow_infinity_range_p): Likewise.
10147 (usable_range_p): Remove as always returning true.
10148 (make_overflow_infinity): Remove.
10149 (negative_overflow_infinity): Likewise.
10150 (positive_overflow_infinity): Likewise.
10151 (avoid_overflow_infinity): Likewise.
10152 (set_value_range): Adjust accordingly.
10153 (set_value_range_to_nonnegative): Likewise, remove now unused
10154 overflow_infinity arg.
10155 (vrp_operand_equal_p): Adjust.
10156 (update_value_range): Likewise.
10157 (range_int_cst_singleton_p): Likewise.
10158 (operand_less_p): Likewise.
10159 (compare_values_warnv): Likewise.
10160 (extract_range_for_var_from_comparison_expr): Likewise.
10161 (vrp_int_const_binop): Likewise.
10162 (zero_nonzero_bits_from_vr): Likewise.
10163 (extract_range_from_multiplicative_op_1): Likewise.
10164 (extract_range_from_binary_expr_1): Likewise.
10165 (extract_range_from_unary_expr): Likewise.
10166 (extract_range_from_comparison): Likewise.
10167 (extract_range_basic): Likewise.
10168 (adjust_range_with_scev): Likewise.
10169 (compare_ranges): Likewise.
10170 (compare_range_with_value): Likewise.
10171 (dump_value_range): Likewise.
10172 (test_for_singularity): Likewise, remove strict_overflow_p parameter
10174 (simplify_cond_using_ranges): Adjust.
10176 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
10178 * brig-builtins.def: Added a builtin for class_f64.
10179 * builtin-types.def: Added a builtin type needed by class_f64.
10181 2017-05-03 Jason Merrill <jason@redhat.com>
10183 * timevar.def: Add TV_CONSTEXPR.
10185 2017-05-03 David Malcolm <dmalcolm@redhat.com>
10187 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
10189 2017-05-03 Martin Jambor <mjambor@suse.cz>
10191 * ipa-prop.c (ipa_update_after_lto_read): Removed.
10192 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
10193 * ipa-cp.c (ipcp_propagate_stage): Do not call
10194 ipa_update_after_lto_read.
10195 * ipa-inline.c (ipa_inline): Likewise.
10197 2017-05-03 Martin Jambor <mjambor@suse.cz>
10199 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
10200 tag. Added a default constructor and a destructor.
10201 (ipa_edge_args_sum_t): New class;
10202 (ipa_edge_args_sum): Declare.
10203 (ipa_edge_args_vector): Remove declaration.
10204 (IPA_EDGE_REF): Use ipa_edge_args_sum.
10205 (ipa_free_edge_args_substructures): Remove declaration.
10206 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
10207 (ipa_edge_args_info_available_for_edge_p): Likewise.
10208 * ipa-prop.c (ipa_edge_args_vector): Removed.
10209 (edge_removal_hook_holder): Likewise.
10210 (edge_duplication_hook_holder): Likewise.
10211 (ipa_edge_args_sum): New variable.
10212 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
10213 ipa_edge_args_vector.
10214 (ipa_free_edge_args_substructures): Likewise.
10215 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
10216 ipa_edge_args_vector.
10217 (ipa_edge_removal_hook): Turned into method
10218 ipa_edge_args_sum_t::remove.
10219 (ipa_edge_duplication_hook): Turned into method
10220 ipa_edge_args_sum_t::duplicate.
10221 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
10222 registering edge hooks.
10223 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
10224 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
10225 ipa_edge_args_sum instead of ipa_edge_args_vector.
10226 * ipa-profile.c (ipa_profile): Likewise.
10228 2017-05-03 Martin Jambor <mjambor@suse.cz>
10230 * symbol-summary.h (function_summary): New method exists.
10231 (function_summary::symtab_removal): Deallocate through release.
10232 (call_summary): New class.
10233 (gt_ggc_mx): New overload.
10234 (gt_pch_nx): Likewise.
10235 (gt_pch_nx): Likewise.
10237 2017-05-03 Jeff Law <law@redhat.com>
10239 PR tree-optimization/78496
10240 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
10241 from simplify_cond_using_ranges. Split off code to walk
10242 backwards through casts into ...
10243 (simplify_cond_using_ranges_2): New function.
10244 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
10245 (execute_vrp): After identifying jump threads, call
10246 simplify_cond_using_ranges_2.
10248 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
10251 * ipa-inline.h (inline_summary): Add ctor.
10252 (create_ggc): Do not use ggc_cleared_alloc.
10254 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
10255 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10257 * gcc.c (handle_braces): Support escaping in switch matching
10259 * doc/invoke.texi (Spec Files): Document it.
10260 Remove superfluous @code markup in items.
10262 2017-05-03 David Malcolm <dmalcolm@redhat.com>
10264 * diagnostic-show-locus.c (struct column_range): New struct.
10265 (get_affected_columns): New function.
10266 (get_printed_columns): New function.
10267 (struct correction): New struct.
10268 (correction::ensure_capacity): New function.
10269 (correction::ensure_terminated): New function.
10270 (struct line_corrections): New struct.
10271 (line_corrections::~line_corrections): New dtor.
10272 (line_corrections::add_hint): New function.
10273 (layout::print_trailing_fixits): Reimplement in terms of the new
10275 (selftest::test_overlapped_fixit_printing): New function.
10276 (selftest::diagnostic_show_locus_c_tests): Call it.
10278 2017-05-03 Nathan Sidwell <nathan@acm.org>
10280 Canonicalize canonical type hashing
10281 * tree.h (type_hash_canon_hash): Declare.
10282 * tree.c (type_hash_list, attribute_hash_list): Move into
10283 type_hash_canon_hash.
10284 (build_type_attribute_qual_variant): Break out hash code calc into
10285 type_hash_canon_hash.
10286 (type_hash_canon_hash): New. Generic type hash computation.
10287 (build_range_type_1, build_array_type_1, build_function_type,
10288 build_method_type_directly, build_offset_type, build_complex_type,
10289 make_vector_type): Call it.
10291 2017-05-03 Richard Biener <rguenther@suse.de>
10293 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
10294 When all DRs have unknown misaligned do not always peel
10295 when there is a store but apply the same costing model as if
10296 there were only loads.
10298 2017-05-03 Richard Biener <rguenther@suse.de>
10301 PR tree-optimization/80492
10302 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
10303 compare_base_decls returning dont-know properly.
10305 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
10307 * config/arm/iterators.md (CCSI): New mode iterator.
10308 (arch): New mode attribute.
10309 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
10310 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
10311 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
10312 code iterator for success result mode.
10313 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
10314 the corresponding new insn generators.
10316 2017-05-03 Bin Cheng <bin.cheng@arm.com>
10319 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10320 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
10322 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
10324 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
10325 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
10326 (DDR_A): Wrap DDR argument in brackets.
10327 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
10328 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
10329 (DDR_REVERSED_P): Likewise.
10331 2017-05-03 Jakub Jelinek <jakub@redhat.com>
10333 PR tree-optimization/79472
10334 * tree-switch-conversion.c (struct switch_conv_info): Add
10335 contiguous_range and default_case_nonstandard fields.
10336 (collect_switch_conv_info): Compute contiguous_range and
10337 default_case_nonstandard fields, don't clear final_bb if
10338 contiguous_range and only the default case doesn't have the required
10340 (check_all_empty_except_final): Set default_case_nonstandard instead
10341 of failing if contiguous_range and the default case doesn't have empty
10343 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
10344 and only the default case doesn't have the required constants. Skip
10346 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
10347 if default_case_nonstandard.
10348 (build_constructors): Build constant 1 just once. Assert that default
10349 values aren't inserted in between cases if contiguous_range. Skip
10351 (build_arrays): Skip virtual phis.
10352 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
10353 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
10354 Handle virtual phis.
10355 (gen_inbound_check): Handle default_case_nonstandard case.
10356 (process_switch): Adjust check_final_bb caller. Call
10357 gather_default_values with the first non-default case instead of
10358 default case if default_case_nonstandard.
10360 2017-05-02 Nathan Sidwell <nathan@acm.org>
10362 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
10363 check. Fix formatting.
10365 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
10367 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
10368 errors when comparing specialized and unspecialized times.
10370 2017-05-02 David Malcolm <dmalcolm@redhat.com>
10372 * diagnostic-show-locus.c
10373 (layout::should_print_annotation_line_p): Make private.
10374 (layout::print_annotation_line): Make private.
10375 (layout::annotation_line_showed_range_p): Make private.
10376 (layout::show_ruler): Make private.
10377 (layout::print_source_line): Make private. Pass in line and
10378 line_width, rather than calling location_get_source_line. Drop
10380 (layout::print_leading_fixits): New method.
10381 (layout::print_any_fixits): Rename to...
10382 (layout::print_trailing_fixits): ...this, and make private.
10383 Don't print newline fixits.
10384 (diagnostic_show_locus): Move logic for printing one row into...
10385 (layout::print_line): ...this new function. Move the
10386 location_get_source_line call and error-handling from
10387 print_source_line to here. Call print_leading_fixits, and rename
10388 print_any_fixits to print_trailing_fixits.
10389 (selftest::test_fixit_insert_containing_newline): Update now that
10390 newlines are partially supported.
10391 (selftest::test_fixit_insert_containing_newline_2): New test.
10392 (selftest::test_fixit_replace_containing_newline): Update comments.
10393 (selftest::diagnostic_show_locus_c_tests): Call the new test.
10394 * edit-context.c (class added_line): New class.
10395 (class edited_line): Describe newline handling in comment.
10396 (edited_line::actually_edited_p): New method.
10397 (edited_line::print_content): Delete redundant decl.
10398 (edited_line::m_predecessors): New field.
10399 (edited_file::print_content): Call edited_line::print_content.
10400 (edited_file::print_diff): Update to support newlines.
10401 (edited_file::print_diff_hunk): Likewise.
10402 (edited_file::print_run_of_changed_lines): New function.
10403 (edited_file::print_diff_line): Convert to...
10404 (print_diff_line): ...this.
10405 (edited_file::get_effective_line_count): New function.
10406 (edited_line::edited_line): Initialize new field m_predecessors.
10407 (edited_line::~edited_line): Clean up m_predecessors.
10408 (edited_line::apply_fixit): Handle newlines.
10409 (edited_line::get_effective_line_count): New function.
10410 (edited_line::print_content): New function.
10411 (edited_line::print_diff_lines): New function.
10412 (selftest::test_applying_fixits_insert_containing_newline): New
10414 (selftest::test_applying_fixits_replace_containing_newline): New
10416 (selftest::insert_line): New function.
10417 (selftest::test_applying_fixits_multiple_lines): Add example of
10419 (selftest::edit_context_c_tests): Call the new tests.
10421 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10423 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
10424 parameter cand. Update dump information.
10425 (get_computation_cost): Update uses.
10427 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10429 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
10430 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
10431 (get_computation_at, rewrite_use_address): Update use of
10432 get_computation_aff.
10434 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10436 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
10437 (get_computation): Delete.
10438 (get_computation_cost): Implement like get_computation_cost_at.
10439 Use get_computation_at.
10440 (get_computation_cost_at): Delete.
10441 (rewrite_use_nonlinear_expr): Use get_computation_at.
10442 (rewrite_use_compare, remove_unused_ivs): Ditto.
10444 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10446 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
10448 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10450 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
10451 (ivopts_global_cost_for_size): Rename parameter and update uses.
10452 (iv_ca_recount_cost): Update uses.
10453 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
10454 candidates seperately in n_invs and n_cands.
10455 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
10457 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10459 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
10460 (find_inv_vars_cb): New.
10461 (find_depends): Renamed to ...
10462 (find_inv_vars): ... this.
10463 (add_candidate_1, force_var_cost): Call find_inv_vars.
10464 (split_address_cost, determine_group_iv_cost_cond): Ditto.
10466 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10468 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
10469 inv_vars. Add inv_exprs.
10470 (struct iv_cand): Rename depends_on to inv_vars.
10471 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
10472 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
10473 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
10474 (dump_cand): Dump inv_vars.
10475 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
10476 (record_invariant, find_depends, add_candidate_1): Ditto.
10477 (set_group_iv_cost, force_var_cost): Ditto.
10478 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
10479 (get_computation_cost_at, get_computation_cost): Ditto.
10480 (determine_group_iv_cost_generic): Ditto.
10481 (determine_group_iv_cost_address): Ditto.
10482 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
10483 (determine_group_iv_costs): Ditto.
10484 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
10485 (iv_ca_set_remove_invariants): Renamed to ...
10486 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
10487 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
10488 (iv_ca_set_add_invariants): Renamed to ...
10489 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
10490 (iv_ca_set_cp): Use iv_ca_set_add_invs.
10491 (iv_ca_has_deps): Support inv_vars and inv_exprs.
10492 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
10493 (create_new_ivs): Remove useless dump.
10495 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10497 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
10499 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
10500 (iv_ca_set_no_cp, create_new_iv): Ditto.
10502 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10504 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
10506 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10508 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
10509 function tree_check2.
10511 2017-05-02 Martin Liska <mliska@suse.cz>
10513 * doc/gcov.texi: Add missing preposition.
10514 * gcov.c (function_info::function_info): Properly fill up
10515 all member variables.
10517 2017-05-02 Tamar Christina <tamar.christina@arm.com>
10519 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
10521 2017-05-02 Tamar Christina <tamar.christina@arm.com>
10523 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
10525 2017-05-02 Martin Liska <mliska@suse.cz>
10528 * lto-streamer-in.c (lto_read_tree_1): Remove
10529 LTO_STREAMER_DEBUG.
10530 * lto-streamer.c (struct tree_hash_entry): Likewise.
10531 (struct tree_entry_hasher): Likewise.
10532 (tree_entry_hasher::hash): Likewise.
10533 (tree_entry_hasher::equal): Likewise.
10534 (lto_streamer_init): Likewise.
10535 (lto_orig_address_map): Likewise.
10536 (lto_orig_address_get): Likewise.
10537 (lto_orig_address_remove): Likewise.
10538 * lto-streamer.h: Likewise.
10539 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
10540 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
10542 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
10544 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
10545 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
10546 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
10547 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
10548 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
10549 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
10550 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
10551 (mm_maskz_sub_ss): New intrinsics.
10552 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
10553 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
10554 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
10555 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
10556 (__builtin_ia32_subss_mask_round): New builtins.
10557 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
10558 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
10559 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
10561 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
10562 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
10564 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
10567 2017-05-02 Martin Jambor <mjambor@suse.cz>
10569 PR tree-optimization/78687
10570 * tree-sra.c (access): New field parent.
10571 (process_subtree_disqualification): New function.
10572 (disqualify_candidate): Call it.
10573 (build_accesses_from_assign): Reset write flag if creating an
10575 (build_access_subtree): Fill in parent field and also prpagate
10576 down grp_write flag.
10577 (create_artificial_child_access): New parameter set_grp_write, set
10578 grp_write to its value.
10579 (propagate_subaccesses_across_link): Also propagate grp_write flag
10581 (propagate_all_subaccesses): Push the closest parent back to work
10582 queue if add_access_to_work_queue returned true.
10584 2017-05-02 Richard Biener <rguenther@suse.de>
10586 * common.opt (fstrict-overflow): Alias negative to fwrapv.
10587 * doc/invoke.texi (fstrict-overflow): Remove all traces of
10588 -fstrict-overflow documentation.
10589 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
10590 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
10591 flag_strict_overflow.
10592 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
10593 * lto-opts.c (lto_write_options): Do not stream it.
10594 * lto-wrapper.c (merge_and_complain): Do not handle it.
10595 * opts.c (default_options_table): Do not set -fstrict-overflow.
10596 (finish_options): Likewise do not clear it when sanitizing.
10597 * simplify-rtx.c (simplify_const_relational_operation): Do not
10598 test flag_strict_overflow.
10600 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10602 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
10603 using enabled attribute.
10604 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
10605 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
10606 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
10607 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
10608 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
10609 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
10610 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
10611 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
10612 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
10613 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
10615 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10617 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
10619 2017-05-02 Richard Biener <rguenther@suse.de>
10621 PR tree-optimization/80591
10623 2017-04-10 Richard Biener <rguenther@suse.de>
10625 * tree-ssa-structalias.c (find_func_aliases): Properly handle
10628 2017-05-02 Richard Biener <rguenther@suse.de>
10630 PR tree-optimization/80549
10631 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
10632 (cleanup_tree_cfg_noloop): Create forwarders to known loop
10633 headers if they do not have a preheader.
10635 2017-05-02 Martin Liska <mliska@suse.cz>
10638 * common.opt: Fix typo.
10639 * doc/invoke.texi: Likewise.
10641 2017-05-01 Jan Beulich <jbeulich@suse.com>
10643 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
10644 swapping, add (x,x,m,x,n) alternative.
10646 2017-05-01 Nathan Sidwell <nathan@acm.org>
10648 * calls.c (combine_pending_stack_adjustment_and_call): Remove
10649 unnecessary unadjusted_alignment check.
10651 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10654 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
10656 * gimplify.c (gimplify_cilk_detach): New function.
10657 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
10658 * tree-core.h: Document EXPR_CILK_SPAWN.
10659 * tree.h (EXPR_CILK_SPAWN): Define.
10661 2017-05-01 David Malcolm <dmalcolm@redhat.com>
10663 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
10664 to use new fixit_hint representation, using the "replace" logic.
10665 (get_line_span_for_fixit_hint): Likewise.
10666 (layout::print_any_fixits): Likewise.
10667 (selftest::test_one_liner_many_fixits): Rename to...
10668 (selftest::test_one_liner_many_fixits_1): ...this, and update
10669 comment and expected output to reflect that the multiple fix-it
10670 hints are now consolidated into one insertion.
10671 (selftest::test_one_liner_many_fixits_2): New test.
10672 (selftest::test_diagnostic_show_locus_one_liner): Update for
10674 (selftest::test_fixit_consolidation): Update for fix-it API
10676 * diagnostic.c (print_parseable_fixits): Likewise.
10677 * edit-context.c (edited_line::m_line_events): Convert from
10678 auto_vec <line_event *> to auto_vec <line_event>.
10679 (class line_event): Convert from abstract base class to a concrete
10680 class, taking over the role of replace_event.
10681 (class insert_event): Delete.
10682 (class replace_event): Rename to class line_event. Convert to
10684 (edit_context::add_fixits): Reimplement.
10685 (edit_context::apply_insert): Delete.
10686 (edit_context::apply_replace): Rename to...
10687 (edit_context::apply_fixit): ...this. Convert to half-open range.
10688 (edited_file::apply_insert): Delete.
10689 (edited_file::apply_replace): Rename to...
10690 (edited_file::apply_fixit): ...this.
10691 (edited_line::~edited_line): Drop deletion of events.
10692 (edited_line::apply_insert): Delete.
10693 (edited_line::apply_replace): Rename to...
10694 (edited_line::apply_fixit): ...this. Convert to half-open range.
10695 Update for change to type of m_line_events.
10696 * edit-context.h (edit_context::apply_insert): Delete.
10697 (edit_context::apply_replace): Rename to...
10698 (edit_context::apply_fixit): ...this.
10700 2017-05-01 Martin Sebor <msebor@redhat.com>
10702 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
10705 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
10708 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
10709 __get_cpuid_max returns 0.
10710 (__get_cpuid_count): Ditto.
10712 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
10714 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
10715 replacement expression is another instance of one of its arguments.
10717 2017-05-01 Jakub Jelinek <jakub@redhat.com>
10720 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
10721 check for stack push/pop autoinc.
10722 * config/i386/i386.c (ix86_agi_dependent): Return false
10723 if the only reason why modified_in_p returned true is that
10724 addr is SP based and set_insn is a push or pop.
10726 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10728 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
10731 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10734 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
10735 (account_size_time): Use two predicates - exec_pred and
10737 (evaluate_conditions_for_known_args): Compute both clause and
10739 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
10740 (inline_summary_t::duplicate): Update.
10741 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
10743 (compute_inline_parameters): Likewise.
10744 (estimate_edge_size_and_time): Update caluclation of time.
10745 (estimate_node_size_and_time): Compute both time and nonspecialized
10747 (estimate_ipcp_clone_size_and_time): Update.
10748 (inline_merge_summary): Update.
10749 (do_estimate_edge_time): Update.
10750 (do_estimate_edge_size): Update.
10751 (do_estimate_edge_hints): Update.
10752 (inline_read_section, inline_write_summary): Stream both new predicates.
10753 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
10755 (compute_inlined_call_time): Cleanup.
10756 (big_speedup_p): Update.
10757 (edge_badness): Update.
10758 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
10759 (size_time_entry): Replace predicate by exec_predicate and
10760 nonconst_predicate.
10761 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
10762 (estimate_edge_time): Return also nonspec_time.
10763 (reset_edge_growth_cache): Update.
10765 2017-04-29 Jakub Jelinek <jakub@redhat.com>
10767 PR rtl-optimization/80491
10768 * ifcvt.c (noce_process_if_block): When looking for x setter
10769 with missing else_bb, don't check only the insn right before
10770 cond_earliest, but look for the last insn that x is modified in
10771 within the same bb.
10773 PR rtl-optimization/80491
10774 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
10776 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
10778 PR tree-optimization/80487
10779 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
10781 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10783 PR tree-optimization/79697
10784 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
10785 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
10786 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
10788 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
10789 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
10791 2017-04-28 Martin Sebor <msebor@redhat.com>
10793 PR tree-optimization/80523
10794 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
10795 (init_target_to_host_charmap, target_to_host, target_strtol10): New
10797 (maybe_warn, format_directive, parse_directive): Use new functions.
10798 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
10800 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
10802 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
10804 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
10806 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
10807 target_header_dir): Set correctly.
10808 * configure: Regenerated.
10809 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
10810 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
10811 instead of SYSTEM_HEADER_DIR.
10813 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
10815 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
10816 (estimate_local_effects): Likewise.
10817 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
10818 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
10819 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
10820 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
10821 do_estimate_edge_time, estimate_edge_time): Likewise.
10822 * ipa-inline-analysis.c (estimate_node_size_and_time,
10823 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
10824 (estimate_time_after_inlining): Remove.
10826 2017-04-28 Martin Liska <mliska@suse.cz>
10828 * doc/gcov.texi: Enhance documentation of gcov.
10830 2017-04-28 Martin Liska <mliska@suse.cz>
10832 * doc/gcov.texi: Sort options in alphabetic order.
10833 * doc/gcov-dump.texi: Likewise.
10834 * doc/gcov-tool.texi: Likewise.
10835 * gcov.c (print_usage): Likewise.
10836 * gcov-dump.c (print_usage): Likewise.
10837 * gcov-tool.c (print_merge_usage_message): Likewise.
10838 (print_rewrite_usage_message): Likewise.
10839 (print_overlap_usage_message): Likewise.
10841 2017-04-28 Martin Liska <mliska@suse.cz>
10843 PR gcov-profile/53915
10844 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
10846 2017-04-28 Martin Liska <mliska@suse.cz>
10848 PR gcov-profile/79891
10849 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
10850 is marked by compiler as living on a line.
10851 (get_cycles_count): Remove usage of the union.
10852 (output_intermediate_file): Likewise.
10853 (find_source): Fix GNU coding style.
10854 (accumulate_line_counts): Remove old non-all block mode.
10855 (output_lines): Remove usage of the union.
10856 * profile.c (output_location): Include all BBs, even if
10857 belonging to a same line (and file) as a previous BB.
10859 2017-04-28 Martin Liska <mliska@suse.cz>
10861 * gcov.c (process_args): Handle new argument 'w'.
10862 (read_graph_file): Assign ID to BBs.
10863 (output_branch_count): Display BB # if verbose flag is set.
10864 (output_lines): Likewise for arcs.
10865 (print_usage): Add '--verbose' option help.
10866 * doc/gcov.texi: Document --verbose (-w) option.
10868 2017-04-28 Martin Liska <mliska@suse.cz>
10870 * gcov.c (struct block_location_info): New struct.
10871 (process_file): Fill up the new structure.
10872 (read_graph_file): Replace usage of encoding by the newly added
10874 (add_line_counts): Likewise.
10875 (accumulate_line_counts): Remove usage of the union.
10876 (function_info::function_info): New function.
10877 (function_info::~function_info): Likewise.
10878 (process_file): Call delete instead of release_function.
10879 (release_function): Release the function.
10880 (release_structures): Call delete instead of release_function.
10881 (solve_flow_graph): Replace usage of num_blocks.
10882 (find_exception_blocks): Likewise.
10883 (output_lines): Fix GNU coding style.
10885 2017-04-28 Martin Liska <mliska@suse.cz>
10888 * coverage.c (coverage_remove_note_file): New function.
10889 * coverage.h: Declare the function.
10890 * toplev.c (finalize): Clean if an error has been seen.
10892 2017-04-28 Martin Liska <mliska@suse.cz>
10894 PR gcov-profile/80031
10895 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
10896 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
10897 * gcov.c (read_graph_file): Read just number of blocks.
10898 * profile.c (branch_prob): Do not stream 0 flags per a basic
10901 2017-04-28 Martin Liska <mliska@suse.cz>
10903 * gcov-dump.c (tag_*): Add new argument to declarations.
10904 (dump_gcov_file): Likewise.
10905 (tag_blocks): Add and use new argument depth.
10906 (tag_arcs): Likewise.
10907 (tag_lines): Likewise.
10908 (tag_counters): Likewise.
10909 (tag_summary): Likewise.
10910 (dump_working_sets): Use depth to do a proper indentation.
10912 2017-04-28 Jakub Jelinek <jakub@redhat.com>
10915 * cgraph.h (symtab_node::debug_symtab): No longer inline.
10916 * symtab.c (symtab_node::debug_symtab): Move definition here.
10918 2017-04-28 Richard Biener <rguenther@suse.de>
10920 * lto-streamer.h (LTO_major_version): Bump to 7.
10922 2017-04-28 Richard Biener <rguenther@suse.de>
10924 * tree-vrp.c (assert_info): New struct.
10925 (add_assert_info): New helper.
10926 (register_edge_assert_for_2): Refactor to add asserts to a vector
10928 (register_edge_assert_for_1): Likewise.
10929 (register_edge_assert_for): Likewise.
10930 (finish_register_edge_assert_for): New helper actually registering
10931 asserts where live on edge.
10932 (find_conditional_asserts): Adjust.
10933 (find_switch_asserts): Likewise.
10934 (evrp_dom_walker::try_find_new_range): Generalize.
10935 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
10937 2017-04-27 Marek Polacek <polacek@redhat.com>
10940 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
10941 arg10 and arg11 to itype.
10943 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
10945 * doc/extend.texi (Object Size Checking): Improve grammar.
10947 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
10950 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
10951 that the logic for permitting reciprocal estimates matches that
10954 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10957 * tree.c (type_cache_hasher::equal): Only compare
10958 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
10959 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
10960 non-aggregate element types.
10961 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
10962 about the flag on ARRAY_TYPEs in the comment, formatting fix.
10964 2017-04-27 Richard Biener <rguenther@suse.de>
10966 PR middle-end/80533
10967 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
10968 stripping ARRAY_REFs from MEM_EXPR make sure we're not
10969 keeping a reference to a trailing array.
10971 2017-04-27 Richard Biener <rguenther@suse.de>
10973 PR middle-end/80539
10974 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
10975 being in loop-closed SSA form conservatively.
10976 (chrec_fold_multiply_poly_poly): Likewise.
10978 2017-04-27 Tamar Christina <tamar.christina@arm.com>
10980 PR middle-end/79665
10981 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
10982 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
10984 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10987 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
10988 (aarch64_function_arg_alignment): Return unsigned int again, but still
10989 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
10990 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
10991 Don't emit -Wpsabi note.
10992 (aarch64_function_arg_boundary): Likewise.
10993 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
10996 2017-04-26 Nathan Sidwell <nathan@acm.org>
10998 * tree.h (crc32_unsigned_n): Declare.
10999 (crc32_unsigned, crc32_unsigned): Make inline.
11000 * tree.c (crc32_unsigned_bits): Replace with ...
11001 (crc32_unsigned_n): ... this.
11002 (crc32_unsigned, crc32_byte): Remove.
11003 (crc32_string): Remove unnecessary braces.
11005 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
11007 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
11008 * ipa-inline-analysis.c (MAX_TIME): Remove.
11009 (account_size_time): Use sreal for time.
11010 (dump_inline_summary): Update.
11011 (estimate_function_body_sizes): Update.
11012 (estimate_edge_size_and_time): Update.
11013 (estimate_calls_size_and_time): Update.
11014 (estimate_node_size_and_time): Update.
11015 (inline_merge_summary): Update.
11016 (inline_update_overall_summary): Update.
11017 (estimate_time_after_inlining): Update.
11018 (inline_read_section): Update.
11019 (inline_write_summary): Update.
11020 * ipa-inline.c (compute_uninlined_call_time): Update.
11021 (compute_inlined_call_time): Update.
11022 (recursive_inlining): Update.
11023 (inline_small_functions): Update.
11024 (dump_overall_stats): Update.
11025 * ipa-inline.h: Include sreal.h.
11026 (size_time_entry): Turn time to sreal.
11027 (inline_summary): Turn self_time nad time to sreal.
11029 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
11031 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
11033 (sreal::stream_out, sreal::stream_in): New.
11034 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
11036 2017-04-25 Jakub Jelinek <jakub@redhat.com>
11038 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
11041 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
11044 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
11046 (dimode_scalar_chain::compute_convert_gain): Ditto.
11047 (dimode_scalar_chain::make_vector_copies): Ditto.
11048 (dimode_scalar_chain::convert_reg): Ditto.
11049 (dimode_scalar_chain::convert_insn): Ditto.
11050 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
11051 (VI248_AVX512BW_1): New mode iterator.
11052 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
11053 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
11056 2017-04-25 Martin Sebor <msebor@redhat.com>
11058 PR tree-optimization/80497
11059 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
11060 constants are representable in HOST_WIDE_INT.
11061 (parse_directive): Ditto.
11063 2017-04-25 Martin Sebor <msebor@redhat.com>
11066 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
11067 (new_zero_array): Adjust signature.
11068 (dom_info::dom_init): Used unsigned rather that size_t.
11069 (dom_info::dom_info): Same.
11071 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11072 Jakub Jelinek <jakub@redhat.com>
11075 * config/arm/arm.c: Include gimple.h.
11076 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
11077 returns negative, increment ncrn only if it returned positive.
11078 (arm_needs_doubleword_align): Return int instead of bool,
11079 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
11080 members, but if there is any such non-FIELD_DECL
11081 > PARM_BOUNDARY aligned decl, return -1 instead of false.
11082 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
11083 returns negative, increment nregs only if it returned positive.
11084 (arm_setup_incoming_varargs): Likewise.
11085 (arm_function_arg_boundary): Emit -Wpsabi note if
11086 arm_needs_doubleword_align returns negative, return
11087 DOUBLEWORD_ALIGNMENT only if it returned positive.
11089 2017-04-25 Marek Polacek <polacek@redhat.com>
11092 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
11093 first argument to type.
11095 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
11098 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
11099 type checks to test for compatibility instead of equality.
11101 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11102 Jakub Jelinek <jakub@redhat.com>
11105 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
11107 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
11108 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
11109 the alignment computation, but return their maximum in warn_alignment.
11110 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
11111 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
11113 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
11114 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
11117 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11119 * config/arc/simdext.md (dmpyh): Fix typo.
11121 2017-04-25 Richard Biener <rguenther@suse.de>
11123 PR tree-optimization/80492
11124 * alias.c (compare_base_decls): Handle registers with asm
11125 specification conservatively.
11126 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
11127 compare_base_decls returning dont-know properly.
11129 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11131 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
11132 (legitimate_offset_address_p): New function.
11133 (arc_legitimate_address_p): Use above function.
11135 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11137 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
11139 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11141 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
11142 ACCH registers whenever they are available.
11144 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11146 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
11147 double regs fix when not used.
11149 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11151 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
11153 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
11154 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
11156 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11158 * config/arc/arc.c (arc_output_addsi): Check for h-register class
11159 when emitting short ADD instructions.
11161 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11163 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
11165 (cmpsi_cc_c_insn): Likewise.
11166 (cbranchsi4_scratch): Compute proper instruction length using
11167 compact_hreg_operand.
11168 * config/arc/predicates.md (compact_hreg_operand): New predicate.
11170 2017-04-25 Richard Biener <rguenther@suse.de>
11172 PR middle-end/80509
11173 * passes.c (pass_manager::pass_manager): Initialize
11174 m_name_to_pass_map.
11176 2017-04-25 Richard Biener <rguenther@suse.de>
11178 PR tree-optimization/79201
11179 * tree-ssa-sink.c (statement_sink_location): Handle calls.
11181 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11184 * config/s390/vector.md: Split MEM->GPR vector moves for
11185 non-s_operand addresses.
11187 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11190 * config/s390/predicates.md (reload_const_wide_int_operand): New
11192 * config/s390/s390.md ("movti"): Remove d/P alternative.
11193 ("movti_bigconst"): New pattern definition.
11195 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11198 * s390-protos.h (s390_expand_cs_hqi): Removed.
11199 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
11200 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
11201 modes as well as CCZ1mode and CCZmode.
11202 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
11203 signature of s390_emit_compare_and_swap.
11204 (s390_expand_cs_hqi): Likewise, make static.
11205 (s390_expand_cs_tdsi): Generate an explicit compare before trying
11206 compare-and-swap, in some cases.
11207 (s390_expand_cs): Wrapper function.
11208 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
11210 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
11211 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
11212 patterns for small and large integers. Forbid symref memory operands.
11213 Move expander to s390.c. Require cc register.
11214 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
11215 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
11216 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
11217 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
11218 symref memory operands. Remove CC mode and call s390_match_ccmode
11220 ("atomic_exchange<mode>"): Allow and implement all integer modes.
11222 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11224 * config/s390/s390.md (define_peephole2): New peephole to help
11225 combining the load-and-test pattern with volatile memory.
11227 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11229 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
11230 with CCZmode for TARGET_Z196.
11232 2017-04-25 Jakub Jelinek <jakub@redhat.com>
11234 PR rtl-optimization/80501
11235 * combine.c (make_compound_operation_int): Set subreg_code to SET
11236 even for AND with mask of the sign bit of mode.
11238 PR rtl-optimization/80500
11239 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
11240 sum's initial value.
11242 2017-04-25 Julian Brown <julian@codesourcery.com>
11243 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11245 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
11247 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
11249 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
11251 2017-04-25 Julian Brown <julian@codesourcery.com>
11252 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11254 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
11255 (thunderx2t99_sha): New Reservation.
11257 2017-04-25 Julian Brown <julian@codesourcery.com>
11258 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11260 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
11261 type for 1-element load.
11263 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
11265 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
11267 2017-04-24 Martin Jambor <mjambor@suse.cz>
11269 PR tree-optimization/80293
11270 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
11271 char arrays not totally scalarizable if it is false.
11272 (analyze_all_variable_accesses): Pass correct value in the new
11273 parameter. Add a statistics counter.
11275 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
11277 PR middle-end/79931
11278 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
11280 2017-04-24 Richard Biener <rguenther@suse.de>
11282 PR tree-optimization/80494
11283 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
11284 out for complex types.
11286 2017-04-24 Richard Biener <rguenther@suse.de>
11288 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
11289 * tree-ssa-sccvn.c (print_scc): Print SCC size.
11290 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
11291 (DFS): Adjust and never fail.
11292 (sccvn_dom_walker::fail): Remove.
11293 (sccvn_dom_walker::before_dom_children): Adjust.
11294 (run_scc_vn): Likewise and never fail.
11295 * tree-ssa-pre.c (pass_pre::execute): Adjust.
11296 (pass_fre::execute): Likewise.
11298 2017-04-24 Richard Biener <rguenther@suse.de>
11300 PR tree-optimization/79725
11301 * tree-ssa-sink.c (statement_sink_location): Return whether
11302 failure reason was zero uses. Move that check later.
11303 (sink_code_in_bb): Deal with zero uses by removing the stmt
11306 2017-04-24 Richard Biener <rguenther@suse.de>
11309 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
11310 pointer-based references.
11312 2017-04-24 Richard Biener <rguenther@suse.de>
11315 * pass_manager.h (pass_manager::operator new): Remove.
11316 (pass_manager::operator delete): Likewise.
11317 * passes.c (pass_manager::operator new): Remove.
11318 (pass_manager::operator delete): Likewise.
11319 (pass_manager::pass_manager): Zero individual pass members.
11321 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
11324 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
11325 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
11326 Check "XEXP (src, 1)" operand here.
11327 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
11328 Check "XEXP (src, 1)" operand here.
11329 (dimode_scalar_chain::make_vector_copies): Detect count register
11330 of a shift instruction. Zero extend count register from QImode
11331 to DImode to satisfy vector shift pattern count operand predicate.
11332 Substitute vector shift count operand with a DImode copy.
11333 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
11336 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
11338 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
11339 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
11340 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
11341 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
11342 (UNSPEC_NOREX_MEM): Remove definition.
11344 2017-04-21 Richard Biener <rguenther@suse.de>
11346 PR tree-optimization/79547
11347 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
11348 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
11349 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
11350 without any constraints.
11352 2017-04-21 Richard Biener <rguenther@suse.de>
11354 PR tree-optimization/78847
11355 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
11357 2017-04-21 Richard Biener <rguenther@suse.de>
11359 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
11360 (build_distinct_type_copy): Likewise.
11361 (build_variant_type_copy): Likewise.
11362 * tree.c (build_qualified_type): Pass down mem-stat info.
11363 (build_distinct_type_copy): Likewise.
11364 (build_variant_type_copy): Likewise.
11366 2017-04-21 Richard Biener <rguenther@suse.de>
11368 PR tree-optimization/80237
11369 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
11371 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
11372 for a simplified result.
11374 2016-04-21 Richard Biener <rguenther@suse.de>
11376 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
11377 sth as strict as a simple_iv but a chrec without symbols and an
11378 operand defined in the loop we are peeling (and not some subloop).
11379 (propagate_constants_for_unrolling): Propagate all constants.
11381 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
11384 * config/i386/i386.c (print_reg): Remove assert for disalowed
11385 regno values, call output_operand_lossage instead.
11387 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
11390 * config/i386/constraints.md (Yc): New register constraint.
11391 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
11392 Use Yc constraint for alternative 2 of operand 0. Remove
11393 preferred_for_speed attribute.
11395 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
11397 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
11398 lastprivate clauses in SIMT case.
11400 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
11402 * doc/invoke.texi (-Wextra-semi): Document new warning option.
11404 2017-04-20 Richard Biener <rguenther@suse.de>
11406 PR tree-optimization/57796
11407 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
11408 as N scalar stores.
11409 (vect_model_load_cost): Cost gathers as N scalar loads.
11411 2017-04-20 Richard Biener <rguenther@suse.de>
11413 * ggc-page.c (ggc_allocated_p): Rename to ...
11414 (safe_lookup_page_table_entry): ... this and return the lookup
11416 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
11418 2017-04-20 Richard Biener <rguenther@suse.de>
11420 PR tree-optimization/80453
11421 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
11422 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
11423 from the conditions.
11424 (vn_phi_eq): Pass them down.
11425 (vn_phi_lookup): Record them.
11426 (vn_phi_insert): Likewise.
11428 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
11430 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
11431 uninitialized variable warning to avoid buffer overrun.
11433 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
11436 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
11437 is suppressed for '{ 0 }' in C.
11439 2017-04-20 Jakub Jelinek <jakub@redhat.com>
11441 * BASE-VER: Set to 8.0.0.
11443 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
11445 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
11446 priority .init_array and .fini_array section with SECTION_NOTYPE
11449 2017-04-20 Jakub Jelinek <jakub@redhat.com>
11451 PR middle-end/80423
11452 * tree.h (build_array_type): Add typeless_storage default argument.
11453 * tree.c (type_cache_hasher::equal): Also compare
11454 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
11455 (build_array_type): Add typeless_storage argument, set
11456 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
11458 (build_nonshared_array_type): Adjust build_array_type_1 caller.
11459 (build_array_type): Likewise. Add typeless_storage argument.
11461 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
11462 Jakub Jelinek <jakub@redhat.com>
11464 PR tree-optimization/80426
11465 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
11466 operation on symbolic operands, also compute the overflow for the
11467 invariant part when the operation degenerates into a negation.
11469 2017-04-19 Jakub Jelinek <jakub@redhat.com>
11472 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
11473 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
11476 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
11478 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
11481 * config/avr/avr.c (tree.h): Include it.
11482 (cgraph.h): Include it.
11483 (avr_encode_section_info): Don't warn for uninitialized progmem
11484 variable if it's just an alias.
11486 2017-04-19 Richard Biener <rguenther@suse.de>
11489 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
11490 when needed by AutoPGO.
11492 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
11495 * doc/lto.texi: Remove an extra 'that'.
11497 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
11499 PR rtl-optimization/80429
11500 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
11501 are only used in debug insns.
11503 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
11504 Vladimir Makarov <vmakarov@redhat.com>
11506 * config/sparc/predicates.md (input_operand): Add comment. Return
11507 true for any memory operand when LRA is in progress.
11508 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
11510 2017-04-18 Jeff Law <law@redhat.com>
11513 * mips.md ({return,simple_return}_internal): Do not overwrite
11516 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11518 PR tree-optimization/80443
11519 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
11520 instead of adding 1, subtract -1 and similarly instead of subtracting
11523 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
11525 PR rtl-optimization/80357
11526 * haifa-sched.c (tmp_bitmap): New variable.
11527 (model_recompute): Handle duplicate use records.
11528 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
11529 (free_global_sched_pressure_data): Free it.
11531 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
11534 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
11535 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
11536 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
11537 instead of SYSTEM_HEADER_DIR.
11539 2017-04-18 Jeff Law <law@redhat.com>
11541 PR middle-end/80422
11542 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
11543 predecessors after walking up the insn chain.
11545 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11548 * dwarf2out.c (modified_type_die): Try harder not to emit internal
11549 sizetype type into debug info.
11551 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
11554 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
11555 unneeded test for TARGET_UPPER_REGS_SF.
11556 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
11558 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11561 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
11562 instead of gsi_after_labels.
11564 2017-04-18 Jeff Law <law@redhat.com>
11566 * regcprop.c (maybe_mode_change): Avoid creating copies of the
11570 2017-04-13 Jeff Law <law@redhat.com>
11571 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11572 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11574 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
11577 * config/avr/avr.c (intl.h): Include it.
11578 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
11580 2017-04-18 Martin Liska <mliska@suse.cz>
11582 PR gcov-profile/78783
11583 * gcov-tool.c (gcov_output_files): Validate that destination
11584 file is either removed by the tool or by a user.
11586 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
11587 Guy Benyei <guybe@mellanox.com>
11589 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
11590 block, and do not negate it, the stored id is already negative.
11592 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
11594 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
11596 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
11599 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
11600 masks of options that should be turned off if the VSX vector
11601 options are turned off.
11602 (OTHER_P8_VECTOR_MASKS): Likewise.
11603 (OTHER_VSX_VECTOR_MASKS): Likewise.
11604 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
11605 rs6000_disable_incompatible_switches to validate no type switches
11607 (rs6000_incompatible_switch): New function to disallow turning on
11608 other vector options if -mno-vsx, -mno-power8-vector, or
11609 -mno-power9-vector are specified.
11611 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11613 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
11615 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11617 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
11618 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
11619 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
11620 (ARG_POINTER_CFA_OFFSET): Likewise.
11622 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11624 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
11625 conditions to take advantage of various optimizations.
11627 2017-04-13 Jeff Law <law@redhat.com>
11629 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11630 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11631 (zero_extendsidi2_dext): Likewise.
11633 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11636 * fold-const.c (fold_ternary_loc): Revert
11637 use op0 instead of fold_convert_loc (loc, type, arg0) part of
11640 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
11642 PR rtl-optimization/80343
11643 * lra-remat.c (update_scratch_ops): Assign original hard reg to
11644 new scratch pseudo.
11646 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
11649 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
11650 to ubsan_encode_value.
11652 2017-04-13 Jeff Law <law@redhat.com>
11654 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
11655 appearing in DEBUG_INSNs.
11657 2017-04-13 Martin Liska <mliska@suse.cz>
11659 PR gcov-profile/80413
11660 * gcov-io.c (gcov_write_string): Copy to buffer just when
11661 allocated size is greater than zero.
11663 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11666 * dwarf2out.c (decls_for_scope): Ignore declarations of
11667 current_function_decl in BLOCK_NONLOCALIZED_VARS.
11669 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
11672 * ipa-visibility.c (non_local_p): Fix typos.
11673 (localize_node): When localizing symbol in same comdat group,
11674 dissolve the group only when we know external symbols are going
11676 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
11678 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11680 PR tree-optimization/79390
11681 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
11682 order does not result in usable sequence, retry with reversed operand
11688 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
11689 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
11690 op0 instead of fold_convert_loc (loc, type, arg0).
11692 2017-04-12 Jeff Law <law@redhat.com>
11694 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
11695 has a delay slot in the generated code.
11697 * config/cris/cris.md (cris_preferred_reload_class): Return
11698 GENNONACR_REGS rather than GENERAL_REGS.
11700 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11703 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
11704 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
11705 signedness of the result type.
11707 2017-04-12 Richard Biener <rguenther@suse.de>
11708 Jeff Law <law@redhat.com>
11710 PR tree-optimization/80359
11711 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
11712 trim stores to TARGET_MEM_REFs.
11714 2017-04-12 Richard Biener <rguenther@suse.de>
11716 PR tree-optimization/79390
11717 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
11718 threading case even more.
11720 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
11723 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
11724 for quad_address_p for TImode, instead of just not indexed_address.
11726 2017-04-12 Richard Biener <rguenther@suse.de>
11727 Bernd Edlinger <bernd.edlinger@hotmail.de>
11729 PR middle-end/79671
11730 * alias.c (component_uses_parent_alias_set_from): Handle
11731 TYPE_TYPELESS_STORAGE.
11732 (get_alias_set): Likewise.
11733 * tree-core.h (tree_type_common): Add typeless_storage flag.
11734 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
11735 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
11736 for types containing members with TYPE_TYPELESS_STORAGE.
11737 (place_field): Likewise.
11738 (layout_type): Likewise for ARRAY_TYPE.
11739 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
11740 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
11741 TYPE_TYPELESS_STORAGE.
11742 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
11744 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11747 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
11748 first argument to type.
11750 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11754 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
11755 CONST0_RTX (mode) rather than const0_rtx where appropriate.
11756 (rs6000_expand_binop_builtin): Likewise.
11757 (rs6000_expand_ternop_builtin): Likewise; also add missing
11758 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
11759 vshasigma built-ins.
11760 * doc/extend.texi: Document that vec_xxpermdi's third argument
11761 must be a constant.
11763 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
11765 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
11766 Use shift_const cost parameter when calculating gain of STV shifts.
11768 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
11770 PR rtl-optimization/70478
11771 * lra-constraints.c (process_alt_operands): Check memory for
11772 disfavoring memory insn operand.
11774 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11776 PR middle-end/80100
11777 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
11778 left shift in unsigned HOST_WIDE_INT type.
11780 PR rtl-optimization/80385
11781 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
11782 (not (neg X)) into (plus X -1) for complex or non-integral modes.
11785 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
11786 if they have any depend clauses.
11788 2017-04-11 Martin Liska <mliska@suse.cz>
11791 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
11792 * ipa-split.c (split_function): Create a local comdat symbol
11793 if caller is in a comdat group.
11795 2017-04-11 Martin Liska <mliska@suse.cz>
11798 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
11801 2017-04-11 Martin Sebor <msebor@redhat.com>
11803 PR middle-end/80364
11804 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
11805 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
11807 (directive::set_width, directive::set_precision, format_character):
11809 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
11812 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
11815 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
11816 conflict, set target->arch_name instead of target->cpu_name.
11818 2017-04-11 Richard Biener <rguenther@suse.de>
11820 PR tree-optimization/80374
11821 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
11822 build_zero_cst, remove fold_convertible_p check again.
11824 2017-04-11 Martin Liska <mliska@suse.cz>
11827 * ubsan.c (instrument_object_size): Do not instrument register
11830 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11833 * config/i386/i386-builtin-types.def
11834 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
11835 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
11836 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
11837 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
11838 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
11839 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
11840 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
11841 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
11842 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
11843 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
11844 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
11845 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
11846 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
11847 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
11848 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
11849 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
11850 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
11851 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
11852 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
11853 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
11854 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
11855 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
11856 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
11857 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
11858 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
11859 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
11860 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
11861 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
11862 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
11863 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
11864 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
11865 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
11866 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
11867 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
11868 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
11869 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
11870 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
11871 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
11872 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
11873 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
11874 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
11875 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
11876 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
11877 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
11878 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
11879 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
11881 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
11882 flag to second_arg_count, handle 4 argument function type _COUNT
11883 aliases, handle second_arg_count on second argument rather than last.
11885 2017-04-10 Jeff Law <law@redhat.com>
11887 PR tree-optimization/80374
11888 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
11889 record anything if we can not convert integer_zero_node to the
11892 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
11895 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11896 Enhance special handling given to the TARGET_P9_MINMAX option in
11897 relation to certain other options.
11899 2017-04-10 Bin Cheng <bin.cheng@arm.com>
11901 PR tree-optimization/80153
11902 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
11903 remove POINTER_PLUS_EXPR's base part directly, rather than through
11906 2017-04-10 Richard Biener <rguenther@suse.de>
11907 Bin Cheng <bin.cheng@arm.com>
11909 PR tree-optimization/80153
11910 * tree-affine.c (aff_combination_to_tree): Get base pointer from
11911 the first element of pointer type aff_tree. Build result expr in
11913 (add_elt_to_tree): Convert to type unconditionally. Remove other
11914 fold_convert calls.
11915 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
11916 (rewrite_use_nonlinear_expr): Check invariant using iv information.
11918 2017-04-10 Richard Biener <rguenther@suse.de>
11920 * tree-ssa-structalias.c (find_func_aliases): Properly handle
11923 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
11925 PR rtl-optimization/70478
11926 * lra-constraints.c (curr_small_class_check): New.
11927 (update_and_check_small_class_inputs): New.
11928 (process_alt_operands): Update curr_small_class_check. Disfavor
11929 alternative insn memory operands. Check available regs for small
11932 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
11935 * config/mips/mips.opt (-mvirt): Update description.
11936 * doc/invoke.texi (-mvirt): Likewise.
11938 2017-04-10 Richard Biener <rguenther@suse.de>
11940 PR middle-end/80362
11941 * fold-const.c (fold_binary_loc): Look at unstripped ops when
11942 looking for NEGATE_EXPR in -A / -B to A / B folding.
11944 2017-04-10 Martin Liska <mliska@suse.cz>
11946 PR gcov-profile/80224
11947 * gcov.c (print_usage): Fix usage string.
11948 (get_gcov_intermediate_filename): Remove.
11949 (output_gcov_file): Use both for normal and intermediate format.
11950 (generate_results): Do not initialize special file for
11951 intermediate format.
11953 2017-04-10 Richard Biener <rguenther@suse.de>
11955 PR tree-optimization/80304
11956 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
11959 2017-04-10 Nathan Sidwell <nathan@acm.org>
11962 * config/rs6000/rs6000.c (rs6000_vector_type): New.
11963 (rs6000_init_builtins): Use it.
11965 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11967 * config/arm/arm.md (<mrc>): Add mode to SET source.
11968 (<mrrc>): Likewise.
11970 2017-04-10 Richard Biener <rguenther@suse.de>
11972 PR middle-end/80344
11973 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
11975 2017-04-10 Jakub Jelinek <jakub@redhat.com>
11978 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
11979 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
11980 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
11981 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
11982 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
11983 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
11984 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
11985 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
11986 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
11987 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
11988 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
11989 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
11990 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
11991 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
11992 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
11993 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
11994 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
11995 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
11996 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
11997 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
11998 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
11999 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
12000 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
12002 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
12004 PR rtl-optimization/70478
12005 * lra-constraints.c: Reverse the last patch.
12007 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
12009 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
12010 Add comment for WCHAR_T.
12012 2017-04-08 Martin Liska <mliska@suse.cz>
12015 2017-04-07 Martin Liska <mliska@suse.cz>
12018 * ipa-split.c (split_function): Add function part to a same comdat
12021 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12024 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
12026 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
12028 * rs6000/rs6000.c (vec_load_pendulum): Rename...
12029 (vec_pairing): ...to this.
12030 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
12031 (rs6000_sched_init): Adjust for name change.
12032 (struct rs6000_sched_context): Likewise.
12033 (rs6000_init_sched_context): Likewise.
12034 (rs6000_set_sched_context): Likewise.
12036 2017-04-07 Jakub Jelinek <jakub@redhat.com>
12042 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
12044 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
12045 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
12046 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
12048 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
12050 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
12052 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
12054 PR rtl-optimization/70703
12055 * ira-color.c (update_conflict_hard_regno_costs): Use
12056 int64_t instead of HOST_WIDE_INT.
12058 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
12060 PR rtl-optimization/70478
12061 * lra-constraints.c (process_alt_operands): Disfavor alternative
12062 insn memory operands.
12064 2017-04-07 Jeff Law <law@redhat.com>
12066 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
12067 CALL and NOTE_INSN_CALL_ARG_LOCATION.
12069 2017-04-07 Martin Liska <mliska@suse.cz>
12072 * config/aarch64/aarch64.c (aarch64_process_target_attr):
12073 Show error message instead of an ICE.
12075 2017-04-07 Martin Liska <mliska@suse.cz>
12078 * ipa-split.c (split_function): Add function part to a same comdat
12081 2017-04-07 Richard Biener <rguenther@suse.de>
12083 PR middle-end/80341
12084 * tree.c (get_unwidened): Also handle ! for_type case for
12086 * convert.c (do_narrow): Split out from ...
12087 (convert_to_integer_1): ... here. Do not pass final truncation
12088 type to get_unwidened for TRUNC_DIV_EXPR.
12090 2017-04-07 Richard Biener <rguenther@suse.de>
12092 * tree-affine.c (wide_int_ext_for_comb): Take type rather
12094 (aff_combination_const): Adjust.
12095 (aff_combination_scale): Likewise.
12096 (aff_combination_add_elt): Likewise.
12097 (aff_combination_add_cst): Likewise.
12098 (aff_combination_convert): Likewise.
12099 (add_elt_to_tree): Likewise. Remove unused argument.
12100 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
12102 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
12104 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
12106 * config/arm/arm.c (arm_default_short_enums): Use
12107 ARM_DEFAULT_SHORT_ENUMS.
12108 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
12110 2017-04-06 Jakub Jelinek <jakub@redhat.com>
12113 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
12114 members with redundant out-of-class redeclaration.
12116 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
12119 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
12120 * config/i386/i386.md (*zero_extendsidi2):
12121 Add (?*x,*x) and (?*v,*v) alternatives.
12123 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
12126 * config/i386/i386.c (ix86_expand_builtin)
12127 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
12128 mode from insn data. Convert operands to insn operand mode.
12129 Copy operands that don't satisfy insn predicate to a register.
12131 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
12133 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
12136 2017-04-06 Richard Biener <rguenther@suse.de>
12138 PR tree-optimization/80334
12139 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
12140 preserve alignment of accesses.
12142 2017-04-06 Richard Biener <rguenther@suse.de>
12144 PR tree-optimization/80262
12145 * tree-sra.c (build_ref_for_offset): Preserve address-space
12147 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
12148 Drop useless address-space information on MEM_REF offsets.
12150 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
12152 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
12154 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
12156 PR rtl-optimization/70703
12157 * ira-color.c (update_conflict_hard_regno_costs): Use
12158 HOST_WIDE_INT instead of long.
12160 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
12163 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
12164 not defined for x86_64 target. Add -mmmx target option when __SSE2__
12166 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
12167 for x86_64 target. Handle -m3dnowa option.
12169 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
12171 PR rtl-optimization/70703
12172 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
12173 (update_conflict_hard_regno_costs): Use long instead of unsigned
12174 arithmetic for cost calculation.
12176 2017-04-05 Jakub Jelinek <jakub@redhat.com>
12177 Bernd Edlinger <bernd.edlinger@hotmail.de>
12180 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
12183 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
12186 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
12187 ptr_mode with Pmode throughout.
12188 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
12189 into probe_stack_range and use DImode.
12191 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
12194 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
12195 call_eh_return is true.
12197 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12199 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
12200 Initialize last_match_fntype_index.
12202 2017-04-05 Jakub Jelinek <jakub@redhat.com>
12205 * tree-nvr.c: Include internal-fn.h.
12206 (pass_return_slot::execute): Ignore internal calls without
12209 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12210 Richard Biener <rguenther@suse.de>
12213 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
12214 captures used multiple times, except for the last use.
12215 * generic-match-head.c: Include gimplify.h.
12217 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12219 PR tree-optimization/79390
12220 * target.h (struct noce_if_info): Declare.
12221 * targhooks.h (default_noce_conversion_profitable_p): Declare.
12222 * target.def (noce_conversion_profitable_p): New target hook.
12223 * ifcvt.h (struct noce_if_info): New type, moved from ...
12224 * ifcvt.c (struct noce_if_info): ... here.
12225 (noce_conversion_profitable_p): Renamed to ...
12226 (default_noce_conversion_profitable_p): ... this. No longer
12228 (noce_try_store_flag_constants, noce_try_addcc,
12229 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
12230 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
12231 instead of noce_conversion_profitable_p.
12232 * config/i386/i386.c: Include ifcvt.h.
12233 (ix86_option_override_internal): Don't override
12234 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
12235 (ix86_noce_conversion_profitable_p): New function.
12236 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
12237 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
12238 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
12239 * doc/tm.texi: Regenerated.
12241 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12243 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
12246 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
12249 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
12250 instructions for small multiply cores.
12252 2017-04-04 Jeff Law <law@redhat.com>
12254 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
12256 (mips_expand_vec_perm_const): Initialize elements in orig_perm
12257 that are not set by the loop over the elements.
12259 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12262 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
12263 int mode, convert_modes it to mode as unsigned, otherwise use
12264 lowpart_subreg to mode rather than SImode.
12265 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
12266 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
12267 Use DImode instead of SImode for the shift count operand.
12268 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
12271 2017-04-04 Richard Biener <rguenther@suse.de>
12273 PR middle-end/80281
12274 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
12275 arithmetic done for the negate or the plus. Simplify.
12276 (A - (-B) -> A + B): Likewise.
12277 * fold-const.c (split_tree): Make sure to not negate pointers.
12279 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
12281 PR rtl-optimization/60818
12282 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
12283 a compare of comparisons with the thing compared if this results
12284 in a different machine mode.
12286 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
12288 * alias.c (base_alias_check): Fix typo in comment.
12289 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
12290 * cgraphunit.c (symbol_table::compile): Likewise.
12291 * collect2.c (maybe_run_lto_and_relink): Likewise.
12292 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
12293 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
12294 * config/avr/avr.c (avr_map_op_t): Likewise.
12295 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
12296 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
12297 * config/epiphany/epiphany.md (movcc): Likewise.
12298 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
12299 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
12301 * config/mips/mips.c (mips_save_restore_reg): Likewise.
12302 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
12303 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
12304 * config/sh/sh.c (sh_rtx_costs): Likewise.
12305 * fold-const.c (fold_truth_andor): Likewise.
12306 * genautomata.c (collapse_flag): Likewise.
12307 * gengtype.h (struct type::u::s): Likewise.
12308 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
12309 * input.c (FORMAT_AMOUNT): Likewise.
12310 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
12311 (known_aggs_to_agg_replacement_list): Likewise.
12312 * ipa-inline-analysis.c: Likewise.
12313 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
12314 * ipa-polymorphic-call.c
12315 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
12316 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
12317 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
12319 * modulo-sched.c (apply_reg_moves): Likewise.
12320 * omp-expand.c (build_omp_regions_1): Likewise.
12321 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
12322 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
12323 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
12324 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
12325 * value-prof.c: Likewise.
12326 * var-tracking.c (val_reset): Likewise.
12328 2017-04-03 Richard Biener <rguenther@suse.de>
12330 PR tree-optimization/80275
12331 * fold-const.c (split_address_to_core_and_offset): Handle
12334 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
12336 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
12337 descriptors is at least equal to that of functions.
12339 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
12341 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
12343 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
12346 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
12347 (mov<IMOD4:mode>): New expander.
12348 (*mov<IMOD4:mode>_internal): New insn and split pattern.
12350 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
12352 PR rtl-optimization/79405
12353 * fwprop.c (propagations_left): New variable.
12354 (forward_propagate_into): Decrement it.
12355 (fwprop_init): Initialize it.
12356 (fw_prop): If the variable has reached zero, stop propagating.
12357 (fwprop_addr): Ditto.
12359 2017-03-31 Jakub Jelinek <jakub@redhat.com>
12362 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
12363 a FUNCTION_DECL, pass it as decl instead of origin to
12366 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
12368 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
12371 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
12374 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
12375 TARGET_VSX_SMALL_INTEGER.
12377 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12379 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
12380 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
12382 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
12384 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
12385 extraction from odd-numbered MSA register.
12387 2017-03-31 Jakub Jelinek <jakub@redhat.com>
12389 PR middle-end/80173
12390 * expmed.c (store_bit_field_1): Don't attempt to create
12391 a word subreg out of hard registers wider than word if they
12392 have HARD_REGNO_NREGS of 1 for their mode.
12394 PR middle-end/80163
12395 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
12396 conversions to integer types wider than word and pointer.
12399 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
12400 (rtx_equal_for_cselib_p): Pass 0 to it.
12401 * cselib.c (cselib_hasher::equal): Likewise.
12402 (rtx_equal_for_cselib_1): Add depth argument. If depth
12403 is 128, don't look up VALUE locs and punt. Increment
12404 depth in recursive calls when walking VALUE locs.
12406 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
12408 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
12409 (make_gcov_file_name): Use the canonical path name for generating
12411 (read_line): Fix handling of files with ascii null bytes.
12413 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
12415 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
12416 to initialise a vector register instead
12417 of using a const_int.
12419 2017-03-30 Jakub Jelinek <jakub@redhat.com>
12421 PR translation/80189
12422 * gimplify.c (omp_default_clause): Use %qs instead of %s in
12423 diagnostic messages.
12425 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
12428 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
12429 (dfp_diex_<mode>): Update mode of operand 1.
12430 * doc/extend.texi (dxex, dxexq): Document change to return type.
12431 (diex, diexq): Document change to argument type.
12433 2017-03-30 Martin Jambor <mjambor@suse.cz>
12436 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
12437 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
12438 it reflects the signature changes performed at the callee side.
12439 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
12440 to cgraph_build_function_type_skip_args.
12441 (build_function_decl_skip_args): Adjust call to the above function.
12443 2017-03-30 Jakub Jelinek <jakub@redhat.com>
12446 * config/i386/sse.md
12447 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
12448 register as dest whenever it is a MEM not rtx_equal_p to the
12449 corresponding dup operand, and when forcing into reg move the
12450 reg into the memory afterwards.
12451 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
12452 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
12453 for the force_reg mode.
12454 (avx512vl_vextractf128<mode>): Use register as dest either
12455 always when a MEM, or when it is a MEM not rtx_equal_p to the
12456 corresponding dup operand, or even not when it is a CONST_VECTOR
12457 depending on the mode and lo vs. hi.
12458 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
12460 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
12461 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
12462 Likewise. Require that operands[2] is even.
12463 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
12464 Remove extraneous parens. Require that operands[2] is a multiple
12466 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
12467 operands[0] is a MEM if <mask_applied>, the predicates/constraints
12468 disallow memory then.
12470 2017-03-30 Richard Biener <rguenther@suse.de>
12472 PR tree-optimization/77498
12473 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
12474 to non-constants over backedges.
12476 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
12478 PR rtl-optimization/80233
12479 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
12480 as last_combined_insn. Do not test for BARRIER_P separately.
12482 2017-03-29 Andreas Schwab <schwab@suse.de>
12485 * calls.c (prepare_call_address): Convert funexp to Pmode before
12486 copying to temp reg.
12488 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12490 PR tree-optimization/80158
12491 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
12492 Handle possible future case of more than one alternate
12494 (replace_rhs_if_not_dup): Likewise.
12495 (replace_one_candidate): Likewise.
12497 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
12499 PR rtl-optimization/80193
12500 * ira.c (ira): Do not check allocation for LRA.
12502 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
12504 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
12505 (nvptx_output_simt_exit): Declare.
12506 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
12507 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
12508 (init_softstack_frame): Move initialization of crtl->is_leaf to...
12509 (nvptx_declare_function_name): ...here. Emit declaration of local
12510 memory space buffer for omp_simt_enter insn.
12511 (nvptx_output_unisimt_switch): New.
12512 (nvptx_output_softstack_switch): New.
12513 (nvptx_output_simt_enter): New.
12514 (nvptx_output_simt_exit): New.
12515 * config/nvptx/nvptx.h (struct machine_function): New fields
12516 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
12517 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
12518 (UNSPECV_SIMT_EXIT): Ditto.
12519 (omp_simt_enter_insn): New insn.
12520 (omp_simt_enter): New expansion.
12521 (omp_simt_exit): New insn.
12522 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
12524 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
12525 (expand_GOMP_SIMT_ENTER_ALLOC): New.
12526 (expand_GOMP_SIMT_EXIT): New.
12527 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
12528 (GOMP_SIMT_ENTER_ALLOC): Ditto.
12529 (GOMP_SIMT_EXIT): Ditto.
12530 * target-insns.def (omp_simt_enter): New insn.
12531 (omp_simt_exit): Ditto.
12532 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
12534 (lower_rec_simd_input_clauses): Implement SIMT privatization.
12535 (lower_rec_input_clauses): Likewise.
12536 (lower_lastprivate_clauses): Handle SIMT privatization.
12538 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
12539 (ompdevlow_adjust_simt_enter): New.
12540 (find_simtpriv_var_op): New.
12541 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
12542 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
12544 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
12545 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
12546 (copy_decl_for_dup_finish): Ditto.
12548 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
12550 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
12553 * config/i386/i386.c (ix86_option_override_internal): Always
12554 allow -mpreferred-stack-boundary=3 for 64-bit targets.
12556 2017-03-28 Bin Cheng <bin.cheng@arm.com>
12558 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
12560 2017-03-28 Bin Cheng <bin.cheng@arm.com>
12562 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
12563 mark new edge's irreducible flag accordign to it.
12564 (vect_do_peeling): Check loop preheader edge's irreducible flag
12565 and pass it to function slpeel_add_loop_guard.
12567 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
12569 PR tree-optimization/80218
12570 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
12571 Update block frequencies and counts.
12573 2017-03-28 Richard Biener <rguenther@suse.de>
12575 PR tree-optimization/78644
12576 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
12577 of a simplification result we may not use it at all.
12579 2017-03-28 Richard Biener <rguenther@suse.de>
12582 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
12583 without arguments, generate default definition of a SSA name.
12585 2017-03-28 Richard Biener <rguenther@suse.de>
12587 PR middle-end/80222
12588 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
12589 TYPE_REF_CAN_ALIAS_ALL references.
12590 * fold-const.c (fold_indirect_ref_1): Likewise.
12592 2017-03-28 Martin Liska <mliska@suse.cz>
12595 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
12596 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
12598 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12599 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
12601 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
12602 (EXTRA_SPECS): Define.
12603 (SUBTARGET_EXTRA_SPECS): Likewise.
12604 (SUBTARGET_CPP_SPEC): Likewise.
12605 * config/arc/elf.h (EXTRA_SPECS): Renamed to
12606 SUBTARGET_EXTRA_SPECS.
12607 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
12609 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12611 * config/arc/simdext.md (vst64_insn): Update pattern.
12612 (vld32wh_insn): Likewise.
12613 (vld32wl_insn): Likewise.
12614 (vld64_insn): Likewise.
12615 (vld32_insn): Likewise.
12617 2017-03-28 Marek Polacek <polacek@redhat.com>
12620 * fold-const.c (fold_comparison): Use protected_set_expr_location
12621 instead of SET_EXPR_LOCATION.
12623 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
12625 * tree.c (add_expr): Avoid name lookup warning.
12627 2017-03-27 Jeff Law <law@redhat.com>
12629 PR tree-optimization/80216
12630 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
12631 function name. Limit recursion depth.
12632 (record_temporary_equivalences): Corresponding changes.
12634 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
12636 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
12639 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12642 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
12644 * cfgcleanup.c (reg_note_cfa_p): New array.
12645 (insns_have_identical_cfa_notes): New function.
12646 (old_insns_match_p): Don't cross-jump in between /f
12647 and non-/f instructions. If both i1 and i2 are frame related,
12648 verify all CFA notes, their order and content.
12650 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
12653 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
12654 HImode and SImode with zero extend to DImode to one insn.
12655 (bswap<mode>2_extenddi): Likewise.
12656 (bswapsi2_extenddi): Likewise.
12657 (bswaphi2_extendsi): Likewise.
12658 (bswaphi2): Combine bswap HImode and SImode into one insn.
12659 Separate memory insns from swapping register.
12660 (bswapsi2): Likewise.
12661 (bswap<mode>2): Likewise.
12662 (bswaphi2_internal): Delete, no longer used.
12663 (bswapsi2_internal): Likewise.
12664 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
12665 store, and gpr<-gpr swap insns.
12666 (bswap<mode>2_store): Likewise.
12667 (bswaphi2_reg): Register only splitter, combine with the splitter.
12668 (bswaphi2 splitter): Likewise.
12669 (bswapsi2_reg): Likewise.
12670 (bswapsi2 splitter): Likewise.
12671 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
12672 the insns into load, store, and register/register insns.
12673 (bswapdi2_ldbrx): Likewise.
12674 (bswapdi2_load): Likewise.
12675 (bswapdi2_store): Likewise.
12676 (bswapdi2_reg): Likewise.
12678 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
12680 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
12681 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
12683 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12686 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
12688 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
12689 special handling for target option conflicts between dform
12690 options (-mpower9-dform, -mpower9-dform-vector,
12691 -mpower9-dform-scalar) and -mno-direct-move.
12693 2017-03-27 Richard Biener <rguenther@suse.de>
12695 PR tree-optimization/80181
12696 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
12698 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12700 * config/arc/predicates.md (move_double_src_operand): Replace the
12701 call to move_double_src_operand with a call to address_operand.
12703 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12705 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
12706 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
12707 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
12709 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12711 * config/arc/predicates.md (long_immediate_loadstore_operand):
12712 Consider scaled addresses cases.
12714 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12716 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
12717 restored when in interrupt.
12718 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
12719 doesn't have delay slot.
12721 2017-03-27 Richard Biener <rguenther@suse.de>
12724 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
12725 inlined thunk clones.
12727 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12730 * asan.c (instrument_derefs): Copy over last operand from
12731 original COMPONENT_REF to the new COMPONENT_REF with
12732 DECL_BIT_FIELD_REPRESENTATIVE.
12733 * ubsan.c (instrument_object_size): Likewise.
12735 2017-03-27 Richard Biener <rguenther@suse.de>
12737 PR tree-optimization/80170
12738 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
12739 sure DR/SCEV didnt fold in constants we do not see when looking
12740 at the reference base alignment.
12742 2017-03-27 Richard Biener <rguenther@suse.de>
12744 PR middle-end/80171
12745 * gimple-fold.c (fold_ctor_reference): Properly guard against
12746 NULL return value from canonicalize_constructor_val.
12748 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
12751 * config/i386/i386.c (ix86_expand_builtin)
12752 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
12753 flags reg setting and flags reg using instructions.
12754 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
12755 clobbering instructions to zero extend op2.
12757 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
12759 * doc/install.texi (Configuration) <--with-aix-soname>:
12760 Update link to AIX ld.
12762 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
12764 PR rtl-optimization/80160
12765 PR rtl-optimization/80159
12766 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
12767 reg_alternate_class into account.
12769 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
12772 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
12773 to consider in curr_insn_transform.
12775 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12777 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
12778 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
12779 and emit_mode_inner.
12781 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12783 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
12784 argument to the overloaded builtin variants. Use the new flag to
12785 deprecate certain builtin variants.
12786 * config/s390/s390-builtin-types.def: Add new builtin types.
12787 * config/s390/s390-builtins.h: Support new flags field for
12788 overloaded builtins.
12789 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
12790 (s390_macro_to_expand): Enable vector float data type.
12791 (s390_cpu_cpp_builtins_internal): Indicate support of the new
12792 builtins by incrementing the __VEC__ version number.
12793 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
12795 (s390_resolve_overloaded_builtin): Emit error messages depending
12796 on the builtin flags.
12797 * config/s390/s390.c (s390_expand_builtin): Support additional
12798 flags argument. Change error message to match the messages
12799 emitted in s390-c.c.
12800 * config/s390/s390.md: New UNSPEC_* constants.
12801 (op_type): Add new instruction types.
12802 * config/s390/vecintrin.h: Add new builtins and test data class
12804 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
12805 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
12806 (VEC_INEXACT, VEC_NOINEXACT): New constants.
12807 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
12808 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
12809 ("vec_mergel<mode>"): V_HW -> VEC_HW.
12811 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
12812 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
12813 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
12814 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
12816 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
12817 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
12818 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
12819 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
12821 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
12822 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
12823 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
12824 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
12825 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
12826 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
12827 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
12829 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
12830 ("vec_scatter_element<V_HW_4:mode>_DI")
12831 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
12832 ("vec_fpint<mode>", "vflls")
12833 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
12834 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
12835 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
12836 ("*vec_cmphe<mode>_cc"): ... these.
12838 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
12839 mode constant instead of magic value.
12841 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12843 * config/s390/s390.c (s390_expand_vec_compare): Support other
12844 vector floating point modes than just V2DF.
12845 (s390_expand_vcond): Likewise.
12846 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
12847 (s390_cannot_change_mode_class): Prevent mode changes between TF
12848 and V1TF in vector registers.
12849 * config/s390/s390.md (DF, SF): New mode attributes.
12850 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
12851 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
12852 SFmode support for VRs.
12853 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
12855 (VFT, VF_HW): New mode iterators.
12856 (vw, sdx): New mode attributes.
12857 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
12858 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
12859 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
12860 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
12861 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
12862 also the new vector floating point modes. Renaming to ...
12864 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
12865 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
12866 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
12867 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
12868 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
12869 ("vec_unordered<mode>"): ... these.
12871 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
12872 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
12873 ("*vec_extendv2df"): New insn definitions.
12875 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12877 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
12878 ("mulditi3_2", "*muldi3_sign"): New patterns.
12879 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
12880 rename the pattern definition.
12882 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12884 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
12886 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
12888 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12890 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
12891 instruction if possible.
12892 * config/s390/vector.md (vec_halfnumelts): New mode
12894 ("*vec_vllezlf<mode>"): New pattern.
12896 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12898 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
12899 ("popcountv4si2", "popcountv2di2"): Rename to ...
12900 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
12901 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
12903 ("popcount<mode>2_vxe"): New pattern.
12905 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12907 * common/config/s390/s390-common.c (processor_flags_table): Add
12909 * config.gcc: Add arch12.
12910 * config/s390/driver-native.c (s390_host_detect_local_cpu):
12911 Default to arch12 for unknown CPU model numbers.
12912 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
12913 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
12914 PROCESSOR_max sanity check.
12915 * config/s390/s390-opts.h (enum processor_type): Add
12917 * config/s390/s390.c (processor_table): Add arch12.
12918 (s390_expand_builtin): Add check for B_VXE flag.
12919 (s390_issue_rate): Add PROCESSOR_ARCH12.
12920 (s390_get_sched_attrmask): Likewise.
12921 (s390_get_unit_mask): Likewise.
12922 (s390_sched_score): Enable z13 scheduling for arch12.
12923 (s390_sched_reorder): Likewise.
12924 (s390_sched_variable_issue): Likewise.
12925 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
12927 (s390_tune_attr): Use z13 scheduling also for arch12.
12928 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
12929 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
12930 (TARGET_VXE_P): New macros.
12931 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
12932 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
12933 * config/s390/s390.opt: Add arch12 as processor_type.
12935 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12937 * config/s390/s390.md
12938 ("fixuns_truncdddi2", "fixuns_trunctddi2")
12939 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
12940 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
12942 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
12943 Rename expanders to ...
12945 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
12946 ("fixuns_truncdddi2_emu"): ... these.
12948 ("fixuns_trunc<mode>si2_emu"): New expander.
12950 ("*fixuns_truncdfdi2_z13"): Rename to ...
12951 ("*fixuns_truncdfdi2_vx"): ... this.
12953 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12955 * config/s390/2964.md: Remove the single element vector compare
12956 instructions which are no longer used.
12957 * config/s390/s390.c (s390_select_ccmode): Remove handling of
12959 (s390_canonicalize_comparison): Remove handling of DFmode
12961 (s390_expand_vec_compare_scalar): Remove function.
12962 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
12963 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
12965 ("*cmp<mode>_ccs"): Add wfcdb instruction.
12967 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12969 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
12971 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
12972 will anyway by matched by mov<mode>_64dfp.
12974 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12976 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
12977 vlef/vstef. Add missing operand to vleif.
12979 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12981 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
12982 pair for all vector types with 64 bit elements.
12983 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
12984 * config/s390/vector.md (V_HW_64): ... here.
12985 (V_128_NOSINGLE): New mode iterator.
12986 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
12987 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
12988 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
12989 ("*vec_load_pairv2di"): Change to ...
12990 ("*vec_load_pair<mode>"): ... this one.
12992 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12994 * config/s390/constraints.md: Add comments.
12995 (jKK): Reject element sizes > 8 bytes.
12996 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
12998 * config/s390/s390.md: Add the s_operand checks formerly in
12999 s390_split_ok_p to various splitters where they are still
13001 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
13002 for 128 bit vectors. Plus two splitters.
13004 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13006 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
13009 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13012 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
13013 error if the boundary argument is not constant.
13015 2017-03-24 Jakub Jelinek <jakub@redhat.com>
13017 PR rtl-optimization/80112
13018 * loop-doloop.c (doloop_condition_get): Don't check condition
13019 if cmp isn't SET with IF_THEN_ELSE src.
13021 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13023 PR tree-optimization/80158
13024 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
13025 replacing a candidate statement, also replace it for the
13026 candidate's alternate interpretation.
13027 (replace_rhs_if_not_dup): Likewise.
13028 (replace_one_candidate): Likewise.
13030 2017-03-24 Richard Biener <rguenther@suse.de>
13032 PR tree-optimization/80167
13033 * graphite-isl-ast-to-gimple.c
13034 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
13036 (translate_isl_ast_to_gimple::get_rename): Likewise.
13038 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
13040 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
13041 handling of certain combinations of target options, including the
13042 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
13043 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
13045 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13048 * config/arm/arm.md (*load_multiple): Add reload_completed to
13049 matching condition.
13051 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13052 Richard Biener <rguenth@suse.de>
13054 PR tree-optimization/79908
13055 PR tree-optimization/80136
13056 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
13057 been cast away, gimplify_and_add suffices.
13059 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
13061 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
13063 2017-03-23 Richard Biener <rguenther@suse.de>
13065 PR tree-optimization/80032
13066 * gimplify.c (gimple_push_cleanup): Forced unconditional
13067 cleanups still have to go to the conditional_cleanups
13070 2017-03-22 Jakub Jelinek <jakub@redhat.com>
13072 PR tree-optimization/80072
13073 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
13075 (next_operand_entry_id): Change type to unsigned int.
13076 (sort_by_operand_rank): Make sure to return the right return value
13077 even if unsigned fields are bigger than INT_MAX.
13078 (struct oecount): Change cnt and id type to unsigned int.
13079 (oecount_hasher::equal): Formatting fix.
13080 (oecount_cmp): Make sure to return the right return value
13081 even if unsigned fields are bigger than INT_MAX.
13082 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
13085 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
13086 TREE_READONLY on result if writing it more than once.
13089 * doc/invoke.texi (-fsanitize=thread): Document that with
13090 -fnon-call-exceptions atomics are not able to throw
13094 * tsan.c: Include tree-eh.h.
13095 (instrument_builtin_call): Call maybe_clean_eh_stmt or
13096 maybe_clean_or_replace_eh_stmt where needed.
13097 (instrument_memory_accesses): Add cfg_changed argument.
13098 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
13099 if it returned true.
13100 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
13102 PR rtl-optimization/63191
13103 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
13104 wrapper function, moved the whole old content into ...
13105 (ix86_delegitimize_address_1): ... this. New inline function.
13106 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
13107 true as last argument instead of ix86_delegitimize_address.
13109 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
13111 * config/aarch64/aarch64.c (generic_branch_cost): Copy
13112 cortexa57_branch_cost.
13114 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
13116 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
13118 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13121 * doc/md.texi (Constraints): Document wA constraint.
13122 * config/rs6000/constraints.md (wA): New.
13123 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
13124 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
13125 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
13126 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
13128 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
13131 * gimplify.c (is_oacc_declared): New function.
13132 (oacc_default_clause): Use it to set default flags for acc declared
13133 variables inside parallel regions.
13134 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
13135 declared variables.
13136 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
13137 declare attribute to any decl as necessary.
13139 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
13142 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
13143 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
13144 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
13145 (arm_arch_lpae): This.
13146 * config/arm/arm.c (arm_arch7ve): Rename into ...
13147 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
13148 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
13151 2017-03-22 Martin Liska <mliska@suse.cz>
13154 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
13155 error message instead of an ICE.
13157 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13159 * doc/extend.texi (6.11 Additional Floating Types): Revise.
13161 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
13163 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
13165 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
13168 2017-03-21 Martin Sebor <msebor@redhat.com>
13170 * doc/extend.texi: Use "cannot" instead of "can't."
13171 * doc/hostconfig.texi: Same.
13172 * doc/install.texi: Same.
13173 * doc/invoke.texi: Same.
13174 * doc/loop.texi: Same.
13175 * doc/md.texi: Same.
13176 * doc/objc.texi: Same.
13177 * doc/rtl.texi: Same.
13178 * doc/tm.texi: Same.
13179 * doc/tm.texi.in: Same.
13180 * doc/trouble.texi: Same.
13182 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
13185 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
13186 (collect_checksum_attributes): Set it.
13187 (die_checksum_ordered): Use it.
13189 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13191 PR tree-optimization/79908
13192 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
13193 change: For a VA_ARG whose LHS has been cast away, use
13194 force_gimple_operand to construct the side effects.
13196 2017-03-21 David Malcolm <dmalcolm@redhat.com>
13198 PR translation/80001
13199 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
13200 more amenable to translation.
13201 (oacc_loop_auto_partitions): Likewise.
13203 2017-03-21 Marek Polacek <polacek@redhat.com>
13204 Martin Sebor <msebor@redhat.com>
13206 PR tree-optimization/80109
13207 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
13208 on INTEGRAL_TYPE_P.
13210 2017-03-21 Jakub Jelinek <jakub@redhat.com>
13211 Segher Boessenkool <segher@kernel.crashing.org>
13214 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
13215 check reg_used_between_p between insn and one of succ or succ2
13216 depending on if succ is artificial insn not inserted into insn
13219 2017-03-21 Martin Liska <mliska@suse.cz>
13221 PR gcov-profile/80081
13222 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
13223 * doc/gcc.texi: Include gcov-dump stuff.
13224 * doc/gcov-dump.texi: New file.
13226 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
13228 PR rtl-optimization/79150
13229 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
13230 conditional jump, if the jump is the last insn of the loop.
13232 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13233 Richard Biener <rguenth@suse.de>
13235 PR tree-optimization/79908
13236 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
13237 been cast away, use force_gimple_operand to construct the side
13240 2017-03-21 Martin Liska <mliska@suse.cz>
13242 PR libfortran/79956
13243 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
13246 2017-03-21 Brad Spengler <spender@grsecurity.net>
13249 * plugin.c (htab_hash_plugin): New function.
13250 (add_new_plugin): Use it and adjust.
13251 (parse_plugin_arg_opt): Adjust.
13252 (init_one_plugin): Likewise.
13254 2017-03-21 Richard Biener <rguenther@suse.de>
13256 PR tree-optimization/80032
13257 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
13258 if set force the cleanup to happen unconditionally.
13259 (gimplify_target_expr): Push inserted clobbers with force_uncond
13260 to avoid them being removed by control-dependent DCE.
13262 2017-03-21 Richard Biener <rguenther@suse.de>
13264 PR tree-optimization/80122
13265 * tree-inline.c (copy_bb): Do not expans va-arg packs or
13266 va_arg_pack_len when the inlined call stmt requires pack
13268 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
13270 2017-03-21 Jakub Jelinek <jakub@redhat.com>
13273 * tsan.c (instrument_builtin_call): If the memory model argument
13274 is not a constant, assume it is valid.
13277 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
13280 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
13282 PR rtl-optimization/79910
13283 * combine.c (can_combine_p): Do not allow combining an I0 or I1
13284 if its dest is used by an insn before I2 (other than the combined
13285 insns themselves, which are properly handled already).
13287 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
13290 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
13292 * combine.c (record_used_regs): New static function.
13293 (try_combine): Handle situations where there is an additional
13294 instruction between I2 and I3 which needs to have a LOG_LINK
13298 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
13300 * combine.c (try_combine): Delete redundant i1 test. Call
13301 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
13303 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13306 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
13307 alternatives 13/14.
13309 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13311 PR tree-optimization/80054
13312 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
13313 the optimization if a PHI or any of its arguments is not dominated
13314 by the candidate's basis. Use gphi* rather than gimple* as
13316 (replace_profitable_candidates): Clean up a gimple* variable that
13317 should be a gphi* variable.
13319 2017-03-20 Martin Sebor <msebor@redhat.com>
13322 * doc/extend.texi (attribute constructor): Document present limitation.
13324 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
13327 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
13328 __POWER9_VECTOR__ #ifdef control, change template definition to
13329 use Power9-specific built-in function.
13330 (vec_any_eq): Likewise.
13331 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
13332 to control outcomes from this test.
13333 (vector_ae_<mode>p): For VEC_F modes, likewise.
13335 2017-03-20 Ian Lance Taylor <iant@google.com>
13337 * config/i386/i386.c (ix86_function_regparm): Save an extra
13338 register for -fsplit-stack with DECL_STATIC_CHAIN.
13340 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
13343 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
13344 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
13346 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
13348 * config/riscv/riscv.c (riscv_print_operand): Use "fence
13350 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
13353 2017-03-20 Marek Polacek <polacek@redhat.com>
13356 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
13358 2017-03-20 Richard Biener <rguenther@suse.de>
13360 PR tree-optimization/80113
13361 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
13362 allocate extra SSA name for PHI def.
13363 (add_close_phis_to_outer_loops): Likewise.
13364 (add_close_phis_to_merge_points): Likewise.
13365 (copy_loop_close_phi_args): Likewise.
13366 (copy_cond_phi_nodes): Likewise.
13368 2017-03-20 Martin Liska <mliska@suse.cz>
13370 PR middle-end/79753
13371 * tree-chkp.c (chkp_build_returned_bound): Do not build
13372 returned bounds for a LHS that's not a BOUNDED_P type.
13374 2017-03-20 Martin Liska <mliska@suse.cz>
13378 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
13379 COMPLEX_CST and VECTOR_CST.
13381 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13384 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
13385 target operand. A new splitter adds the clobber statement in case
13386 the target operand is dead anyway.
13388 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
13390 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
13391 to age-old versions of binutils and glibc.
13393 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
13395 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
13397 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
13399 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
13401 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
13403 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
13404 requirement for binutils 2.13.
13406 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
13408 * combine.c (try_combine): Delete redundant i1 test. Call
13409 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
13411 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
13413 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
13414 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
13416 <riscv64-*-elf>: Re-arrange section
13417 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
13418 <riscv32-*-linux>: Likewise.
13419 <riscv64-*-elf>: Likewise
13420 <riscv64-*-linux>: Likewise.
13422 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
13425 * aarch64.opt(verbose-cost-dump): Fix typo.
13427 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
13430 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
13431 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
13433 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
13435 * reload.c (find_reloads): When reloading a nonoffsettable address,
13436 use RELOAD_OTHER for it and its address reloads.
13438 PR rtl-optimization/79910
13439 * combine.c (record_used_regs): New static function.
13440 (try_combine): Handle situations where there is an additional
13441 instruction between I2 and I3 which needs to have a LOG_LINK
13444 2017-03-17 Jeff Law <law@redhat.com>
13446 PR tree-optimization/71437
13447 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
13448 conditional in the hash table first.
13449 (vrp_dom_walker::before_dom_children): Extract condition from
13450 ASSERT_EXPR. Record condition, its inverion and any implied
13451 conditions as well.
13453 2017-03-17 Marek Polacek <polacek@redhat.com>
13454 Markus Trippelsdorf <markus@trippelsdorf.de>
13456 PR tree-optimization/80079
13457 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
13460 2017-03-17 Richard Biener <rguenther@suse.de>
13462 PR middle-end/80075
13463 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
13464 Properly verify the LHS before the RHS possibly claims to be
13466 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
13469 2017-03-17 Martin Jambor <mjambor@suse.cz>
13471 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
13472 (List of -O2 options): Likewise.
13473 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
13476 2017-03-17 Tom de Vries <tom@codesourcery.com>
13478 * gcov-dump.c (print_usage): Print bug_report_url.
13480 2017-03-17 Richard Biener <rguenther@suse.de>
13482 PR middle-end/80050
13483 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
13484 (parser::peek): Likewise.
13486 2017-03-17 Richard Biener <rguenther@suse.de>
13488 PR tree-optimization/80048
13489 * sese.c (free_sese_info): Properly release rename_map and
13490 copied_bb_map elements.
13492 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
13494 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
13495 Add linked-list forward and backlinks. Insert on
13496 construction, remove on destruction.
13497 (class pass_store_merging): Add m_stores_head field.
13498 (pass_store_merging::terminate_and_process_all_chains):
13499 Iterate over m_stores_head list.
13500 (pass_store_merging::terminate_all_aliasing_chains):
13502 (pass_store_merging::execute): Check for debug stmts first.
13503 Push new chains onto the m_stores_head stack.
13505 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
13508 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
13509 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
13510 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
13512 2017-03-16 Jeff Law <law@redhat.com>
13514 PR tree-optimization/71437
13515 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
13516 member function. Implementation moved into after_dom_children
13517 member function and into the threader's thread_outgoing_edges
13519 (dom_opt_dom_walker::after_dom_children): Simplify by moving
13520 some code into new thread_outgoing_edges.
13521 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
13522 definition. Simplify marker handling (do it here). Assume we always
13523 have the available expression and the const/copies tables.
13524 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
13526 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
13527 * tree-vrp.c (equiv_stack): No longer file scoped.
13528 (vrp_dom_walker): New class.
13529 (vrp_dom_walker::before_dom_children): New member function.
13530 (vrp_dom_walker::after_dom_children): Likewise.
13531 (identify_jump_threads): Setup domwalker. Use it rather than
13532 walking edges in a random order by hand. Simplify setup/finalization.
13533 (finalize_jump_threads): Remove.
13534 (vrp_finalize): Do not call identify_jump_threads here.
13535 (execute_vrp): Do it here instead and call thread_through_all_blocks
13538 PR tree-optimization/71437
13539 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
13541 (simplify_stmt_for_jump_threading): Add basic_block argument. All
13543 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
13544 (dom_opt_dom_walker::thread_across_edge): Remove
13545 handle_dominating_asserts argument. All callers changed.
13546 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
13547 changes. Remove calls to lhs_of_dominating_assert. Other
13548 uses of handle_dominating_asserts turn into unconditional code
13549 (simplify_control_stmt_condition_1): Likewise.
13550 (simplify_control_stmt_condition): Likewise.
13551 (thread_through_normal_block, thread_across_edge): Likewise.
13552 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
13553 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
13554 object if it is not an SSA_NAME.
13555 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
13556 before calling into the VRP specific simplifiers.
13557 (identify_jump_threads): Remove handle_dominating_asserts
13560 2017-03-16 Jakub Jelinek <jakub@redhat.com>
13563 * tree-diagnostic.c (default_tree_printer): No longer static.
13564 * tree-diagnostic.h (default_tree_printer): New prototype.
13566 2017-03-16 Tamar Christina <tamar.christina@arm.com>
13568 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
13569 Change ins into fmov.
13571 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13573 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
13574 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
13575 Use h_con constraint for operand 1.
13576 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
13577 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
13579 2017-03-15 Jeff Law <law@redhat.com>
13581 PR tree-optimization/71437
13582 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
13583 (record_temporary_equivalences): Use it.
13585 PR tree-optimization/71437
13586 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
13587 tree-ssa-scopedtables.
13588 (lookup_avail_expr, build_and_record_new_cond): Likewise.
13589 (record_conditions, record_cond, vuse_eq): Likewise.
13590 (record_edge_info): Adjust to API tweak of record_conditions.
13591 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
13592 (record_temporary_equivalences, optimize_stmt): Likewise.
13593 (eliminate_redundant_computations): Likewise.
13594 (record_equivalences_from_stmt): Likewise.
13595 * tree-ssa-scopedtables.c: Include options.h and params.h.
13596 (vuse_eq): New function, moved from tree-ssa-dom.c
13597 (build_and_record_new_cond): Likewise.
13598 (record_conditions): Likewise. Accept vector of conditions rather
13599 than edge_equivalence structure for first argument.
13600 for the first argument.
13601 (avail_exprs_stack::lookup_avail_expr): New member function, moved
13602 from tree-ssa-dom.c.
13603 (avail_exprs_stack::record_cond): Likewise.
13604 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
13605 from tree-ssa-dom.c.
13606 (avail_exprs_stack): Add new member functions lookup_avail_expr
13608 (record_conditions): Declare.
13610 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
13613 * lra-constraints.c (process_alt_operands): Increase reject for
13614 reloading an input/output operand.
13616 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
13619 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
13620 insns to convert from signed/unsigned char/short to IEEE 128-bit
13622 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
13624 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
13627 * config/i386/i386.c (ix86_vector_duplicate_value): Create
13628 subreg of inner mode for values already in registers.
13630 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
13632 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
13633 iteration reg is used after the loop.
13635 2017-03-14 Martin Sebor <msebor@redhat.com>
13637 PR tree-optimization/79800
13638 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
13639 precision in negative-positive range.
13640 (format_floating): Call non-const overload with adjusted precision.
13642 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
13645 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
13648 2017-03-14 Martin Sebor <msebor@redhat.com>
13650 PR middle-end/80020
13651 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
13652 * builtins.def (aligned_alloc): Use it.
13655 * Makefile.in (GTFILES): Add calls.c.
13656 * calls.c: Include "gt-calls.h".
13658 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
13660 PR rtl-optimization/79728
13661 * regs.h (struct target_regs): New field
13662 x_contains_allocatable_regs_of_mode.
13663 (contains_allocatable_regs_of_mode): New macro.
13664 * reginfo.c (init_reg_sets_1): Initialize it, and change
13665 contains_reg_of_mode so it includes global regs as well.
13666 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
13667 rather than contains_regs_of_mode.
13669 2017-03-14 Martin Liska <mliska@suse.cz>
13671 * doc/invoke.texi: Document options that can't be combined with
13672 -fcheck-pointer-bounds.
13674 2017-03-14 Martin Liska <mliska@suse.cz>
13676 PR middle-end/79831
13677 * doc/invoke.texi (-Wchkp): Document the option.
13679 2017-03-14 Martin Liska <mliska@suse.cz>
13681 * Makefile.in: Install gcov-dump.
13683 2017-03-14 Martin Liska <mliska@suse.cz>
13685 * multiple_target.c (expand_target_clones): Bail out for
13686 an invalid attribute.
13688 2017-03-14 Richard Biener <rguenther@suse.de>
13690 * alias.c (struct alias_set_entry): Pack properly.
13691 * cfgloop.h (struct loop): Likewise.
13692 * cse.c (struct set): Likewise.
13693 * ipa-utils.c (struct searchc_env): Likewise.
13694 * loop-invariant.c (struct invariant): Likewise.
13695 * lra-remat.c (struct cand): Likewise.
13696 * recog.c (struct change_t): Likewise.
13697 * rtl.h (struct address_info): Likewise.
13698 * symbol-summary.h (function_summary): Likewise.
13699 * tree-loop-distribution.c (struct partition): Likewise.
13700 * tree-object-size.c (struct object_size_info): Likewise.
13701 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
13702 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
13703 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
13704 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
13705 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
13706 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
13707 (struct _stmt_vec_info): Likewise.
13709 2017-03-14 Martin Liska <mliska@suse.cz>
13712 * multiple_target.c (create_dispatcher_calls): Check that
13713 a target can create a function dispatcher.
13715 2017-03-14 Martin Liska <mliska@suse.cz>
13718 * multiple_target.c (expand_target_clones): Drop local.local
13719 flag for default implementation.
13721 2017-03-14 Richard Biener <rguenther@suse.de>
13723 PR tree-optimization/80030
13724 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
13726 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
13728 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
13729 gcc_fallthrough() instead of __attribute__((fallthrough));
13731 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13733 * doc/gcc.texi: Remove "up" link to (DIR).
13734 * doc/gccint.texi: Ditto.
13736 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13738 * doc/install.texi (Specific) <avr>: Remove reference to
13741 2017-03-13 Jeff Law <law@redhat.com>
13743 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
13744 attribute rather than comments.
13746 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
13747 match_scratch operand is highest.
13749 2017-03-13 Martin Liska <mliska@suse.cz>
13751 PR middle-end/78339
13752 * ipa-pure-const.c (warn_function_noreturn): If the declarations
13753 is a CHKP clone, use original declaration.
13755 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13757 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
13758 (arc_conditional_register_usage): Use a different allocation order
13759 when optimizing for size.
13760 * common/config/arc/arc-common.c (arc_option_optimization_table):
13761 Section anchors default on when optimizing for size.
13763 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13765 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
13767 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13769 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
13770 * config/arc/arc.md (cpu_facility): Add cd variant.
13771 (*movqi_insn): Add code density variant.
13772 (*movhi_insn): Likewise.
13773 (*movqi_insn): Likewise.
13774 (*addsi3_mixed): Likewise.
13775 (subsi3_insn): Likewise.
13777 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13779 * config/arc/arc.md (movsi_cond_exec): Update constraint.
13781 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13783 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
13784 expressions with MINUS and UNARY ops.
13786 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13789 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
13791 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
13792 between vec_select and vector argument.
13793 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
13794 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
13795 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
13796 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
13797 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
13798 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
13800 2017-03-13 Richard Biener <rguenther@suse.de>
13803 * params.def (vect-max-peeling-for-alignment): Fix typo.
13805 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13807 * doc/install.texi (Specific) <mips-*-*>: Remove description of
13808 issue that only occurred with binutils below 2.18.
13810 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13812 * doc/install.texi (Specific) <cris-axis-elf>: No longer
13813 refer to binutils 2.11/2.12 minimum.
13815 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13817 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
13818 ftp.kernel.org and simplify binutils requirement.
13820 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
13822 * doc/invoke.texi (Warning Options): Fix spelling of link-time
13824 (Optimize Options): Ditto. Also remove redundancy.
13826 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13828 PR translation/79848
13829 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
13831 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
13832 to G_ to avoid double translation.
13834 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13836 PR translation/79923
13837 * auto-profile.c (get_combined_location): Convert leading
13838 character of diagnostics to lower case and remove trailing period.
13839 (read_profile): Likewise for various diagnostics.
13840 * config/arm/arm.c (arm_option_override): Remove trailing period
13841 from various diagnostics.
13842 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
13843 (msp430_expand_delay_cycles): Likewise.
13845 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13848 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
13849 full command-line argument, rather than just "str".
13850 (aarch64_validate_march): Likewise.
13851 (aarch64_validate_mtune): Likewise.
13853 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
13855 PR rtl-optimization/78911
13856 * lra-assigns.c (must_not_spill_p): New function.
13857 (spill_for): Use it.
13859 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13861 PR tree-optimization/79981
13862 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
13863 ATOMIC_COMPARE_EXCHANGE ifn result.
13864 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
13865 IFN_ATOMIC_COMPARE_EXCHANGE.
13867 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13870 * opts.c (parse_sanitizer_options): Add missing question mark to
13871 "did you mean" message.
13873 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13875 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
13877 (VMULEUH_UNS): Likewise.
13878 (VMULOUB_UNS): Likewise.
13879 (VMULOUH_UNS): Likewise.
13880 * config/rs6000/rs6000.c (builtin_function_type): Remove
13881 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
13883 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13886 * read-rtl-function.c (function_reader::read_rtx_operand): Update
13887 x with result of extra_parsing_for_operand_code_0.
13888 (function_reader::extra_parsing_for_operand_code_0): Convert
13889 return type from void to rtx, returning x. When reading
13890 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
13891 larger size containing struct block_symbol.
13893 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
13895 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
13896 -mfloat128-hardware without -m64.
13898 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
13901 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
13902 entries to the case statement that marks unsigned arguments to
13903 overloaded functions.
13905 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
13907 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
13908 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
13910 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
13913 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
13914 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
13916 2017-03-10 Martin Liska <mliska@suse.cz>
13920 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
13921 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
13924 2017-03-10 Olivier Hainque <hainque@adacore.com>
13926 * tree-switch-conversion (array_value_type): Start by resetting
13927 candidate type to it's main variant.
13929 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13931 PR rtl-optimization/79909
13932 * combine.c (try_combine): Use simplify_replace_rtx on individual
13933 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
13934 of the whole CALL_INSN_FUNCTION_USAGE.
13936 PR tree-optimization/79972
13937 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
13938 get_range_info on SSA_NAMEs. Formatting fixes.
13940 2017-03-10 Richard Biener <rguenther@suse.de>
13941 Jakub Jelinek <jakub@redhat.com>
13943 PR tree-optimization/77975
13944 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
13945 edge to be constant.
13946 (get_val_for): For constant x return it. Formatting fix.
13947 (loop_niter_by_eval): Avoid pointless looping if the next iteration
13948 would use the same bases as the current one.
13950 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13952 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
13953 instead of vec_select for V1TImode.
13954 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
13956 (VSX_LE_128): Add V1TI to this mode iterator.
13957 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
13958 (*vsx_le_perm_store_<mode>): Likewise.
13959 (pre-reload splitter for VSX stores): Likewise.
13960 (post-reload splitter for VSX stores): Likewise.
13961 (*vsx_xxpermdi2_le_<mode>): Likewise.
13962 (*vsx_lxvd2x2_le_<mode>): Likewise.
13963 (*vsx_stxvd2x2_le_<mode>): Likewise.
13965 2017-03-09 Michael Eager <eager@eagercon.com>
13967 Correct failures with --enable-checking=yes,rtl.
13969 * config/microblaze/microblaze.c (microblaze_expand_shift):
13970 Replace GET_CODE test with CONST_INT_P and INTVAL test with
13971 test for const0_rtx.
13972 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
13973 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
13975 2017-03-09 Richard Biener <rguenther@suse.de>
13977 PR tree-optimization/79977
13978 * graphite-scop-detection.c (scop_detection::merge_sese):
13979 Handle the case of extra exits to blocks dominating the entry.
13981 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
13983 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13986 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
13988 PR rtl-optimization/79949
13989 * lra-constraints.c (process_alt_operands): Check memory when
13990 trying to predict a cycle. Print about the overall increase.
13992 2017-03-09 Richard Biener <rguenther@suse.de>
13994 PR middle-end/79971
13995 * gimple-expr.c (useless_type_conversion_p): Preserve
13996 TYPE_SATURATING for fixed-point types.
13998 2017-03-09 Richard Biener <rguenther@suse.de>
14001 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
14002 alignment of BLKmode params.
14004 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14007 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
14008 (VALL_NO_V2Q): Likewise.
14010 * config/aarch64/aarch64-simd.md
14011 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
14013 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
14014 VALL_NO_V2Q mode iterator.
14015 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
14017 2017-03-09 Martin Liska <mliska@suse.cz>
14019 PR tree-optimization/79631
14020 * tree-chkp-opt.c (chkp_is_constant_addr): Call
14021 tree_int_cst_sign_bit just for INTEGER constants.
14023 2017-03-09 Martin Liska <mliska@suse.cz>
14027 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
14030 2017-03-09 Marek Polacek <polacek@redhat.com>
14033 * tree.c (inchash::add_expr): Handle TREE_VEC.
14035 2017-03-09 Martin Liska <mliska@suse.cz>
14038 (chkp_narrow_size_and_offset): New function.
14039 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
14040 (void chkp_parse_bit_field_ref): New function.
14041 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
14042 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
14044 2017-03-09 Martin Liska <mliska@suse.cz>
14047 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
14048 (chkp_find_bounds_1): Remove gcc_unreachable.
14050 2017-03-09 Jakub Jelinek <jakub@redhat.com>
14053 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
14054 BUILT_IN_SYNC*, determine the access type from the size suffix and
14055 always build a MEM_REF with that type. Handle forgotten
14056 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
14059 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
14060 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
14061 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
14062 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
14063 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
14064 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
14065 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
14066 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
14067 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
14068 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
14069 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
14070 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
14071 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
14072 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
14073 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
14074 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
14075 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
14076 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
14077 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
14078 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
14079 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
14080 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
14081 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
14082 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
14083 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
14084 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
14085 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
14086 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
14087 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
14088 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
14089 definitions outside of __OPTIMIZE__ guarded section.
14092 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
14093 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
14094 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
14095 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
14098 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14100 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
14101 ("vfenez<mode>"): Add missing constraints.
14103 2017-03-08 Martin Sebor <msebor@redhat.com>
14106 * config/nds32/nds32.c (nds32_option_override):
14107 Fix misspelled diagnostic.
14109 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14112 * gimplify.c (gimplify_omp_for): Replace index var in outer
14113 taskloop statement with an artificial variable and add
14114 OMP_CLAUSE_PRIVATE clause for it.
14116 2017-03-08 Richard Biener <rguenther@suse.de>
14118 PR tree-optimization/79955
14119 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
14120 for accesses that are completely outside of the variable.
14122 2017-03-08 Andrew Haley <aph@redhat.com>
14124 PR tree-optimization/79943
14125 * tree-ssa-loop-split.c (compute_new_first_bound): When
14126 calculating the new upper bound, (END-BEG) should be added, not
14129 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14131 * config/avr/avr.md (setmemhi): Make sure match_dup
14132 operand number comes before match_scratch.
14134 2017-03-08 Richard Biener <rguenther@suse.de>
14136 PR tree-optimization/79920
14137 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
14138 with ncopies == 1 to ...
14139 (vect_transform_slp_perm_load): ... here. Properly compute
14140 all element loads by iterating VF times over the group. Do
14141 not handle ncopies (computed in a broken way) in
14142 vect_create_mask_and_perm.
14144 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14147 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
14148 is a uniform vector, use uniform_vector_p return value instead of
14149 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
14151 2017-03-07 Marek Polacek <polacek@redhat.com>
14153 PR middle-end/79809
14154 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
14155 (alloca_call_type): Likewise.
14157 2017-03-07 Martin Liska <mliska@suse.cz>
14159 * gcov.c (process_args): Put comment to correct location.
14161 2017-03-07 Martin Liska <mliska@suse.cz>
14163 PR middle-end/68270
14164 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
14165 Use array_at_struct_end_p instead of DECL_CHAIN (field).
14166 (chkp_narrow_bounds_for_field): Likewise.
14167 (chkp_parse_array_and_component_ref): Pass one more argument to
14170 2017-03-07 Richard Biener <rguenther@suse.de>
14172 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
14175 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
14177 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
14178 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
14180 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14183 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
14184 to end of description.
14185 (PARAM_MAX_STORES_TO_MERGE): Likewise.
14187 2017-03-07 Jakub Jelinek <jakub@redhat.com>
14189 PR rtl-optimization/79901
14190 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
14192 (*avx512f_<code><mode>3<mask_name>): ... this.
14193 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
14194 iterator instead of VI8_AVX2_AVX512BW.
14196 PR rtl-optimization/79901
14197 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
14198 min/max expander, expand it using expand_vec_cond_expr.
14201 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
14204 2017-03-06 Jakub Jelinek <jakub@redhat.com>
14207 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
14208 to void * for PCH reasons.
14209 * dwarf2out.c (output_loc_operands, output_die): Cast
14210 v.val_vec.array to unsigned char *.
14212 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
14215 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
14218 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
14220 PR rtl-optimization/79571
14221 * lra-constraints.c (process_alt_operands): Calculate static
14222 reject and subtract it from overall when only addresses will be
14225 2017-03-06 Julia Koval <julia.koval@intel.com>
14228 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
14229 incoming stack boundary to 128 for 64-bit targets.
14231 2017-03-06 Richard Biener <rguenther@suse.de>
14233 PR tree-optimization/79894
14234 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
14235 to NULL after folding it.
14237 2017-03-06 Richard Biener <rguenther@suse.de>
14239 PR tree-optimization/79824
14240 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
14241 check disabling peeling for gaps.
14243 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
14245 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
14246 attributes): Document gettimeofday.
14248 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
14250 * config/s390/s390.c (s390_option_override_internal): Set
14251 PARAM_MIN_VECT_LOOP_BOUND
14253 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
14255 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
14256 * config/s390/s390.md: Likewise.
14258 2017-03-06 Jakub Jelinek <jakub@redhat.com>
14261 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
14262 (<avx2_avx512>_perm<mode>): Rename to ...
14263 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
14265 (<avx512>_perm<mode>_mask): Rename to ...
14266 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
14268 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
14269 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
14270 instead of VI8F_256_512.
14271 (avx512f_perm<mode>): New define_expand.
14272 (avx512f_perm<mode>_mask): Likewise.
14273 (avx512f_perm<mode>_1<mask_name>): New define_insn.
14274 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
14276 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14278 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
14279 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
14281 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
14283 2017-03-06 Martin Liska <mliska@suse.cz>
14286 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
14287 when having a SSA NAME w/o VAR_DECL assigned to it.
14289 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14291 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
14292 msa_dpsub_<su>_d): Fix MODE for vec_select.
14294 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14296 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
14298 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
14300 2017-03-06 Richard Biener <rguenther@suse.de>
14302 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
14303 * plugin.c (register_plugin_info): Likewise.
14304 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
14306 2017-03-05 Jakub Jelinek <jakub@redhat.com>
14308 * config/i386/sse.md (sse_storehps, sse_storelps,
14309 avx_<castmode><avxsizesuffix>_<castmode>,
14310 avx512f_<castmode><avxsizesuffix>_<castmode>,
14311 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
14312 in condition that at least one operand is not a MEM.
14314 2017-03-03 Jakub Jelinek <jakub@redhat.com>
14316 PR middle-end/79805
14317 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
14318 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
14320 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
14321 gimple_call_nothrow_p flag based on whether original builtin can throw.
14322 If it can, emit following stmts on the fallthrough edge.
14323 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
14324 don't create new bb if inserting just debug stmts on the edge, try to
14325 insert them on the fallthru bb or just reset debug stmts.
14327 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
14330 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
14331 restore recog_data (including the operand rtxes inside it) around
14332 the call to get_insn_template.
14334 2017-03-03 Martin Sebor <msebor@redhat.com>
14336 PR tree-optimization/79699
14337 * context.c (context::~context): Free MPFR caches to avoid
14338 a memory leak on program exit.
14340 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14342 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
14343 Use wide_int::ulow () instead of .elt (0).
14345 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
14347 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
14348 (*pushxf): Limit oF constraint to 32bit targets and add oC
14349 constraint for 64bit targets.
14350 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
14351 (*pushdf): Change rmF constraint to rmC.
14353 2017-03-03 Martin Liska <mliska@suse.cz>
14355 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
14356 Remove unused variable.
14358 2017-03-03 Jakub Jelinek <jakub@redhat.com>
14361 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
14362 is a memory operand, increase num_memory.
14363 (ix86_expand_args_builtin): Likewise.
14365 2017-03-03 Jan Hubicka <jh@suse.cz>
14368 * ipa-devirt.c (maybe_record_node): Properly handle
14369 __cxa_pure_virtual visibility.
14371 2017-03-03 Martin Liska <mliska@suse.cz>
14373 PR tree-optimization/79803
14374 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
14376 (pass_loop_prefetch::execute): Disabled optimization if an
14377 assumption about L1 cache size is not met.
14379 2017-03-03 Martin Liska <mliska@suse.cz>
14381 PR rtl-optimization/79574
14382 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
14383 (hash_scan_set): Likewise.
14384 (dump_hash_table): Likewise.
14385 (hoist_code): Likewise.
14387 2017-03-03 Richard Biener <rguenther@suse.de>
14389 * fixed-value.c (fixed_from_string): Restore use of elt (1)
14390 in place of uhigh ().
14391 (fixed_convert_from_real): Likewise.
14393 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
14396 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
14398 2017-03-03 Richard Biener <rguenther@suse.de>
14400 PR middle-end/79818
14401 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
14402 TYPE_OVERFLOW_UNDEFINED check.
14404 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14406 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
14408 (vector_ae_<mode>_p): Likewise.
14409 (vector_nez_<mode>_p): Likewise.
14410 (vector_ne_v2di_p): Likewise.
14411 (vector_ae_v2di_p): Likewise.
14412 (vector_ne_<mode>_p): Likewise.
14413 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
14415 (vsx_tsqrt<mode>2_fe): Likewise.
14417 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
14420 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
14422 2017-03-02 Jakub Jelinek <jakub@redhat.com>
14424 PR rtl-optimization/79780
14425 * cprop.c (one_cprop_pass): When second and further conditional trap
14426 in a single basic block is turned into an unconditional trap, turn it
14427 into a deleted note to avoid RTL verification failures.
14429 2017-03-02 Richard Biener <rguenther@suse.de>
14431 * fold-const.c (const_binop): Use ulow () instead of elt (0).
14433 2017-03-02 Richard Biener <rguenther@suse.de>
14435 PR tree-optimization/79345
14437 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
14438 param and abort the walk, returning -1 if it is hit.
14439 (walk_aliased_vdefs): Take a limit param and pass it on.
14440 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
14441 defaulting to 0 and return a signed int.
14442 * tree-ssa-uninit.c (struct check_defs_data): New struct.
14443 (check_defs): New helper.
14444 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
14445 about uninitialized memory.
14446 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
14447 bogus uninitialized warning.
14448 (fixed_convert_from_real): Likewise.
14450 2017-03-02 Bin Cheng <bin.cheng@arm.com>
14452 PR tree-optimization/66768
14453 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
14454 iv_use if base object can't be determined.
14456 2017-03-02 Jakub Jelinek <jakub@redhat.com>
14458 PR tree-optimization/79345
14459 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
14460 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
14461 (get_pattern_stats): Initialize it.
14462 * genemit.c (gen_expand): Verify match_scratch numbers come after
14463 match_operand/match_dup numbers.
14464 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
14465 match_scratch numbers.
14466 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
14468 * config/s390/s390.md (trunctdsd2): Likewise.
14470 2017-03-02 Richard Biener <rguenther@suse.de>
14472 * wide-int.h (wide_int_storage::operator=): Implement in terms
14475 2017-03-02 Richard Biener <rguenther@suse.de>
14477 PR tree-optimization/79777
14478 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
14479 the to insert expression to sth existing.
14481 2017-03-01 Martin Sebor <msebor@redhat.com>
14483 PR middle-end/79692
14484 * gimple-ssa-sprintf.c
14485 (directive::known_width_and_precision): New function.
14486 (format_integer): Use it.
14487 (get_mpfr_format_length): Consider the full range of precision
14488 when computing %g output with the # flag. Set the likely byte
14489 count to 3 rather than 1 when precision is indeterminate.
14490 (format_floating): Correct the lower bound of precision.
14492 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14494 * doc/invoke.texi: Document default code model for 64-bit Linux.
14496 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14499 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
14500 udiv rather than div since input pattern is unsigned.
14502 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
14504 * config/i386/i386.c (print_reg): Warn for values of
14505 unsupported size in integer register.
14507 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
14510 * config/rs6000/predicates.md (current_file_function_operand): Do
14511 not allow self calls to be local if the function is replaceable.
14513 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
14516 * config/rs6000/altivec.h (vec_ctz and others): Change the
14517 preprocessor macro that controls conditional compilation from
14518 _ARCH_PWR9 to __POWER9_VECTOR__.
14519 (vec_all_ne): Change parameterization of __altivec_scalar_pred
14520 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
14521 control (instead of _ARCH_PWR9 control) so that template
14522 definition uses power9-specific function.
14523 (vec_any_eq): Likewise.
14524 (vec_all_ne): Change macro definition to use a power9-specific
14525 expansion under #ifdef __POWER9_VECTOR__ control (instead of
14526 _ARCH_PWR9 control).
14527 (vec_any_eq) Likewise.
14528 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
14529 expansion for CMPNEF to remove support for xvcmpnesp instruction.
14530 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
14531 support for xvcmpnedp instruction.
14532 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
14533 macro expansion so that Power9 implementation of vec_all_ne does
14534 not use the AltiVec predicate framework.
14535 (VCMPNEH_P): Likewise.
14536 (VCMPNEW_P): Likewise.
14537 (VCMPNED_P): Likewise.
14538 (VCMPNEFP_P): Likewise.
14539 (VCMPNEDP_P): Likewise.
14540 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
14541 implementation of vec_any_eq to not use AltiVec predicate
14543 (VCMPAEH_P): Likewise.
14544 (VCMPAEW_P): Likewise.
14545 (VCMPAED_P): Likewise.
14546 (VCMPAEFP_P): Likewise.
14547 (VCMPAEDP_P): Likewise.
14548 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
14549 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
14550 not use the AltiVec predicate framework.
14551 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
14552 of vec_any_eq to not use AltiVec predicate framework.
14553 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
14554 support for predefined __POWER9_VECTOR__ macro to indicate that
14555 Power9 instruction selection is enabled.
14556 (altivec_overloaded_builtins): Remove extraneous
14557 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
14558 function argument types RS6000_BTI_bool_V16QI and
14559 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
14560 entry for overloaded function argument types RS6000_BTI_bool_V4SI
14561 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
14562 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
14563 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
14564 Power9 for implementations of vec_cmpne. Change the signature for
14565 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
14566 (representing vec_all_ne) to remove the previously described first
14567 argument of type RS6000_BTI_INTSI, as this was an artifact of
14568 reliance on the AltiVec predicate framework, which is no longer
14569 used in the implementation of these functions. Add
14570 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
14571 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
14572 since, unlike the AltiVec predicate framework implementation, we
14573 do not share function descriptors between vec_alle and vec_anyeq.
14574 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
14575 set of modes that receive special treatment even when
14576 TARGET_P9_VECTOR is true. The special treatment emits code that
14577 does not depend on Power9 instructions.
14578 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
14579 define_expand to not rely on AltiVec predicate framework.
14580 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
14582 (vector_ne_v2di_p): Change this define_expand to not rely on
14583 AltiVec predicate framework.
14584 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
14586 (vector_ne_<mode>_p): Change this define_expand to not rely on
14587 AltiVec predicate framework.
14588 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
14590 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
14591 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
14592 define_insn pattern.
14593 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
14594 define_insn pattern because the xvcmpne<VSs>. instruction is not
14596 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
14597 instruction is not supported.
14599 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14601 * config/nvptx/nvptx.c: Include intl.h.
14603 2017-03-01 Martin Jambor <mjambor@suse.cz>
14606 * ipa-prop.h (ipa_bits): Removed field known.
14607 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
14608 to pointers. Adjusted their comments to warn about their sharing.
14609 (ipcp_transformation_summary): Change bits to a vector of pointers.
14610 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
14611 (ipa_get_ipa_bits_for_value): Declare.
14612 * tree-vrp.h (value_range): Mark as GTY((for_user)).
14613 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
14614 (ipa_bits_hash_table): Likewise.
14615 (ipa_vr_ggc_hash_traits): Likewise.
14616 (ipa_vr_hash_table): Likewise.
14617 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
14618 being pointers and vr_known being removed.
14619 (ipa_set_jf_unknown): Likewise.
14620 (ipa_get_ipa_bits_for_value): New function.
14621 (ipa_set_jfunc_bits): Likewise.
14622 (ipa_get_value_range): New overloaded functions.
14623 (ipa_set_jfunc_vr): Likewise.
14624 (ipa_compute_jump_functions_for_edge): Use the above functions to
14625 construct bits and vr parts of jump functions.
14626 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
14627 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14629 (ipcp_grow_transformations_if_necessary): Also allocate
14630 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14632 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
14633 them. Fix too long lines.
14634 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
14635 vr_known being removed.
14636 (ipa_read_jump_function): Use new setter functions to construct bits
14637 and vr parts of jump functions or set them to NULL.
14638 (write_ipcp_transformation_info): Adjust for bits being pointers.
14639 (read_ipcp_transformation_info): Likewise.
14640 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
14642 Include gt-ipa-prop.h.
14643 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
14645 (ipcp_store_bits_results): Likewise.
14646 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
14647 Do not write to existing jump functions but use a temporary instead.
14649 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14652 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
14653 attempt to use its first operand as BIT_FIELD_REF base.
14655 2017-03-01 Richard Biener <rguenther@suse.de>
14657 PR middle-end/79721
14658 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
14659 interpolating formula in wrapping arithmetic.
14660 (chrec_apply): Convert chrec_evaluate return value to wanted type.
14662 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14664 PR tree-optimization/79734
14665 * tree-vect-generic.c (expand_vector_condition): Optimize
14666 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
14667 Handle VEC_COND_EXPR where comparison has different inner width from
14668 type's inner width.
14670 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
14672 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
14673 markup, and similar issues. Remove @opindex entries for things
14674 that aren't options. Add missing -mmpy-option entries.
14676 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14678 PR tree-optimization/79737
14679 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
14680 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
14681 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
14682 instead of byte_size. Formatting fix.
14683 (shift_bytes_in_array_right): Formatting fix.
14685 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
14688 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
14689 condition on optimize for the leaf function test.
14691 2017-02-28 Martin Liska <mliska@suse.cz>
14694 * read-rtl-function.c (function_reader::handle_unknown_directive):
14695 Bail out when one uses -flto.
14697 2017-02-28 Martin Liska <mliska@suse.cz>
14699 * common.opt: Replace space with tabular for options of <number>
14701 * config/i386/i386.opt: Show <number> value for
14702 -mlarge-data-threshold.
14703 * opts.c (print_filtered_help): Do not display number in hexadecimal
14706 2017-02-28 Martin Liska <mliska@suse.cz>
14708 * common.opt: Fix --help=option -Q for options which are of
14711 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
14713 * config/i386/i386.c (print_reg): Error out for values
14714 of 8-bit size in invalid integer register.
14716 2017-02-28 Martin Sebor <msebor@redhat.com>
14718 PR tree-optimization/79691
14719 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
14721 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14724 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
14725 gcc_unreachable with output_operand_lossage.
14727 2017-02-28 Richard Biener <rguenther@suse.de>
14729 PR tree-optimization/79740
14730 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
14732 (visit_nary_op): Insert the nary into the hashtable if we
14733 pattern-matched sth.
14734 * tree-ssa-pre.c (eliminate_insert): Robustify.
14736 2017-02-28 Richard Biener <rguenther@suse.de>
14738 PR middle-end/79731
14739 * fold-const.c (decode_field_reference): Reject out-of-bound
14742 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14744 * config/i386/i386.c: Include intl.h.
14745 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
14746 instead of just cond ? "..." : "...".
14747 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
14748 * coverage.c (read_counts_file): Likewise.
14749 * omp-offload.c: Include intl.h.
14750 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
14751 of just cond ? "..." : "...".
14752 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
14753 of just cond ? "..." : "...".
14755 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
14758 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
14760 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
14761 'tune for' CPU name.
14762 * config/arm/arm-cpu-data.h: Regenerated.
14764 2017-02-28 Richard Biener <rguenther@suse.de>
14766 PR tree-optimization/79732
14767 * tree-inline.c (expand_call_inline): Do not shadow var.
14769 2017-02-28 Richard Biener <rguenther@suse.de>
14771 PR tree-optimization/79723
14772 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
14773 address-space properly.
14775 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
14777 * doc/optinfo.texi (Optimization groups): Fix option used for
14779 * doc/invoke.texi (-fopt-info): Document "omp".
14780 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
14781 (OPTGROUP_ALL): Add OPTGROUP_OMP.
14782 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
14783 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
14784 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
14786 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
14788 * dumpfile.c (optgroup_options): Instead of "openmp", associate
14789 OPTGROUP_OMP with "omp".
14791 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
14794 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
14795 for arithmetic shift of unsigned V2DI.
14797 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
14799 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
14800 arc/linux.h headers.
14801 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
14802 (LINK_SPEC): Likewise.
14803 (ARC_TLS_EXTRA_START_SPEC): Likewise.
14804 (EXTRA_SPECS): Likewise.
14805 (STARTFILE_SPEC): Likewise.
14806 (ENDFILE_SPEC): Likewise.
14807 (LIB_SPEC): Likewise.
14808 (TARGET_SDATA_DEFAULT): Likewise.
14809 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
14810 (MULTILIB_DEFAULTS): Likewise.
14811 (DWARF2_UNWIND_INFO): Likewise.
14812 * config/arc/big.h: New file.
14813 * config/arc/elf.h: Likewise.
14814 * config/arc/linux.h: Likewise.
14815 * config/arc/t-uClibc: Remove.
14817 2017-02-27 Bin Cheng <bin.cheng@arm.com>
14819 PR tree-optimization/77536
14820 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
14821 (tree_transform_and_unroll_loop): Use above function to compute the
14822 estimated niter of unrolled loop and use it when scaling profile.
14823 Also use count info rather than frequency if it's non-zero.
14824 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
14825 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
14826 (vect_transform_loop): Call above function.
14828 2017-02-27 Richard Biener <rguenther@suse.de>
14830 PR tree-optimization/45397
14831 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
14832 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
14833 (visit_nary_op): Add pattern matching for CSEing sign-changed
14834 or truncated operations with wider ones.
14836 2017-02-27 Richard Biener <rguenther@suse.de>
14838 PR tree-optimization/79690
14839 * tree-vect-stmts.c (vectorizable_store): Use vector type
14840 built from the DR with address-space.
14842 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
14844 * doc/invoke.texi (Optimize Options): Refine the description
14845 of asan-use-after-return.
14847 2017-02-25 Alan Modra <amodra@gmail.com>
14849 PR rtl-optimization/79584
14850 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
14851 base, not ad->base_term, the reg within base. Remove assertion
14852 that ad->base == ad->base_term. Replace gen_int_mode using
14853 bogus mode with const0_rtx.
14855 2017-02-25 Jakub Jelinek <jakub@redhat.com>
14857 PR middle-end/79396
14858 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
14859 FMA_EXPR like tcc_binary or tcc_unary.
14861 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
14864 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
14866 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
14867 (output_loc_operands): Handle DW_OP_call_ref and
14868 DW_OP_GNU_variable_value.
14869 (struct variable_value_struct): New type.
14870 (struct variable_value_hasher): Likewise.
14871 (variable_value_hash): New variable.
14872 (string_types): Remove.
14873 (copy_loc_descr): New function.
14874 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
14875 (prepend_loc_descr_to_each): New function.
14876 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
14877 instead of add_loc_descr_to_each if the first argument is single
14878 location list and the second has multiple.
14879 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
14880 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
14881 when looking for variable value which doesn't have other location info.
14882 (loc_list_from_tree): Formatting fix.
14883 (gen_array_type_die): Simplify DW_AT_string_length handling.
14884 (adjust_string_types): Remove.
14885 (gen_subprogram_die): Don't call adjust_string_types nor test/set
14886 string_types. Call resolve_variable_values.
14887 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
14888 (resolve_addr_in_expr): Likewise. Add A argument.
14889 (copy_deref_exprloc): Remove deref argument. Adjust for the
14890 original expression being DW_OP_GNU_variable_value with optionally
14891 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
14892 optionally after it.
14893 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
14894 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
14895 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
14896 (variable_value_hasher::hash, variable_value_hasher::equal): New
14898 (resolve_variable_value_in_expr, resolve_variable_value,
14899 resolve_variable_values, note_variable_value_in_expr,
14900 note_variable_value): New functions.
14901 (dwarf2out_early_finish): Call note_variable_value on all toplevel
14904 2017-02-24 Jakub Jelinek <jakub@redhat.com>
14907 * opts.h (handle_generated_option): Add GENERATED_P argument.
14908 * opts-common.c (handle_option): Adjust function comment.
14909 (handle_generated_option): Add GENERATED_P argument, pass it to
14911 (control_warning_option): Pass false to handle_generated_option
14913 * opts.c (maybe_default_option): Pass true to handle_generated_option
14915 * optc-gen.awk: Likewise.
14917 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14919 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
14920 a REG, look at the REG it is a SUBREG of.
14921 (splitter for cmpeqsi_t): Ditto.
14923 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14925 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
14926 the special USEs with the pattern of the insn, not the insn itself.
14928 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
14931 * doc/invoke.texi: Document -mload-store-pairs.
14933 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14934 Sandra Loosemore <sandra@codesourcery.com>
14936 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
14937 argument isn't a CONST_INT.
14938 (nios2_alternate_compare_const): Assert op is a CONST_INT.
14939 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
14940 (nios2_validate_compare): Bypass alternate compare logic if *op2
14941 is not a CONST_INT.
14942 (ldstwm_operation_p): Return false if first_base is not a REG or
14943 if first_offset is not a CONST_INT.
14945 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14947 * config/cris/cris.md: Use correct operand in a define_peephole2.
14949 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14951 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
14953 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14955 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
14956 this_insn if it is an INSN or JUMP_INSN.
14957 (force_offsettable): Look at base, not at addr.
14958 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
14959 on things that aren't necessarily CONST_INTs.
14961 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
14963 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
14964 -mfpmath=sse is the default also for x86-32 targets with SSE2
14965 instruction set when @option{-ffast-math} is enabled
14967 2017-02-24 Jeff Law <law@redhat.com>
14969 PR rtl-optimizatoin/79286
14970 * ira.c (update_equiv_regs): Drop may_trap_p exception to
14973 2017-02-24 Richard Biener <rguenther@suse.de>
14975 PR tree-optimization/79389
14976 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
14979 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
14981 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
14982 function comment to reflect reality.
14983 (loop_exits_before_overflow): Fix typo in function description.
14985 2017-02-24 Richard Biener <rguenther@suse.de>
14987 PR tree-optimization/79389
14988 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
14989 properly that a threading opportunity exists. Detect conditional
14990 copy/constant propagation opportunities.
14992 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
14994 * config/visium/visium.md (type): Add trap.
14995 (b): New mode attribute.
14996 (*btst): Rename into...
14997 (*btst<mode>): ...this and adjust.
14998 (*cbranchsi4_btst_insn): Rename into...
14999 (*cbranch<mode>4_btst_insn): ...this and adjust.
15000 (trap): New define_insn.
15002 2017-02-23 Jakub Jelinek <jakub@redhat.com>
15004 PR tree-optimization/79389
15005 * ifcvt.c (struct noce_if_info): Add rev_cond field.
15006 (noce_reversed_cond_code): New function.
15007 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
15008 reversed_comparison_code. Formatting fix.
15009 (noce_try_store_flag): Test rev_cond != NULL in addition to
15010 reversed_comparison_code.
15011 (noce_try_store_flag_constants): Likewise.
15012 (noce_try_store_flag_mask): Likewise.
15013 (noce_try_addcc): Use rev_cond if non-NULL instead of
15014 reversed_comparison_code.
15015 (noce_try_cmove_arith): Likewise. Formatting fixes.
15016 (noce_try_minmax, noce_try_abs): Clear rev_cond.
15017 (noce_find_if_block): Initialize rev_cond.
15018 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
15019 instead of false as last argument never attempt to reverse it
15022 2017-02-23 Bin Cheng <bin.cheng@arm.com>
15024 PR tree-optimization/79663
15025 * tree-predcom.c (combine_chains): Process refs in reverse order
15026 only for ZERO length chains, and add explaining comment.
15028 2017-02-23 Jeff Law <law@redhat.com>
15030 PR tree-optimization/79578
15031 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
15032 in call to operand_equal_p.
15034 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
15037 * config/i386/cpuid.h: Fix another undefined behavior.
15039 2017-02-23 Richard Biener <rguenther@suse.de>
15041 PR tree-optimization/79683
15042 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
15043 vector types for data-refs.
15045 2017-02-23 Martin Liska <mliska@suse.cz>
15047 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
15049 2017-02-23 Jakub Jelinek <jakub@redhat.com>
15051 PR middle-end/79665
15052 * internal-fn.c (get_range_pos_neg): Moved to ...
15053 * tree.c (get_range_pos_neg): ... here. No longer static.
15054 * tree.h (get_range_pos_neg): New prototype.
15055 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
15056 are known to be in between 0 and signed maximum inclusive, try to
15057 expand both unsigned and signed divmod and use the cheaper one from
15060 2017-02-22 Jeff Law <law@redhat.com>
15062 PR tree-optimization/79578
15063 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
15064 to compare base operands.
15066 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
15069 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
15070 gpc_reg_operand instead of fpr_reg_operand.
15072 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
15074 * config/mips/mips.c (mips_return_in_memory): Force FP
15075 vector types to be returned in memory for o32 ABI.
15077 2017-02-22 Jakub Jelinek <jakub@redhat.com>
15079 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
15080 instead of DW_TAG_member for static data member declarations and don't
15081 set no_linkage_name for static inline data members.
15082 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
15085 2017-02-22 Martin Liska <mliska@suse.cz>
15087 * doc/invoke.texi: Replace inequality signs with square brackets
15090 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15092 PR tree-optimization/68644
15093 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
15095 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
15098 * lra-constraints.c (simplify_operand_subreg): Handle
15099 WORD_REGISTER_OPERATIONS targets.
15101 2017-02-22 Jakub Jelinek <jakub@redhat.com>
15104 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
15105 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
15106 elimination by swapping fld*.
15108 2017-02-22 Richard Biener <rguenther@suse.de>
15110 PR tree-optimization/79673
15111 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
15112 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
15113 irrelevant address-space qualifiers and avoiding a
15114 ADDR_SPACE_CONVERT_EXPR from fold_convert.
15116 2017-02-22 Richard Biener <rguenther@suse.de>
15118 PR tree-optimization/79666
15119 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
15120 to not symbolically negate if that may introduce undefined
15123 2017-02-22 Martin Liska <mliska@suse.cz>
15126 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
15127 * data-streamer-out.c (streamer_write_gcov_count_stream):
15129 * value-prof.c (stream_out_histogram_value): Make assert more
15130 precise based on type of counter.
15132 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
15135 * config/i386/i386.md (standard_x87sse_constant_load splitter):
15136 Use nonimmediate_operand instead of memory_operand for operand 1.
15137 (float-extend standard_x87sse_constant_load splitter): Ditto.
15139 2017-02-21 Jeff Law <law@redhat.com>
15141 PR tree-optimization/79621
15142 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
15143 blocks with edges to themselves.
15145 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15148 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
15149 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
15150 Use gimple_call_builtin_p.
15153 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
15154 on temporarily removed DEBUG_INSNs.
15156 PR tree-optimization/79649
15157 * tree-loop-distribution.c (classify_partition): Give up on
15158 non-generic address space loads/stores.
15160 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
15162 * doc/loop.texi (Loop manipulation): Remove nonexistent
15163 tree_ssa_loop_version from the documentation.
15164 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
15166 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15169 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
15170 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
15171 * config/rs6000/rs6000.c: Include except.h.
15172 (rs6000_expand_split_stack_prologue): Call
15173 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
15175 2017-02-21 Martin Jambor <mjambor@suse.cz>
15178 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
15179 have been analyzed.
15181 2017-02-21 Martin Jambor <mjambor@suse.cz>
15183 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
15184 for backward compatibility only.
15185 * doc/invoke.texi (Option Summary): Remove all references to
15186 -fipa-cp-alignment.
15188 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
15192 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15194 * lra-constraints.c (curr_insn_transform): Handle
15195 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
15197 2017-02-21 Martin Liska <mliska@suse.cz>
15199 * config/i386/i386.opt: Replace -masm-dialect with -masm.
15201 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
15203 PR translation/79638
15204 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
15206 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
15209 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
15210 (arm_function_ok_for_sibcall): Return false for an indirect call by
15211 descriptor if all the argument registers are used.
15212 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
15213 alignment of the function.
15215 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15217 PR tree-optimization/61441
15218 * simplify-rtx.c (simplify_const_unary_operation): For
15219 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
15220 the sNaN unmodified.
15222 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
15224 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
15225 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
15226 instead of SYSTEM_HEADER_DIR.
15228 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
15229 Martin Liška <mliska@suse.cz>
15231 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
15232 Fix typos and grammar, use active voice, and clarify.
15234 2017-02-20 Marek Polacek <polacek@redhat.com>
15236 PR middle-end/79537
15237 * gimplify.c (gimplify_expr): Handle unused *&&L;.
15240 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
15242 2017-02-20 Jakub Jelinek <jakub@redhat.com>
15245 * config/i386/i386.c (ix86_expand_builtin): Handle
15246 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
15247 ix86_builtins_isa[fcode].isa as a requirement of those
15248 flags and any other flag in the bitmask.
15249 (ix86_init_mmx_sse_builtins): Use 0 instead of
15250 ~OPTION_MASK_ISA_64BIT as mask.
15251 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
15252 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
15253 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
15254 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
15256 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15259 * lra-constraints.c (split_reg): Check requested split mode
15260 is supported by the register.
15262 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15264 * lra-constraints.c (simplify_operand_subreg): Remove early
15267 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15270 * lra-constraints.c (curr_insn_transform): Tighten condition
15271 for converting SUBREG reloads from OP_OUT to OP_INOUT.
15273 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15276 * lra-constraints.c (curr_insn_transform): Handle
15277 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
15279 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
15282 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
15284 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
15286 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
15289 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
15292 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15294 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
15297 2017-02-18 Jakub Jelinek <jakub@redhat.com>
15299 * final.c (last_columnnum, override_columnnum): New variables.
15300 (final_start_function): Set last_columnnum, pass it to begin_prologue
15301 hook and pass 0 to dwarf2out_begin_prologue.
15302 (final_scan_insn): Update override_columnnum. Pass last_columnnum
15303 to source_line debug hook.
15304 (notice_source_line): Compute last_columnnum and for debug_column_info
15305 return true on column changes.
15306 * debug.h (struct gcc_debug_hooks): Add column argument to
15307 source_line and begin_prologue hooks.
15308 (debug_nothing_int_charstar_int_bool): Remove prototype.
15309 (debug_nothing_int_int_charstar,
15310 debug_nothing_int_int_charstar_int_bool): New prototypes.
15311 (dwarf2out_begin_prologue): Add column argument.
15312 * debug.c (do_nothing_debug_hooks): Adjust source_line and
15313 begin_prologue hooks.
15314 (debug_nothing_int_charstar_int_bool): Remove.
15315 (debug_nothing_int_int_charstar,
15316 debug_nothing_int_int_charstar_int_bool): New functions.
15317 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
15318 through to dwarf2out_source_line.
15319 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
15320 (dwarf2out_source_line): Add column argument, emit it if requested.
15321 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
15323 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
15324 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
15325 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
15326 through to dwarf2out_begin_prologue.
15327 (vmsdbgout_source_line): Add column argument, pass it through to
15328 dwarf2out_source_line.
15329 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
15330 dbxout_source_line caller.
15331 (dbxout_source_line): Add column argument.
15333 * common.opt (gno-column-info, gcolumn-info): New options.
15334 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
15335 (check_die): Also test for multiple DW_AT_decl_column attributes.
15336 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
15337 DW_AT_decl_column if requested.
15338 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
15340 (gen_variable_die): Likewise.
15341 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
15342 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
15345 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
15346 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
15347 (ix86_handle_option): Handle OPT_m3dnowa.
15348 * doc/invoke.texi (-m3dnowa): Document.
15349 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
15350 -m3dnowa instead of -m3dnow -march=athlon.
15353 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
15354 instead of gcc_assert for K, r and R code checks. Formatting fixes.
15356 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15359 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
15360 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
15361 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
15362 generator for vsx_xxpermdi_<mode>_be.
15363 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
15364 force big-endian semantics.
15365 (vsx_xxpermdi_<mode>_be): New define_expand with same
15366 implementation as previous version of vsx_xxpermdi_<mode>.
15368 2017-02-17 Jakub Jelinek <jakub@redhat.com>
15370 PR tree-optimization/79327
15371 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
15372 variable, its initialization and use.
15374 2017-02-17 Julia Koval <julia.koval@intel.com>
15376 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
15377 (OPTION_MASK_ISA_PKU_UNSET): New.
15378 (ix86_handle_option): Handle -mrdpid.
15379 * config/i386/cpuid.h (bit_RDPID): New.
15380 * config/i386/driver-i386.c (host_detect_local_cpu):
15381 Detect RDPID feature.
15382 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
15383 * config/i386/i386-c.c (ix86_target_macros_internal):
15385 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
15386 (ix86_valid_target_attribute_inner_p): Add "rdpid".
15387 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
15388 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
15389 * config/i386/i386.md (define_insn "rdpid"): New.
15390 * config/i386/i386.opt Add -mrdpid.
15391 * config/i386/immintrin.h (_rdpid_u32): New.
15393 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
15395 PR rtl-optimization/79541
15396 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
15397 instead of transforming it into USE.
15399 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
15401 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
15402 If HONOR_SNANS (SFmode) force the input to a register.
15403 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
15404 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
15405 an frsp or similar insn.
15407 2017-02-17 Martin Liska <mliska@suse.cz>
15409 PR rtl-optimization/79577
15410 * params.def (selsched-max-sched-times): Increase minimum to 1.
15412 2017-02-17 Martin Liska <mliska@suse.cz>
15414 PR rtl-optimization/79574
15415 * gcse.c (want_to_gcse_p): Prevent integer overflow.
15417 2017-02-17 Martin Liska <mliska@suse.cz>
15419 PR tree-optimization/79529
15420 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
15421 ssa_defined_default_def_p to handle cases which are implicitly
15423 * tree-ssa.c (ssa_defined_default_def_p): New function.
15424 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
15425 which are implicitly defined.
15426 * tree-ssa.h (ssa_defined_default_def_p): Declare.
15428 2017-02-17 Richard Biener <rguenther@suse.de>
15430 PR middle-end/79576
15431 * params.def (max-ssa-name-query-depth): Limit to 10.
15433 2017-02-17 Richard Biener <rguenther@suse.de>
15435 PR tree-optimization/79552
15436 * tree-ssa-structalias.c (visit_loadstore): Properly verify
15439 2017-02-17 Richard Biener <rguenther@suse.de>
15442 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
15444 2017-02-17 Marek Polacek <polacek@redhat.com>
15446 PR middle-end/79536
15447 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
15448 (fold_negate_expr): New wrapper.
15450 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
15452 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
15453 Correct terminology and de-emphasize pre-standard behavior.
15455 2017-02-16 Alan Modra <amodra@gmail.com>
15457 PR rtl-optimization/79286
15458 * ira.c (def_dominates_uses): New function.
15459 (update_equiv_regs): Don't create an equivalence for insns that
15460 may trap where the register def does not dominate the use.
15462 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
15464 PR rtl-optimization/78127
15465 * lra.c (lra): Call lra_eliminate before finish the loop after
15468 2017-02-16 Richard Biener <rguenther@suse.de>
15470 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
15472 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
15473 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
15474 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
15475 (isl_val_int_from_wi): New function.
15476 (extract_affine_gmp): Rename to ...
15477 (extract_affine_wi): ... this, take a widest_int.
15478 (extract_affine_int): Just wrap extract_affine_wi.
15479 (add_param_constraints): Use isl_val_int_from_wi.
15480 (add_loop_constraints): Likewise, and extract_affine_wi.
15482 2017-02-15 Jeff Law <law@redhat.com>
15484 PR middle-end/79521
15485 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
15486 ira_init_register_move_cost_if_necessary.
15488 2017-02-15 Martin Sebor <msebor@redhat.com>
15490 PR middle-end/32003
15491 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
15492 removed in a prior commit.
15494 2017-02-15 Bin Cheng <bin.cheng@arm.com>
15496 PR tree-optimization/79347
15497 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
15498 counters during peeling.
15500 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
15502 * Makefile.in (site.exp): Remove "set ISLVER".
15504 2017-02-15 Jakub Jelinek <jakub@redhat.com>
15507 * real.c (real_from_integer): Call real_convert even for decimal.
15509 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
15512 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
15514 2017-02-14 Andrew Pinski <apinski@cavium.com>
15516 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
15517 cores and change the partno/implementer to be correct.
15518 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
15519 the 'B" as the implementer.
15520 * config/aarch64/aarch64-tune.md: Regenerate.
15522 2017-02-14 Carl Love <cel@us.ibm.com>
15524 * config/rs6000/rs6000.c: Add case statement entry to make the
15525 xvcvuxdsp built-in argument unsigned.
15526 * config/rs6000/vsx.md: Fix the source and return operand types so they
15527 match the instruction definitions from the ISA document. Fix typo
15528 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
15531 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
15534 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
15535 member early_clobber_alts.
15536 * lra-lives.c (reg_early_clobber_p): New.
15537 (process_bb_lives): Use it.
15538 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
15539 (debug_operand_data): Initialize early_clobber_alts.
15540 (setup_operand_alternative): Set up early_clobber_alts.
15541 (collect_non_operand_hard_regs): Ditto. Pass early clobber
15542 alternatives to new_insn_reg.
15543 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
15545 (lra_update_insn_regno_info): Pass the new arg.
15547 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15549 PR middle-end/79505
15550 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
15551 (new_oacc_loop_raw): Don't clear already cleared fields.
15554 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
15555 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
15556 _mm512_prefetch_i64gather_ps): New inline functions and macros.
15558 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
15561 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
15563 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
15566 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
15567 the extra instruction to the right place to store 128-bit constant
15570 2017-02-14 Martin Sebor <msebor@redhat.com>
15572 PR middle-end/79448
15573 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
15574 warning for strings of unknown length.
15576 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
15578 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
15580 2017-02-14 Jeff Law <law@redhat.com>
15583 * ira-costs.c (scan_one_insn): Initialize register move costs
15584 for pseudos seen in USE/CLOBBER insns.
15586 PR tree-optimization/79095
15587 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
15588 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
15589 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
15590 if the operands are known to be not equal, then the resulting range
15592 (intersect_ranges): If the new range is ~[0,0] and the old range is
15593 wide, then prefer ~[0,0].
15594 * tree-vrp.c (overflow_comparison_p_1): New function.
15595 (overflow_comparison_p): New function.
15596 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
15597 if NAME is used in an overflow test.
15598 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
15599 overflow check that can be expressed as an equality test, then adjust
15600 ops to be that equality test.
15602 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15604 * config/s390/s390-builtin-types.def: Remove flags argument.
15605 * config/s390/s390.c (s390_init_builtins): Likewise.
15607 2017-02-14 Martin Liska <mliska@suse.cz>
15609 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
15610 vector. Fix trailing white spaces.
15612 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
15614 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
15617 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15619 PR rtl-optimization/68664
15620 * config/arm/arm.c (arm_sched_can_speculate_insn):
15621 New function. Declare prototype.
15622 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15624 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15626 PR rtl-optimization/68664
15627 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
15629 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15631 2017-02-14 Amit Pawar <amit.pawar@amd.com>
15633 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
15634 max skip bytes for function, loop and jump.
15636 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15638 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
15639 ABS_EXPR for gimple dump.
15641 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15644 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
15646 PR tree-optimization/79408
15647 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
15648 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
15649 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
15650 also if rhs1 is INTEGER_CST.
15652 2017-02-14 Richard Biener <rguenther@suse.de>
15654 PR middle-end/79432
15655 * tree-into-ssa.c (insert_phi_nodes): When the function can
15656 have abnormal edges rewrite SSA names with broken use-def
15657 dominance out of SSA and register them for PHI insertion.
15659 2017-02-13 Martin Sebor <msebor@redhat.com>
15661 PR middle-end/79496
15662 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
15663 clearing info.nowrite flag when snprintf size argument is a range.
15665 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15667 * cprop.c (cprop_jump): Add missing space in string literal.
15668 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
15669 (get_constraint_for_component_ref): Likewise.
15670 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
15671 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
15672 * lra-constraints.c (process_alt_operands): Likewise.
15673 * ipa-inline.c (inline_small_functions): Likewise.
15674 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
15675 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15676 * trans-mem.c (diagnose_tm_1_op): Likewise.
15677 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
15678 (grid_parallel_clauses_gridifiable): Likewise.
15680 * config/nvptx/mkoffload.c (process): Add space in between
15683 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
15684 "MOD4_SSE_REGS" and "ALL_REGS".
15686 * spellcheck.c (test_data): Add , in between "foo" and "food".
15688 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15691 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
15692 boundary crossing check and subsequent code generation agree.
15694 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15696 * config/aarch64/aarch64.c (has_memory_op): Delete.
15697 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
15700 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15702 PR rtl-optimization/79388
15703 PR rtl-optimization/79450
15704 * combine.c (distribute_notes): When removing TEM_INSN for which
15705 corresponding dest has last value recorded, invalidate that last
15708 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15710 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
15711 of explicit '@'. Add missing assembly comment marker on branch costs
15714 2017-02-13 Nathan Sidwell <nathan@acm.org>
15716 * gengtype-lex.l (<in_struct>): Add '/'.
15718 2017-02-13 Martin Liska <mliska@suse.cz>
15721 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
15723 2017-02-13 Richard Biener <rguenther@suse.de>
15725 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
15727 * configure: Re-generate.
15728 * config.in: Likewise.
15729 * graphite-dependences.c: Simplify as if
15730 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
15731 * graphite-isl-ast-to-gimple.c: Likewise.
15732 * graphite-optimize-isl.c: Likewise.
15733 * graphite-poly.c: Likewise.
15734 * graphite-sese-to-poly.c: Likewise.
15735 * graphite.h: Likewise.
15736 * toplev.c: Include isl/version.h and use isl_version () for
15737 printing the ISL version.
15738 * doc/install.texi: Update ISL requirement.
15740 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15742 * doc/standards.texi (Standards): Update reference to
15745 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15747 * doc/extend.texi (Named Address Spaces): sourceware.org now
15749 * doc/install.texi (Binaries): Ditto.
15752 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15754 * doc/cpp.texi: Replace "stringify"/"stringification" with C
15755 standard terminology "stringize"/"stringizing" throughout.
15756 * doc/cppinternals.texi: Likewise.
15758 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15760 * doc/extend.texi: Fix some spelling mistakes and typos.
15761 * doc/invoke.texi: Likewise.
15763 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15766 * params.def (inline-min-speedup) Change from 10 to 8.
15768 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15770 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
15773 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15776 * ipa-inline-analysis.c (get_minimal_bb): New function.
15777 (record_modified): Use it.
15778 (remap_edge_change_prob): Handle also ancestor functions.
15780 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
15782 * doc/contrib.texi (Contributors): Remove broken link into
15783 the Mauve CVS repository.
15785 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15787 PR middle-end/79454
15788 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
15789 result computation whenever lhs doesn't have vector mode, not
15790 just when it has BLKmode.
15792 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15794 * doc/makefile.texi (profiledbootstrap): Refer to the
15795 installation instructions only in textual form.
15797 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15800 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
15802 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15804 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
15805 (Specific): Update mingw-w64 reference.
15807 (Specific): Remove broken link to Renesas RX processor.
15809 2017-02-10 Richard Biener <rguenther@suse.de>
15811 * toplev.c (process_options): Do not mention obsolete graphite
15812 options when printing sorry message about missing graphite support.
15813 Mention -floop-nest-optimize.
15815 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
15817 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
15818 (vtst_p16): Likewise.
15819 (vtstq_p8): Likewise.
15820 (vtstq_p16): Likewise.
15822 (vtstq_p64): Likewise.
15823 * config/arm/arm_neon.h (vgetq_lane_p64): New.
15824 (vset_lane_p64): New.
15825 (vsetq_lane_p64): New.
15827 2017-02-10 Jakub Jelinek <jakub@redhat.com>
15829 PR tree-optimization/79411
15830 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
15831 stmt operands are SSA_NAMEs used in abnormal phis.
15832 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
15835 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15838 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
15841 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15843 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
15845 2017-02-09 Jakub Jelinek <jakub@redhat.com>
15847 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
15851 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
15852 not arbitrary TREE_CONSTANT.
15855 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
15856 "omp declare target link" attribute unless is_global_var.
15857 * omp-offload.c (find_link_var_op): Likewise.
15859 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
15860 Chung-Lin Tang <cltang@codesourcery.com>
15862 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
15864 (gimplify_adjust_omp_clauses): Don't delete TILE.
15865 (gimplify_omp_for): Deal with TILE.
15866 * internal-fn.c (expand_GOACC_TILE): New function.
15867 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
15869 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
15870 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
15872 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
15873 avoid DIV for outermost collapse var.
15874 (expand_oacc_for): Insert tile element loop as needed. Adjust.
15875 Remove out of date comments, fix whitespace.
15876 * omp-general.c (omp_extract_for_data): Deal with tiling.
15877 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
15878 adjust OLF_DIM_BASE value.
15879 (struct omp_for_data): Add tiling field.
15880 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
15881 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
15882 for auto loops. Remove default auto determining, moved to
15883 oacc_loop_fixed_partitions.
15884 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
15885 stmts, add e_mask field.
15886 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
15887 (oacc_thread_numbers): Use oacc_dim_call.
15888 (oacc_xform_tile): New.
15889 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
15890 (finish_oacc_loop): Adjust for ifns vector.
15891 (oacc_loop_discover_walk): Append loop abstraction sites to list,
15892 add case for GOACC_TILE fns.
15893 (oacc_loop_xform_loop): Delete.
15894 (oacc_loop_process): Iterate over call list directly, and add
15895 handling for GOACC_TILE fns.
15896 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
15898 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
15899 vector partitioning to outer loops. Assign 2 partitions to loops
15900 when available. Add TILE handling.
15901 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
15902 (execite_oacc_device_lower): Process GOACC_TILE fns,
15903 ignore unknown specs.
15904 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
15905 * tree.c (omp_clause_num_ops): Adjust TILE ops.
15906 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
15908 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
15910 * configure.ac (ACX_BUGURL): Update.
15911 * configure: Regenerate.
15913 2017-02-09 Richard Biener <rguenther@suse.de>
15915 PR tree-optimization/69823
15916 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
15917 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
15919 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15921 * config/arc/arc-c.def: Add __NPS400__ definition.
15922 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
15923 (TARGET_NPS400): Define.
15925 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15927 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
15929 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
15930 pointer, arch_info.
15931 (arc_cpu_types): Fill the arch_info field with a pointer into the
15932 arc_arch_types table.
15933 (arc_selected_cpu): Declare.
15934 * config/arc/arc.c (arc_selected_cpu): Make global.
15935 (arc_selected_arch): Delete.
15936 (arc_base_cpu): Delete.
15937 (arc_override_options): Remove references to deleted variables,
15938 update access to arch information.
15939 (ARC_OPT): Update access to arch information.
15940 (ARC_OPTX): Likewise.
15941 * config/arc/arc.h (arc_base_cpu): Remove declaration.
15942 (TARGET_ARC600): Update access to arch information.
15943 (TARGET_ARC601): Likewise.
15944 (TARGET_ARC700): Likewise.
15945 (TARGET_EM): Likewise.
15946 (TARGET_HS): Likewise.
15947 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
15950 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
15953 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
15954 condition/operands for integer GE/LE/GEU/LEU operations.
15956 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
15958 PR translation/79397
15959 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
15962 2017-02-08 Martin Jambor <mjambor@suse.cz>
15965 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
15966 whether allocation happened.
15967 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
15968 nothing was allocated.
15970 2017-02-08 Jakub Jelinek <jakub@redhat.com>
15972 PR tree-optimization/79408
15973 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
15974 constant, but SSA_NAME with a known integer range, use the minimum
15975 of that range instead of op1 to determine if modulo can be replaced
15976 with its first operand.
15978 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15980 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
15982 2017-02-08 Richard Biener <rguenther@suse.de>
15984 PR tree-optimization/71824
15985 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
15986 Check all loops contained in the merged region.
15988 2017-02-07 Andrew Pinski <apinski@cavium.com>
15990 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
15992 2017-02-07 Andrew Pinski <apinski@cavium.com>
15994 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
15995 (thunderxt88): Likewise.
15996 (thunderxt81): Disable LSE and change v8.1 to v8.
15997 (thunderxt83): Likewise.
15999 2017-02-07 Jakub Jelinek <jakub@redhat.com>
16000 Richard Biener <rguenther@suse.de>
16002 PR middle-end/79399
16003 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
16004 type from int to size_t.
16005 * ira-costs.c (struct_costs_size): Change type from int to size_t.
16007 2017-02-07 Jakub Jelinek <jakub@redhat.com>
16009 PR rtl-optimization/79386
16010 * cprop.c (bypass_conditional_jumps): Initialize
16011 bypass_last_basic_block already before splitting bbs after
16012 unconditional traps...
16013 (bypass_conditional_jumps): ... rather than here.
16016 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
16017 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
16018 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
16019 fix -masm=intel patterns.
16021 2017-02-07 Richard Biener <rguenther@suse.de>
16023 PR tree-optimization/79256
16024 PR middle-end/79278
16025 * builtins.c (get_object_alignment_2): Use min_align_of_type
16026 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
16027 and ADJUST_FIELD_ALIGN.
16029 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
16031 * doc/tm.texi: Regenerate.
16032 * stor-layout.c (layout_decl): Adjust.
16033 (update_alignment_for_field): Likewise.
16034 (place_field): Likewise.
16035 (min_align_of_type): Likewise.
16036 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
16037 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
16038 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
16039 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
16040 * config/frv/frv.c (frv_adjust_field_align): Likewise.
16041 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
16042 * config/i386/i386.c (x86_field_alignment): Likewise.
16043 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
16044 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
16045 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
16046 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
16047 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
16048 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
16052 2017-01-30 Richard Biener <rguenther@suse.de>
16054 PR tree-optimization/79256
16055 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
16056 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
16059 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
16061 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
16062 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
16063 builtins to SImode and emit a zero-extend, if necessary.
16065 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
16067 * docs/invoke.texi (RISC-V Options): Alphabetize.
16069 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
16071 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
16074 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
16076 * config/riscv/riscv.c: New file.
16077 * common/config/riscv/riscv-common.c: Likewise.
16078 * config.gcc: Likewise.
16079 * config/riscv/constraints.md: Likewise.
16080 * config/riscv/elf.h: Likewise.
16081 * config/riscv/generic.md: Likewise.
16082 * config/riscv/linux.h: Likewise.
16083 * config/riscv/multilib-generator: Likewise.
16084 * config/riscv/peephole.md: Likewise.
16085 * config/riscv/pic.md: Likewise.
16086 * config/riscv/predicates.md: Likewise.
16087 * config/riscv/riscv-builtins.c: Likewise.
16088 * config/riscv/riscv-c.c: Likewise.
16089 * config/riscv/riscv-ftypes.def: Likewise.
16090 * config/riscv/riscv-modes.def: Likewise.
16091 * config/riscv/riscv-opts.h: Likewise.
16092 * config/riscv/riscv-protos.h: Likewise.
16093 * config/riscv/riscv.h: Likewise.
16094 * config/riscv/riscv.md: Likewise.
16095 * config/riscv/riscv.opt: Likewise.
16096 * config/riscv/sync.md: Likewise.
16097 * config/riscv/t-elf-multilib: Likewise.
16098 * config/riscv/t-linux: Likewise.
16099 * config/riscv/t-linux-multilib: Likewise.
16100 * config/riscv/t-riscv: Likewise.
16101 * configure.ac: Likewise.
16102 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
16103 Waterman as RISC-V maintainers.
16104 * doc/install.texi: Add RISC-V entries.
16105 * doc/invoke.texi: Add RISC-V options section.
16106 * doc/md.texi: Add RISC-V constraints section.
16107 * configure: Regenerated.
16109 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
16112 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
16113 false values to be constant vectors with all 0 or all 1 bits set.
16114 (vcondu<mode><mode>): Likewise.
16115 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
16117 (fpmask_comparison_operator): Update comment.
16118 (vecint_comparison_operator): New predicate.
16119 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
16120 vector conditionals when the true and false values are constant
16121 vectors with all 0 bits or all 1 bits set.
16123 2017-02-06 Martin Sebor <msebor@redhat.com>
16125 PR tree-optimization/79376
16126 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
16128 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
16130 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
16131 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
16132 to simplify split condition.
16134 2017-02-06 Jakub Jelinek <jakub@redhat.com>
16136 * omp-expand.c (oxpand_omp_atomic_fetch_op,
16137 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
16140 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
16142 PR rtl-optimization/68664
16143 * target.def (can_speculate_insn): New hook.
16144 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
16145 * doc/tm.texi: Regenerate.
16146 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
16147 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
16148 (rs6000_sched_can_speculate_insn): New function.
16150 2017-02-06 Jakub Jelinek <jakub@redhat.com>
16152 PR tree-optimization/79284
16153 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
16154 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
16155 vectorizable_mask_load_store, vectorizable_operation,
16156 vect_is_simple_cond, get_same_sized_vectype): Use it instead
16157 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
16158 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
16159 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
16160 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
16161 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
16162 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
16163 is_gimple_assign (stmt). Replace another such test with
16164 is_gimple_assign (stmt).
16166 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
16169 * config/avr/avr.c (rtl-iter.h): Include it.
16170 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
16171 (avr_legitimate_combined_insn): ...and implementation.
16173 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
16175 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
16176 * config/s390/s390.c (s390_const_operand_ok)
16177 (s390_canonicalize_comparison, s390_extract_part)
16178 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
16179 (s390_contiguous_bitmask_p, s390_rtx_costs)
16180 (legitimize_pic_address): Likewise.
16181 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
16182 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
16183 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
16184 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
16185 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
16187 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
16189 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
16190 REGNO($0) == REGNO($1).
16192 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16194 * config/s390/linux.h(SIZE_TYPE): Add comment.
16196 2017-02-06 Julian Brown <julian@codesourcery.com>
16197 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16198 Virendra Pathak <virendra.pathak@broadcom.com>
16200 * config/aarch64/aarch64-cores.def: Change the scheduler
16202 * config/aarch64/aarch64.md: Include thunderx2t99.md.
16203 * config/aarch64/thunderx2t99.md: New file.
16205 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16207 * doc/standards.texi (Go Language): Update link to language
16210 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
16212 * tree-eh.c (lower_resx): Sanitize profile.
16213 (cleanup_empty_eh_move_lp): Likewise.
16215 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
16218 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
16220 * cfgloopmanip.h (loop_version): Update prototype.
16221 * modulo-sched.c (sms_schedule): Update call of loop_version.
16222 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
16223 * tree-parloops.c (gen_parallel_loop): Likewise.
16224 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
16225 * tree-ssa-loop-split.c (split_loop): Likewise.
16226 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
16227 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
16229 2017-02-05 Martin Liska <mliska@suse.cz>
16232 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
16234 (print_operand_address): Initialize a struct to zero.
16236 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16238 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
16239 garbage collector only in textual form.
16241 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16243 * doc/extend.texi (x86 specific memory model extensions for
16244 transactional memory): Simplify a phrase.
16246 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
16249 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
16250 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
16251 (atomic_storedi_1): Likewise.
16253 2017-02-04 Jakub Jelinek <jakub@redhat.com>
16255 PR tree-optimization/79338
16256 * tree-parloops.c (gather_scalar_reductions): Don't call
16257 vect_analyze_loop_form for loop->inner before destroying loop's
16260 2017-02-03 Martin Sebor <msebor@redhat.com>
16262 PR tree-optimization/79327
16263 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
16264 when precision has resulted in leading zeros.
16265 (format_integer): Adjust the likely counter to assume an unknown
16266 argument that may be zero is non-zero.
16268 2017-02-03 Jason Merrill <jason@redhat.com>
16271 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
16272 avoid copying non-taken branch.
16274 2017-02-03 Jakub Jelinek <jakub@redhat.com>
16276 PR tree-optimization/79340
16277 * tree-vect-loop.c (vectorizable_reduction): Release
16278 vec_defs elements after safe_splicing them into other vectors.
16281 PR tree-optimization/79327
16282 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
16283 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
16285 (format_integer): Use wide_int_to_tree instead of build_int_cst
16286 + to_?hwi. If argmin is NULL, just set argmin and argmax to
16287 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
16288 of shortest and longest sequence.
16290 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
16292 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
16293 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
16295 2017-02-03 Walter Lee <walt@tilera.com>
16298 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
16299 after initial stackframe link reg save.
16300 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
16302 2017-02-03 Jakub Jelinek <jakub@redhat.com>
16305 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
16306 wu for stxssp alternative.
16308 2017-02-03 Martin Sebor <msebor@redhat.com>
16310 PR tree-optimization/79352
16311 * gimple-fold.c (get_range_strlen): Add argument.
16312 (get_range_strlen): Change return type to bool.
16313 (get_maxval_strlen): Pass in a dummy argument.
16314 * gimple-fold.h (get_range_strlen): Change return type to bool.
16315 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
16316 * tree.h (array_at_struct_end_p): Add argument.
16317 * tree.c (array_at_struct_end_p): Handle it.
16319 2017-02-03 Martin Liska <mliska@suse.cz>
16322 * multiple_target.c (create_dispatcher_calls): Redirect edge
16323 from a caller of a dispatcher.
16324 (expand_target_clones): Make the clones local.
16325 (ipa_target_clone): Do both target clones and resolvers.
16326 (ipa_dispatcher_calls): Remove the pass.
16327 (pass_dispatcher_calls::gate): Likewise.
16328 (make_pass_dispatcher_calls): Likewise.
16329 * passes.def (pass_target_clone): Put as very first IPA early
16332 2017-02-03 Martin Liska <mliska@suse.cz>
16334 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
16335 in case of a function with ifunc attribute.
16337 2017-02-03 Martin Liska <mliska@suse.cz>
16339 * cgraph.c (cgraph_node::dump): Dump function version info.
16340 * symtab.c (symtab_node::dump_base): Add missing new line.
16342 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
16344 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
16345 (ifcombine_ifandif): Use it.
16347 2017-02-03 Martin Liska <mliska@suse.cz>
16349 * doc/invoke.texi: Document default value for
16350 use-after-scope-direct-emission-threshold.
16352 2017-02-03 Martin Liska <mliska@suse.cz>
16354 PR tree-optimization/79339
16355 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
16356 (format_floating): Likewise.
16358 2017-02-03 Martin Liska <mliska@suse.cz>
16361 * ipa-prop.c (ipa_node_params_t::insert): Remove current
16363 (ipa_node_params_t::remove): Likewise.
16364 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
16365 initialization from removed ipa_node_params_t::insert.
16366 (ipa_node_params::~ipa_node_params): Move from removed
16367 ipa_node_params_t::release.
16368 * symbol-summary.h (symbol_summary::m_released): New member.
16369 Do not release a summary twice. Do not allow to call finalizer
16370 for types of a summary that live in GGC memory.
16372 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16374 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
16377 2017-02-02 Martin Sebor <msebor@redhat.com>
16379 PR middle-end/79275
16380 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
16381 (format_string): Tighten up the range of output for non-constant
16382 strings and correct the expected range for wide non-constant strings.
16384 2017-02-02 Martin Sebor <msebor@redhat.com>
16386 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
16388 PR middle-end/32003
16389 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
16391 (-fdump-tree-@var): Add to index and document how to come up
16392 with pass-specific option and dump file names.
16393 (-fdump-passes): Clarify where to look for output.
16395 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
16397 PR middle-end/77445
16398 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
16399 statistics of the analyzed path; allow threading for speed when
16400 any of BBs along the path are optimized for speed.
16402 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
16404 PR middle-end/78468
16405 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
16406 settings of the virtual registers.
16409 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
16411 * explow.c (get_dynamic_stack_size): Take known alignment of stack
16412 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
16415 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16417 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
16418 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
16420 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16422 * config/s390/s390.md: Add missing comments with the expanded
16424 * config/s390/vector.md: Likewise.
16425 * config/s390/vx-builtins.md: Likewise.
16427 2017-02-02 Jakub Jelinek <jakub@redhat.com>
16430 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
16431 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
16432 conditions on a single line.
16434 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16436 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
16437 __S390_VX__ to __VX__.
16439 2017-02-01 Andrew Pinski <apinski@cavium.com>
16441 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
16442 stmt_info to record_stmt_cost.
16443 (vect_get_known_peeling_cost): Pass stmt_info if known to
16445 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
16446 cpu_vector_cost field into
16447 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
16448 field into vec_int_stmt_cost and vec_fp_stmt_cost.
16449 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
16450 splitting of scalar_stmt_cost and vec_stmt_cost.
16451 (thunderx_vector_cost): Likewise.
16452 (cortexa57_vector_cost): LIkewise.
16453 (exynosm1_vector_cost): Likewise.
16454 (xgene1_vector_cost): Likewise.
16455 (thunderx2t99_vector_cost): Improve after the splitting of the two
16457 (aarch64_builtin_vectorization_cost): Update for the splitting of
16458 scalar_stmt_cost and vec_stmt_cost.
16460 2017-02-01 Torvald Riegel <triegel@redhat.com>
16461 Richard Henderson <rth@redhat.com>
16463 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
16464 conditional on existance of a fast atomic load.
16465 * optabs-query.c (can_atomic_load_p): New function.
16466 * optabs-query.h (can_atomic_load_p): Declare it.
16467 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
16468 no fast atomic load is available for the particular size of access.
16469 (expand_atomic_compare_and_swap): Likewise.
16470 (expand_atomic_load): Likewise.
16471 (expand_atomic_store): Likewise.
16472 (expand_atomic_fetch_op): Likewise.
16473 * testsuite/lib/target-supports.exp
16474 (check_effective_target_sync_int_128): Remove x86 because it provides
16475 no fast atomic load.
16476 (check_effective_target_sync_int_128_runtime): Likewise.
16478 2017-02-01 Richard Biener <rguenther@suse.de>
16480 * graphite.c: Include tree-vectorizer.h for find_loop_location.
16481 (graphite_transform_loops): Provide opt-info for optimized nests.
16482 * tree-parloop.c (parallelize_loops): Provide opt-info for
16483 parallelized loops.
16485 2017-02-01 Richard Biener <rguenther@suse.de>
16487 PR middle-end/79315
16488 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
16489 was not set before.
16491 2017-02-01 Richard Biener <rguenther@suse.de>
16493 PR tree-optimization/71824
16494 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
16495 Verify the loops are valid in the merged SESE region.
16496 (scop_detection::can_represent_loop_1): Check analyzing the
16497 evolution of the number of iterations in the region succeeds.
16499 2017-01-31 Ian Lance Taylor <iant@golang.org>
16501 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
16502 REG_ARGS_SIZE note to 32-bit push insns and call insn.
16504 2017-01-31 David Malcolm <dmalcolm@redhat.com>
16506 PR preprocessor/79210
16507 * input.c (get_substring_ranges_for_loc): Replace line_width
16508 assertion with error-handling.
16510 2017-01-31 Richard Biener <rguenther@suse.de>
16512 PR tree-optimization/77318
16513 * graphite-sese-to-poly.c (extract_affine): Fix assert.
16514 (create_pw_aff_from_tree): Take loop parameter.
16515 (add_condition_to_pbb): Pass loop of the condition to
16516 create_pw_aff_from_tree.
16518 2017-01-31 Jakub Jelinek <jakub@redhat.com>
16520 * config/s390/s390.c (s390_asan_shadow_offset): New function.
16521 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
16523 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
16527 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
16529 (convert_int_to_float128): Likewise.
16530 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
16531 (convert_int_to_float128): Likewise.
16532 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
16533 (UNSPEC_IEEE128_CONVERT): Likewise.
16534 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
16535 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
16536 Use local variables for IBM extended format.
16537 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
16538 (fix_trunc<mode>si2_fprs): Likewise.
16539 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
16540 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
16541 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
16542 to know that we can now have integers of all sizes in vector
16544 (fix<uns>_<mode>di2_hw): Likewise.
16545 (float<uns>_<mode>si2_hw): Likewise.
16546 (fix_<mode>si2_hw): Likewise.
16547 (fixuns_<mode>si2_hw): Likewise.
16548 (float<uns>_<mode>di2_hw): Likewise.
16549 (float_<mode>di2_hw): Likewise.
16550 (float_<mode>si2_hw): Likewise.
16551 (floatuns_<mode>di2_hw): Likewise.
16552 (floatuns_<mode>si2_hw): Likewise.
16553 (xscvqp<su>wz_<mode>): Delete, no longer used.
16554 (xscvqp<su>dz_<mode>): Likewise.
16555 (xscv<su>dqp_<mode>): Likewise.
16556 (ieee128_mfvsrd_64bit): Likewise.
16557 (ieee128_mfvsrd_32bit): Likewise.
16558 (ieee128_mfvsrwz): Likewise.
16559 (ieee128_mtvsrw): Likewise.
16560 (ieee128_mtvsrd_64bit): Likewise.
16561 (ieee128_mtvsrd_32bit): Likewise.
16563 2017-01-31 Martin Liska <mliska@suse.cz>
16566 * ipa-prop.c (ipa_free_all_node_params): Call release method
16567 instead of ~sumbol_summary to not to trigger double times
16570 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
16572 PR tree-optimization/71691
16573 * bitmap.h (class auto_bitmap): New.
16574 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
16575 is_maybe_undefined instead of ssa_undefined_value_p.
16577 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16579 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
16580 __S390_ARCH_LEVEL__ to __ARCH__.
16582 2017-01-31 Jakub Jelinek <jakub@redhat.com>
16584 PR tree-optimization/79267
16585 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
16586 if should_remove_lhs_p is true.
16588 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
16591 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
16592 (add_alignment_attribute): New.
16593 (base_type_die): Add alignment attribute.
16594 (subrange_type_die): Likewise.
16595 (modified_type_die): Likewise.
16596 (gen_array_type_die): Likewise.
16597 (gen_descr_array_type_die: Likewise.
16598 (gen_enumeration_type_die): Likewise.
16599 (gen_subprogram_die): Likewise.
16600 (gen_variable_die): Likewise.
16601 (gen_field_die): Likewise.
16602 (gen_ptr_to_mbr_type_die): Likewise.
16603 (gen_struct_or_union_type_die): Likewise.
16604 (gen_subroutine_type_die): Likewise.
16605 (gen_typedef_die): Likewise.
16606 (base_type_cmp): Compare alignment attribute.
16608 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16611 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
16612 (setb_unsigned) New pattern for setb with CCUNS.
16613 * config/rs6000/rs6000.c (expand_block_compare): Use a different
16614 subfc./subfe sequence to avoid overflow problems. Generate a
16615 shorter sequence with cmpld/setb for power9.
16616 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
16617 for generating subfc. instruction.
16618 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
16619 now uses this instruction.
16621 2017-01-30 Ian Lance Taylor <iant@google.com>
16624 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
16625 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
16627 2017-01-30 Martin Sebor <msebor@redhat.com>
16629 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
16630 Move constant to the right of a relational operator.
16631 (get_mpfr_format_length, format_character, format_string): Ditto.
16632 (should_warn_p, maybe_warn): Same.
16634 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
16636 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
16639 * asan.c (get_translation_unit_decl): Remove function.
16640 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
16642 2017-01-30 Martin Liska <mliska@suse.cz>
16644 PR gcov-profile/79259
16645 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
16646 -fprofile-generate.
16648 2017-01-30 Martin Liska <mliska@suse.cz>
16651 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
16652 Initialize variables with NULL value.
16654 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
16657 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
16659 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
16661 2017-01-30 Richard Biener <rguenther@suse.de>
16663 PR tree-optimization/79276
16664 * tree-vrp.c (process_assert_insertions): Properly adjust common
16665 when removing a duplicate.
16667 * gcc.dg/torture/pr79276.c: New testcase.
16669 2017-01-30 Richard Biener <rguenther@suse.de>
16671 PR tree-optimization/79256
16672 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
16673 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
16675 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
16677 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
16680 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
16681 ("*r<noxa>sbg_<mode>_sll_bitmask")
16682 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
16683 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
16684 Use contiguous_bitmask_nowrap_operand.
16686 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16689 * config/rs6000/altivec.h (vec_xl): Revise #define.
16690 (vec_xst): Likewise.
16692 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
16694 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
16696 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
16698 PR rtl-optimization/79194
16699 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
16700 traps before call to bypass_conditional_jumps.
16702 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16704 PR tree-optimization/71374
16705 * lra-constraints.c (check_conflict_input_operands): New.
16706 (match_reload): Use it.
16708 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16711 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
16712 account to calculate conflict_set.
16714 2017-01-27 Bin Cheng <bin.cheng@arm.com>
16716 PR rtl-optimization/78559
16717 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
16718 other_insn in combine.
16720 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16722 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
16723 uint16_type_node for BT_UINT16.
16725 2017-01-27 David Malcolm <dmalcolm@redhat.com>
16727 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
16728 "RTL Tests" to menu.
16729 (GIMPLE Tests): New node.
16730 (RTL Tests): New node.
16732 2017-01-27 Richard Biener <rguenther@suse.de>
16734 PR tree-optimization/79245
16735 * tree-loop-distribution.c (distribute_loop): Apply cost
16736 modeling also to detected patterns.
16738 2017-01-27 Richard Biener <rguenther@suse.de>
16740 PR tree-optimization/71433
16741 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
16742 (compare_assert_loc): New function.
16743 (process_assert_insertions): Sort and optimize assert locations
16744 to remove duplicates and push down identical assertions on
16745 edges to their destination block.
16747 2017-01-27 Richard Biener <rguenther@suse.de>
16749 PR tree-optimization/79244
16750 * tree-vrp.c (remove_range_assertions): Forcefully propagate
16751 out SSA names even if abnormal.
16753 2017-01-27 Jakub Jelinek <jakub@redhat.com>
16755 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
16756 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
16757 instead of MPFR_RNDN.
16759 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
16762 * arm.c (arm_option_override): Don't call build_target_option_node
16763 until after doing all option overrides.
16764 (arm_valid_target_attribute_tree): Likewise.
16766 2017-01-27 Martin Liska <mliska@suse.cz>
16768 * doc/invoke.texi (-fprofile-arcs): Document profiling support
16769 for {cd}tors and C++ {cd}tors.
16771 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
16773 * config/s390/s390.md ("*setmem_long_and")
16774 ("*setmem_long_and_31z"): Use zero_extend instead of and.
16776 2017-01-26 Martin Sebor <msebor@redhat.com>
16778 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
16781 2017-01-26 Martin Sebor <msebor@redhat.com>
16783 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
16784 HAVE_DFmode before using XFmode or DFmode.
16785 (parse_directive): Avoid using the z length modifier to avoid
16786 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
16788 PR middle-end/78703
16789 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
16790 to accept adjustment as an array.
16791 (get_int_range): New function.
16792 (struct directive): Make width and prec arrays.
16793 (directive::set_width, directive::set_precision): Call get_int_range.
16794 (format_integer, format_floating): Handle width and precision ranges.
16795 (format_string, parse_directive): Same.
16797 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16800 * dwarf2out.c (generate_skeleton_bottom_up): For children with
16801 comdat_type_p set, just clone them, but keep the children in the
16805 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
16806 which have direct callers with -fvar-tracking-assignments enabled
16808 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
16809 inside of type units.
16811 2017-01-26 Martin Sebor <msebor@redhat.com>
16813 PR middle-end/78703
16814 * gimple-ssa-sprintf.c (struct result_range): Add likely and
16816 (struct format_result): Replace number_chars, number_chars_min,
16817 and number_chars_max with a single member of struct result_range.
16819 (format_result::operator+=): Adjust.
16820 (struct fmtresult): Remove bounded. Handle likely and unlikely
16822 (fmtresult::adjust_for_width_or_precision): New function.
16823 (fmtresult:type_max_digits): New function.
16824 (bytes_remaining): Handle likely and unlikely counters.
16825 (min_bytes_remaining): Remove.
16826 (format_percent): Simplify.
16827 (format_integer, format_floating): Set likely and unlikely counters.
16828 (get_string_length, format_character, format_string): Same.
16829 (format_plain, should_warn_p): New function.
16830 (maybe_warn): Call should_warn_p. Update diagnostic messages
16831 and handle those for all directives, including plain strings.
16832 (format_directive): Handle likely and unlikely counters.
16833 Remove unnecessary quoting from diagnostics. Add an informational
16835 (add_bytes): Remove.
16836 (pass_sprintf_length::compute_format_length): Simplify.
16837 (try_substitute_return_value): Handle likely and unlikely counters.
16839 2017-01-26 Carl Love <cel@us.ibm.com>
16841 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
16842 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
16844 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
16847 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
16848 endianess for subregs into account.
16849 * lra-constraints.c (lra_constraints): Do risky transformations
16850 always on the first iteration.
16851 * lra-lives.c (check_pseudos_live_through_calls): Add arg
16852 last_call_used_reg_set.
16853 (process_bb_lives): Define and use last_call_used_reg_set.
16854 * lra.c (lra): Always continue after lra_constraints on the first
16857 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
16859 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
16861 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
16863 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16865 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
16866 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
16867 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
16868 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
16869 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
16870 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
16871 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
16872 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
16873 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
16875 2017-01-26 Marek Polacek <polacek@redhat.com>
16878 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
16879 for the third operand.
16881 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16883 PR middle-end/79236
16884 * omp-low.c (struct omp_context): Add simt_stmt field.
16885 (scan_omp_for): Return omp_context *.
16886 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
16887 context to the _simt_ SIMD stmt.
16888 (lower_omp_for): For combined SIMD with sibling _simt_
16889 SIMD, make sure to use the same decls in _looptemp_
16890 clauses as in the sibling.
16892 2017-01-26 David Sherwood <david.sherwood@arm.com>
16894 PR middle-end/79212
16895 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
16898 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16901 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
16902 emit fld b; fld a; if possible.
16904 * brig-builtins.def: Update copyright years.
16905 * config/arm/arm_acle_builtins.def: Update copyright years.
16907 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16910 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
16911 constraint instead of o for the stxsd instruction.
16913 2017-01-25 Carl Love <cel@us.ibm.com>
16915 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
16916 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
16918 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
16920 * doc/invoke.texi (C++ Dialect Options): Fix typo.
16922 2017-01-25 Richard Biener <rguenther@suse.de>
16924 PR tree-optimization/69264
16925 * target.def (vector_alignment_reachable): Improve documentation.
16926 * doc/tm.texi: Regenerate.
16927 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
16929 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
16930 earlier changes with respect to TYPE_USER_ALIGN.
16931 (vector_alignment_reachable_p): Likewise. Improve dumping.
16933 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16936 * config/arm/arm.md (xordi3): Force constant operand into a register
16939 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16941 * doc/invoke.texi (-fstore-merging): Correct default optimization
16942 levels at which it is enabled.
16943 (-O): Move -fstore-merging from list to...
16946 2017-01-25 Richard Biener <rguenther@suse.de>
16949 * omp-expand.c: Include debug.h.
16950 (expand_omp_taskreg): Make sure to generate early debug before
16951 outlining anything from a function.
16952 (expand_omp_target): Likewise.
16953 (grid_expand_target_grid_body): Likewise.
16955 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
16958 * asan.c (get_translation_unit_decl): New function.
16959 (asan_add_global): Extract modules file name from globals
16960 TRANSLATION_UNIT_DECL name.
16962 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
16965 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
16966 for long calls with APCS frame and VFP.
16968 2017-01-24 David Malcolm <dmalcolm@redhat.com>
16970 * cfg.c (original_copy_tables_initialized_p): New function.
16971 * cfg.h (original_copy_tables_initialized_p): New decl.
16972 * cfgrtl.c (relink_block_chain): Guard the call to
16973 free_original_copy_tables with a call to
16974 original_copy_tables_initialized_p.
16975 * cgraph.h (symtab_node::native_rtl_p): New decl.
16976 * cgraphunit.c (symtab_node::native_rtl_p): New function.
16977 (symtab_node::needed_p): Don't assert for early assembly output
16978 for __RTL functions.
16979 (cgraph_node::finalize_function): Set "force_output" for __RTL
16981 (cgraph_node::analyze): Bail out early for __RTL functions.
16982 (analyze_functions): Update assertion to support __RTL functions.
16983 (cgraph_node::expand): Bail out early for __RTL functions.
16984 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
16986 * function.h (struct function): Update comment for field
16988 * gimple-expr.c: Include "tree-pass.h".
16989 (gimple_has_body_p): Return false for __RTL functions.
16990 * Makefile.in (OBJS): Add run-rtl-passes.o.
16991 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
16993 (gcc::pass_manager::get_clean_slate): New accessor.
16994 * passes.c: Include "insn-addr.h".
16995 (should_skip_pass_p): Add logging. Update logic for running
16996 "expand" to be compatible with both __GIMPLE and __RTL. Guard
16997 property-provider override so it is only done for gimple passes.
16999 (skip_pass): New function.
17000 (execute_one_pass): Call skip_pass when skipping passes.
17001 * read-md.c (md_reader::read_char): Support filtering
17002 the input to a subset of line numbers.
17003 (md_reader::md_reader): Initialize fields
17004 m_first_line and m_last_line.
17005 (md_reader::read_file_fragment): New function.
17006 * read-md.h (md_reader::read_file_fragment): New decl.
17007 (md_reader::m_first_line): New field.
17008 (md_reader::m_last_line): New field.
17009 * read-rtl-function.c (function_reader::create_function): Only
17010 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
17011 curr_properties. Set DECL_INITIAL to a dummy block.
17012 (read_rtl_function_body_from_file_range): New function.
17013 * read-rtl-function.h (read_rtl_function_body_from_file_range):
17015 * run-rtl-passes.c: New file.
17016 * run-rtl-passes.h: New file.
17018 2017-01-24 Jeff Law <law@redhat.com>
17020 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
17023 2017-01-24 Bin Cheng <bin.cheng@arm.com>
17025 PR tree-optimization/79159
17026 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
17027 (record_nonwrapping_iv): Improve boundary using above function if no
17028 value range information.
17030 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
17031 Martin Jambor <mjambor@suse.cz>
17033 * brig-builtins.def: New file.
17034 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
17035 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
17036 (DEF_HSAIL_SAT_BUILTIN): Likewise.
17037 (DEF_HSAIL_INTR_BUILTIN): Likewise.
17038 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
17039 * builtin-types.def (BT_INT8): New.
17040 (BT_INT16): Likewise.
17041 (BT_UINT8): Likewise.
17042 (BT_UINT16): Likewise.
17043 (BT_FN_ULONG): Likewise.
17044 (BT_FN_UINT_INT): Likewise.
17045 (BT_FN_UINT_ULONG): Likewise.
17046 (BT_FN_UINT_LONG): Likewise.
17047 (BT_FN_UINT_PTR): Likewise.
17048 (BT_FN_ULONG_PTR): Likewise.
17049 (BT_FN_INT8_FLOAT): Likewise.
17050 (BT_FN_INT16_FLOAT): Likewise.
17051 (BT_FN_UINT32_FLOAT): Likewise.
17052 (BT_FN_UINT16_FLOAT): Likewise.
17053 (BT_FN_UINT8_FLOAT): Likewise.
17054 (BT_FN_UINT64_FLOAT): Likewise.
17055 (BT_FN_UINT16_UINT32): Likewise.
17056 (BT_FN_UINT32_UINT16): Likewise.
17057 (BT_FN_UINT16_UINT16_UINT16): Likewise.
17058 (BT_FN_INT_PTR_INT): Likewise.
17059 (BT_FN_UINT_PTR_UINT): Likewise.
17060 (BT_FN_LONG_PTR_LONG): Likewise.
17061 (BT_FN_ULONG_PTR_ULONG): Likewise.
17062 (BT_FN_VOID_UINT64_UINT64): Likewise.
17063 (BT_FN_UINT8_UINT8_UINT8): Likewise.
17064 (BT_FN_INT8_INT8_INT8): Likewise.
17065 (BT_FN_INT16_INT16_INT16): Likewise.
17066 (BT_FN_INT_INT_INT): Likewise.
17067 (BT_FN_UINT_FLOAT_UINT): Likewise.
17068 (BT_FN_FLOAT_UINT_UINT): Likewise.
17069 (BT_FN_ULONG_UINT_UINT): Likewise.
17070 (BT_FN_ULONG_UINT_PTR): Likewise.
17071 (BT_FN_ULONG_ULONG_ULONG): Likewise.
17072 (BT_FN_UINT_UINT_UINT): Likewise.
17073 (BT_FN_VOID_UINT_PTR): Likewise.
17074 (BT_FN_UINT_UINT_PTR: Likewise.
17075 (BT_FN_UINT32_UINT64_PTR): Likewise.
17076 (BT_FN_INT_INT_UINT_UINT): Likewise.
17077 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
17078 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
17079 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
17080 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
17081 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
17082 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
17083 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
17084 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
17085 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
17086 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
17087 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
17088 * doc/frontends.texi: List BRIG FE.
17089 * doc/install.texi (Testing): Add BRIG tesring requirements.
17090 * doc/invoke.texi (Overall Options): Mention BRIG.
17091 * doc/standards.texi (Standards): Doucment BRIG HSA version.
17093 2017-01-24 Richard Biener <rguenther@suse.de>
17095 PR translation/79208
17096 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
17098 2017-01-24 Martin Jambor <mjambor@suse.cz>
17101 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
17102 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
17103 and known_contexts.
17105 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
17107 PR middle-end/79123
17108 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
17109 casts from signed to unsigned really don't have a range.
17111 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
17113 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
17114 GMP_RNDx for compatiblity.
17116 2017-01-24 Martin Liska <mliska@suse.cz>
17119 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
17120 that would prevent us to call alloca with -1 as argument.
17122 2017-01-24 Jakub Jelinek <jakub@redhat.com>
17124 * dwarf2out.c (output_compilation_unit_header, output_file_names):
17125 Avoid -Wformat-security warning.
17127 2017-01-23 Andrew Pinski <apinski@cavium.com>
17129 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
17132 2017-01-23 Martin Sebor <msebor@redhat.com>
17134 PR middle-end/78703
17135 * gimple-ssa-sprintf.c (warn_level): New global.
17136 (format_integer): Use it here and throughout the rest of the file.
17137 Use the same switch to compute sign as base.
17138 (maybe_warn): New function.
17139 (format_directive): Factor out warnings into maybe_warn.
17140 Add debugging output. Use warn_level.
17141 (add_bytes): Use warn_level.
17142 (pass_sprintf_length::compute_format_length): Add debugging output.
17143 (try_substitute_return_value): Same.
17144 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
17146 PR middle-end/78703
17147 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
17148 (struct fmtresult, format_integer, format_floating): Adjust.
17149 (fmtresult::fmtresult): Set max correctly in two argument ctor.
17150 (get_string_length, format_string,format_directive): Same.
17151 (pass_sprintf_length::compute_format_length): Same.
17152 (try_substitute_return_value): Simplify slightly.
17154 PR middle-end/78703
17155 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
17156 (fmtresult::operator+=): Outlined.
17157 (struct fmtresult): Add ctors.
17158 (struct conversion_spec): Rename...
17159 (struct directive): ...to this. Add and remove data members.
17160 (directive::set_width, directive::set_precision): New functions.
17161 (format_percent): Use fmtresult ctor.
17162 (get_width_and_precision): Remove.
17163 (format_integer): Make naming changes. Avoid computing width and
17165 (format_floating): Same. Adjust indentation.
17166 (format_character, format_none): New functions.
17167 (format_string): Moved character handling to format_character.
17168 (format_directive): Remove arguments, change return type.
17169 (parse_directive): New function.
17170 (pass_sprintf_length::compute_format_length): Move directive
17171 parsing to parse_directive.
17173 2017-01-23 Jakub Jelinek <jakub@redhat.com>
17175 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
17176 (assign_assembler_name_if_needed): ... this.
17177 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
17178 (assign_assembler_name_if_needed): ... this.
17179 (free_lang_data_in_cgraph): Adjust callers.
17180 * cgraphunit.c (cgraph_node::analyze): Likewise.
17181 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
17184 2017-01-23 Richard Biener <rguenther@suse.de>
17186 PR tree-optimization/79088
17187 PR tree-optimization/79188
17188 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
17189 resetting loop bounds after last path deletion. Reset loop
17190 bounds of the target loop, make code match the comments.
17191 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
17192 Make sure loops need no fixups.
17194 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
17196 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
17197 exponent support with double type for first argument.
17198 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
17199 type returned by __builtin_vec_extract_sig,
17200 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
17201 functions from "vector int" to "vector unsigned int" or from
17202 "vector long long int" to "vector unsigned long long int".
17203 Changed type returned by __builtin_vec_extract_exp,
17204 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
17205 functions from "vector int" to "vector unsigned int" or from
17206 "vector long long int" to "vector unsigned long long int".
17207 Changed return type of __builtin_vec_test_data_class,
17208 __builtin_vec_test_data_class_sp, and
17209 __builtin_vec_test_data_class_dp from "vector int" to
17210 "vector bool int" or from "vector long long int" to "vector bool
17211 long long int" and changed second argument type from "unsigned
17212 int" to "int". Added new overloaded function forms "vector float
17213 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
17214 "vector float __builtin_vec_insert_exp_sp (vector float, vector
17215 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
17216 double, vector unsigned long long int)" and "vector double
17217 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
17218 long int)". Changed return type of
17219 __builtin_scalar_test_data_class and
17220 __builtin_scalar_test_data_class_sp and
17221 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
17222 int" and changed second argument from "unsigned int" to "int".
17223 Changed type returned by __builtin_scalar_test_neg,
17224 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
17225 from "int" to "bool int". Added new overloaded function form
17226 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
17227 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
17228 exponent double-precision with floating point first argument.
17229 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
17230 documentation of scalar_test_data_class, scalar_test_neg,
17231 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
17232 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
17233 vec_test_data_class built-in functions to reflect refinements in
17234 their type signatures.
17236 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
17238 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
17240 (aarch64_elf_asm_destructor): Likewise.
17242 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
17244 PR rtl-optimization/78634
17245 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
17246 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
17247 * ifcvt.c (noce_try_cmove): Add missing cost check.
17249 PR rtl-optimization/71724
17250 * combine.c (if_then_else_cond): Look for situations where it is
17251 beneficial to undo the work of one of the recursive calls.
17253 2017-01-23 Bin Cheng <bin.cheng@arm.com>
17255 PR tree-optimization/70754
17256 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
17257 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
17258 combined stmt before it if not NULL.
17259 (combine_chains): Process refs reversely and compute dominance point
17262 2017-01-23 Martin Liska <mliska@suse.cz>
17264 PR tree-optimization/79196
17265 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
17266 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
17268 (strlen_optimize_stmt): Call the renamed function.
17270 2017-01-23 Michael Matz <matz@suse.de>
17272 PR tree-optimization/78384
17273 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
17275 2017-01-23 Richard Biener <rguenther@suse.de>
17277 PR tree-optimization/79186
17278 * tree-vrp.c (register_new_assert_for): Make sure we've seen
17279 both incoming edges before moving an assert.
17281 2017-01-23 Martin Jambor <mjambor@suse.cz>
17283 * ipa-prop.c (load_from_param_1): Removed.
17284 (load_from_unmodified_param): Bits from load_from_param_1 put back
17286 (load_from_param): Removed.
17287 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
17288 with stmt. Reverted back to use of load_from_unmodified_param.
17290 2017-01-23 Martin Jambor <mjambor@suse.cz>
17293 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
17294 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
17295 field a pointer to garbage collected vector, mark lattices and
17296 ipcp_orig_node with GTY((skip)).
17297 (ipa_get_param_count): Adjust to descriptors being a pointer.
17298 (ipa_get_param): Likewise.
17299 (ipa_get_type): Likewise.
17300 (ipa_get_param_move_cost): Likewise.
17301 (ipa_set_param_used): Likewise.
17302 (ipa_get_controlled_uses): Likewise.
17303 (ipa_set_controlled_uses): Likewise.
17304 (ipa_is_param_used): Likewise.
17305 (ipa_node_params_t): Move into garbage collector. New methods insert
17307 (ipa_node_params_sum): Annotate wth GTY(()).
17308 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
17310 (ipa_load_from_parm_agg): Adjust declaration.
17311 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
17312 * ipa-profile.c (ipa_profile): Likewise.
17313 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
17314 (ipa_populate_param_decls): Make descriptors parameter garbage
17316 (ipa_dump_param): Adjust to descriptors being a pointer.
17317 (ipa_alloc_node_params): Likewise.
17318 (ipa_initialize_node_params): Likewise.
17319 (load_from_param_1): Make descriptors parameter garbage collected.
17320 (load_from_unmodified_param): Likewise.
17321 (load_from_param): Likewise.
17322 (ipa_load_from_parm_agg): Likewise.
17323 (ipa_node_params::~ipa_node_params): Removed.
17324 (ipa_free_all_node_params): Remove call to delete operator.
17325 (ipa_node_params_t::insert): New.
17326 (ipa_node_params_t::remove): Likewise.
17327 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
17328 copy known_csts and known_contexts vectors.
17329 (ipa_read_node_info): Adjust to descriptors being a pointer.
17330 (ipcp_modif_dom_walker): Make m_descriptors field garbage
17332 (ipcp_transform_function): Make descriptors variable garbage
17335 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
17337 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
17338 * config/i386/avx512dqintrin.h: Ditto.
17339 * config/i386/avx512fintrin.h: Ditto.
17340 * config/i386/i386.c: Handle new builtins.
17341 * config/i386/i386-builtin.def: Add new builtins.
17342 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
17343 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
17345 2017-01-23 Jakub Jelinek <jakub@redhat.com>
17346 Martin Liska <mliska@suse.cz>
17348 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
17349 * asan.c (asan_expand_poison_ifn): Support stores and use
17350 appropriate ASAN report function.
17351 * internal-fn.c (expand_ASAN_POISON_USE): New function.
17352 * internal-fn.def (ASAN_POISON_USE): Declare.
17353 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
17354 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
17355 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
17356 ASAN_POISON calls w/o LHS.
17357 * tree-ssa.c (execute_update_addresses_taken): Create clobber
17358 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
17359 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
17360 * gimplify.c (asan_poison_variables): Add attribute
17361 use_after_scope_memory to variables that really needs to live
17363 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
17364 having the attribute.
17366 2017-01-23 Martin Liska <mliska@suse.cz>
17368 * asan.c (create_asan_shadow_var): New function.
17369 (asan_expand_poison_ifn): Likewise.
17370 * asan.h (asan_expand_poison_ifn): New declaration.
17371 * internal-fn.c (expand_ASAN_POISON): Likewise.
17372 * internal-fn.def (ASAN_POISON): New builtin.
17373 * sanopt.c (pass_sanopt::execute): Expand
17374 asan_expand_poison_ifn.
17375 * tree-inline.c (copy_decl_for_dup_finish): Make function
17377 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
17378 * tree-ssa.c (is_asan_mark_p): New function.
17379 (execute_update_addresses_taken): Rewrite local variables
17380 (identified just by use-after-scope as addressable) into SSA.
17382 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
17384 * doc/install.texi (Specific): opensource.apple.com uses https
17385 now. Remove trailing slash.
17387 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
17389 * README.Portability: Remove note on an Irix compatibility issue.
17391 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
17393 * gcov.c (INCLUDE_ALGORITHM): Define.
17394 (INCLUDE_VECTOR): Define.
17395 No longer include <vector> and <algorithm> directly.
17397 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
17399 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
17401 * doc/invoke.texi (Code Gen Options): Ditto.
17403 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
17406 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
17408 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
17410 rtl-optimization/79125
17411 * cprop.c (local_cprop_pass): Handle cases where we make an
17412 unconditional trap.
17414 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
17418 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
17419 read from, for big endian.
17421 2017-01-20 Jiong Wang <jiong.wang@arm.com>
17423 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
17424 register pauth builtins for LP64 only.
17426 2017-01-20 Marek Polacek <polacek@redhat.com>
17429 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
17432 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
17434 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
17436 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
17437 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
17438 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
17440 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17443 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
17444 in big-endian mode when they are not a single duplicated value.
17446 2017-01-20 Richard Biener <rguenther@suse.de>
17448 * BASE-VER: Bump to 7.0.1.
17450 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
17452 * omp-low.c (omplow_simd_context): New struct. Use it...
17453 (lower_rec_simd_input_clauses): ...here and...
17454 (lower_rec_input_clauses): ...here to hold common data. Adjust all
17455 references to idx, lane, max_vf, is_simt.
17457 2017-01-20 Graham Markall <graham.markall@embecosm.com>
17459 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
17462 2017-01-20 Martin Jambor <mjambor@suse.cz>
17464 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
17465 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
17467 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
17468 (GTFILES): Rename hsa.c to hsa-common.c.
17469 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
17470 * hsa-dump.c: Likewise.
17471 * hsa-gen.c: Likewise.
17472 * hsa-regalloc.c: Likewise.
17473 * ipa-hsa.c: Likewise.
17474 * omp-expand.c: Likewise.
17475 * omp-low.c: Likewise.
17476 * toplev.c: Likewise.
17478 2017-01-20 Marek Polacek <polacek@redhat.com>
17481 * doc/invoke.texi: Document -Wduplicated-branches.
17482 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
17483 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
17484 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
17485 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
17486 return 0 only when not OEP_LEXICOGRAPHIC.
17487 (fold_build_cleanup_point_expr): Use the expression
17488 location when building CLEANUP_POINT_EXPR.
17489 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
17490 * tree.c (add_expr): Handle error_mark_node.
17492 2017-01-20 Martin Liska <mliska@suse.cz>
17495 * tree-profile.c (init_ic_make_global_vars): Do not call
17497 (gimple_init_gcov_profiler): Likewise.
17499 2017-01-20 Martin Liska <mliska@suse.cz>
17502 * cgraph.h (maybe_create_reference): Remove argument and
17504 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
17506 * ipa-cp.c (create_specialized_node): Likewise.
17507 * symtab.c (symtab_node::maybe_create_reference): Handle
17508 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
17510 2017-01-20 Martin Liska <mliska@suse.cz>
17512 * read-rtl-function.c (function_reader::create_function): Use
17513 build_decl instread of build_decl_stat.
17515 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
17517 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
17518 * config/i386/avx512dqintrin.h: Ditto.
17519 * config/i386/avx512fintrin.h: Ditto.
17520 * config/i386/i386-builtin-types.def: Add new types.
17521 * config/i386/i386.c: Handle new types.
17522 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
17523 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
17524 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
17525 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
17526 (__builtin_ia32_kshiftridi): New.
17527 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
17529 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
17533 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
17534 define to rs6000_init_stack_protect_guard.
17535 (rs6000_init_stack_protect_guard): New function.
17537 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
17538 Yunqiang Su <yunqiang.su@imgtec.com>
17540 * config.gcc (supported_defaults): Add madd4.
17541 (with_madd4): Add validation.
17542 (all_defaults): Add madd4.
17543 * config/mips/mips.opt (mmadd4): New option.
17544 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17546 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17548 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
17549 (ISA_HAS_FUSED_MADD4): Likewise.
17550 * doc/invoke.texi (-mmadd4): Document the new option.
17551 * doc/install.texi (--with-madd4): Document the new option.
17553 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17555 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
17556 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
17557 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
17558 (aarch64_init_pauth_hint_builtins): New.
17559 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
17560 (aarch64_expand_builtin): Expand new builtins.
17562 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17564 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
17565 * combine-stack-adj.c (no_unhandled_cfa): Handle
17566 REG_CFA_TOGGLE_RA_MANGLE.
17567 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
17568 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
17569 info for return address signing.
17570 (aarch64_expand_epilogue): Likewise.
17572 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17574 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
17575 * config/aarch64/aarch64-protos.h
17576 (aarch64_return_address_signing_enabled): New declaration.
17577 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
17579 (aarch64_expand_prologue): Sign return address before it's pushed onto
17581 (aarch64_expand_epilogue): Authenticate return address fetched from
17583 (aarch64_override_options): Sanity check for ILP32 and ISA level.
17584 (aarch64_attributes): New function attributes for "sign-return-address".
17585 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
17586 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
17587 ("*do_return"): Generate combined instructions according to key index.
17588 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
17589 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
17591 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
17592 * config/aarch64/aarch64.opt (msign-return-address=): New.
17593 * doc/extend.texi (AArch64 Function Attributes): Documents
17594 "sign-return-address=".
17595 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
17597 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
17599 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
17600 overall option summary.
17602 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17604 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
17605 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
17606 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
17607 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
17609 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
17611 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
17612 -mpower9-minmax by default for -mcpu=power9.
17613 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
17614 128-bit floating point.
17616 2017-01-20 Alan Modra <amodra@gmail.com>
17618 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
17619 optimizing for size.
17621 2017-01-20 Alan Modra <amodra@gmail.com>
17624 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
17625 for strcmp and strncmp from corresponding builtin decl.
17627 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17629 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
17630 instead of i386/rtems-64.h.
17631 * config/i386/rtems-64.h: Remove.
17633 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17637 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
17639 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
17641 2017-01-19 Tamar Christina <tamar.christina@arm.com>
17643 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
17644 Change int to HOST_WIDE_INT.
17645 * config/aarch64/aarch64-protos.h
17646 (aarch64_simd_gen_const_vector_dup): Likewise.
17647 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
17649 2017-01-19 David Malcolm <dmalcolm@redhat.com>
17651 * langhooks-def.h (lhd_type_for_size): New decl.
17652 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
17653 * langhooks.c (lhd_type_for_size): New function, taken from
17656 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
17658 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
17659 define_bypass for CR latency.
17660 (power9-cracked-alu): Update bypass latency and remove power9-branch.
17661 (power9-alu2): Add define_bypass for CR latency.
17663 (power9-mul): Update insn latency.
17664 (power9-mul-compare): Update insn latency, bypass latency and remove
17667 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17669 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
17671 * config/aarch64/aarch64.md
17672 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
17673 aarch64_nopcrelative_literal_loads.
17674 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
17676 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
17678 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
17679 TARGET_LOONGSON_3A.
17680 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
17682 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
17685 * config.gcc (supported_defaults): Add lxc1-sxc1.
17686 (with_lxc1_sxc1): Add validation.
17687 (all_defaults): Add lxc1-sxc1.
17688 * config/mips/mips.opt (mlxc1-sxc1): New option.
17689 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17691 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17692 __mips_no_lxc1_sxc1.
17693 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
17694 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
17695 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
17697 2017-01-19 Richard Biener <rguenther@suse.de>
17699 PR tree-optimization/72488
17700 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
17701 sure to restore SSA info.
17702 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
17704 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
17706 PR rtl-optimization/79121
17707 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
17708 of the inner type when shifting an extended value.
17710 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
17713 * symtab.c (symtab_node::equal_address_to): Fix comparing of
17714 interposable aliases.
17716 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
17719 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
17720 Use the evmergelohi instruction.
17721 (mov_si<mode>_e500_subreg4_2_le): Likewise.
17722 (mov_sitf_e500_subreg8_2_be): Likewise.
17723 (mov_sitf_e500_subreg12_2_le): Likewise.
17724 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
17725 (mov_si<mode>_e500_subreg4_2_be): Likewise.
17726 (mov_sitf_e500_subreg8_2_le): Likewise.
17727 (mov_sitf_e500_subreg12_2_be): Likewise.
17729 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17731 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
17732 attribute from vecsimple to vecperm.
17733 (altivec_vbpermq2): Likewise.
17735 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17738 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
17740 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17741 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
17742 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
17743 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
17744 case where N arg is SIZE_MAX.
17745 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
17746 (cmpstrsi): Add pattern.
17748 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17750 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17751 __builtin_vec_revb builtins.
17752 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
17753 built-in functions to support generation of the ISA 3.0 XXBR<x>
17754 vector byte reverse instructions.
17755 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
17756 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
17757 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
17758 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
17759 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
17760 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
17761 (P9V_BUILTIN_VEC_REVB): Likewise.
17762 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
17763 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
17764 (p9_xxbrq_v16qi): Likewise.
17765 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
17766 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
17767 (p9_xxbrh_v8hi): Likewise.
17768 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
17769 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
17770 vec_revb built-in functions.
17772 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
17774 PR rtl-optimization/78952
17775 * config/i386/i386.md (any_extract): New code iterator.
17776 (*insvqi_2): Use any_extract for source operand.
17777 (*insvqi_3): Use any_shiftrt for source operand.
17779 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
17781 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
17783 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
17785 2017-01-18 Matthias Klose <doko@ubuntu.com>
17787 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
17789 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17791 * config/rs6000/altivec.h (vec_bperm): Change #define.
17792 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
17793 (altivec_vbpermq2): New define_insn.
17794 (altivec_vbpermd): Likewise.
17795 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
17796 function interface.
17797 (VBPERMD): Likewise.
17798 (VBPERM): New polymorphic function interface.
17799 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
17800 Add entries for P9V_BUILTIN_VEC_VBPERM.
17801 * doc/extend.texi: Add interfaces for vec_bperm.
17803 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17805 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
17806 first letter of error messages.
17807 (s390_resolve_overloaded_builtin): Likewise.
17808 * config/s390/s390.c (s390_expand_builtin): Likewise.
17809 (s390_invalid_arg_for_unprototyped_fn): Likewise.
17810 (s390_valid_target_attribute_inner_p): Likewise.
17811 * config/s390/s390.md ("tabort"): Likewise.
17813 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
17815 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
17816 (ISA_AVOID_DIV_HILO): New macro.
17817 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
17818 (ISA_HAS_DDIV): Likewise.
17820 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17822 * doc/invoke.texi (fabi-version): Correct number of occurrences.
17824 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17826 * doc/invoke.texi (fabi-version): Spelling fix.
17828 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17831 * doc/invoke.texi (fabi-version): Mention mangling fix for
17834 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17837 * doc/invoke.texi (fabi-version): Document discriminator mangling.
17839 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
17842 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
17843 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
17845 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
17847 (stack_protect_test): Ditto.
17848 * config/rs6000/rs6000.opt (mstack-protector-guard=,
17849 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
17851 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
17852 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
17853 -mstack-protector-guard-offset=.
17854 (RS/6000 and PowerPC Options): Ditto.
17856 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
17858 * config/i386/i386.h (MASK_CLASS_P): New define.
17859 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
17860 there are no registers from different register sets also when
17861 mask registers are used. Update function comment.
17862 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
17863 to (*k/*r) and (*k/*km) alternatives.
17865 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
17867 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
17868 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
17869 (EH_RETURN_HANDLER_RTX): New define.
17870 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
17871 Force frame pointer in EH return functions.
17872 (aarch64_expand_epilogue): Add barrier for eh_return.
17873 (aarch64_final_eh_return_addr): Remove.
17874 (aarch64_eh_return_handler_rtx): New function.
17875 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
17877 (aarch64_eh_return_handler_rtx): New prototype.
17879 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17881 * config/rs6000/altivec.h (vec_rlmi): New #define.
17882 (vec_vrlnm): Likewise.
17883 (vec_rlnm): Likewise.
17884 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
17885 (UNSPEC_VRLNM): Likewise.
17886 (VIlong): New mode iterator.
17887 (altivec_vrl<VI_char>mi): New define_insn.
17888 (altivec_vrl<VI_char>nm): Likewise.
17889 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
17891 (VRLDNM): Likewise.
17892 (RLNM): New polymorphic function entry.
17893 (VRLWMI): New monomorphic function entry.
17894 (VRLDMI): Likewise.
17895 (RLMI): New polymorphic function entry.
17896 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
17897 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
17898 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
17901 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17904 * dwarf2out.c (field_byte_offset): Restore the
17905 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
17906 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
17907 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
17910 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17913 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
17915 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17918 * dwarf2out.c (add_data_member_location_attribute): For constant
17919 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
17920 instead of DW_AT_data_member_location, DW_AT_bit_offset and
17921 DW_AT_byte_size attributes.
17923 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17925 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
17926 after forcing to constant memory when the code model is medium.
17928 2017-01-17 Julia Koval <julia.koval@intel.com>
17931 * config/i386/avx512fintrin.h
17932 (_mm512_i32gather_ps): Change __addr type to void const*.
17933 (_mm512_mask_i32gather_ps): Ditto.
17934 (_mm512_i32gather_pd): Ditto.
17935 (_mm512_mask_i32gather_pd): Ditto.
17936 (_mm512_i64gather_ps): Ditto.
17937 (_mm512_mask_i64gather_ps): Ditto.
17938 (_mm512_i64gather_pd): Ditto.
17939 (_mm512_mask_i64gather_pd): Ditto.
17940 (_mm512_i32gather_epi32): Ditto.
17941 (_mm512_mask_i32gather_epi32): Ditto.
17942 (_mm512_i32gather_epi64): Ditto.
17943 (_mm512_mask_i32gather_epi64): Ditto.
17944 (_mm512_i64gather_epi32): Ditto.
17945 (_mm512_mask_i64gather_epi32): Ditto.
17946 (_mm512_i64gather_epi64): Ditto.
17947 (_mm512_mask_i64gather_epi64): Ditto.
17948 (_mm512_i32scatter_ps): Change __addr type to void*.
17949 (_mm512_mask_i32scatter_ps): Ditto.
17950 (_mm512_i32scatter_pd): Ditto.
17951 (_mm512_mask_i32scatter_pd): Ditto.
17952 (_mm512_i64scatter_ps): Ditto.
17953 (_mm512_mask_i64scatter_ps): Ditto.
17954 (_mm512_i64scatter_pd): Ditto.
17955 (_mm512_mask_i64scatter_pd): Ditto.
17956 (_mm512_i32scatter_epi32): Ditto.
17957 (_mm512_mask_i32scatter_epi32): Ditto.
17958 (_mm512_i32scatter_epi64): Ditto.
17959 (_mm512_mask_i32scatter_epi64): Ditto.
17960 (_mm512_i64scatter_epi32): Ditto.
17961 (_mm512_mask_i64scatter_epi32): Ditto.
17962 (_mm512_i64scatter_epi64): Ditto.
17963 (_mm512_mask_i64scatter_epi64): Ditto.
17964 * config/i386/avx512pfintrin.h
17965 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
17966 (_mm512_mask_prefetch_i32gather_ps): Ditto.
17967 (_mm512_mask_prefetch_i64gather_pd): Ditto.
17968 (_mm512_mask_prefetch_i64gather_ps): Ditto.
17969 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
17970 (_mm512_prefetch_i32scatter_ps): Ditto.
17971 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
17972 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
17973 (_mm512_prefetch_i64scatter_pd): Ditto.
17974 (_mm512_prefetch_i64scatter_ps): Ditto.
17975 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
17976 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
17977 * config/i386/avx512vlintrin.h
17978 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
17979 (_mm_mmask_i32gather_ps): Ditto.
17980 (_mm256_mmask_i32gather_pd): Ditto.
17981 (_mm_mmask_i32gather_pd): Ditto.
17982 (_mm256_mmask_i64gather_ps): Ditto.
17983 (_mm_mmask_i64gather_ps): Ditto.
17984 (_mm256_mmask_i64gather_pd): Ditto.
17985 (_mm_mmask_i64gather_pd): Ditto.
17986 (_mm256_mmask_i32gather_epi32): Ditto.
17987 (_mm_mmask_i32gather_epi32): Ditto.
17988 (_mm256_mmask_i32gather_epi64): Ditto.
17989 (_mm_mmask_i32gather_epi64): Ditto.
17990 (_mm256_mmask_i64gather_epi32): Ditto.
17991 (_mm_mmask_i64gather_epi32): Ditto.
17992 (_mm256_mmask_i64gather_epi64): Ditto.
17993 (_mm_mmask_i64gather_epi64): Ditto.
17994 (_mm256_i32scatter_ps): Change __addr type to void*.
17995 (_mm256_mask_i32scatter_ps): Ditto.
17996 (_mm_i32scatter_ps): Ditto.
17997 (_mm_mask_i32scatter_ps): Ditto.
17998 (_mm256_i32scatter_pd): Ditto.
17999 (_mm256_mask_i32scatter_pd): Ditto.
18000 (_mm_i32scatter_pd): Ditto.
18001 (_mm_mask_i32scatter_pd): Ditto.
18002 (_mm256_i64scatter_ps): Ditto.
18003 (_mm256_mask_i64scatter_ps): Ditto.
18004 (_mm_i64scatter_ps): Ditto.
18005 (_mm_mask_i64scatter_ps): Ditto.
18006 (_mm256_i64scatter_pd): Ditto.
18007 (_mm256_mask_i64scatter_pd): Ditto.
18008 (_mm_i64scatter_pd): Ditto.
18009 (_mm_mask_i64scatter_pd): Ditto.
18010 (_mm256_i32scatter_epi32): Ditto.
18011 (_mm256_mask_i32scatter_epi32): Ditto.
18012 (_mm_i32scatter_epi32): Ditto.
18013 (_mm_mask_i32scatter_epi32): Ditto.
18014 (_mm256_i32scatter_epi64): Ditto.
18015 (_mm256_mask_i32scatter_epi64): Ditto.
18016 (_mm_i32scatter_epi64): Ditto.
18017 (_mm_mask_i32scatter_epi64): Ditto.
18018 (_mm256_i64scatter_epi32): Ditto.
18019 (_mm256_mask_i64scatter_epi32): Ditto.
18020 (_mm_i64scatter_epi32): Ditto.
18021 (_mm_mask_i64scatter_epi32): Ditto.
18022 (_mm256_i64scatter_epi64): Ditto.
18023 (_mm256_mask_i64scatter_epi64): Ditto.
18024 (_mm_i64scatter_epi64): Ditto.
18025 (_mm_mask_i64scatter_epi64): Ditto.
18026 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
18027 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
18028 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
18029 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
18030 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
18031 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
18032 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
18033 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
18034 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
18035 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
18036 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
18037 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
18038 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
18039 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
18040 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
18041 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
18042 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
18043 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
18044 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
18045 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
18046 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
18047 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
18048 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
18049 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
18050 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
18051 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
18052 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
18053 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
18054 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
18055 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
18056 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
18057 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
18058 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
18059 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
18060 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
18061 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
18062 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
18063 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
18064 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
18065 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
18066 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
18067 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
18068 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
18069 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
18070 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
18071 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
18072 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
18073 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
18074 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
18075 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
18076 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
18077 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
18078 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
18079 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
18080 definitions accordingly.
18082 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
18083 Kuan-Lin Chen <kuanlinchentw@gmail.com>
18086 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
18089 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
18092 * ira-conflicts.c (ira_build_conflicts): Update total conflict
18093 hard regs for inner regno.
18095 2017-01-17 Martin Liska <mliska@suse.cz>
18098 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
18099 assumption and add comment.
18101 2017-01-17 Nathan Sidwell <nathan@acm.org>
18103 * ipa-visibility.c (localize_node): New function, broken out of ...
18104 (function_and_variable_visibility): ... here. Call it.
18106 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
18108 PR middle-end/77445
18109 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
18110 correctly set frequency of oudgoing edge.
18111 (duplicate_thread_path): Fix profile updating.
18113 2017-01-17 Jakub Jelinek <jakub@redhat.com>
18116 * configure.ac: Add GCC_BASE_VER.
18117 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
18118 version from BASE-VER file.
18119 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
18120 (gcc.o): Depend on $(BASEVER).
18121 * common.opt (dumpfullversion): New option.
18122 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
18123 * doc/invoke.texi: Document -dumpfullversion.
18124 * doc/install.texi: Document --with-gcc-major-version-only.
18125 * configure: Regenerated.
18127 2017-01-17 Richard Biener <rguenther@suse.de>
18129 PR tree-optimization/71433
18130 * tree-vrp.c (register_new_assert_for): Merge same asserts
18131 on all incoming edges.
18132 (process_assert_insertions_for): Handle insertions at the
18135 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
18137 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
18138 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
18140 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
18143 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
18146 2017-01-17 Alan Modra <amodra@gmail.com>
18149 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
18150 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
18151 symbolic stack limit when pic.
18153 2017-01-16 Martin Sebor <msebor@redhat.com>
18155 PR tree-optimization/78608
18156 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
18158 2017-01-16 Jeff Law <law@redhat.com>
18161 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
18162 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
18163 for several include directories that may be relative to sysroot.
18164 * config/i386/x-mingw32 (gplus_includedir): Define.
18165 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
18166 (native_system_includedir): Likewise.
18167 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
18168 override if TARGET_SYSTEM_ROOT is defined.
18169 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
18171 PR tree-optimization/79090
18172 PR tree-optimization/33562
18173 PR tree-optimization/61912
18174 PR tree-optimization/77485
18175 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
18176 and computed trims into the dump file.
18178 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
18180 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
18182 2017-01-16 Jakub Jelinek <jakub@redhat.com>
18185 * gimplify.c (gimplify_init_constructor): If want_value and
18186 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
18190 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
18191 sequence. Formatting fixes.
18192 (doloop_optimize): Formatting fixes.
18195 * gcc.c (debug_level_greater_than_spec_func): New function.
18196 (static_spec_functions): Add debug-level-gt spec function.
18197 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
18199 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
18200 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
18201 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
18202 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
18203 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
18204 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
18206 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
18208 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
18209 QImode fixups to general and mask registers only.
18211 2017-01-16 Carl Love <cel@us.ibm.com>
18213 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
18214 for built-in functions
18215 vector signed char vec_nabs (vector signed char)
18216 vector signed short vec_nabs (vector signed short)
18217 vector signed int vec_nabs (vector signed int)
18218 vector signed long long vec_nabs (vector signed long long)
18219 vector float vec_nabs (vector float)
18220 vector double vec_nabs (vector double)
18221 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
18223 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
18224 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
18225 * doc/extend.texi: Update the documentation file for the new built-in
18228 2017-01-16 Martin Sebor <msebor@redhat.com>
18230 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
18233 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18235 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
18236 UNSPEC_VSX__XXSPLTD to require special splat handling.
18238 2017-01-16 David Malcolm <dmalcolm@redhat.com>
18241 * system.h: Poison strndup.
18243 2017-01-16 Alan Modra <amodra@gmail.com>
18246 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
18249 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
18251 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
18253 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
18255 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
18256 call recog here. Assert that INSN_CODE (insn) is non-negative.
18258 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
18261 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
18263 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
18264 in the currently scheduled RTL fragment.
18266 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
18268 PR rtl-optimization/78751
18269 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
18272 2017-01-14 Jeff Law <law@redhat.com>
18274 PR tree-optimization/79090
18275 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
18276 variable length stores.
18277 (compute_trims): Delete dead assignment to *trim_tail.
18278 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
18281 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
18283 PR rtl-optimization/78626
18284 PR rtl-optimization/78727
18285 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
18286 of a block, and split such blocks after everything else is finished.
18288 2017-01-14 Alan Modra <amodra@gmail.com>
18291 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
18292 target legitimate_combined_insn.
18293 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
18294 (rs6000_legitimate_combined_insn): New function.
18295 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
18297 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
18298 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
18299 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
18301 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
18303 * doc/frontends.texi (G++ and GCC): Remove references to Java.
18305 2017-01-13 Jeff Law <law@redhat.com>
18307 PR tree-optimization/33562
18308 PR tree-optimization/61912
18309 PR tree-optimization/77485
18310 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
18312 (delete_dead_assignment): Likewise.
18313 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
18314 statement to delete_dead_call and delete_dead_assignment.
18316 2017-01-13 David Malcolm <dmalcolm@redhat.com>
18319 * substring-locations.c (format_warning_va): Strengthen case 1 so
18320 that both endpoints of the substring must be within the format
18321 range for just the substring to be printed.
18323 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
18325 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
18326 * config/i386/i386.c (ix86_target_string): Add missing options
18327 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
18328 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
18329 flags_other and ix86_target_other to flags2_other. Display unknown
18331 (ix86_valid_target_attribute_inner_p): Add missing options and
18332 reorder options by implied ISAs, as in ix86_target_string.
18334 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
18336 * hash-table.h (hash_table::too_empty_p): New function.
18337 (hash_table::expand): Use it.
18338 (hash_table::traverse): Likewise.
18339 (hash_table::empty_slot): Use sizeof (value_type) instead of
18340 sizeof (PTR) to convert bytes to elements. Shrink the table
18341 if the current size is excessive for the current number of
18344 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
18346 * ira-costs.c (record_reg_classes): Break from the inner loop
18347 early once alt_fail is known to be true. Update outer loop
18348 handling accordingly.
18350 2017-01-13 Jeff Law <law@redhat.com>
18352 * tree-ssa-dse.c (decrement_count): New function.
18353 (increment_start_addr, maybe_trim_memstar_call): Likewise.
18354 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
18355 when we know the partially dead statement is a mem* function.
18357 PR tree-optimization/61912
18358 PR tree-optimization/77485
18359 * tree-ssa-dse.c: Include expr.h.
18360 (maybe_trim_constructor_store): New function.
18361 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
18363 PR tree-optimization/33562
18364 PR tree-optimization/61912
18365 PR tree-optimization/77485
18366 * doc/invoke.texi: Document new dse-max-object-size param.
18367 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
18368 * tree-ssa-dse.c: Include params.h.
18369 (dse_store_status): New enum.
18370 (initialize_ao_ref_for_dse): New, partially extracted from
18372 (valid_ao_ref_for_dse, normalize_ref): New.
18373 (setup_live_bytes_from_ref, compute_trims): Likewise.
18374 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
18375 (maybe_trim_partially_dead_store): Likewise.
18376 (maybe_trim_complex_store): Likewise.
18377 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
18378 Track what bytes live from the original store. Return tri-state
18379 for dead, partially dead or live.
18380 (dse_dom_walker): Add constructor, destructor and new private members.
18381 (delete_dead_call, delete_dead_assignment): New extracted from
18383 (dse_optimize_stmt): Make a member of dse_dom_walker.
18384 Use initialize_ao_ref_for_dse.
18386 PR tree-optimization/33562
18387 PR tree-optimization/61912
18388 PR tree-optimization/77485
18389 * sbitmap.h (bitmap_count_bits): Prototype.
18390 (bitmap_clear_range, bitmap_set_range): Likewise.
18391 * sbitmap.c (bitmap_clear_range): New function.
18392 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
18394 2017-01-13 Martin Liska <mliska@suse.cz>
18397 * function.c (set_cfun): Add new argument force.
18398 * function.h (set_cfun): Likewise.
18399 * ipa-inline-transform.c (inline_call): Use the function when
18400 strict alising from is dropped for function we inline to.
18402 2017-01-13 Richard Biener <rguenther@suse.de>
18404 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
18405 for dumping GIMPLE INTEGER_CSTs.
18407 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18409 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
18410 to 201112L since C++17.
18412 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
18415 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
18416 if -fsanitize=kernel-address is present.
18418 2017-01-13 Richard Biener <rguenther@suse.de>
18420 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
18421 as _Literal ( type ) number in case usual suffixes do not
18422 preserve all information.
18424 2017-01-13 Richard Biener <rguenther@suse.de>
18426 PR tree-optimization/77283
18427 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
18428 and ssa-iterators.h.
18429 (is_feasible_trace): Implement a cost model based on joiner
18432 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
18435 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
18436 char or short to __float128/_Float128 directly.
18438 2017-01-12 Martin Sebor <msebor@redhat.com>
18440 to -Wformat-overflow.
18441 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
18442 (min_bytes_remaining): Same.
18443 (get_string_length): Same.
18444 (format_string): Same.
18445 (format_directive): Same.
18447 (pass_sprintf_length::handle_gimple_call): Same.
18449 2017-01-12 Jakub Jelinek <jakub@redhat.com>
18451 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
18452 info.nowrite calls with no lhs that can't throw. Return bool
18453 whether gsi_remove has been called or not.
18454 (pass_sprintf_length::handle_gimple_call): Return bool whether
18455 try_substitute_return_value called gsi_remove. Formatting fix.
18456 (pass_sprintf_length::execute): Don't use gsi_remove if
18457 handle_gimple_call returned true.
18460 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
18461 be removed due to side-effects, don't remove following barrier nor
18462 turn the successor edge into fallthru edge.
18464 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18467 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
18468 element-reversing loads and stores as not swappable.
18470 2017-01-12 Nathan Sidwell <nathan@acm.org>
18471 Nicolai Stange <nicstange@gmail.com>
18473 * combine.c (try_combine): Don't ignore result of overlap checking
18474 loop. Combine overlap & asm check into single loop.
18476 2017-01-12 Richard Biener <rguenther@suse.de>
18478 * tree-pretty-print.c (dump_generic_node): Provide -gimple
18479 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
18481 2017-01-12 Richard Biener <rguenther@suse.de>
18483 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
18484 and TS_TARGET_OPTION directly derive from TS_BASE.
18485 * tree-core.h (tree_optimization_option): Derive from tree_base.
18486 (tree_target_option): Likewise.
18488 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
18490 * config/i386/i386.c (memory_address_length): Increase len
18491 only when rip_relative_addr_p returns false.
18493 2017-01-11 Julia Koval <julia.koval@intel.com>
18495 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
18496 (OPTION_MASK_ISA_SGX_SET): New.
18497 (ix86_handle_option): Handle OPT_msgx.
18498 * config.gcc: Added sgxintrin.h.
18499 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
18500 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
18501 * config/i386/i386.c (ix86_target_string): Add -msgx.
18503 (ix86_option_override_internal): Handle new options.
18504 (ix86_valid_target_attribute_inner_p): Add sgx.
18505 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
18506 * config/i386/i386.opt: Add msgx.
18507 * config/i386/sgxintrin.h: New file.
18508 * config/i386/x86intrin.h: Add sgxintrin.h.
18510 2017-01-11 Jakub Jelinek <jakub@redhat.com>
18513 * fold-const.c (maybe_nonzero_address): Return 1 for function
18515 (tree_single_nonzero_warnv_p): Don't handle function local objects
18519 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
18522 2017-01-11 David Malcolm <dmalcolm@redhat.com>
18525 * opts.c: Include "spellcheck.h"
18526 (struct string_fragment): New struct.
18527 (struct edit_distance_traits<const string_fragment &>): New
18529 (get_closest_sanitizer_option): New function.
18530 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
18532 2017-01-11 Jakub Jelinek <jakub@redhat.com>
18534 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
18536 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
18537 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
18538 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
18539 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
18540 for initial die_offset if dwarf_split_debug_info.
18541 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
18542 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
18544 (output_skeleton_debug_sections): Formatting fix. Use
18545 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
18546 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
18548 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
18550 * config/arm/cortex-a53.md: Add bypasses for
18551 cortex_a53_r2f_cvt.
18552 (cortex_a53_r2f): Only use for transfers.
18553 (cortex_a53_f2r): Likewise.
18554 (cortex_a53_r2f_cvt): Add reservation for conversions.
18555 (cortex_a53_f2r_cvt): Likewise.
18557 2017-01-11 Tamar Christina <tamar.christina@arm.com>
18559 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
18560 to all inlined functions, change static to extern.
18562 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
18565 * config/arm/arm.c (legitimize_pic_address): Handle reference to
18567 (arm_assemble_integer): Likewise.
18569 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
18571 * config.gcc: Use new awk script to check CPU, FPU and architecture
18572 parameters for --with-... options.
18573 * config/arm/parsecpu.awk: New file
18574 * config/arm/arm-cpus.in: New file.
18575 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
18577 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
18579 * config/arm/t-arm: Update dependency rules.
18580 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
18581 of processing .def files.
18582 * config/arm/genopt.sh: Deleted.
18583 * config/arm/gentune.sh: Deleted.
18584 * config/arm/arm-cores.def: Deleted.
18585 * config/arm/arm-arches.def: Deleted.
18586 * config/arm/arm-fpus.def: Deleted.
18587 * config/arm/arm-tune.md: Regenerated.
18588 * config/arm/arm-tables.opt: Regenerated.
18589 * config/arm/arm-cpu.h: New generated file.
18590 * config/arm/arm-cpu-data.h: New generated file.
18591 * config/arm/arm-cpu-cdata.h: New generated file.
18593 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
18596 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
18598 (input_varpool_node): Unpack dynamically_initialized bit.
18600 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
18602 PR rtl-optimization/79032
18603 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
18604 the alignment of the adjusted memory reference against that of MODE,
18605 instead of the alignment of the original memory reference.
18607 2017-01-11 Martin Jambor <mjambor@suse.cz>
18609 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
18611 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
18612 decorated functions.
18614 2017-01-11 Richard Biener <rguenther@suse.de>
18616 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
18617 set range/nonnull info for PHI results. Do not set it on
18618 stmts marked for removal.
18620 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
18622 * expr.c (store_field): In the bitfield case, fetch the return value
18623 from the registers before applying a single big-endian adjustment.
18624 Always do a final load for a BLKmode value not larger than a word.
18626 2017-01-10 David Malcolm <dmalcolm@redhat.com>
18629 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18630 that we correctly handle column numbers greater than
18631 LINE_MAP_MAX_COLUMN_NUMBER.
18633 2017-01-10 Martin Sebor <msebor@redhat.com>
18635 PR middle-end/78245
18636 * gimple-ssa-sprintf.c (get_destination_size): Call
18637 {init,fini}object_sizes.
18638 * tree-object-size.c (addr_object_size): Adjust.
18639 (pass_through_call): Adjust.
18640 (pass_object_sizes::execute): Adjust.
18641 * tree-object-size.h (fini_object_sizes): Declare.
18643 2017-01-10 Martin Sebor <msebor@redhat.com>
18645 PR tree-optimization/78775
18646 * builtins.c (get_size_range): Move...
18647 * calls.c: ...to here.
18648 (alloc_max_size): Accept zero argument.
18649 (operand_signed_p): Remove.
18650 (maybe_warn_alloc_args_overflow): Call get_size_range.
18651 * calls.h (get_size_range): Declare.
18653 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
18655 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
18656 from TI's devices.csv file as of September 2016.
18657 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
18659 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
18661 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
18662 * doc/invoke.texi: Likewise.
18663 * doc/md.texi: Likewise.
18664 * doc/objc.texi: Likewise.
18666 2017-01-10 Joshua Conner <joshconner@google.com>
18668 * config/arm/fuchsia-elf.h: New file.
18669 * config/fuchsia.h: New file.
18670 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
18671 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
18673 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
18675 2016-01-10 Richard Biener <rguenther@suse.de>
18677 PR tree-optimization/79034
18678 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18679 Propagate out degenerate PHIs in the joiner.
18681 2017-01-10 Martin Liska <mliska@suse.cz>
18683 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
18684 (sort_congruence_classes_by_decl_uid): Likewise.
18685 (sort_congruence_class_groups_by_decl_uid): Likewise.
18686 (sem_item_optimizer::merge_classes): Sort class, groups in these
18687 classes and members in the groups by DECL_UID of declarations.
18688 This would make merge operations stable.
18690 2017-01-10 Martin Liska <mliska@suse.cz>
18692 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
18693 usage of m_classes_vec.
18694 (sem_item_optimizer::~sem_item_optimizer): Likewise.
18695 (sem_item_optimizer::get_group_by_hash): Likewise.
18696 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
18697 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
18698 (sem_item_optimizer::verify_classes): Likewise.
18699 (sem_item_optimizer::process_cong_reduction): Likewise.
18700 (sem_item_optimizer::dump_cong_classes): Likewise.
18701 (sem_item_optimizer::merge_classes): Likewise.
18702 * ipa-icf.h (congruence_class_hash): Rename from
18703 congruence_class_group_hash. Remove declaration of m_classes_vec.
18705 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
18707 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
18708 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
18709 * config.gcc: Add avx512vpopcntdqintrin.h.
18710 * config/i386/avx512vpopcntdqintrin.h: New.
18711 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
18712 * config/i386/i386-builtin-types.def: Add new types.
18713 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
18714 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
18715 __builtin_ia32_vpopcountq_v8di_mask): New.
18716 * config/i386/i386-c.c (ix86_target_macros_internal): Define
18717 __AVX512VPOPCNTDQ__.
18718 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
18719 (PTA_AVX512VPOPCNTDQ): Define.
18720 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
18721 TARGET_AVX512VPOPCNTDQ_P): Define.
18722 * config/i386/i386.opt: Add mavx512vpopcntdq.
18723 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
18724 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
18726 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18728 PR middle-end/77484
18729 * predict.def (PRED_CALL): Set to 67.
18731 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
18733 * expr.c (store_field): In the bitfield case, if the value comes from
18734 a function call and is of an aggregate type returned in registers, do
18735 not modify the field mode; extract the value in all cases if the mode
18736 is BLKmode and the size is not larger than a word.
18738 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
18741 * config/i386/cpuid.h: Fix undefined behavior.
18743 2017-01-04 Jeff Law <law@redhat.com>
18745 PR tree-optimization/79007
18746 PR tree-optimization/67955
18747 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
18748 conservative for pt.null when flag_non_call_exceptions is on.
18750 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18752 PR translation/79019
18753 PR translation/79020
18754 * params.def (PARAM_INLINE_MIN_SPEEDUP,
18755 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
18756 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
18758 * config/avr/avr.opt (maccumulate-args): Likewise.
18759 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
18760 * common.opt (freport-bug): Likewise.
18761 * cif-code.def (CIF_FINAL_ERROR): Likewise.
18762 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
18763 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
18764 translatable string.
18765 * config/i386/i386.c (function_value_32): Likewise.
18766 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
18767 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
18769 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
18770 * common/config/msp430/msp430-common.c (msp430_handle_option):
18772 * symtab.c (symtab_node::verify_base): Likewise.
18773 * opts.c (set_debug_level): Likewise.
18774 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
18775 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
18776 missing whitespace to translatable strings.
18777 * config/avr/avr.md (bswapsi2): Fix typo in comment.
18778 * config/sh/superh.h: Likewise.
18779 * config/i386/xopintrin.h: Likewise.
18780 * config/i386/znver1.md: Likewise.
18781 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
18782 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
18783 * double-int.h (struct double_int): Likewise.
18784 * double-int.c (div_and_round_double): Likewise.
18785 * wide-int.cc: Likewise.
18786 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
18787 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
18788 * cfgcleanup.c (crossjumps_occured): Renamed to ...
18789 (crossjumps_occurred): ... this.
18790 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
18793 PR tree-optimization/78899
18794 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
18795 returning bool return struct loop *, NULL for failure and the new
18797 (versionable_outer_loop_p): Don't version outer loop if it has
18798 dont_vectorized bit set.
18799 (tree_if_conversion): When versioning outer loop, ensure
18800 tree_if_conversion is performed also on the inner loop of the
18801 non-vectorizable outer loop copy.
18802 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
18803 LOOP_VECTORIZED in inner loop of the scalar outer loop and
18804 prevent vectorization of it.
18805 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
18806 the outer loop vectorization of the non-scalar version is attempted
18807 before vectorization of the inner loop in scalar version. If
18808 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
18809 vectorization of its inner loop.
18810 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
18811 has 2 inner loops, rename also on edges from bb whose single pred
18812 is outer_loop->header. Fix typo in function comment.
18814 2017-01-09 Martin Sebor <msebor@redhat.com>
18817 * asan.c (asan_emit_stack_protection): Increase local buffer size
18818 to avoid snprintf truncation warning.
18820 2017-01-09 Andrew Pinski <apinski@cavium.com>
18822 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
18823 to reference thunderx2t99 for the tuning structure
18824 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
18826 (thunderx2t99_extra_costs): This.
18827 * config/aarch64/aarch64-tune.md: Regenerate.
18828 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
18829 (vulcan_addrcost_table): This.
18830 (vulcan_regmove_cost): Rename to ...
18831 (thunderx2t99_regmove_cost): This.
18832 (vulcan_vector_cost): Rename to ...
18833 (thunderx2t99_vector_cost): this.
18834 (vulcan_branch_cost): Rename to ...
18835 (thunderx2t99_branch_cost): This.
18836 (vulcan_tunings): Rename to ...
18837 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
18838 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
18840 2017-01-09 Martin Jambor <mjambor@suse.cz>
18844 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
18845 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
18846 (propagate_vr_accross_jump_function): Use the above function for all
18847 value range computations for pass-through jump functions and type
18848 converasion from explicit value range values.
18849 (ipcp_propagate_stage): Do not attempt to deduce types of formal
18850 parameters from TYPE_ARG_TYPES.
18851 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
18852 (ipa_write_node_info): Stream type of the actual argument.
18853 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
18855 2017-01-09 Martin Liska <mliska@suse.cz>
18858 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
18859 (lookup_compiler): Do not show error message with have_E.
18861 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18863 PR tree-optimization/78938
18864 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
18865 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
18866 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
18867 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
18870 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18872 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
18875 2017-01-09 Richard Biener <rguenther@suse.de>
18877 PR tree-optimization/78997
18878 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
18879 name condition properly.
18881 2017-01-09 Richard Biener <rguenther@suse.de>
18884 * dwarf2out.c (is_cxx): New overload with context.
18885 (is_naming_typedef_decl): Use it.
18887 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18889 * invoke.texi (Option Summary): Correct spacing in option lists
18890 and add line breaks to fix over-long lines.
18892 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18894 PR middle-end/17660
18896 * extend.texi (Common Variable Attributes): Add xref to GCC
18897 Internals manual to explain mode attribute keywords.
18899 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18902 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
18903 and Preprocessor Options.
18904 (Options for Linking): Document -pthread here....
18905 (RS/6000 and PowerPC Options): ...not here.
18906 (Solaris 2 Options): ...or here.
18907 * doc/cppopts.texi: Document -pthread.
18909 2017-01-08 Martin Sebor <msebor@redhat.com>
18911 PR middle-end/77708
18912 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
18913 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
18914 New member functions.
18915 (format_directive): Used them.
18917 (pass_sprintf_length::handle_gimple_call): Same.
18918 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
18919 to avoid truncation for any argument.
18920 (extract_affine_mul): Same.
18921 * tree.c (get_file_function_name): Same.
18923 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18925 PR middle-end/77484
18926 * predict.def (PRED_INDIR_CALL): Set to 86.
18928 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18930 PR preprocessor/54124
18931 * doc/cppopts.texi: Reformat -d subtable to list the full name
18932 of the options. Add cross-reference to the docs for the general
18933 compiler -d options.
18934 * doc/invoke.texi (Developer Options): Add cross-reference to the
18935 preprocessor-specific -d option documentation.
18937 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18939 PR preprocessor/13498
18940 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
18941 redudant material, and reflect new command-line options.
18942 (System Headers): Likewise.
18944 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18946 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
18947 -isystem, and -idirafter. Copy-edit.
18948 * doc/cppopts.texi: Copy-edit. Remove contradiction about
18949 default for -ftrack-macro-expansion. Delete obsolete and
18950 badly-formatted implementation details about -fdebug-cpp output.
18951 * doc/cppwarnopts.texi: Copy-edit.
18953 2017-01-07 David Malcolm <dmalcolm@redhat.com>
18956 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18957 that the transition from a max line width >= 1<<10 to narrower
18958 lines works correctly.
18960 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
18962 * doc/options.texi (PerFunction): New.
18963 * opt-functions.awk (switch_flags): Map both Optimization and
18964 PerFunction to CL_OPTIMIZATION.
18965 * opth-gen.awk: Test for PerFunction flag along with
18967 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
18968 it only when the latter is present. Skip those that don't in
18969 the hash function generator.
18970 * common.opt (fvar-tracking): Mark as PerFunction instead of
18972 (fvar-tracking-assignments): Likewise.
18973 (fvar-tracking-assignments-toggle): Likewise.
18974 (fvar-tracking-uninit): Likewise.
18976 2017-01-07 Jakub Jelinek <jakub@redhat.com>
18978 PR translation/79018
18979 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
18982 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
18985 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
18986 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
18987 TARGET_LONG_JUMP_TABLE_OFFSETS.
18988 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
18989 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
18990 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
18991 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
18992 * config/m68k/m68k.md (tablejump expander): Likewise.
18993 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
18994 TARGET_LONG_JUMP_TABLE_OFFSETS.
18995 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
18996 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
18998 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
18999 David Holsgrove <david.holsgrove@xilinx.com>
19001 * common/config/microblaze/microblaze-common.c
19002 (TARGET_EXCEPT_UNWIND_INFO): Remove.
19003 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
19005 * config/microblaze/microblaze.c (microblaze_must_save_register)
19006 (microblaze_expand_epilogue, microblaze_return_addr): Handle
19008 (microblaze_eh_return): New function.
19009 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
19010 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
19011 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
19012 * config/microblaze/microblaze.md (eh_return): New pattern.
19014 2017-01-06 Jakub Jelinek <jakub@redhat.com>
19016 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
19017 GCC_DIAGNOSTIC_STRINGIFY): Define.
19019 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
19021 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19023 * config/arm/arm.md (<mcrr>): New.
19025 * config/arm/arm.c (arm_arch5te): New.
19026 (arm_option_override): Set arm_arch5te.
19027 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
19029 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
19030 (arm_mcrr_qualifiers): ... this. New.
19031 (MRRC_QUALIFIERS): Define to...
19032 (arm_mrrc_qualifiers): ... this. New.
19033 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
19035 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
19036 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
19037 (MRRCI, mrrc, MRRC): New.
19038 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
19039 VUNSPEC_MRRC2): New.
19041 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19043 * config/arm/arm.md (<mcr>): New.
19045 * config/arm/arm.c (arm_coproc_builtin_available): Add
19046 support for mcr, mrc, mcr2 and mrc2.
19047 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
19048 (arm_mcr_qualifiers): ... this. New.
19049 (MRC_QUALIFIERS): Define to ...
19050 (arm_mrc_qualifiers): ... this. New.
19051 (MCR_QUALIFIERS): Define to ...
19052 (arm_mcr_qualifiers): ... this. New.
19053 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
19055 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
19056 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
19057 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
19058 VUNSPEC_MRC2): New.
19060 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19062 * config/arm/arm.md (*ldc): New.
19066 * config/arm/arm.c (arm_coproc_builtin_available): Add
19067 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
19068 (arm_coproc_ldc_stc_legitimate_address): New.
19069 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
19070 'qualifier_const_pointer'.
19071 (LDC_QUALIFIERS): Define to...
19072 (arm_ldc_qualifiers): ... this. New.
19073 (STC_QUALIFIERS): Define to...
19074 (arm_stc_qualifiers): ... this. New.
19075 * config/arm/arm-protos.h
19076 (arm_coproc_ldc_stc_legitimate_address): New.
19077 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
19078 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
19079 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
19080 stc2, stcl, stc2l): New.
19081 * config/arm/constraints.md (Uz): New.
19082 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
19083 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
19084 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
19085 VUNSPEC_STC2L): New.
19087 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19089 * config/arm/arm.md (<cdp>): New.
19090 * config/arm/arm.c (neon_const_bounds): Rename this ...
19091 (arm_const_bounds): ... this.
19092 (arm_coproc_builtin_available): New.
19093 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
19094 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
19095 (CDP_QUALIFIERS): Define to...
19096 (arm_cdp_qualifiers): ... this. New.
19098 (arm_expand_builtin_args): Add case for 6 arguments.
19099 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
19100 (arm_const_bounds): ... this.
19101 (arm_coproc_builtin_available): New.
19102 * config/arm/arm_acle.h (__arm_cdp): New.
19104 * config/arm/arm_acle_builtins.def (cdp): New.
19106 * config/arm/iterators.md (CDPI,CDP,cdp): New.
19107 * config/arm/neon.md: Rename all 'neon_const_bounds' to
19108 'arm_const_bounds'.
19109 * config/arm/types.md (coproc): New.
19110 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
19111 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
19112 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
19113 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
19115 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19117 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
19118 (UBINOP_QUALIFIERS): New.
19120 (acle_builtin_data): New. Change comment.
19121 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
19122 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
19123 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
19124 arm_acle_builtins.def.
19125 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
19126 (arm_init_acle_builtins): New.
19127 (CRC32_BUILTIN): Remove.
19128 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
19129 crc32cb, crc32ch and crc32cw.
19130 (arm_init_crc32_builtins): Remove.
19131 (arm_init_builtins): Use arm_init_acle_builtins rather
19132 than arm_init_crc32_builtins.
19133 (arm_expand_acle_builtin): New.
19134 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
19135 * config/arm/arm_acle_builtins.def: New.
19137 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19139 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
19140 (arm_builtin_datum): ... this.
19141 (arm_init_neon_builtin): Rename to ...
19142 (arm_init_builtin): ... this. Add a new parameters PREFIX
19143 and USE_SIG_IN_NAME.
19144 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
19145 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
19146 'arm_builtin_datum'.
19147 (arm_init_vfp_builtins): Likewise.
19148 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
19149 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
19150 (arm_expand_neon_args): Rename to ...
19151 (arm_expand_builtin_args): ... this. Rename builtin_arg
19152 enum values and differentiate between ARG_BUILTIN_MEMORY
19153 and ARG_BUILTIN_NEON_MEMORY.
19154 (arm_expand_neon_builtin_1): Rename to ...
19155 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
19156 values, arm_expand_builtin_args and add bool parameter NEON.
19157 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
19158 (arm_expand_vfp_builtin): Likewise.
19159 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
19161 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19163 PR middle-end/77484
19164 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
19165 * predict.c (tree_estimate_probability_bb): Reverse direction of
19166 polymorphic call predictor.
19168 2017-01-06 David Malcolm <dmalcolm@redhat.com>
19170 * passes.c (execute_one_pass): Split out pass-skipping logic into...
19171 (determine_pass_name_match): ...this new function and...
19172 (should_skip_pass_p): ...this new function.
19174 2017-01-06 Nathan Sidwell <nathan@acm.org>
19176 * ipa-visibility.c (function_and_variable_visibility): Reformat
19177 comments and long lines. Remove extrneous if.
19178 * symtab.c (symtab_node::make_decl_local): Fix code format.
19179 (symtab_node::set_section_for_node): Fix comment typo.
19181 2017-01-06 Martin Liska <mliska@suse.cz>
19184 * lra-constraints.c: Rename invariant to lra_invariant.
19185 * predict.c (set_even_probabilities): Initialize e to NULL.
19187 2017-01-05 Martin Sebor <msebor@redhat.com>
19189 PR tree-optimization/78910
19190 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
19191 (format_integer): Correct off-by-one error in the handling
19192 of precision with negative numbers in signed conversions..
19194 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
19196 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
19198 2017-01-05 Jakub Jelinek <jakub@redhat.com>
19200 PR tree-optimization/71016
19201 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
19202 factor_out_conditional_conversion. Formatting fix.
19203 (factor_out_conditional_conversion): Add cond_stmt argument.
19204 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
19205 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
19208 2017-01-05 David Malcolm <dmalcolm@redhat.com>
19210 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
19211 read-rtl-function.o, and selftest-rtl.o.
19212 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
19213 (selftest::aarch64_test_loading_full_dump): New function.
19214 (selftest::aarch64_run_selftests): New function.
19215 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
19216 selftest::aarch64_run_selftests.
19217 * config/i386/i386.c
19218 (selftest::ix86_test_loading_dump_fragment_1): New function.
19219 (selftest::ix86_test_loading_call_insn): New function.
19220 (selftest::ix86_test_loading_full_dump): New function.
19221 (selftest::ix86_test_loading_unspec): New function.
19222 (selftest::ix86_run_selftests): Call the new functions.
19223 * emit-rtl.c (maybe_set_max_label_num): New function.
19224 * emit-rtl.h (maybe_set_max_label_num): New decl.
19225 * function.c (instantiate_decls): Guard call to
19226 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
19227 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
19229 * gensupport.c (gen_reader::gen_reader): Pass "false"
19230 for new "compact" param of rtx_reader.
19231 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
19232 rather than an empty string for NULL strings.
19233 * read-md.c: Potentially include config.h rather than bconfig.h.
19234 Wrap include of errors.h with #ifdef GENERATOR_FILE.
19235 (have_error): New global, copied from errors.c.
19236 (md_reader::read_name): Rename to...
19237 (md_reader::read_name_1): ...this, adding "out_loc" param,
19238 and converting "missing name or number" to returning false, rather
19240 (md_reader::read_name): Reimplement in terms of read_name_1.
19241 (md_reader::read_name_or_nil): New function.
19242 (md_reader::read_string): Handle "(nil)" by returning NULL.
19243 (md_reader::md_reader): Add new param "compact".
19244 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
19245 (md_reader::read_file): New method.
19246 * read-md.h (md_reader::md_reader): Add new param "compact".
19247 (md_reader::read_file): New method.
19248 (md_reader::is_compact): New accessor.
19249 (md_reader::read_name): Convert return type from void to file_location.
19250 (md_reader::read_name_or_nil): New decl.
19251 (md_reader::read_name_1): New decl.
19252 (md_reader::m_compact): New field.
19253 (noop_reader::noop_reader): Pass "false" for new "compact" param
19255 (rtx_reader::rtx_reader): Add new "compact" param.
19256 (rtx_reader::read_rtx_operand): Make virtual and convert return
19257 type from void to rtx.
19258 (rtx_reader::read_until): New decl.
19259 (rtx_reader::handle_any_trailing_information): New virtual function.
19260 (rtx_reader::postprocess): New virtual function.
19261 (rtx_reader::finalize_string): New virtual function.
19262 (rtx_reader::m_in_call_function_usage): New field.
19263 (rtx_reader::m_reuse_rtx_by_id): New field.
19264 * read-rtl-function.c: New file.
19265 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
19266 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
19267 (selftest::verify_three_block_rtl_cfg): New decl.
19268 * read-rtl-function.h: New file.
19269 * read-rtl.c: Potentially include config.h rather than bconfig.h.
19270 For host, include function.h, memmodel.h, and emit-rtl.h.
19271 (one_time_initialization): New function.
19272 (struct compact_insn_name): New struct.
19273 (compact_insn_names): New array.
19274 (find_code): Handle insn codes in compact dumps.
19275 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
19276 (bind_subst_iter_and_attr): Likewise.
19277 (add_condition_to_string): Likewise.
19278 (add_condition_to_rtx): Likewise.
19279 (apply_attribute_uses): Likewise.
19280 (add_current_iterators): Likewise.
19281 (apply_iterators): Likewise.
19282 (initialize_iterators): Guard usage of apply_subst_iterator with
19283 #ifdef GENERATOR_FILE.
19284 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
19285 (md_reader::read_mapping): Likewise.
19286 (add_define_attr_for_define_subst): Likewise.
19287 (add_define_subst_attr): Likewise.
19288 (read_subst_mapping): Likewise.
19289 (check_code_iterator): Likewise.
19290 (rtx_reader::read_rtx): Likewise. Move one-time initialization
19292 (one_time_initialization): New function.
19293 (rtx_reader::read_until): New method.
19294 (read_flags): New function.
19295 (parse_reg_note_name): New function.
19296 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
19297 Handle reuse_rtx ids.
19298 Wrap iterator lookup within #ifdef GENERATOR_FILE.
19299 Add parsing support for RTL dumps, mirroring the special-cases in
19300 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
19301 values, and calling handle_any_trailing_information.
19302 (rtx_reader::read_rtx_operand): Convert return type from void
19303 to rtx, returning return_rtx. Handle case 'e'. Call
19304 finalize_string on XSTR and XTMPL fields.
19305 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
19306 "(nil)" values were omitted. Call the postprocess vfunc on the
19308 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
19309 class ctor. Initialize m_in_call_function_usage. Call
19310 one_time_initialization.
19311 * rtl-tests.c (selftest::test_uncond_jump): Call
19312 set_new_first_and_last_insn.
19313 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
19314 * selftest-rtl.c: New file.
19315 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
19316 (selftest::get_insn_by_uid): New decl.
19317 * selftest-run-tests.c (selftest::run_tests): Call
19318 read_rtl_function_c_tests.
19319 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
19320 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
19323 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
19325 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
19326 operands in a special way. Assert that pos+len <= mode precision.
19328 2017-01-05 Jakub Jelinek <jakub@redhat.com>
19330 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
19331 3 argument Alias with unlimited for the negative form.
19332 (fno-vect-cost-model): Removed.
19334 2017-01-05 Martin Liska <mliska@suse.cz>
19336 * hsa-gen.c (gen_hsa_divmod): New function.
19337 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
19339 2017-01-05 Martin Liska <mliska@suse.cz>
19342 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
19345 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19347 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
19348 small constant length operands.
19350 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19352 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
19353 between loop iterations.
19355 2017-01-05 Martin Liska <mliska@suse.cz>
19358 * gimplify.c (gimplify_decl_expr): Compare to
19359 asan_poisoned_variables instread of checking flags.
19360 (gimplify_target_expr): Likewise.
19361 (gimplify_expr): Likewise.
19362 (gimplify_function_tree): Conditionally initialize
19363 asan_poisoned_variables.
19365 2017-01-04 Jeff Law <law@redhat.com>
19367 PR tree-optimizatin/78812
19368 * rtl.h (contains_mem_rtx_p): Prototype.
19369 * ifcvt.c (containts_mem_rtx_p): Move from here to...
19370 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
19371 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
19372 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
19373 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
19375 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19377 * input.c (assert_char_at_range): Default-initialize actual_range.
19379 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19381 * df-scan.c (df_ref_create_structure): Make regno unsigned,
19382 to match the caller.
19384 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19386 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
19387 insns after final jump in test to emit dummy move.
19389 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19391 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
19392 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
19394 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19396 * multiple_target.c (create_dispatcher_calls): Init e_next.
19397 * tree-ssa-loop-split.c (split_loop): Init border.
19398 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
19401 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
19406 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
19407 (altivec_register_operand): Do not return true if the operand
19408 contains a SUBREG mixing SImode and SFmode.
19409 (vsx_register_operand): Likewise.
19410 (vsx_reg_sfsubreg_ok): New predicate.
19411 (vfloat_operand): Do not return true if the operand contains a
19412 SUBREG mixing SImode and SFmode.
19413 (vint_operand): Likewise.
19414 (vlogical_operand): Likewise.
19415 (gpc_reg_operand): Likewise.
19416 (int_reg_operand): Likewise.
19417 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
19418 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
19419 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
19421 (rs6000_emit_move_si_sf_subreg): New helper function.
19422 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
19423 fixup SUBREGs involving SImode and SFmode.
19424 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
19425 numbers for the new peephole2 optimization.
19426 (peephole2 for SFmode unions): New peephole2 to optimize cases in
19427 the GLIBC math library that do AND/IOR/XOR operations on single
19428 precision floating point.
19429 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
19430 target macros to say whether we need to avoid SUBREGs mixing
19432 (TARGET_ALLOW_SF_SUBREG): Likewise.
19433 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
19434 (UNSPEC_SI_FROM_SF): Likewise.
19435 (iorxor): Change spacing.
19436 (and_ior_xor): New iterator for AND, IOR, and XOR.
19437 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
19438 (movdi_from_sf_zero_ext): Likewise.
19439 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
19440 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
19441 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
19442 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
19443 (fms<mode>4): Likewise.
19444 (fnma<mode>4): Likewise.
19445 (fnms<mode>4): Likewise.
19446 (nfma<mode>4): Likewise.
19447 (nfms<mode>4): Likewise.
19449 2017-01-04 Marek Polacek <polacek@redhat.com>
19452 * doc/invoke.texi: Document -Wpointer-compare.
19454 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19456 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
19459 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
19460 descriptions for -gdwarf-5 and emit them as uleb128 instead of
19463 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
19466 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
19467 documentation of the powerpc_popcntb_ok attribute.
19468 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19469 code to issue warning messages if a requested CPU configuration is
19470 not supported by the binary (assembler and loader) toolchain.
19471 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
19472 made to define a built-in function that has been disabled.
19473 (paired_init_builtins): Add assertion to prevent ICE if attempt is
19474 made to define a built-in function that has been disabled.
19475 (altivec_init_builtins): Add comment explaining why definition
19476 of the DST built-in functions is not preceded by an assertion
19477 check. Add assertions to prevent ICE if attempts are made to
19478 define an altivec predicate or an abs* built-in function that has
19480 (htm_init_builtins): Add comment explaining why definition of the
19481 htm built-in functions is not preceded by an assertion check.
19483 2017-01-04 Jeff Law <law@redhat.com>
19485 PR tree-optimizatin/67955
19486 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
19487 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
19488 the points-to solution does not include pt_null. Use DECL_PT_UID
19491 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
19493 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
19494 Use gen_int_mode instead of gen_lopwart for const_int operands.
19496 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19498 PR tree-optimization/71563
19499 * match.pd: Simplify X << Y into X if Y is known to be 0 or
19500 out of range value - has low bits known to be zero.
19502 2017-01-04 Alan Modra <amodra@gmail.com>
19504 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
19505 * configure: Regenerate.
19506 * config.in: Regenerate.
19508 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19511 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
19512 a substring of the message, but strcmp with the whole message. Ifdef
19513 ENABLE_NLS, translate the message first using dgettext.
19515 2017-01-03 Jeff Law <law@redhat.com>
19517 PR tree-optimizatin/78856
19518 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
19519 (mark_threaded_blocks): Remove code to truncate thread paths that
19520 cross multiple loop headers. Instead invalidate the cached loop
19521 iteration information and handle case of a thread path walking
19522 into an irreducible region.
19524 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
19527 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
19528 assertions. Add support for doing the signbit if the IEEE 128-bit
19529 floating point value is in a GPR.
19530 * config/rs6000/rs6000.md (Fsignbit): Delete.
19531 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
19532 Update the length attribute if the value is in a GPR.
19533 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
19534 the sign or zero extension instruction, since the value is always 0/1.
19535 (signbit<mode>2_dm2): Delete using <Fsignbit>.
19538 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
19539 extracting SImode to a GPR register so that we can generate a
19540 store, limit the vector to be in a traditional Altivec register
19541 for the vextuwrx instruction.
19543 2017-01-03 Ian Lance Taylor <iant@google.com>
19545 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
19547 2017-01-03 Martin Sebor <msebor@redhat.com>
19549 PR tree-optimization/78696
19550 * gimple-ssa-sprintf.c (format_floating): Correct handling of
19551 precision. Use MPFR for %f for greater fidelity. Correct handling
19553 (pass_sprintf_length::compute_format_length): Set width and precision
19554 specified by asrerisk to void_node for vararg functions.
19555 (try_substitute_return_value): Adjust dump output.
19557 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
19559 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
19561 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
19563 * doc/invoke.texi (SPARC options): Document -mlra as the default.
19564 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
19565 -mlra/-mno-lra was passed to the compiler.
19567 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
19569 PR rtl-optimization/65618
19570 * emit-rtl.c (try_split): Move initialization of "before" and
19571 "after" to just before the call to emit_insn_after_setloc.
19573 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
19575 * doc/md.texi (Standard Names): Remove reference to Java frontend.
19577 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
19579 * dwarf2out.c (gen_enumeration_type_die): When
19580 -gno-strict-dwarf, add a DW_AT_encoding attribute.
19582 2017-01-03 Jakub Jelinek <jakub@redhat.com>
19584 PR tree-optimization/78965
19585 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
19586 Change first argument from const call_info & to call_info &. For %n
19587 set info.nowrite to false.
19589 PR middle-end/78901
19590 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
19591 possibly throwing calls.
19593 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
19594 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
19595 and fns handling, rather than in a separate case SSA_NAME.
19597 2017-01-02 Jeff Law <law@redhat.com>
19599 * config/darwin-driver.c (darwin_driver_init): Const-correctness
19600 fixes for first_period and second_period variables.
19602 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
19605 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
19606 (*insvqi_1): New insn pattern.
19607 (*insvqi_1_mem_rex64): Ditto.
19608 (*insvqi_2): Ditto.
19609 (*insvqi_3): Rename from *insvqi.
19611 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
19613 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
19615 * doc/cfg.texi (Edges): Remove reference to Java.
19616 (Maintaining the CFG): Ditto.
19618 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19620 PR middle-end/77674
19621 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
19622 transparent aliases.
19624 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19626 PR middle-end/77484
19627 * predict.def (PRED_CALL): Update hitrate.
19628 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
19629 * predict.c (tree_estimate_probability_bb): Split CALL predictor
19630 into direct/indirect/polymorphic variants.
19632 2017-01-01 Jakub Jelinek <jakub@redhat.com>
19634 Update copyright years.
19636 * gcc.c (process_command): Update copyright notice dates.
19637 * gcov-dump.c (print_version): Ditto.
19638 * gcov.c (print_version): Ditto.
19639 * gcov-tool.c (print_version): Ditto.
19640 * gengtype.c (create_file): Ditto.
19641 * doc/cpp.texi: Bump @copying's copyright year.
19642 * doc/cppinternals.texi: Ditto.
19643 * doc/gcc.texi: Ditto.
19644 * doc/gccint.texi: Ditto.
19645 * doc/gcov.texi: Ditto.
19646 * doc/install.texi: Ditto.
19647 * doc/invoke.texi: Ditto.
19649 Copyright (C) 2017 Free Software Foundation, Inc.
19651 Copying and distribution of this file, with or without modification,
19652 are permitted in any medium without royalty provided the copyright
19653 notice and this notice are preserved.