1 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3 * tree.c (make_tree_vec_stat): Remove _stat from name.
4 (grow_tree_vec_stat): Likewise.
5 * tree.h (make_tree_vec_stat): Adjust prototype.
6 (grow_tree_vec_stat): Likewise.
7 (make_tree_vec): Remove macro.
8 (grow_tree_vec): Likewise.
10 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12 * fold-const.c (fold_build1_stat_loc): Adjust.
13 (fold_build2_stat_loc): Likewise.
14 (fold_build3_stat_loc): Likewise.
15 * tree.c (build0_stat): Remove _stat from name.
16 (build1_stat): Likewise.
17 (build2_stat): Likewise.
18 (build3_stat): Likewise.
19 (build4_stat): Likewise.
20 (build5_stat): Likewise.
21 * tree.h (build1_loc): Remove macro, and rename _stat function
23 (build2_loc): Likewise.
24 (build3_loc): Likewise.
25 (build4_loc): Likewise.
26 (build5_loc): Likewise.
28 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30 * tree.c (make_int_cst_stat): Remove _stat from name.
31 * tree.h (make_int_cst_stat): Adjust prototype.
32 (make_int_cst): Remove macro.
34 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
36 * tree.c (make_tre_binfo_stat): Remove _stat from name.
37 * tree.h (make_tree_binfo_stat): Adjust prototype.
38 (make_tree_binfo): Remove.
40 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
42 * tree.c (copy_node_stat): Rename to copy_node.
43 (build_distinct_type_copy): Adjust.
44 * tree.h (copy_node_stat): Adjust prototype.
45 (copy_node): Remove macro.
47 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
49 * tree.c (make_node_stat): rename to make_node.
50 (build_tree_list_stat): Adjust.
51 (build0_stat): Likewise.
52 (build2_stat): Likewise.
53 (build3_stat): Likewise.
54 (build4_stat): Likewise.
55 (build5_stat): Likewise.
56 (build_decl_stat): Likewise.
57 * tree.h (make_node_stat): Adjust prototype.
58 (make_node): remove macro.
60 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
62 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
63 (PPC_FEATURE2_SCV): Likewise.
64 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
66 2017-07-28 Tamar Christina <tamar.christina@arm.com>
68 * config/aarch64/aarch64.c
69 (aarch64_internal_mov_immediate): Add new special pattern.
70 * config/aarch64/aarch64.md (*movdi_aarch64):
71 Add reg/32bit const mov case.
73 2017-07-28 Tamar Christina <tamar.christina@arm.com>
74 Richard Sandiford <richard.sandiford@linaro.org>
76 * config/aarch64/aarch64.md (mov<mode>): Generalize.
77 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
78 Add integer and movi cases.
79 (movi-split-hf-df-sf split, fp16): New.
80 (enabled): Added TARGET_FP_F16INST.
81 * config/aarch64/iterators.md (GPF_HF): New.
82 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
84 2017-07-28 Tamar Christina <tamar.christina@arm.com>
86 * config/aarch64/aarch64.c
87 (aarch64_simd_container_mode): Add prototype.
88 (aarch64_expand_mov_immediate): Add HI support.
89 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
90 (aarch64_can_const_movi_rtx_p): New.
91 (aarch64_preferred_reload_class):
92 Remove restrictions of using FP registers for certain SIMD operations.
93 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
94 (aarch64_valid_floating_const): Add integer move validation.
95 (aarch64_simd_imm_scalar_p): Remove.
96 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
97 (aarch64_legitimate_constant_p): Expand list of supported cases.
98 * config/aarch64/aarch64-protos.h
99 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
100 (aarch64_reinterpret_float_as_int): New.
101 (aarch64_simd_imm_scalar_p): Remove.
102 * config/aarch64/constraints.md (Uvi): New.
103 (Dd): Split into Ds and new Dd.
104 * config/aarch64/aarch64.md (*movsi_aarch64):
106 (*movdi_aarch64): Add SIMD mov case.
108 2017-07-28 Bin Cheng <bin.cheng@arm.com>
110 * tree-predcom.c: (struct chain): Handle store-store chain in which
111 stores for elimination only store loop invariant values.
112 (execute_pred_commoning_chain): Ditto.
113 (prepare_initializers_chain_store_elim): Ditto.
114 (prepare_finalizers): Ditto.
115 (is_inv_store_elimination_chain): New function.
116 (initialize_root_vars_store_elim_1): New function.
118 2017-07-28 Bin Cheng <bin.cheng@arm.com>
120 * tree-predcom.c: Revise general description of the pass.
121 (enum chain_type): New enum type for store elimination.
122 (struct chain): New field supporting store elimination.
123 (struct component): Ditto.
124 (dump_chain): Dump store-stores chain.
125 (release_chain): Release resources.
126 (split_data_refs_to_components): Compute and create component
127 contains only stores for elimination.
128 (get_chain_last_ref_at): New function.
129 (make_invariant_chain): Initialization.
130 (make_rooted_chain): Specify chain type in parameter and record it.
131 (add_looparound_copies): Skip for store-stores chain.
132 (determine_roots_comp): Compute type of chain and pass it to
134 (initialize_root_vars_store_elim_2): New function.
135 (finalize_eliminated_stores): New function.
136 (remove_stmt): Handle store for elimination.
137 (execute_pred_commoning_chain): Execute predictive commoning on
139 (determine_unroll_factor): Skip unroll for store-stores chain.
140 (prepare_initializers_chain_store_elim): New function.
141 (prepare_initializers_chain): Hanlde store-store chain.
142 (prepare_finalizers_chain, prepare_finalizers): New function.
143 (tree_predictive_commoning_loop): Return integer value indicating
144 if loop is unrolled or lcssa form is corrupted.
145 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
147 2017-07-28 Bin Cheng <bin.cheng@arm.com>
149 * tree-predcom.c (initialize_root): Delete.
150 (execute_pred_commoning_chain): Initialize root vars and replace
151 reference of non-combined chain directly, rather than call above
154 2017-07-28 Bin Cheng <bin.cheng@arm.com>
156 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
157 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
159 2017-07-28 Bin Cheng <bin.cheng@arm.com>
161 * tree-predcom.c (struct chain): New field init_seq.
162 (release_chain): Release init_seq.
163 (prepare_initializers_chain): Record intialization stmts in above
165 (insert_init_seqs): New function.
166 (tree_predictive_commoning_loop): Call insert_init_seqs.
168 2017-07-28 Bin Cheng <bin.cheng@arm.com>
170 * tree-predcom.c (determine_roots_comp): Skip trivial components.
172 2017-07-28 Richard Biener <rguenther@suse.de>
174 * match.pd: Remove superfluous :c.
175 * genmatch.c (simplify::id): Add member.
176 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
178 (current_id): New global.
179 (dt_node::parent): Move from ...
180 (dt_operand::parent): ... here. Add for_id member.
181 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
182 (decision_tree::find_node): Relax order requirement when
183 merging DT_TRUE nodes to ones inbetween the current simplify
184 and the one we try to merge with. Add diagnostic whenever
185 we need to enforce pattern order by not merging.
186 (decision_tree::insert): Set current_id.
187 (decision_tree::print_node): Dump parent node and for_id.
188 (parser::last_id): Add member.
189 (parser::push_simplify): Assign unique id.
190 (parser::parser): Initialize last_id.
192 2017-07-28 Martin Liska <mliska@suse.cz>
195 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
196 gimple_build_debug_bind.
198 2017-07-28 Richard Biener <rguenther@suse.de>
200 PR tree-optimization/81502
201 * match.pd: Add pattern combining BIT_INSERT_EXPR with
203 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
205 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
206 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
207 for BIT_FIELD_REF args.
208 * fold-const.c (make_bit_field_ref): Likewise.
209 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
211 2017-07-28 Jakub Jelinek <jakub@redhat.com>
214 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
215 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
216 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
217 Or it into SANITIZER_UNDEFINED.
218 * ubsan.c: Include gimple-fold.h and varasm.h.
219 (ubsan_expand_ptr_ifn): New function.
220 (instrument_pointer_overflow): New function.
221 (maybe_instrument_pointer_overflow): New function.
222 (instrument_object_size): Formatting fix.
223 (pass_ubsan::execute): Call instrument_pointer_overflow
224 and maybe_instrument_pointer_overflow.
225 * internal-fn.c (expand_UBSAN_PTR): New function.
226 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
227 * sanitizer.def (__ubsan_handle_pointer_overflow,
228 __ubsan_handle_pointer_overflow_abort): New builtins.
229 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
230 * internal-fn.def (UBSAN_PTR): New internal function.
231 * opts.c (sanitizer_opts): Add pointer-overflow.
232 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
233 * fold-const.c (build_range_check): Compute pointer range check in
234 integral type if pointer arithmetics would be needed. Formatting
237 2017-07-28 Martin Liska <mliska@suse.cz>
240 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
241 parameters that are of a variable-length.
243 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
245 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
247 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
248 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
249 (CRT_CALL_STATIC_FUNCTION): Likewise.
250 (ASM_DEFAULT_SPEC): New define.
251 (ASM_SPEC32): Likewise.
252 (ASM_SPEC64): Likewise.
253 (ASM_SPEC_COMMON): Likewise.
254 (ASM_SPEC): Likewise.
255 (INVALID_64BIT): Likewise.
256 (LINK_OS_DEFAULT_SPEC): Likewise.
257 (LINK_OS_SPEC32): Likewise.
258 (LINK_OS_SPEC64): Likewise.
259 (POWERPC_LINUX): Likewise.
260 (PTRDIFF_TYPE): Likewise.
261 (RESTORE_FP_PREFIX): Likewise.
262 (RESTORE_FP_SUFFIX): Likewise.
263 (SAVE_FP_PREFIX): Likewise.
264 (SAVE_FP_SUFFIX): Likewise.
265 (SIZE_TYPE): Likewise.
266 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
267 (TARGET_64BIT): Likewise.
268 (TARGET_64BIT): Likewise.
269 (TARGET_AIX): Likewise.
270 (WCHAR_TYPE_SIZE): Likewise.
271 (WCHAR_TYPE): Undefine.
272 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
273 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
274 (CPP_OS_RTEMS_SPEC): Delete.
275 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
276 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
278 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
280 2017-07-28 Jakub Jelinek <jakub@redhat.com>
282 PR tree-optimization/81578
283 * tree-parloops.c (build_new_reduction): Bail out if
284 reduction_code isn't one of the standard OpenMP reductions.
285 Move the details printing after that decision.
287 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
289 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
290 related to reload_in_progress.
291 (splat_input_operand): Likewise.
292 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
294 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
296 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
297 (TARGET_INSTANTIATE_DECLS): Likewise.
298 (legitimate_indexed_address_p): Delete reload_in_progress code.
299 (rs6000_debug_legitimate_address_p): Likewise.
300 (rs6000_eliminate_indexed_memrefs): Likewise.
301 (rs6000_emit_le_vsx_store): Likewise.
302 (rs6000_emit_move_si_sf_subreg): Likewise.
303 (rs6000_emit_move): Likewise.
304 (register_to_reg_type): Likewise.
305 (rs6000_pre_atomic_barrier): Likewise.
306 (rs6000_machopic_legitimize_pic_address): Likewise.
307 (rs6000_allocate_stack_temp): Likewise.
308 (rs6000_address_for_fpconvert): Likewise.
309 (rs6000_address_for_altivec): Likewise.
310 (rs6000_secondary_memory_needed_rtx): Delete function.
311 (rs6000_check_sdmode): Likewise.
312 (rs6000_alloc_sdmode_stack_slot): Likewise.
313 (rs6000_instantiate_decls): Likewise.
314 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
315 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
316 Delete reload_in_progress.
317 (*vec_reload_and_plus_<mptrsize>): Likewise.
318 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
319 (vsx_div_v2di): Likewise.
320 (vsx_udiv_v2di): Likewise.
322 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
324 * config/rs6000/rs6000.opt (mlra): Replace with stub.
325 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
326 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
327 (rs6000_debug_reg_global): Delete print of LRA status.
328 (rs6000_option_override_internal): Delete dead LRA related code.
329 (rs6000_lra_p): Delete function.
330 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
332 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
334 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
335 * config/riscv/rtems.h: New file.
337 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
338 Sudakshina Das <sudi.das@arm.com>
340 * config/aarch64/aarch64.md
341 (define_split for and<mode>3nr_compare): Move
342 non aarch64_logical_operand to a register.
343 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
344 register immediate operand to a register.
345 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
347 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
350 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
352 2017-07-27 Richard Biener <rguenther@suse.de>
354 PR tree-optimization/81573
355 PR tree-optimization/81494
356 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
357 multi defuse cycle case.
359 2017-07-27 Richard Biener <rguenther@suse.de>
361 PR tree-optimization/81571
362 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
365 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
367 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
368 earlier and only if MASK_FPU is set. Adjust formatting.
370 2017-07-27 Martin Liska <mliska@suse.cz>
372 * opt-functions.awk: Add validation of value of Init.
373 * optc-gen.awk: Pass new argument.
375 2017-07-27 Martin Liska <mliska@suse.cz>
377 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
380 2017-07-27 Martin Liska <mliska@suse.cz>
382 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
383 BBs and edges seen by autoFDO.
385 2017-07-27 Richard Biener <rguenther@suse.de>
387 PR tree-optimization/81502
388 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
389 with incompatible but same sized type.
390 (execute_update_addresses_taken): Likewise.
392 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
394 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
395 flag_tree_loop_vectorize rather than flag_tree_vectorize.
397 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
400 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
401 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
402 Change s_operand to memory_operand.
404 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
406 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
407 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
408 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
409 Emit instructions rather than returning an expression. Handle TFmode
410 and KFmode by casting to TImode.
411 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
412 (rs6000_emit_le_vsx_store): Likewise.
413 * config/rs6000/vsx.md (VSX_TI): New iterator.
414 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
415 (*vsx_le_undo_permute_<mode>): Likewise.
416 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
417 emit the split sequence.
418 (*vsx_le_perm_store_<mode>): Likewise.
420 2017-07-27 Jakub Jelinek <jakub@redhat.com>
422 PR tree-optimization/81555
423 PR tree-optimization/81556
424 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
425 if true, force CHANGED for the recursive invocation.
426 (reassociate_bb): Remember original length of ops array, pass
427 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
429 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
430 attributes for noipa attribute. For naked attribute use
431 lookup_attribute first before lookup_attribute_spec.
432 * final.c (rest_of_handle_final): Disable IPA RA for functions with
434 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
435 for functions with noipa attribute.
436 (cgraph_externally_visible_p): Return true for functions with noipa
438 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
439 for functions with noipa attribute.
440 * doc/extend.texi: Document noipa function attribute.
441 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
442 also for functions with noipa attribute.
443 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
445 2017-07-26 Andrew Pinski <apinski@cavium.com>
447 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
448 vec_unalign_load_cost and vec_unalign_store_cost.
450 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
452 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
453 -mvsx-small-integer option.
454 (ISA_3_0_MASKS_IEEE): Likewise.
455 (OTHER_VSX_VECTOR_MASKS): Likewise.
456 (POWERPC_MASKS): Likewise.
457 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
458 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
459 code, only testing for DImode being allowed in non-VSX floating
461 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
462 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
464 (rs6000_option_override_internal): Delete -mvsx-small-integer.
465 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
466 TARGET_P8_VECTOR test.
467 (rs6000_secondary_reload_simple_move): Likewise.
468 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
469 since TARGET_P9_VECTOR was already tested.
470 (rs6000_opt_masks): Remove -mvsx-small-integer.
471 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
472 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
474 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
475 test for TARGET_VEXTRACTUB was used, and that uses
477 (p9 extract splitter): Likewise.
478 (vsx_extract_<mode>_di_p9): Likewise.
479 (vsx_extract_<mode>_store_p9): Likewise.
480 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
481 for TARGET_P9_VECTOR was used. Delete code that is now dead with
482 the elimination of TARGET_VSX_SMALL_INTEGER.
483 (vsx_extract_<mode>_p8): Likewise.
484 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
485 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
486 (vsx_set_<mode>_p9): Likewise.
487 (vsx_set_v4sf_p9): Likewise.
488 (vsx_set_v4sf_p9_zero): Likewise.
489 (vsx_insert_extract_v4sf_p9): Likewise.
490 (vsx_insert_extract_v4sf_p9_2): Likewise.
491 * config/rs6000/rs6000.md (sign extend splitter): Change
492 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
493 (floatsi<mode>2_lfiwax_mem): Likewise.
494 (floatunssi<mode>2_lfiwzx_mem): Likewise.
495 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
496 since a test for TARGET_P9_VECTOR was used.
497 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
498 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
499 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
500 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
501 TARGET_P8_VECTOR test.
502 (fix_trunc<mode>si2_stfiwx): Likewise.
503 (fix_trunc<mode>si2_internal): Likewise.
504 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
505 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
507 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
508 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
509 TARGET_P8_VECTOR test.
510 (fixuns_trunc<mode>si2_stfiwx): Likewise.
511 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
512 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
514 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
515 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
516 since a test for TARGET_P9_VECTOR was used.
517 (splitter for loading small constants): Likewise.
519 2017-07-26 Andrew Pinski <apinski@cavium.com>
521 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
524 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
527 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
528 (fp_valid_at): Likewise.
530 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
532 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
533 (qdf24xx_addrcost_table): Likewise.
534 (cortexa57_tunings): Update to use generic_branch_cost.
535 (cortexa72_tunings): Likewise.
536 (cortexa73_tunings): Likewise.
537 (qdf24xx_tunings): Likewise.
539 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
541 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
542 (thunderx2t99_branch_cost): Likewise.
543 (cortexa35_tunings): Update to use generic_branch_cost.
544 (cortexa53_tunings): Likewise.
545 (cortexa57_tunings): Likewise.
546 (cortexa72_tunings): Likewise.
547 (cortexa73_tunings): Likewise.
548 (thunderx2t99_tunings): Likewise.
550 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
552 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
553 (sparc_option_override): Honour MASK_FSMULD.
554 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
555 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
556 * config/sparc/sparc.opt (mfsmuld): New option.
557 * doc/invoke.texi (mfsmuld): Document option.
559 2017-07-26 Marek Polacek <polacek@redhat.com>
562 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
564 2017-07-26 Richard Biener <rguenther@suse.de>
566 * gimple-match-head.c (do_valueize): Return OP if valueize
568 (get_def): New helper to get at the def stmt of a SSA name
570 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
571 do_valueize to get at the def stmt.
572 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
574 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
577 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
579 2017-07-26 Martin Liska <mliska@suse.cz>
582 * function.c (expand_function_start): Make expansion of
583 nonlocal_goto_save_area after parm_birth_insn.
585 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
587 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
588 from all CPU target flags enable members.
590 2017-07-26 Richard Biener <rguenther@suse.de>
592 * genmatch.c (dt_simplify::gen): Make iterator vars const.
593 (decision_tree::gen): Make 'type' const.
594 (write_predicate): Likewise.
596 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
598 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
599 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
600 (rs6000_option_override_internal): Likewise.
601 (rs6000_expand_vector_set): Likewise.
602 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
603 (TARGET_UPPER_REGS_SF): Likewise.
604 (TARGET_UPPER_REGS_DI): Likewise.
605 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
606 (TARGET_DIRECT_MOVE_64BIT): Likewise.
607 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
608 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
609 (Splitters for DI constants in Altivec registers): Likewise.
610 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
611 (vsx_set_v4sf_p9): Likewise.
612 (vsx_set_v4sf_p9_zero): Likewise.
613 (vsx_insert_extract_v4sf_p9): Likewise.
614 (vsx_insert_extract_v4sf_p9_2): Likewise.
616 2017-07-25 Carl Love <cel@us.ibm.com>
618 * doc/extend.texi: Update the built-in documentation file for the
619 existing built-in functions
620 vector signed char vec_cnttz (vector signed char);
621 vector unsigned char vec_cnttz (vector unsigned char);
622 vector signed short vec_cnttz (vector signed short);
623 vector unsigned short vec_cnttz (vector unsigned short);
624 vector signed int vec_cnttz (vector signed int);
625 vector unsigned int vec_cnttz (vector unsigned int);
626 vector signed long long vec_cnttz (vector signed long long);
627 vector unsigned long long vec_cnttz (vector unsigned long long);
629 2017-07-25 Andrew Pinski <apinski@cavium.com>
631 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
632 accesses where the use is for the first operand of a BIT_INSERT.
634 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
637 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
638 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
640 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
642 * config/i386/gstabs.h: Delete.
643 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
645 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
647 * config/i386/i386.c (ix86_decompose_address): Do not check for
648 register RTX when looking at index_reg or base_reg.
649 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
651 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
653 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
654 to update EH info here.
656 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
658 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
660 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
662 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
664 2017-07-25 Torsten Duwe <duwe@suse.de>
666 * common.opt: Introduce -fpatchable-function-entry
667 command line option, and its variables function_entry_patch_area_size
668 and function_entry_patch_area_start.
669 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
670 including a two-value parser.
671 * target.def (print_patchable_function_entry): New target hook.
672 * targhooks.h (default_print_patchable_function_entry): New function.
673 * targhooks.c (default_print_patchable_function_entry): Likewise.
674 * toplev.c (process_options): Switch off IPA-RA if
675 patchable function entries are being generated.
676 * varasm.c (assemble_start_function): Look at the
677 patchable-function-entry command line switch and current
678 function attributes and maybe generate NOP instructions by
679 calling the print_patchable_function_entry hook.
680 * doc/extend.texi: Document patchable_function_entry attribute.
681 * doc/invoke.texi: Document -fpatchable_function_entry
683 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
685 * doc/tm.texi: Re-generate.
687 2017-07-25 Jakub Jelinek <jakub@redhat.com>
690 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
691 TARGET_AVX512DQ rather than TARGET_AVX512BW.
693 2017-07-25 Tamar Christina <tamar.christina@arm.com>
695 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
697 2017-07-25 Richard Biener <rguenther@suse.de>
699 PR tree-optimization/81455
700 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
701 not walk in cycles when looking for guards.
703 2017-07-25 Richard Biener <rguenther@suse.de>
705 PR tree-optimization/81529
706 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
707 when optimizing backedge uses.
709 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
711 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
713 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
714 to dl_section_ref. On AIX, append an expression to subtract
715 the size of the section length to dl_section_ref.
717 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
719 * configure.ac: If any of the config.* scripts fail, exit 1.
720 * configure: Regenerate.
722 2017-07-25 Richard Biener <rguenther@suse.de>
725 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
726 of immediate uses, be more verbose on errors.
728 2017-07-25 Richard Biener <rguenther@suse.de>
730 PR tree-optimization/81510
731 * tree-vect-loop.c (vect_is_simple_reduction): When the
732 reduction stmt is not inside the loop bail out.
734 2017-07-25 Richard Biener <rguenther@suse.de>
736 PR tree-optimization/81303
737 * tree-vect-loop-manip.c (vect_loop_versioning): Build
738 profitability check against LOOP_VINFO_NITERSM1.
740 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
742 * domwalk.c (cmp_bb_postorder): Simplify.
743 (sort_bbs_postorder): New function. Use it...
744 (dom_walker::walk): ...here to optimize common cases.
746 2017-07-25 Martin Liska <mliska@suse.cz>
749 * ipa-visibility.c (function_and_variable_visibility): Make the
750 redirection just on target that supports aliasing.
751 Fix GNU coding style.
753 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
756 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
758 * config/arm/rtems.h: Likewise.
759 * config/bfin/rtems.h: Likewise.
760 * config/i386/rtemself.h: Likewise.
761 * config/lm32/rtems.h: Likewise.
762 * config/m32c/rtems.h: Likewise.
763 * config/m68k/rtemself.h: Likewise.
764 * config/microblaze/rtems.h: Likewise.
765 * config/mips/rtems.h: Likewise.
766 * config/moxie/rtems.h: Likewise.
767 * config/nios2/rtems.h: Likewise.
768 * config/powerpcspe/rtems.h: Likewise.
769 * config/rs6000/rtems.h: Likewise.
770 * config/rtems.h: Likewise.
771 * config/sh/rtems.h: Likewise.
772 * config/sh/rtemself.h: Likewise.
773 * config/sparc/rtemself.h: Likewise.
775 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
778 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
779 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
780 * tree-ssa-structalias.c (alias_get_name): Same.
782 2017-07-25 Bin Cheng <bin.cheng@arm.com>
785 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
786 instructions if no du chain is found.
788 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
790 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
792 2017-07-25 Richard Biener <rguenther@suse.de>
795 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
798 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
800 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
802 (ISA_2_7_MASKS_SERVER): Likewise.
803 (ISA_3_0_MASKS_IEEE): Likewise.
804 (OTHER_P8_VECTOR_MASKS): Likewise.
805 (OTHER_VSX_VECTOR_MASKS): Likewise.
806 (POWERPC_MASKS): Likewise.
807 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
808 duplicate list of options.
809 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
810 explicit -mupper-regs options.
811 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
812 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
813 alias for -mupper-regs-df.
814 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
815 (rs6000_init_hard_regno_mode_ok): Likewise.
816 (rs6000_option_override_internal): Likewise.
817 (rs6000_opt_masks): Likewise.
818 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
819 options in terms of whether -mvsx or -mpower8-vector was used.
820 (TARGET_UPPER_REGS_DI): Likewise.
821 (TARGET_UPPER_REGS_SF): Likewise.
822 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
823 -mupper-regs-* options.
825 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
827 * passes.c (emergency_dump_function): Print some empty lines and a
828 header before the RTL dump.
830 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
832 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
834 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
837 * config/aarch64/aarch64.c (aarch64_classify_symbol):
838 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
840 2017-07-24 Carl Love <cel@us.ibm.com>
842 * config/rs6000/rs6000-c.c: Add support for built-in functions
843 vector float vec_extract_fp32_from_shorth (vector unsigned short);
844 vector float vec_extract_fp32_from_shortl (vector unsigned short);
845 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
846 vec_extract_fp_from_shortl): Add defines for the two builtins.
847 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
848 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
850 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
851 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
852 * doc/extend.texi: Update the built-in documentation file for the
853 new built-in function.
855 2017-07-24 Jakub Jelinek <jakub@redhat.com>
858 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
860 * doc/generic.texi: Likewise.
861 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
862 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
864 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
866 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
867 (aarch64_mls_elt_merge<mode>): Likewise.
869 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
871 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
874 2017-07-23 Michael Collison <michael.collison@arm.com>
876 * config/arm/arm.c (arm_option_override): Deprecate
877 use of -mstructure-size-boundary.
878 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
879 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
881 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
884 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
885 Reduce cost estimate for direct moves.
887 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
890 * config/i386/i386.c (ix86_option_override_internal): Disable
891 BMI, BMI2 and TBM instructions for -m16.
893 2017-07-21 Carl Love <cel@us.ibm.com>
895 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
896 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
897 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
898 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
899 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
900 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
901 VMULOSW): New enum "unspec" values.
902 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
903 altivec_vmulosw): New patterns.
904 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
905 VMULOSW): Add definitions.
907 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
909 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
911 * config/aarch64/aarch64-options-extensions.def (rdma); New.
912 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
913 (AARCH64_FL_V8_1): Renumber.
914 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
915 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
916 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
917 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
918 rdma to feature modifiers list.
920 2017-07-21 Yury Gribov <tetra2005@gmail.com>
923 * ipa-visibility (function_and_variable_visibility): Convert
924 recursive PLT call to direct call if appropriate.
926 2017-07-21 Andrew Pinski <apinski@cavium.com>
928 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
929 operand 1 to see if the types precision matches.
930 * fold-const.c (operand_equal_p): Likewise.
932 2017-07-21 Richard Biener <rguenther@suse.de>
934 PR tree-optimization/81303
935 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
936 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
937 (vect_peeling_hash_get_lowest_cost): Adjust.
938 (vect_enhance_data_refs_alignment): Likewise. Use
939 vect_get_peeling_costs_all_drs to compute the penalty for no
940 peeling to match up costs.
942 2017-07-21 Richard Biener <rguenther@suse.de>
944 PR tree-optimization/81500
945 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
946 we didn't identify a reduction path.
948 2017-07-21 Tom de Vries <tom@codesourcery.com>
949 Cesar Philippidis <cesar@codesourcery.com>
951 PR gcov-profile/81442
952 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
955 2017-07-21 Tom de Vries <tom@codesourcery.com>
958 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
960 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
961 nvptx_override_options_after_change.
963 2017-07-21 Ulrich Drepper <drepper@redhat.com>
965 * dwarf2out.c (output_file_names): Avoid double testing for
968 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
970 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
972 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
974 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
976 (try_crossjump_to_edge): Do not punt on partitioned functions.
978 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
980 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
981 Put all BBs reachable only via paths crossing cold region to cold
983 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
985 2016-07-21 Richard Biener <rguenther@suse.de>
987 PR tree-optimization/81303
988 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
989 into account prologue and epilogue iterations when raising
990 min_profitable_iters to sth at least covering one vector iteration.
992 2017-07-21 Tamar Christina <tamar.christina@arm.com>
994 * config/arm/arm.c (arm_test_cpu_arch_dat):
997 2017-07-20 Nathan Sidwell <nathan@acm.org>
1000 * tree.h (TYPE_METHODS): Delete.
1001 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
1002 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
1003 (dbxout_type_methods): Scan TYPE_FIELDS.
1004 (dbxout_type): Don't check TYPE_METHODS here.
1005 * function.c (use_register_for_decl): Always ignore register for
1006 class types when not optimizing.
1007 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
1008 * tree.c (free_lang_data_in_type): Stitch out member functions and
1009 templates from TYPE_FIELDS.
1010 (build_distinct_type_copy, verify_type_variant,
1011 verify_type): Member fns are on TYPE_FIELDS.
1012 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
1013 * tree-pretty-print.c (dump_generic_node): Likewise.
1015 2017-07-20 Jakub Jelinek <jakub@redhat.com>
1018 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
1019 V2TImode and V4TImode.
1020 (ix86_expand_vector_extract): Likewise.
1021 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
1022 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
1023 (ssescalarmode): Handle V4TImode and V2TImode.
1024 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
1025 (*vec_extractv2ti, *vec_extractv4ti): New insns.
1026 (VEXTRACTI128_MODE): New mode iterator.
1027 (splitter for *vec_extractv?ti first element): New.
1028 (VEC_INIT_MODE): New mode iterator.
1029 (vec_init<mode>): Consolidate 3 expanders into one using
1030 VEC_INIT_MODE mode iterator.
1032 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
1034 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
1035 non_spilled_static_chain_regno_p.
1037 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
1039 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
1041 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
1043 * bb-reorder.c (connect_traces): Allow copying of blocks within
1046 2017-07-20 Richard Biener <rguenther@suse.de>
1048 * gimple.h (gimple_phi_result): Add gphi * overload.
1049 (gimple_phi_result_ptr): Likewise.
1050 (gimple_phi_arg): Likewise. Adjust index assert to only
1051 allow actual argument accesses rather than all slots available
1053 (gimple_phi_arg_def): Add gphi * overload.
1054 * tree-phinodes.c (make_phi_node): Initialize only actual
1056 (resize_phi_node): Clear memory not covered by old node,
1057 do not initialize excess argument slots.
1058 (reserve_phi_args_for_new_edge): Initialize new argument slot
1061 2017-07-20 Bin Cheng <bin.cheng@arm.com>
1063 PR tree-optimization/81388
1065 2016-07-21 Bin Cheng <bin.cheng@arm.com>
1067 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
1068 by removing computation of may_be_zero.
1070 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1071 Tom de Vries <tom@codesourcery.com>
1074 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
1075 when gimple level profile disagrees with what RTL expander did.
1077 2017-07-20 Richard Biener <rguenther@suse.de>
1079 PR tree-optimization/61171
1080 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
1081 (vect_analyze_stmt): Add slp instance parameter.
1082 (vectorizable_reduction): Likewise.
1083 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
1084 (vect_is_simple_reduction): Deal with chains not detected
1085 as SLP reduction chain, specifically not properly associated
1086 chains containing a mix of plus/minus.
1087 (get_reduction_op): Remove.
1088 (get_initial_defs_for_reduction): Simplify, pass in whether
1089 this is a reduction chain, pass in the SLP node for the PHIs.
1090 (vect_create_epilog_for_reduction): Get the SLP instance as
1092 (vectorizable_reduction): Get the SLP instance as arg.
1093 During analysis remember the SLP node with the PHIs in the
1094 instance. Simplify getting at the vectorized reduction PHIs.
1095 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
1096 through SLP instance.
1097 (vect_slp_analyze_operations): Likewise.
1098 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
1099 (vect_transform_stmt): Likewise.
1101 2017-07-20 Tom de Vries <tom@codesourcery.com>
1103 PR tree-optimization/81489
1104 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
1105 read of phi arg location to before loop that modifies phi.
1107 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
1109 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
1112 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1115 * except.c (execute): Fix ordering issue.
1117 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
1119 PR rtl-optimization/81423
1120 * combine.c (make_compound_operation_int): Don't try to optimize
1121 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
1123 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
1125 PR rtl-optimization/81423
1126 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
1127 with a constant that is -1 in the truncated to mode.
1129 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1131 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
1132 (determine_unlikely_bbs): ... here.
1133 * predict.h (propagate_unlikely_bbs_forward): Declare.
1134 * cfgexpand.c (pass_expand::execute): Use it.
1135 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
1137 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
1138 propagate_unlikely_bbs_forward.
1140 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1143 * except.c (maybe_add_nop_after_section_switch): New function.
1146 2017-07-19 Tom de Vries <tom@codesourcery.com>
1148 * gimple.h (gimple_phi_set_arg): Make assert more strict.
1150 2017-07-19 Tom de Vries <tom@codesourcery.com>
1152 * gimple.h (gimple_phi_arg): Make assert more strict.
1154 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
1156 * config.gcc (powerpc*-*-*): Add mmintrin.h.
1157 * config/rs6000/mmintrin.h: New file.
1158 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
1160 2017-07-19 Jakub Jelinek <jakub@redhat.com>
1162 PR tree-optimization/81346
1163 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
1165 2017-07-19 Tom de Vries <tom@codesourcery.com>
1167 * config/nvptx/nvptx.md (VECIM): Add V2DI.
1169 2017-07-19 Tom de Vries <tom@codesourcery.com>
1171 * config/nvptx/nvptx-modes.def: Add V2DImode.
1172 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
1173 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
1174 (nvptx_output_mov_insn): Handle lack of mov.b128.
1175 (nvptx_print_operand): Handle 'H' and 'L' codes.
1176 (nvptx_vector_mode_supported): Allow V2DImode.
1177 (nvptx_preferred_simd_mode): New function.
1178 (nvptx_data_alignment): New function.
1179 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
1180 nvptx_preferred_simd_mode.
1181 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
1183 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
1185 2017-07-19 Tom de Vries <tom@codesourcery.com>
1187 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
1188 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
1189 (nvptx_vector_mode_supported): New function. Allow V2SImode.
1190 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
1191 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
1192 (mov<VECIM>_insn): New define_insn.
1193 (define_expand "mov<VECIM>): New define_expand.
1195 2017-07-19 Tom de Vries <tom@codesourcery.com>
1197 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
1199 2017-07-19 Jakub Jelinek <jakub@redhat.com>
1201 PR tree-optimization/81346
1202 * fold-const.h (fold_div_compare, range_check_type): Declare.
1203 * fold-const.c (range_check_type): New function.
1204 (build_range_check): Use range_check_type.
1205 (fold_div_compare): No longer static, rewritten into
1206 a match.pd helper function.
1207 (fold_comparison): Don't call fold_div_compare here.
1208 * match.pd (X / C1 op C2): New optimization using fold_div_compare
1211 2017-07-19 Nathan Sidwell <nathan@acm.org>
1213 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
1214 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
1215 * tree.c (find_decls_types_r, verify_type): Use
1216 TYPE_{MIN,MAX}_VALUE_RAW.
1217 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
1218 (hash_tree): Likewise.
1219 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
1221 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
1224 2017-07-18 Tom de Vries <tom@codesourcery.com>
1227 * omp-expand.c (expand_omp_for_static_chunk): Handle
1228 equal-argument loop exit phi.
1230 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
1233 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
1234 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
1235 operand 2 predicate.
1236 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
1237 operand 2 predicate.
1238 (ror,rol -> rorx splitters): Use const_int_operand as
1239 operand 2 predicate.
1241 2017-06-18 Richard Biener <rguenther@suse.de>
1243 PR tree-optimization/81410
1244 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
1245 the gap in the ! slp_perm SLP case after each group.
1247 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1250 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
1253 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1256 * predict.c (set_even_probabilities): Cleanup; do not affect
1257 probabilities that are already known.
1258 (combine_predictions_for_bb): Call even when count is set.
1260 2017-07-18 Nathan Sidwell <nathan@acm.org>
1262 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
1265 2017-07-18 Bin Cheng <bin.cheng@arm.com>
1268 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
1269 optimization for loop niter analysis.
1271 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
1274 * config/avr/avr.c (avr_optimize_casesi): Don't use
1275 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
1277 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1279 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
1280 body_cost_vec from _vect_peel_extended_info.
1281 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
1282 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
1285 2017-07-18 Bin Cheng <bin.cheng@arm.com>
1287 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
1289 2017-07-18 Richard Biener <rguenther@suse.de>
1291 PR tree-optimization/80620
1292 PR tree-optimization/81403
1293 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
1294 info when re-using a VN table entry.
1296 2017-07-18 Richard Biener <rguenther@suse.de>
1298 PR tree-optimization/81418
1299 * tree-vect-loop.c (vectorizable_reduction): Properly compute
1300 vectype_in. Verify that with lane-reducing reduction operations
1301 we have a single def-use cycle.
1303 2017-07-17 Carl Love <cel@us.ibm.com>
1305 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
1307 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1308 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
1309 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
1310 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
1311 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
1312 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
1313 VMULOSW): New enum "unspec" values.
1314 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
1315 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
1316 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
1317 altivec_vmulosw): New patterns.
1318 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
1319 VMULOSW): Add definitions.
1320 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
1322 * config/alpha/alpha.c: Include predict.h.
1324 2017-07-17 Yury Gribov <tetra2005@gmail.com>
1326 * tree-vrp.c (compare_assert_loc): Fix comparison function
1327 to return predictable results.
1329 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1331 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
1334 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
1335 * doc/invoke.texi (mexpand-adddi): Update text.
1337 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1339 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
1340 that also clobbers the CC register. The old expand code is moved
1342 (*arc_clzsi2): ... here.
1343 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
1344 the CC register. The old expand code is moved to ...
1345 (arc_ctzsi2): ... here.
1347 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1349 * config/arc/arc.opt (mindexed-loads): Use initial value
1350 TARGET_INDEXED_LOADS_DEFAULT.
1351 (mauto-modify-reg): Use initial value
1352 TARGET_AUTO_MODIFY_REG_DEFAULT.
1353 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
1354 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
1355 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
1356 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
1358 2017-07-17 Martin Liska <mliska@suse.cz>
1361 * opts.c (finish_options): Do not allow -fgnu-tm
1362 w/ -fsanitize={kernel-,}address. Say sorry.
1364 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1367 * tree-loop-distribution.c (classify_partition): Only assert on
1368 numer of iterations.
1369 (merge_dep_scc_partitions): Delete prameter. Update function call.
1370 (distribute_loop): Remove code handling loop with unknown niters.
1371 (pass_loop_distribution::execute): Skip loop with unknown niters.
1373 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1376 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
1377 function sort_partitions_by_post_order.
1379 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1381 PR tree-optimization/81374
1382 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
1383 the max index of basic blocks, rather than number of basic blocks.
1385 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1387 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
1389 (arc_legitimate_pic_operand_p): Likewise.
1390 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
1392 (arc_needs_pcl_p): Likewise.
1393 (arc_legitimate_pc_offset_p): Likewise.
1394 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
1395 function is also used in constrains.md.
1396 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
1397 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
1398 PLUS. Only return true/false in known cases, otherwise assert.
1399 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
1400 is already called in arc_legitimate_constant_p.
1401 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
1403 (LEGITIMATE_PIC_OPERAND_P): Use
1404 arc_raw_symbolic_reference_mentioned_p function.
1405 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
1410 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1411 Andrew Burgess <andrew.burgess@embecosm.com>
1413 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
1414 (arc_return_address_register): New function.
1415 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
1416 (arc_handle_fndecl_attribute): Add naked attribute.
1417 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
1418 (TARGET_WARN_FUNC_RETURN): Likewise.
1419 (arc_allocate_stack_slots_for_args): New function.
1420 (arc_warn_func_return): Likewise.
1421 (machine_function): Change type fn_type.
1422 (arc_compute_function_type): Consider new naked function type,
1423 change function return type.
1424 (arc_must_save_register): Adapt to handle new
1425 arc_compute_function_type's return type.
1426 (arc_expand_prologue): Likewise.
1427 (arc_expand_epilogue): Likewise.
1428 (arc_return_address_regs): Delete.
1429 (arc_return_address_register): New function.
1430 (arc_epilogue_uses): Use above function.
1431 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
1432 (arc_function_type): Change encoding, add naked type.
1433 (ARC_INTERRUPT_P): Change to handle the new encoding.
1434 (ARC_FAST_INTERRUPT_P): Likewise.
1435 (ARC_NORMAL_P): Define.
1436 (ARC_NAKED_P): Likewise.
1437 (arc_compute_function_type): Delete prototype.
1438 * config/arc/arc.md (in_ret_delay_slot): Use
1439 arc_return_address_register function.
1440 (simple_return): Likewise.
1441 (p_return_i): Likewise.
1443 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1445 PR tree-optimization/81428
1446 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
1447 can't be built for those types.
1449 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
1451 Remove stuff dead since r239246.
1453 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
1454 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
1455 (avr_inform_devices): Remove dead stuff.
1457 2017-07-17 Tamar Christina <tamar.christina@arm.com>
1459 * config/arm/arm_neon.h: Fix softp typo.
1461 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1463 PR tree-optimization/81365
1464 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
1465 aggregate moves onto bb predecessor edges, make sure there are no
1466 loads that could alias the lhs in between the start of bb and the
1469 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
1472 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
1473 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
1474 [LSHIFTRT, outer_code = TRUNCATE]: Same.
1476 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1478 PR tree-optimization/81396
1479 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
1480 (init_symbolic_number): Initialize it to 1.
1481 (perform_symbolic_merge): Add n_ops from both operands into the new
1483 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
1484 without base_addr as useless if they need more than one operation.
1485 (bswap_replace): Handle !bswap case for NULL base_addr.
1487 2017-07-17 Tom de Vries <tom@codesourcery.com>
1490 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
1493 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
1495 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
1496 conditional builtin define __FIX_LEON3FT_B2BST.
1498 2017-07-17 Daniel Cederman <cederman@gaisler.com>
1500 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
1501 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
1504 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1506 PR rtl-optimization/81424
1507 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
1508 to remove potential trapping from operands if -fnon-call-exceptions.
1510 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1512 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
1513 profile_proability for scalling.
1514 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
1516 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1518 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
1520 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1522 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
1523 fixpoint arithmetics.
1525 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1527 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
1528 fixpoint arithmetics.
1530 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1532 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
1533 fixpoint arithmetics.
1535 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1537 * profile-count.h (profile_probability::from_reg_br_prob_note,
1538 profile_probability::to_reg_br_prob_note): New functions.
1539 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
1540 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
1541 * predict.c (probability_reliable_p): Update.
1542 (edge_probability_reliable_p): Update.
1543 (br_prob_note_reliable_p): Update.
1544 (invert_br_probabilities): Update.
1545 (add_reg_br_prob_note): New function.
1546 (combine_predictions_for_insn): Update.
1547 * asan.c (asan_clear_shadow): Update.
1548 * cfgbuild.c (compute_outgoing_frequencies): Update.
1549 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
1550 (update_br_prob_note): Update.
1551 (rtl_verify_edges): Update.
1552 (purge_dead_edges): Update.
1553 (fixup_reorder_chain): Update.
1554 * emit-rtl.c (try_split): Update.
1555 * ifcvt.c (cond_exec_process_insns): Update.
1556 (cond_exec_process_if_block): Update.
1557 (dead_or_predicable): Update.
1558 * internal-fn.c (expand_addsub_overflow): Update.
1559 (expand_neg_overflow): Update.
1560 (expand_mul_overflow): Update.
1561 * loop-doloop.c (doloop_modify): Update.
1562 * loop-unroll.c (compare_and_jump_seq): Update.
1563 * optabs.c (emit_cmp_and_jump_insn_1): Update.
1564 * predict.h: Update.
1565 * reorg.c (mostly_true_jump): Update.
1567 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
1568 * config/alpha/alpha.c (emit_unlikely_jump): Update.
1569 * config/arc/arc.c: (emit_unlikely_jump): Update.
1570 * config/arm/arm.c: (emit_unlikely_jump): Update.
1571 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
1572 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
1573 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
1574 (ix86_print_operand): Update.
1575 (ix86_split_fp_branch): Update.
1576 (predict_jump): Update.
1577 * config/ia64/ia64.c (ia64_print_operand): Update.
1578 * config/mmix/mmix.c (mmix_print_operand): Update.
1579 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
1580 (rs6000_expand_split_stack_prologue): Update.
1581 * config/rs6000/rs6000.c: Update.
1582 * config/s390/s390.c (s390_expand_vec_strlen): Update.
1583 (s390_expand_vec_movstr): Update.
1584 (s390_expand_cs_tdsi): Update.
1585 (s390_expand_split_stack_prologue): Update.
1586 * config/sh/sh.c (sh_print_operand): Update.
1587 (expand_cbranchsi4): Update.
1588 (expand_cbranchdi4): Update.
1589 * config/sparc/sparc.c (output_v9branch): Update.
1590 * config/spu/spu.c (get_branch_target): Update.
1591 (ea_load_store_inline): Update.
1592 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
1593 * config/tilepro/tilepro.c: Update.
1595 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1597 * gimplify.c (mostly_copy_tree_r): Revert latest change.
1598 (gimplify_save_expr): Likewise.
1600 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1602 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
1604 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1606 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
1608 * timevar.def (TV_IPA_FNSUMMARY): Define.
1610 2017-07-16 Daniel Cederman <cederman@gaisler.com>
1612 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
1613 to back store errata sensitive sequence from being generated.
1614 (sqrtdf2_fix): Likewise.
1616 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1618 * tree-ssa-threadupdate.c (compute_path_counts,
1619 update_joiner_offpath_counts): Use profile_probability.
1621 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
1624 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1626 * config/arm/arm-c.c (arm_cpu_builtins): Define
1627 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1629 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
1631 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1632 array entries to represent __ieee128 versions of the
1633 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
1634 scalar_extract_sig, and scalar_insert_exp built-in functions.
1635 (altivec_resolve_overloaded_builtin): Add special case handling
1636 for the __builtin_scalar_insert_exp function, as represented by
1637 the P9V_BUILTIN_VEC_VSIEDP constant.
1638 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
1639 exponent support for __ieee128 argument.
1640 (VSESQP): Add scalar extract signature support for __ieee128
1642 (VSTDCNQP): Add scalar test negative support for __ieee128
1644 (VSIEQP): Add scalar insert exponent support for __int128 argument
1645 with __ieee128 result.
1646 (VSIEQPF): Add scalar insert exponent support for __ieee128
1647 argument with __ieee128 result.
1648 (VSTDCQP): Add scalar test data class support for __ieee128
1650 (VSTDCNQP): Add overload support for scalar test negative with
1652 (VSTDCQP): Add overload support for scalar test data class
1654 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
1656 (UNSPEC_VSX_SIEXPQP): New constant.
1657 (xsxexpqp): New insn for VSX scalar extract exponent quad
1659 (xsxsigqp): New insn for VSX scalar extract significand quad
1661 (xsiexpqpf): New insn for VSX scalar insert exponent quad
1662 precision with floating point argument.
1663 (xststdcqp): New expand for VSX scalar test data class quad
1665 (xststdcnegqp): New expand for VSX scalar test negative quad
1667 (xststdcqp): New insn to match expansions for VSX scalar test data
1668 class quad precision and VSX scalar test negative quad precision.
1669 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
1670 special case operand checking to enforce that second operand of
1671 VSX scalar test data class with quad precision argument is a 7-bit
1673 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
1674 prototypes and descriptions of __ieee128 versions of
1675 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
1676 scalar_test_data_class, and scalar_test_neg built-in functions.
1678 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1680 PR tree-optimization/81162
1681 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
1682 replace a negate with an add.
1684 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
1686 * doc/invoke.texi (arm/-mcpu): Document +crypto.
1688 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1690 * config/arm/arm-c.c (arm_cpu_builtins): Define
1691 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1693 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1695 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
1696 (armv8-r): Set ARM Cortex-R52 as default CPU.
1697 * config/arm/arm-tables.opt: Regenerate.
1698 * config/arm/arm-tune.md: Regenerate.
1699 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
1701 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
1702 extension for -mcpu=cortex-r52.
1704 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1706 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
1707 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
1708 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
1709 (fp-armv8): Define it as FP_ARMv8 only.
1710 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
1711 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
1713 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
1714 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
1715 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
1716 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
1717 than TARGET_FPU_ARMV8.
1718 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
1719 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
1720 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
1722 * config/arm/neon.md (neon_vrint): Likewise.
1723 (neon_vcvt): Likewise.
1724 (neon_<fmaxmin_op><mode>): Likewise.
1725 (<fmaxmin><mode>3): Likewise.
1726 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
1727 * config/arm/predicates.md (arm_cond_move_operator): Check against
1728 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
1730 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
1732 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
1735 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1737 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
1738 loop in comment with memset.
1740 2017-07-14 Martin Liska <mliska@suse.cz>
1742 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
1743 * dwarf2out.c (is_java): Remove the function.
1744 (output_pubname): Remove usage of the function.
1745 (lower_bound_default): Remove usage of DW_LANG_Java.
1746 (gen_compile_unit_die): Likewise.
1747 * gcc.c: Remove compiler defaults for .java and .zip files.
1748 * gimple-expr.c (remove_suffix): Change as there's no longer
1749 extension than 4-letter one.
1750 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
1751 (gimplify_save_expr): Likewise.
1752 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
1753 as it's possible even for other languages than Java.
1754 * langhooks.h (struct lang_hooks): Remove Java from a comment.
1755 * lto-opts.c (lto_write_options): Remove reference to Java.
1756 * opts.c (strip_off_ending): Update file extension handling.
1757 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
1758 * tree-eh.c (lower_resx): Likewise.
1759 * tree.c (free_lang_data_in_type): Remove dead code.
1760 (find_decls_types_r): Likewise.
1761 (build_common_builtin_nodes): Remove Java from a comment.
1762 (verify_type): Remove dead code.
1763 * varasm.c (assemble_external): Remove Java from a comment.
1765 2017-07-14 Martin Liska <mliska@suse.cz>
1767 * opts.c (finish_options): Add quotes.
1768 (common_handle_option): Likewise.
1770 2017-07-14 Martin Liska <mliska@suse.cz>
1772 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
1773 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
1775 * dwarf2out.c (lower_bound_default): Do not handle
1777 (gen_compile_unit_die): Likewise.
1778 * gcc.c: Remove default extension binding for GNU Pascal.
1779 * stmt.c: Remove Pascal language from a comment.
1780 * xcoffout.c: Likewise.
1782 2017-07-13 David Malcolm <dmalcolm@redhat.com>
1785 * diagnostic-show-locus.c (fixit_cmp): New function.
1786 (layout::layout): Sort m_fixit_hints.
1787 (column_range::column_range): Assert that the values are valid.
1788 (struct char_span): New struct.
1789 (correction::overwrite): New method.
1790 (struct source_line): New struct.
1791 (line_corrections::add_hint): Add assertions. Reimplement memcpy
1792 calls in terms of classes source_line and char_span, and
1793 correction::overwrite.
1794 (selftest::test_overlapped_fixit_printing_2): New function.
1795 (selftest::diagnostic_show_locus_c_tests): Call it.
1797 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
1799 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
1800 early if there is no lhs.
1802 2017-07-13 Martin Liska <mliska@suse.cz>
1804 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
1805 (gen_reference_type_die): Likewise.
1806 * stor-layout.c: Remove Pascal-related comment.
1808 2017-07-13 Martin Liska <mliska@suse.cz>
1810 * opts.c (finish_options): Add quotes to error messages.
1811 (parse_sanitizer_options): Likewise.
1813 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
1815 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
1817 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
1819 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
1821 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
1823 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
1825 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
1827 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1830 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
1831 provides the hardware capability bits, define the macro
1832 __BUILTIN_CPU_SUPPORTS__.
1833 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
1834 if GLIBC does not provide the hardware capability bits. Add a
1835 gcc_unreachable call if the built-in cpu function is neither
1836 __builtin_cpu_is nor __builtin_cpu_supports.
1837 (rs6000_get_function_versions_dispatcher): Change the warning
1838 that an old GLIBC is used which does not export the capability
1839 bits to be an error.
1840 * doc/extend.texi (target_clones attribute): Document the
1841 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
1842 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
1843 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
1844 the macros defined by GCC if the newer GLIBC is available.
1846 2017-07-12 Jeff Law <law@redhat.com>
1848 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
1849 remaining includes slightly.
1850 * config/riscv/riscv-builtins.c: Include profile-count.h.
1852 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1855 * config/avr/avr.c (avr_set_current_function): In diagnostic
1856 messages: Quote keywords and (parts of) identifiers.
1857 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
1860 2017-07-12 Carl Love <cel@us.ibm.com>
1862 * config/rs6000/rs6000-c.c: Add support for built-in functions
1863 vector bool char vec_revb (vector bool char);
1864 vector bool short vec_revb (vector short char);
1865 vector bool int vec_revb (vector bool int);
1866 vector bool long long vec_revb (vector bool long long);
1867 * doc/extend.texi: Update the built-in documentation file for the
1868 new built-in functions.
1870 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1872 * config/s390/s390.md: Remove movcc splitter.
1874 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1876 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
1877 load/store on condition.
1879 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1882 * config/avr/avr.c (avr_encode_section_info)
1883 [progmem && !TREE_READONLY]: Error if progmem object needs
1886 2017-07-11 Michael Collison <michael.collison@arm.com>
1888 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
1891 2017-07-11 Carl Love <cel@us.ibm.com>
1893 * config/rs6000/rs6000-c.c: Add support for builtins
1894 vector unsigned int vec_parity_lsbb (vector signed int);
1895 vector unsigned int vec_parity_lsbb (vector unsigned int);
1896 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
1897 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
1898 vector unsigned long long vec_parity_lsbb (vector signed long long);
1899 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
1900 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
1901 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
1902 * doc/extend.texi: Update the built-in documentation file for the
1903 new built-in functions.
1905 2017-07-11 David Malcolm <dmalcolm@redhat.com>
1907 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
1908 (layout::m_primary_loc): New field.
1909 (layout::layout): Initialize new field. Move location filtering
1910 logic from here to...
1911 (layout::maybe_add_location_range): ...this new method. Add
1912 support for filtering to just the lines already specified by other
1914 (layout::will_show_line_p): New method.
1915 (gcc_rich_location::add_location_if_nearby): New method.
1916 (selftest::test_add_location_if_nearby): New test function.
1917 (selftest::diagnostic_show_locus_c_tests): Call it.
1918 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
1921 2017-07-11 Tom de Vries <tom@codesourcery.com>
1923 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
1924 (bb_first_real_insn): New function.
1925 (nvptx_single): Add extra initialization of broadcasted condition
1928 2017-07-11 Nathan Sidwell <nathan@acm.org>
1930 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
1932 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
1934 * doc/extend.texi (AVR Function Attributes): Remove weblink to
1935 Binutils doc as TEXI will mess them up.
1936 * doc/invoke.texi (AVR Options): Same here.
1938 2017-07-11 Daniel Cederman <cederman@gaisler.com>
1940 * config/sparc/sparc.opt (mfix-ut700): New option.
1941 (mfix-gr712rc): Likewise.
1942 (sparc_fix_b2bst): New variable.
1943 * doc/invoke.texi (SPARC options): Document them.
1944 (ARM options): Fix warnings.
1945 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
1946 instructions to prevent sequences that can trigger the store-store
1947 errata for certain LEON3FT processors.
1948 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
1949 (sparc_option_override): Set sparc_fix_b2bst appropriately.
1950 * config/sparc/sparc.md (fix_b2bst): New attribute.
1951 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
1953 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
1956 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
1958 (*rsqrtsf2_sse): Ditto.
1960 (div<mode>3): Macroize insn from divdf3 and divsf3
1961 using MODEF mode iterator.
1963 2017-07-10 Martin Sebor <msebor@redhat.com>
1965 PR tree-optimization/80397
1966 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
1967 instead of testing for equality to INTEGER_TYPE.
1969 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
1971 * config.gcc: Remove uclibc from arc target spec.
1973 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
1975 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
1977 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1980 * lto-wrapper.c (remove_option): New function.
1981 (merge_and_complain): Merge PIC/PIE options more realistically.
1983 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
1985 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
1989 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
1990 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
1991 * config.in: Regenerate.
1992 * configure: Regenerate.
1993 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
1994 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
1995 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
1996 (TARGET_GASISR_PROLOGUES): ...target mask.
1997 * common/config/avr/avr-common.c
1998 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
1999 Set -mgas-isr-prologues.
2000 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
2001 INSERT_PASS_BEFORE for it.
2002 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
2003 * config/avr/avr.c (avr_option_override)
2004 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
2005 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
2006 (avr_attribute_table) <no_gccisr>: Add new function attribute.
2007 (avr_set_current_function) <is_no_gccisr>: Init machine field.
2008 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
2010 (make_avr_pass_pre_proep): New function.
2011 (emit_push_sfr) <treg>: Add argument to function and use it
2013 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
2014 and set machine->gasisr.yes.
2015 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
2016 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
2017 __gcc_isr.n_pushed to .L__stack_usage.
2018 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
2019 (avr_asm_final_postscan_insn): ...this new static function.
2020 * config/avr/avr.h (machine_function)
2021 <is_no_gccisr, use_L__stack_usage>: New fields.
2022 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
2023 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
2024 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
2025 (gasisr, *gasisr): New expander and insn.
2026 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
2027 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
2028 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
2030 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
2032 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
2035 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
2037 Move jump-tables out of .text again.
2040 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
2041 (ASM_OUTPUT_ADDR_VEC): New function.
2042 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
2043 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
2044 INSN_ADDRESSes as asm comment.
2045 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
2046 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
2047 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
2048 * config/avr/avr.md (*tablejump): Adjust comment.
2049 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
2050 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
2052 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
2053 (avr_output_addr_vec): New proto.
2054 (avr_log_t) <insn_addresses>: New field.
2056 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
2059 * config/i386/i386.c (ix86_function_arg_advance): Set
2060 outgoing_args_on_stack to true if there are outgoing arguments
2062 (ix86_function_arg): Likewise.
2063 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
2064 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
2065 * config/i386/i386.h (machine_function): Add
2066 outgoing_args_on_stack.
2068 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
2070 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
2072 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
2074 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
2076 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
2077 (REAL_H): Remove $(MACHMODE_H).
2078 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
2080 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
2081 $(MACHMODE_H) and double-int.h.
2082 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
2084 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
2087 2017-07-07 Andrew Pinski <apinski@cavium.com>
2089 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
2090 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
2092 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
2094 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
2095 Add warning if GCC was not configured to link against a GLIBC that
2096 exports the hardware capability bits.
2097 (make_resolver_func): Make resolver function private and not a
2098 COMDAT function. Create the name with clone_function_name instead
2099 of make_unique_name.
2102 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
2103 correct operand in doing the split.
2105 2017-07-07 Carl Love <cel@us.ibm.com>
2107 * config/rs6000/rs6000-c: Add support for built-in function
2108 vector unsigned short vec_pack_to_short_fp32 (vector float,
2110 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
2111 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
2112 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
2113 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
2114 (convert_4f32_8i16): Add define_expand.
2115 * doc/extend.texi: Update the built-in documentation file for the
2116 new built-in function.
2118 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2120 * config/sparc/m8.md: New file.
2121 * config/sparc/sparc.md: Include m8.md.
2123 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2125 * config/sparc/sparc.opt: New option -mvis4b.
2126 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
2127 (sparc_option_override): Handle VIS4B.
2128 (enum sparc_builtins): Define
2129 SPARC_BUILTIN_DICTUNPACK{8,16,32},
2130 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
2131 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
2132 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
2133 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
2134 (check_constant_argument): New function.
2135 (sparc_vis_init_builtins): Define builtins
2136 __builtin_vis_dictunpack{8,16,32},
2137 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
2138 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
2139 __builtin_vis_fpcmpde{8,16,32}shl and
2140 __builtin_vis_fpcmpur{8,16,32}shl.
2141 (sparc_expand_builtin): Check that the constant operands to
2142 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
2143 constant and in range.
2144 * config/sparc/sparc-c.c (sparc_target_macros): Handle
2146 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
2147 (SPARC_IMM5_P): Likewise.
2148 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
2149 (enabled): Handle vis4b.
2150 (UNSPEC_DICTUNPACK): New unspec.
2151 (UNSPEC_FPCMPSHL): Likewise.
2152 (UNSPEC_FPUCMPSHL): Likewise.
2153 (UNSPEC_FPCMPDESHL): Likewise.
2154 (UNSPEC_FPCMPURSHL): Likewise.
2155 (cpu_feature): New CPU feature `vis4b'.
2156 (dictunpack{8,16,32}): New insns.
2157 (FPCSMODE): New mode iterator.
2158 (fpcscond): New code iterator.
2159 (fpcsucond): Likewise.
2160 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
2161 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
2162 (fpcmpde{8,16,32}{si,di}shl): Likewise.
2163 (fpcmpur{8,16,32}{si,di}shl): Likewise.
2164 * config/sparc/constraints.md: Define constraints `q' for unsigned
2165 2-bit integer constants and `t' for unsigned 5-bit integer
2167 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
2169 (imm5_operand_dictunpack16): Likewise.
2170 (imm5_operand_dictunpack32): Likewise.
2171 (imm2_operand): Likewise.
2172 * doc/invoke.texi (SPARC Options): Document -mvis4b.
2173 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
2174 ditunpack* and fpcmp*shl builtins.
2176 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2178 * config.gcc: Handle m8 in --with-{cpu,tune} options.
2179 * config.in: Add HAVE_AS_SPARC6 define.
2180 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
2182 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
2184 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
2185 (CPP_CPU_SPEC): Handle m8.
2186 (ASM_CPU_SPEC): Likewise.
2187 * config/sparc/sparc-opts.h (enum processor_type): Add
2189 * config/sparc/sparc.c (m8_costs): New struct.
2190 (sparc_option_override): Handle TARGET_CPU_m8.
2191 (sparc32_initialize_trampoline): Likewise.
2192 (sparc64_initialize_trampoline): Likewise.
2193 (sparc_issue_rate): Likewise.
2194 (sparc_register_move_cost): Likewise.
2195 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
2196 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
2197 (ASM_CPU64_DEFAULT_SPEC): Likewise.
2198 (CPP_CPU_SPEC): Handle M8.
2199 (ASM_CPU_SPEC): Likewise.
2200 (AS_M8_FLAG): Define.
2201 * config/sparc/sparc.md: Add m8 to the cpu attribute.
2202 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
2203 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
2205 * configure: Regenerate.
2206 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
2209 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2211 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
2213 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
2214 ("*movdi_insn_sp32"): Do not set v3pipe.
2215 ("*movsi_insn"): Likewise.
2216 ("*movdi_insn_sp64"): Likewise.
2217 ("*movsf_insn"): Likewise.
2218 ("*movdf_insn_sp32"): Likewise.
2219 ("*movdf_insn_sp64"): Likewise.
2220 ("*zero_extendsidi2_insn_sp64"): Likewise.
2221 ("*sign_extendsidi2_insn"): Likewise.
2222 ("*mov<VM32:mode>_insn"): Likewise.
2223 ("*mov<VM64:mode>_insn_sp64"): Likewise.
2224 ("*mov<VM64:mode>_insn_sp32"): Likewise.
2225 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
2226 ("<vlop:code><VL:mode>3"): Likewise.
2227 ("*not_<vlop:code><VL:mode>3"): Likewise.
2228 ("*nand<VL:mode>_vis"): Likewise.
2229 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
2230 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
2231 ("one_cmpl<VL:mode>2"): Likewise.
2232 ("faligndata<VM64:mode>_vis"): Likewise.
2233 ("alignaddrsi_vis"): Likewise.
2234 ("alignaddrdi_vis"): Likweise.
2235 ("alignaddrlsi_vis"): Likewise.
2236 ("alignaddrldi_vis"): Likewise.
2237 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
2238 ("bmaskdi_vis"): Likewise.
2239 ("bmasksi_vis"): Likewise.
2240 ("bshuffle<VM64:mode>_vis"): Likewise.
2241 ("cmask8<P:mode>_vis"): Likewise.
2242 ("cmask16<P:mode>_vis"): Likewise.
2243 ("cmask32<P:mode>_vis"): Likewise.
2244 ("pdistn<P:mode>_vis"): Likewise.
2245 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
2247 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2249 * config/sparc/sparc.md ("subtype"): New insn attribute.
2250 ("*wrgsr_sp64"): Set insn subtype.
2251 ("*rdgsr_sp64"): Likewise.
2252 ("alignaddrsi_vis"): Likewise.
2253 ("alignaddrdi_vis"): Likewise.
2254 ("alignaddrlsi_vis"): Likewise.
2255 ("alignaddrldi_vis"): Likewise.
2256 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
2257 ("fexpand_vis"): Likewise.
2258 ("fpmerge_vis"): Likewise.
2259 ("faligndata<VM64:mode>_vis"): Likewise.
2260 ("bshuffle<VM64:mode>_vis"): Likewise.
2261 ("cmask8<P:mode>_vis"): Likewise.
2262 ("cmask16<P:mode>_vis"): Likewise.
2263 ("cmask32<P:mode>_vis"): Likewise.
2264 ("fchksm16_vis"): Likewise.
2265 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
2266 ("fmean16_vis"): Likewise.
2267 ("fp<plusminus_insn>64_vis"): Likewise.
2268 ("<plusminus_insn>v8qi3"): Likewise.
2269 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
2270 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
2271 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
2272 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
2273 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
2274 ("*movqi_insn"): Likewise.
2275 ("*movhi_insn"): Likewise.
2276 ("*movsi_insn"): Likewise.
2277 ("movsi_pic_gotdata_op"): Likewise.
2278 ("*movdi_insn_sp32"): Likewise.
2279 ("*movdi_insn_sp64"): Likewise.
2280 ("movdi_pic_gotdata_op"): Likewise.
2281 ("*movsf_insn"): Likewise.
2282 ("*movdf_insn_sp32"): Likewise.
2283 ("*movdf_insn_sp64"): Likewise.
2284 ("*zero_extendhisi2_insn"): Likewise.
2285 ("*zero_extendqihi2_insn"): Likewise.
2286 ("*zero_extendqisi2_insn"): Likewise.
2287 ("*zero_extendqidi2_insn"): Likewise.
2288 ("*zero_extendhidi2_insn"): Likewise.
2289 ("*zero_extendsidi2_insn_sp64"): Likewise.
2290 ("ldfsr"): Likewise.
2291 ("prefetch_64"): Likewise.
2292 ("prefetch_32"): Likewise.
2293 ("tie_ld32"): Likewise.
2294 ("tie_ld64"): Likewise.
2295 ("*tldo_ldub_sp32"): Likewise.
2296 ("*tldo_ldub1_sp32"): Likewise.
2297 ("*tldo_ldub2_sp32"): Likewise.
2298 ("*tldo_ldub_sp64"): Likewise.
2299 ("*tldo_ldub1_sp64"): Likewise.
2300 ("*tldo_ldub2_sp64"): Likewise.
2301 ("*tldo_ldub3_sp64"): Likewise.
2302 ("*tldo_lduh_sp32"): Likewise.
2303 ("*tldo_lduh1_sp32"): Likewise.
2304 ("*tldo_lduh_sp64"): Likewise.
2305 ("*tldo_lduh1_sp64"): Likewise.
2306 ("*tldo_lduh2_sp64"): Likewise.
2307 ("*tldo_lduw_sp32"): Likewise.
2308 ("*tldo_lduw_sp64"): Likewise.
2309 ("*tldo_lduw1_sp64"): Likewise.
2310 ("*tldo_ldx_sp64"): Likewise.
2311 ("*mov<VM32:mode>_insn"): Likewise.
2312 ("*mov<VM64:mode>_insn_sp64"): Likewise.
2313 ("*mov<VM64:mode>_insn_sp32"): Likewise.
2315 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2317 * config/sparc/sparc.md ("type"): New insn type viscmp.
2318 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
2320 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
2321 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
2322 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
2323 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
2325 ("n7_vis_logical_11cycle"): Likewise.
2326 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
2327 * config/sparc/niagara2.md ("niag3_vis": Likewise.
2328 * config/sparc/niagara.md ("niag_vis"): Likewise.
2329 * config/sparc/ultra3.md ("us3_fga"): Likewise.
2330 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
2332 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2334 * config/sparc/sparc.md: New instruction type `bmask'.
2335 (bmaskdi_vis): Use the `bmask' type.
2336 (bmasksi_vis): Likewise.
2337 * config/sparc/ultra3.md (us3_array): Likewise.
2338 * config/sparc/niagara7.md (n7_array): Likewise.
2339 * config/sparc/niagara4.md (n4_array): Likewise.
2340 * config/sparc/niagara2.md (niag2_vis): Likewise.
2341 (niag3_vis): Likewise.
2342 * config/sparc/niagara.md (niag_vis): Likewise.
2344 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2346 * ipa-comdats.c: Remove optimize check from gate.
2347 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
2348 for functions not optimized.
2349 (ipa_fn_summary_read): Skip optimize check.
2350 (ipa_fn_summary_write): Likewise.
2351 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
2353 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
2355 (can_inline_edge_p): Check flag_pcc_struct_return for match.
2356 (check_callers): Give up on caller which is not optimized.
2357 (inline_small_functions): Likewise.
2358 (ipa_inline): Do not give up when not optimizing.
2359 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
2360 away unoptimizes cdtors.
2361 (whole_program_function_and_variable_visibility): Do
2362 ipa_discover_readonly_nonaddressable_vars in LTO mode.
2363 * ipa.c (process_references): Do not check optimize.
2364 (symbol_table::remove_unreachable_nodes): Update optimize check.
2365 (set_writeonly_bit): Update optimize check.
2366 (pass_ipa_cdtor_merge::gate): Do not check optimize.
2367 (pass_ipa_single_use::gate): Remove.
2369 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2371 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
2372 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
2373 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
2374 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
2375 permute_load, permute_store, adjust_extract, adjust_splat,
2376 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
2377 replace_swap_with_copy, dump_swap_insn_table,
2378 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
2379 recombine_lvx_pattern, recombine_stvx_pattern,
2380 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
2381 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
2382 to file rs6000-p8swap.c.
2383 * config/rs6000/rs6000-p8swap.c: New file.
2384 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
2385 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
2386 and rs6000*-*-* targets.
2388 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2390 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
2392 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2394 * lto-wrapper.c (merge_and_complain): Do not merge
2395 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
2396 fsigned_zeros, ftrapping_math, fwrapv.
2397 (append_compiler_options): Do not track these options.
2398 (append_linker_options): Likewie
2400 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2402 * cgraphunit.c (cgraph_node::finalize_function): When
2403 !flag_toplevel_reorde set no_reorder flag.
2404 (varpool_node::finalize_decl): Likewise.
2405 (symbol_table::compile): Drop no toplevel reorder path.
2407 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2409 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
2410 edges; zero probability is not better than uninitialized.
2412 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
2414 * asan.h (asan_sanitize_allocas_p): Declare.
2415 * asan.c (asan_sanitize_allocas_p): New function.
2416 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
2417 (handle_builtin_alloca): Likewise.
2418 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
2419 if !asan_sanitize_allocas_p.
2420 * params.def (asan-instrument-allocas): Add new option.
2421 * params.h (ASAN_PROTECT_ALLOCAS): Define.
2422 * opts.c (common_handle_option): Disable allocas sanitization for
2425 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
2427 * asan.c: Include gimple-fold.h.
2428 (get_last_alloca_addr): New function.
2429 (handle_builtin_stackrestore): Likewise.
2430 (handle_builtin_alloca): Likewise.
2431 (asan_emit_allocas_unpoison): Likewise.
2432 (get_mem_refs_of_builtin_call): Add new parameter, remove const
2433 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
2434 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
2435 (instrument_builtin_call): Pass gimple iterator to
2436 get_mem_refs_of_builtin_call.
2437 (last_alloca_addr): New global.
2438 * asan.h (asan_emit_allocas_unpoison): Declare.
2439 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
2440 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
2441 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
2442 if function calls alloca.
2443 * gimple-fold.c (replace_call_with_value): Remove static keyword.
2444 * gimple-fold.h (replace_call_with_value): Declare.
2445 * internal-fn.c: Include asan.h.
2446 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
2447 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
2449 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2451 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
2452 (C_SELFTEST_FLAGS): New.
2453 (CPP_SELFTEST_FLAGS): New.
2454 (SELFTEST_DEPS): New, from deps of s-selftest.
2455 (C_SELFTEST_DEPS): New, from deps of s-selftest.
2456 (CPP_SELFTEST_DEPS): New.
2457 (selftest): Add dependency on s-selftest-c++.
2458 (s-selftest): Rename to...
2459 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
2460 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
2461 than SELFTEST_FLAGS.
2462 (selftest-gdb): Rename to...
2463 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
2465 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
2466 (selftest-valgrind): Rename to...
2467 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
2469 (selftest-valgrind): Reintroduce as an alias for
2470 selftest-c-valgrind.
2471 (s-selftest-c++): New.
2472 (selftest-c++-gdb): New.
2473 (selftest-c++-valgrind): New.
2475 2017-07-06 Olivier Hainque <hainque@adacore.com>
2477 * gcc.c (process_command): When deciding if undefined variables
2478 should be ignored when processing specs, accept "gcc -v" as well.
2480 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2482 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
2483 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
2485 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
2487 * config/arm/arm-cpus.in (armv8-r): Add new entry.
2488 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
2489 * config/arm/arm-tables.opt: Regenerate.
2490 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
2492 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
2494 2017-07-06 Carl Love <cel@us.ibm.com>
2496 * ChangeLog: Clean up from mid air collision
2498 2017-07-06 Carl Love <cel@us.ibm.com>
2500 * config/rs6000/rs6000-c.c: Add support for built-in functions
2501 vector signed int vec_subc (vector signed int, vector signed int);
2502 vector signed __int128 vec_subc (vector signed __int128,
2503 vector signed __int128);
2504 vector unsigned __int128 vec_subc (vector unsigned __int128,
2505 vector unsigned __int128);
2506 vector signed int vec_sube (vector signed int, vector signed int,
2508 vector unsigned int vec_sube (vector unsigned int,
2509 vector unsigned int,
2510 vector unsigned int);
2511 vector signed __int128 vec_sube (vector signed __int128,
2512 vector signed __int128,
2513 vector signed__int128);
2514 vector unsigned __int128 vec_sube (vector unsigned __int128,
2515 vector unsigned __int128,
2516 vector unsigned __int128);
2517 vector signed int vec_subec (vector signed int, vector signed int,
2519 vector unsigned int vec_subec (vector unsigned int,
2520 vector unsigned int,
2521 vector unsigned int);
2522 vector signed __int128 vec_subec (vector signed __int128,
2523 vector signed __int128,
2524 vector signed__int128);
2525 vector unsigned __int128 vec_subec (vector unsigned __int128,
2526 vector unsigned __int128,
2527 vector unsigned __int128);
2528 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
2529 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
2530 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
2531 BU_ALTIVEC_OVERLOAD_X definitions.
2532 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
2533 * doc/extend.texi: Update the built-in documentation file for the new
2536 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2539 * diagnostic-show-locus.c (layout::layout): Use start and finish
2540 spelling location for the start and finish of each range.
2541 * genmatch.c (linemap_client_expand_location_to_spelling_point):
2542 Add unused aspect param.
2543 * input.c (expand_location_1): Add "aspect" param, and use it
2544 to access the correct part of the location.
2545 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
2547 (expand_location_to_spelling_point): Likewise.
2548 (linemap_client_expand_location_to_spelling_point): Add "aspect"
2549 param, and pass it to expand_location_1.
2551 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
2553 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
2554 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
2555 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
2556 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
2557 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
2558 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
2559 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
2560 _mm_maskz_getmant_ss): New intrinsics.
2561 (__builtin_ia32_getexpss128_mask): Changed to ...
2562 __builtin_ia32_getexpss128_round ... this.
2563 (__builtin_ia32_getexpsd128_mask): Changed to ...
2564 __builtin_ia32_getexpsd128_round ... this.
2565 * config/i386/i386-builtin-types.def
2566 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
2567 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
2568 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
2569 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
2570 __builtin_ia32_getmantss_mask_round): New builtins.
2571 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
2572 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
2573 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
2574 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
2575 * config/i386/sse.md
2576 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
2577 avx512f_sgetexp<mode><mask_scalar_name>
2578 <round_saeonly_scalar_name> ... this.
2579 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
2580 %0, %1, %2<round_saeonly_op3>}): Changed to ...
2581 vgetexp<ssescalarmodesuffix>
2582 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2583 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
2584 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
2585 avx512f_vgetmant<mode><mask_scalar_name>
2586 <round_saeonly_scalar_name> ... this.
2587 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
2588 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
2589 vgetmant<ssescalarmodesuffix>
2590 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
2591 %0<mask_scalar_operand4>, %1, %2
2592 <round_saeonly_scalar_mask_op4>, %3} ... this.
2593 * config/i386/subst.md (mask_scalar_operand4,
2594 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
2595 round_saeonly_scalar_nimm_predicate): New subst attributes.
2597 2017-07-06 Julia Koval <julia.koval@intel.com>
2599 * config/i386/i386.c (ix86_erase_embedded_rounding):
2600 Remove code for old rounding pattern.
2602 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
2604 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
2606 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
2608 * doc/sourcebuild.texi (Test Directives, Variants of
2609 dg-require-support): Add documentation for dg-require-stack-check.
2611 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
2613 * config/i386/subst.md (mask_scalar, round_scalar,
2614 round_saeonly_scalar): New meta-templates.
2615 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
2616 round_scalar_mask_operand3, round_scalar_mask_op3,
2617 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
2618 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
2619 round_saeonly_scalar_constraint,
2620 round_saeonly_scalar_prefix): New subst attribute.
2621 * config/i386/sse.md
2622 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
2623 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
2624 <round_scalar_name> ... this.
2625 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
2626 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
2627 <round_scalar_name> ... this.
2628 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
2629 <sse>_vm<code><mode>3<mask_scalar_name>
2630 <round_saeonly_scalar_name> ... this.
2631 (v<plusminus_mnemonic><ssescalarmodesuffix>
2632 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2633 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2634 v<plusminus_mnemonic><ssescalarmodesuffix>
2635 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2636 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2637 (v<multdiv_mnemonic><ssescalarmodesuffix>
2638 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2639 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2640 v<multdiv_mnemonic><ssescalarmodesuffix>
2641 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2642 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2643 (v<maxmin_float><ssescalarmodesuffix>
2644 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
2645 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
2646 v<maxmin_float><ssescalarmodesuffix>
2647 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2648 %0<mask_scalar_operand3>, %1, %<iptr>2
2649 <round_saeonly_scalar_mask_op3>} ... this.
2651 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
2653 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
2654 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
2656 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
2657 Alan Hayward <alan.hayward@arm.com>
2658 David Sherwood <david.sherwood@arm.com>
2660 * combine.c (simplify_if_then_else): Remove "enum" before
2662 * compare-elim.c (can_eliminate_compare): Likewise.
2663 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
2665 (aarch64_lookup_simd_builtin_type): Likewise.
2666 (aarch64_simd_builtin_type): Likewise.
2667 (aarch64_init_simd_builtin_types): Likewise.
2668 (aarch64_simd_expand_args): Likewise.
2669 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
2671 (aarch64_reverse_mask): Likewise.
2672 (aarch64_simd_emit_reg_reg_move): Likewise.
2673 (aarch64_gen_adjusted_ldpstp): Likewise.
2674 (aarch64_ccmp_mode_to_code): Likewise.
2675 (aarch64_operands_ok_for_ldpstp): Likewise.
2676 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2677 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
2679 (aarch64_min_divisions_for_recip_mul): Likewise.
2680 (aarch64_reassociation_width): Likewise.
2681 (aarch64_get_condition_code_1): Likewise.
2682 (aarch64_simd_emit_reg_reg_move): Likewise.
2683 (aarch64_simd_attr_length_rglist): Likewise.
2684 (aarch64_reverse_mask): Likewise.
2685 (aarch64_operands_ok_for_ldpstp): Likewise.
2686 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2687 (aarch64_gen_adjusted_ldpstp): Likewise.
2688 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
2690 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
2691 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
2692 (arm_lookup_simd_builtin_type): Likewise.
2693 (arm_simd_builtin_type): Likewise.
2694 (arm_init_simd_builtin_types): Likewise.
2695 (arm_expand_builtin_args): Likewise.
2696 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
2697 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
2698 (ft32_setup_incoming_varargs): Likewise.
2699 (ft32_function_arg): Likewise.
2700 (ft32_function_arg_advance): Likewise.
2701 (ft32_pass_by_reference): Likewise.
2702 (ft32_arg_partial_bytes): Likewise.
2703 (ft32_valid_pointer_mode): Likewise.
2704 (ft32_addr_space_pointer_mode): Likewise.
2705 (ft32_addr_space_legitimate_address_p): Likewise.
2706 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
2708 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
2709 (ix86_emit_outlined_ms2sysv_restore): Likewise.
2710 (iamcu_alignment): Likewise.
2711 (canonicalize_vector_int_perm): Likewise.
2712 (ix86_noce_conversion_profitable_p): Likewise.
2713 (ix86_mpx_bound_mode): Likewise.
2714 (ix86_operands_ok_for_move_multiple): Likewise.
2715 * config/microblaze/microblaze-protos.h
2716 (microblaze_expand_conditional_branch_reg): Likewise.
2717 * config/microblaze/microblaze.c
2718 (microblaze_expand_conditional_branch_reg): Likewise.
2719 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
2721 (rs6000_reassociation_width): Likewise.
2722 (rs6000_invalid_binary_op): Likewise.
2723 (fusion_p9_p): Likewise.
2724 (emit_fusion_p9_load): Likewise.
2725 (emit_fusion_p9_store): Likewise.
2726 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
2728 (riscv_hard_regno_mode_ok_p): Likewise.
2729 (riscv_address_insns): Likewise.
2730 (riscv_split_symbol): Likewise.
2731 (riscv_legitimize_move): Likewise.
2732 (riscv_function_value): Likewise.
2733 (riscv_hard_regno_nregs): Likewise.
2734 (riscv_expand_builtin): Likewise.
2735 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
2736 (riscv_build_integer): Likewise.
2737 (riscv_split_integer): Likewise.
2738 (riscv_legitimate_constant_p): Likewise.
2739 (riscv_cannot_force_const_mem): Likewise.
2740 (riscv_regno_mode_ok_for_base_p): Likewise.
2741 (riscv_valid_base_register_p): Likewise.
2742 (riscv_valid_offset_p): Likewise.
2743 (riscv_valid_lo_sum_p): Likewise.
2744 (riscv_classify_address): Likewise.
2745 (riscv_legitimate_address_p): Likewise.
2746 (riscv_address_insns): Likewise.
2747 (riscv_load_store_insns): Likewise.
2748 (riscv_force_binary): Likewise.
2749 (riscv_split_symbol): Likewise.
2750 (riscv_force_address): Likewise.
2751 (riscv_legitimize_address): Likewise.
2752 (riscv_move_integer): Likewise.
2753 (riscv_legitimize_const_move): Likewise.
2754 (riscv_legitimize_move): Likewise.
2755 (riscv_address_cost): Likewise.
2756 (riscv_subword): Likewise.
2757 (riscv_output_move): Likewise.
2758 (riscv_canonicalize_int_order_test): Likewise.
2759 (riscv_emit_int_order_test): Likewise.
2760 (riscv_function_arg_boundary): Likewise.
2761 (riscv_pass_mode_in_fpr_p): Likewise.
2762 (riscv_pass_fpr_single): Likewise.
2763 (riscv_pass_fpr_pair): Likewise.
2764 (riscv_get_arg_info): Likewise.
2765 (riscv_function_arg): Likewise.
2766 (riscv_function_arg_advance): Likewise.
2767 (riscv_arg_partial_bytes): Likewise.
2768 (riscv_function_value): Likewise.
2769 (riscv_pass_by_reference): Likewise.
2770 (riscv_setup_incoming_varargs): Likewise.
2771 (riscv_print_operand): Likewise.
2772 (riscv_elf_select_rtx_section): Likewise.
2773 (riscv_save_restore_reg): Likewise.
2774 (riscv_for_each_saved_reg): Likewise.
2775 (riscv_register_move_cost): Likewise.
2776 (riscv_hard_regno_mode_ok_p): Likewise.
2777 (riscv_hard_regno_nregs): Likewise.
2778 (riscv_class_max_nregs): Likewise.
2779 (riscv_memory_move_cost): Likewise.
2780 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
2781 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
2782 (rl78_addr_space_address_mode): Likewise.
2783 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2785 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2786 (rs6000_reassociation_width): Likewise.
2787 (rs6000_invalid_binary_op): Likewise.
2788 (fusion_p9_p): Likewise.
2789 (emit_fusion_p9_load): Likewise.
2790 (emit_fusion_p9_store): Likewise.
2791 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
2792 (ok_for_simple_move_operands): Likewise.
2793 (ok_for_simple_move_strict_operands): Likewise.
2794 (ok_for_simple_arith_logic_operands): Likewise.
2795 (visium_legitimize_reload_address): Likewise.
2796 (visium_select_cc_mode): Likewise.
2797 (output_cbranch): Likewise.
2798 (visium_split_double_move): Likewise.
2799 (visium_expand_copysign): Likewise.
2800 (visium_expand_int_cstore): Likewise.
2801 (visium_expand_fp_cstore): Likewise.
2802 * config/visium/visium.c (visium_pass_by_reference): Likewise.
2803 (visium_function_arg): Likewise.
2804 (visium_function_arg_advance): Likewise.
2805 (visium_libcall_value): Likewise.
2806 (visium_setup_incoming_varargs): Likewise.
2807 (visium_legitimate_constant_p): Likewise.
2808 (visium_legitimate_address_p): Likewise.
2809 (visium_legitimize_address): Likewise.
2810 (visium_secondary_reload): Likewise.
2811 (visium_register_move_cost): Likewise.
2812 (visium_memory_move_cost): Likewise.
2813 (prepare_move_operands): Likewise.
2814 (ok_for_simple_move_operands): Likewise.
2815 (ok_for_simple_move_strict_operands): Likewise.
2816 (ok_for_simple_arith_logic_operands): Likewise.
2817 (visium_function_value_1): Likewise.
2818 (rtx_ok_for_offset_p): Likewise.
2819 (visium_legitimize_reload_address): Likewise.
2820 (visium_split_double_move): Likewise.
2821 (visium_expand_copysign): Likewise.
2822 (visium_expand_int_cstore): Likewise.
2823 (visium_expand_fp_cstore): Likewise.
2824 (visium_split_cstore): Likewise.
2825 (visium_select_cc_mode): Likewise.
2826 (visium_split_cbranch): Likewise.
2827 (output_cbranch): Likewise.
2828 (visium_print_operand_address): Likewise.
2829 * expmed.c (flip_storage_order): Likewise.
2830 * expmed.h (emit_cstore): Likewise.
2831 (flip_storage_order): Likewise.
2832 * genrecog.c (validate_pattern): Likewise.
2833 * hsa-gen.c (gen_hsa_addr): Likewise.
2834 * internal-fn.c (expand_arith_overflow): Likewise.
2835 * ira-color.c (allocno_copy_cost_saving): Likewise.
2836 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2837 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
2838 (process_invariant_for_inheritance): Likewise.
2839 * lra-eliminations.c (move_plus_up): Likewise.
2840 * omp-low.c (lower_oacc_reductions): Likewise.
2841 * simplify-rtx.c (simplify_subreg): Likewise.
2842 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
2843 (TARGET_CHKP_BOUND_MODE): Likewise..
2844 * targhooks.c (default_chkp_bound_mode): Likewise.
2845 (default_setup_incoming_vararg_bounds): Likewise.
2846 * targhooks.h (default_chkp_bound_mode): Likewise.
2847 (default_setup_incoming_vararg_bounds): Likewise.
2848 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
2849 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
2850 (have_whole_vector_shift): Likewise.
2851 * tree-vect-stmts.c (vectorizable_load): Likewise.
2852 * doc/tm.texi: Regenerate.
2854 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2856 Graceful degrade if Binutils PR21472 is not available.
2859 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
2860 .rodata in flash test fails.
2861 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
2862 * confgure: Regenerate.
2863 * config.in: Regenerate.
2864 * config/avr/avr.c (avr_asm_named_section)
2865 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
2866 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
2867 (avr_asm_init_sections): Same.
2869 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2871 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
2872 (fma<VH:mode>4_intrinsic): Likewise.
2873 (*fmsub<VCVTF:mode>4): Likewise.
2874 (*fmsub<VH:mode>4_intrinsic): Likewise.
2876 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2879 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
2880 Don't depend on "optimize > 0".
2881 (out_movhi_r_mr, out_movqi_mr_r): Same.
2882 (out_movhi_mr_r, out_movqi_r_mr): Same.
2883 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
2884 io_address_operand on "optimize > 0".
2886 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2888 * tree-loop-distribution.c: Add general explanantion on the pass.
2889 (generate_loops_for_partition): Mark distributed loop.
2890 (pg_add_dependence_edges): New parameter. Handle alias data
2891 dependence specially and record it in the parameter if asked.
2892 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
2893 (init_partition_graph_vertices, add_partition_graph_edge): New.
2894 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
2895 (free_partition_graph_vdata, build_partition_graph): New.
2896 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
2897 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
2898 (data_ref_segment_size, latch_dominated_by_data_ref): New.
2899 (compute_alias_check_pairs, version_loop_by_alias_check): New.
2900 (version_for_distribution_p, finalize_partitions): New.
2901 (distribute_loop): Handle alias data dependence specially. Factor
2902 out loop fusion code as functions and call these functions.
2904 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2906 * tree-loop-distribution.c (classify_partition): New parameter and
2907 better handle reduction statement.
2908 (rdg_build_partitions): Revise comment.
2909 (distribute_loop): Compute statements in all partitions and pass it
2910 to classify_partition.
2912 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2914 * tree-loop-distribution.c (enum partition_type): New.
2915 (struct partition): New field type.
2916 (partition_merge_into): Add parameter. Update partition type.
2917 (data_dep_in_cycle_p, update_type_for_merge): New functions.
2918 (build_rdg_partition_for_vertex): Compute partition type.
2919 (rdg_build_partitions): Dump partition type.
2920 (distribute_loop): Update calls to partition_merge_into.
2922 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2924 * tree-loop-distribution.c (struct ddr_hasher): New.
2925 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
2927 (classify_partition): Call get_data_dependence.
2928 (pg_add_dependence_edges): Ditto.
2929 (distribute_loop): Release data dependence hash table.
2931 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2933 * tree-loop-distribution.c (ref_base_address): Delete.
2934 (similar_memory_accesses): Rename ...
2935 (share_memory_accesses): ... to this. Check if partitions access
2936 the same memory reference.
2937 (distribute_loop): Call share_memory_accesses.
2939 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2941 * tree-loop-distribution.c (struct partition): New field recording
2943 (partition_alloc, partition_free): Init and release data refs.
2944 (partition_merge_into): Merge data refs.
2945 (build_rdg_partition_for_vertex): Collect data refs for partition.
2946 (pg_add_dependence_edges): Change parameters from vector to bitmap.
2948 (distribute_loop): Remve data refs from vertice data of partition
2951 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2953 * tree-loop-distribution.c (params.h): Include header file.
2954 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
2955 (datarefs_vec): New global var.
2956 (create_rdg_vertices): Use datarefs_vec directly.
2957 (free_rdg): Don't free data references.
2958 (build_rdg): Update use. Don't free data references.
2959 (distribute_loop): Compute global variable for data references.
2960 Bail out if there are too many data references.
2962 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2964 * tree-loop-distribution.c (loop_nest): New global var.
2965 (build_rdg): Use loop directly, rather than loop nest.
2966 (pg_add_dependence_edges): Remove loop nest parameter. Use global
2968 (distribute_loop): Compute global variable loop nest. Update use.
2970 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2972 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
2973 (partition_merge_into): New parameter. Dump reason for fusion.
2974 (distribute_loop): Update use of partition_merge_into.
2976 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2978 * tree-loop-distribution.c (bb_top_order_index): New.
2979 (bb_top_order_index_size, bb_top_order_cmp): New.
2980 (stmts_from_loop): Use topological order.
2981 (pass_loop_distribution::execute): Compute and release topological
2982 order for basic blocks.
2984 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2986 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
2989 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2991 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
2992 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
2993 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
2994 * internal-fn.def (LOOP_DIST_ALIAS): New.
2995 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
2996 (fold_loop_internal_call): ... this.
2997 (vect_loop_dist_alias_call): New function.
2998 (set_uid_loop_bbs): Call fold_loop_internal_call.
2999 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
3002 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
3005 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
3006 Require dead FLAGS_REG at the beginning of a peephole.
3008 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
3011 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
3012 arguments in the call to __builtin_ia32_sbb_u32.
3013 (_subborrow_u64): Swap _X and _Y arguments in the call to
3014 __builtin_ia32_sbb_u64.
3016 2017-07-04 Jakub Jelinek <jakub@redhat.com>
3019 * tree-vrp.c (compare_assert_loc): Turn into a function template
3020 with stable template parameter. Only test if a->e is NULL,
3021 !a->e == !b->e has been verified already. Use e == NULL or
3022 e != NULL instead of e or ! e tests. If stable is true, don't use
3023 iterative_hash_expr, on the other side allow a or b or both NULL
3024 and sort the NULLs last.
3025 (process_assert_insertions): Sort using compare_assert_loc<false>
3026 instead of compare_assert_loc, later sort using
3027 compare_assert_loc<true> before calling process_assert_insertions_for
3028 in a loop. Use break instead of continue once seen NULL pointer.
3030 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
3032 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
3033 Cortex-R7 and Cortex-R8 processors.
3035 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
3037 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
3038 uninitialized while src is not.
3040 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
3042 * common/config/arm/arm-common.c: Adjust include path for
3044 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
3045 (arm-cpu.h): Create in build directory. Adjust dependency rules.
3046 (arm-cpu-data.h): Likewise.
3047 (arm-cpu-cdata.h): Likewise.
3048 * config/arm/arm-cpu.h: Delete.
3049 * config/arm/arm-cpu-cdata.h: Delete.
3050 * config/arm/arm-cpu-data.h: Delete.
3052 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
3054 * config/arm/arm-cpus.in (cortex-a55): New.
3055 (cortex-a75): Likewise.
3056 (cortex-a75.cortex-a55): Likewise.
3057 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
3059 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
3060 * config/arm/arm-cpu-cdata.h: Regenerate.
3061 * config/arm/arm-cpu-data.h: Regenerate.
3062 * config/arm/arm-cpu.h: Regenerate.
3063 * config/arm/arm-tables.opt: Regenerate.
3064 * config/arm/arm-tune.md: Regenerate.
3066 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
3068 * haifa-sched.c (sched_create_recovery_edges): Update profile.
3070 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
3072 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
3075 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
3077 PR tree-optimization/81292
3078 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
3079 full_string_p, also call adjust_related_strinfos if the adjustment
3080 is simple, otherwise invalidate related strinfos.
3082 2017-07-04 Martin Liska <mliska@suse.cz>
3085 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
3086 newly created variable as DECL_IGNORED_P.
3088 2017-07-04 Martin Liska <mliska@suse.cz>
3091 * ipa-inline.c (inline_small_functions):
3092 Use xstrdup_for_dump.
3094 2017-07-04 Tom de Vries <tom@codesourcery.com>
3096 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
3098 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
3101 * config/darwin.c (darwin_function_switched_text_sections):
3104 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
3106 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
3108 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
3110 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
3112 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3114 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
3115 min_profitable_iters, and th as inclusive lower bounds.
3116 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
3117 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
3118 for min_profitable_iters and min_profitable_estimate.
3119 (vect_transform_loop): Treat th as an inclusive lower bound.
3120 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
3122 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
3125 * config/darwin.c (darwin_function_switched_text_sections):
3126 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
3127 in two pieces, and suppress the use of buf.
3129 2017-07-03 Nathan Sidwell <nathan@acm.org>
3131 * hash-table.h (hash_table_mod1): Fix indentation.
3133 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3136 * predict.c (force_edge_cold): Be more careful about propagation
3138 * profile-count.h (profile_probability::guessed,
3139 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
3141 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
3143 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
3145 * doc/invoke.texi (rcpc architecture extension): Document it.
3147 2017-07-03 Richard Biener <rguenther@suse.de>
3149 PR tree-optimization/60510
3150 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
3151 the scalar reduction PHI and use it.
3152 (vectorizable_reduction): Properly guard the single_defuse_cycle
3153 path for non-SLP reduction chains where we cannot use it.
3154 Rework reduc_def/index and vector type deduction. Rework
3155 vector operand gathering during reduction op code-gen.
3156 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
3157 chains dissolve the chain and leave it to non-SLP reduction
3160 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3162 * tree-data-ref.h (dr_alignment): Declare.
3163 * tree-data-ref.c (dr_alignment): New function.
3164 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
3165 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
3167 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
3169 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3171 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
3172 and base_misalignment fields.
3173 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
3174 * tree-data-ref.c: Include builtins.h.
3175 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
3176 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
3177 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
3178 * tree-vect-data-refs.c: Include tree-cfg.h.
3179 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
3180 fields instead of calculating an alignment here.
3181 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
3182 innermost_loop_behavior fields.
3184 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3186 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
3188 (DR_STEP_ALIGNMENT): New macro.
3189 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
3190 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
3191 (create_data_ref): Print it.
3192 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
3193 to tell whether the step preserves vector (mis)alignment.
3194 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
3195 Move the check for an integer step and generalise to all INTEGER_CST.
3196 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
3197 Print the outer step alignment.
3199 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3201 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
3202 with offset_alignment.
3203 (DR_ALIGNED_TO): Delete.
3204 (DR_OFFSET_ALIGNMENT): New macro.
3205 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
3206 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
3207 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
3208 (create_data_ref): Likewise.
3209 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
3210 (vect_analyze_data_refs): Likewise.
3211 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
3212 creating dummy innermost behavior.
3214 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3216 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
3217 with a "innermost_loop_behavior *" and refeence tree.
3218 * tree-data-ref.c (dr_analyze_innermost): Likewise.
3219 (create_data_ref): Update call accordingly.
3220 * tree-predcom.c (find_looparound_phi): Likewise.
3222 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3224 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
3225 fields with dr_wrt_vec_loop.
3226 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
3227 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
3228 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
3229 (vect_dr_behavior): New function.
3230 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
3231 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
3232 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
3233 track whether the step preserves the misalignment.
3234 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
3235 Use vect_dr_behavior.
3236 (vect_setup_realignment): Update call accordingly.
3237 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
3238 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
3239 call to vect_create_addr_base_for_vector_ref.
3240 (vect_create_cond_for_align_checks): Likewise.
3241 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
3242 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
3243 (vect_recog_mask_conversion_pattern): Likewise.
3244 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
3245 (new_stmt_vec_info): Remove redundant zeroing.
3247 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
3249 * common/config/arm/arm-common.c (arm_be8_option): New function.
3250 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
3251 (ISA_ARMv6): Add isa_bit_be8.
3252 * config/arm/arm.h (arm_be8_option): Add prototype.
3253 (BE8_SPEC_FUNCTION): New define.
3254 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
3255 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
3256 (mlittle-endian): Similarly.
3257 (mbe8, mbe32): New options.
3258 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
3259 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
3261 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3263 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
3265 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3267 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
3268 (cleanup_tree_cfg_bb): Use it.
3269 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
3271 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
3273 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3276 * loop-doloop.c (add_test): Update profile.
3278 2017-07-03 Martin Liska <mliska@suse.cz>
3281 * sanopt.c (rewrite_usage_of_param): New function.
3282 (sanitize_rewrite_addressable_params): Likewise.
3283 (pass_sanopt::execute): Call rewrite_usage_of_param.
3285 2017-07-03 Richard Biener <rguenther@suse.de>
3287 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
3288 back to using VIEW_CONVERT_EXPR.
3290 2017-07-03 Martin Liska <mliska@suse.cz>
3293 * doc/extend.texi: Document when a resolver function is
3294 generated for target_clones.
3296 2017-07-03 Martin Liska <mliska@suse.cz>
3298 * asan.c (asan_emit_stack_protection): Unpoison just red zones
3299 and shadow memory of auto variables which are subject of
3300 use-after-scope sanitization.
3301 (asan_expand_mark_ifn): Add do set only when is_poison.
3303 2016-07-03 Richard Biener <rguenther@suse.de>
3305 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
3307 (vect_force_simple_reduction): Record reduction def -> phi mapping.
3308 (vectorizable_reduction): Perform reduction PHI creation when
3309 visiting a reduction PHI and adjust and simplify code generation
3310 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
3311 (vect_transform_loop): Visit reduction PHIs.
3312 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
3313 defs into the SLP tree.
3314 (vect_build_slp_tree): Reduction defs terminate the recursion.
3315 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
3317 (vect_get_vec_defs_for_stmt_copy): Export.
3318 (vect_get_vec_defs): Likewise.
3319 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
3321 (vect_get_vec_defs_for_stmt_copy): Declare.
3322 (vect_get_vec_defs): Likewise.
3324 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3326 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
3327 parameter with a "loop" parameter and use it instead of the
3328 loop containing DR_STMT. Don't check simple_iv when doing
3329 BB analysis. Describe the two analysis modes in the comment.
3331 2017-07-03 Tom de Vries <tom@codesourcery.com>
3333 PR tree-optimization/69468
3334 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
3335 (find_same_succ_bb): Handle ignore_edge_flags.
3337 2017-07-03 Tom de Vries <tom@codesourcery.com>
3339 PR tree-optimization/81192
3340 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
3342 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
3344 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
3346 2017-07-03 Tom de Vries <tom@codesourcery.com>
3348 PR tree-optimization/81192
3349 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
3350 BB_SAME_SUCC (bb) == NULL.
3352 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3354 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
3357 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3359 * dumpfile.c: Include profile-count.h
3360 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
3362 (insert_cond_bb): Update profile.
3363 * tree-cfg.h (insert_cond_bb): Update prototype.
3364 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
3365 * tree-dump.c: Do not include tree-cfg.
3367 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3369 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
3371 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3373 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
3376 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3378 * tree-complex.c (expand_complex_div_wide): update profile.
3380 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3381 Alan Hayward <alan.hayward@arm.com>
3382 David Sherwood <david.sherwood@arm.com>
3384 * Makefile.in (MACHMODE_H): Remove insn-modes.h
3385 (CORETYPES_H): New define.
3386 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
3387 (insn-modes-inline.h, s-modes-inline-h): New rules.
3388 (generated_files): Add insn-modes-inline.h.
3389 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
3390 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
3391 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
3392 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
3393 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
3394 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
3395 (build/gencodes.o, build/genconditions.o): Likewise.
3396 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
3397 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
3398 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
3399 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
3400 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
3401 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
3402 * coretypes.h: Include everything up to real.h for generators.
3403 Include insn-modes.h first. Include wide-int-print.h after
3404 wide-int.h. Include insn-modes-inline.h and then machmode.h.
3405 * machmode.h: Don't include insn-modes.h here.
3406 * function-tests.c: Remove includes of signop.h, machmode.h,
3407 double-int.h and wide-int.h.
3409 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
3411 * optc-save-gen.awk: Likewise.
3412 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
3413 * godump.c: Remove include of wide-int-print.h.
3414 * pretty-print.h: Likewise.
3415 * wide-int-print.cc: Likewise.
3416 * wide-int.cc: Likewise.
3417 * hash-map-tests.c: Remove include of signop.h.
3418 * hash-set-tests.c: Likewise.
3419 * rtl-tests.c: Likewise.
3420 * mkconfig.sh: Remove include of machmode.h.
3421 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
3423 (emit_insn_modes_inline_h): ...this new function. Emit the code
3424 into an insn-modes-inline.h header file, adding appropriate
3425 include guards and end comments.
3426 (emit_insn_modes_c_header): Remove include of machmode.h.
3427 (emit_min_insn_modes_c_header): Include coretypes.h rather than
3429 (main): Handle -i flag and call emit_insn_modes_inline_h when
3432 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3434 * tree-ssa-strlen.c (strinfo): Rename the length field to
3435 nonzero_chars. Add a full_string_p field.
3436 (compare_nonzero_chars, zero_length_string_p): New functions.
3437 (get_addr_stridx): Add an offset_out parameter.
3438 Use compare_nonzero_chars.
3439 (get_stridx): Update accordingly. Use compare_nonzero_chars.
3440 (new_strinfo): Update after above changes to strinfo.
3441 (set_endptr_and_length): Set full_string_p.
3442 (get_string_length): Update after above changes to strinfo.
3443 (unshare_strinfo): Update call to new_strinfo.
3444 (maybe_invalidate): Likewise.
3445 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
3446 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
3447 as a uhwi instead of an shwi. Update after above changes to
3448 strinfo and new_strinfo.
3449 (zero_length_string): Assert that chainsi contains full strings.
3450 Use zero_length_string_p. Update call to new_strinfo.
3451 (adjust_related_strinfos): Update after above changes to strinfo.
3452 Copy full_string_p from origsi.
3453 (adjust_last_stmt): Use zero_length_string_p.
3454 (handle_builtin_strlen): Update after above changes to strinfo and
3455 new_strinfo. Install the lhs as the string length if the previous
3456 entry didn't describe a full string.
3457 (handle_builtin_strchr): Update after above changes to strinfo
3459 (handle_builtin_strcpy): Likewise.
3460 (handle_builtin_strcat): Likewise.
3461 (handle_builtin_malloc): Likewise.
3462 (handle_pointer_plus): Likewise.
3463 (handle_builtin_memcpy): Likewise. Track nonzero characters
3464 that aren't necessarily followed by a nul terminator.
3465 (handle_char_store): Likewise.
3467 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3469 PR tree-optimization/80769
3470 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
3471 for malloc and calloc. Document the new invariant that all related
3472 strinfos have delayed lengths or none do.
3473 (verify_related_strinfos): Move earlier in file.
3474 (set_endptr_and_length): New function, split out from...
3475 (get_string_length): ...here. Also set the lengths of related
3477 (zero_length_string): Assert that chainsi has known (rather than
3479 (adjust_related_strinfos): Likewise.
3481 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3483 PR tree-optimization/81136
3484 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
3485 assert that two references with the same misalignment have the same
3486 compile-time misalignment if those compile-time misalignments
3489 2017-07-01 Andi Kleen <ak@linux.intel.com>
3491 * print-tree.c (print_node): Print all attributes.
3493 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3495 * cfg.c (scale_bbs_frequencies): New function.
3496 * cfg.h (scale_bbs_frequencies): Declare it.
3497 * cfgloopanal.c (single_likely_exit): Cleanup.
3498 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
3500 (scale_loop_profile): Likewise.
3501 (loop_version): Likewise.
3502 (create_empty_loop_on_edge): Update.
3503 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
3504 scale_loop_frequencies, scale_loop_profile, loopify,
3505 loop_version): Update prototypes.
3506 * modulo-sched.c (sms_schedule): Update.
3507 * predict.c (unlikely_executed_edge_p): Also check probability.
3508 (probably_never_executed_edge_p): Fix typo.
3509 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3510 * tree-parloops.c (gen_parallel_loop): Update.
3511 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
3512 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3513 * tree-ssa-loop-split.c (split_loop): Update.
3514 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3515 * tree-vect-loop-manip.c (vect_do_peeling): Update.
3516 (vect_loop_versioning): Update.
3517 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3519 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3521 * trans-mem.c (split_bb_make_tm_edge): Update profile.
3523 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3525 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
3526 to keep profile consistent.
3528 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3530 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
3531 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
3532 * profile-count.h (max_safe_multiplier): Make unsigned.
3533 (profile_count::guessed_zero): New.
3535 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3537 * bb-reorder.c (fix_up_crossing_landing_pad,
3538 fix_crossing_conditional_branches): Use make_single_succ_edge
3539 to keep profile consistent.
3541 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3543 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
3546 2017-07-01 Jakub Jelinek <jakub@redhat.com>
3549 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
3550 the right scopes, make sure cond_jump isn't preserved between multiple
3551 iterations. Search for fallthru edge whenever there are 3+ edges and
3552 use find_fallthru_edge for it.
3554 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3556 Patch by Alexander Monakov <amonakov@ispras.ru>
3557 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
3558 probabilities consistently.
3560 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3562 * pa.c (pa_expand_compare_and_swap_loop): Update call of
3563 emit_cmp_and_jump_insns.
3565 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3568 * tree-inline.c (expand_call_inline): Combine profile statuses.
3570 2017-06-30 Andrew Pinski <apinski@cavium.com>
3572 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
3573 fold_stmt returned true.
3575 2017-06-30 Nathan Sidwell <nathan@acm.org>
3577 * ggc.h (empty_string): Delete.
3578 * cfgexpand.c (expand_asm_stmt): Use plain "".
3579 * optabs.c (expand_asm_memory_barrier): Likewise.
3580 * stringpool.c (empty_string): Delete.
3581 (digit_vector, digit_string): Delete.
3582 (ggc_alloc_string): Use plain "", don't optimize single digit
3583 strings. Use ggc_alloc_atomic.
3585 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
3587 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
3588 comparison set and one other set, use the cost of the non-comparison
3591 2017-06-30 Nathan Sidwell <nathan@acm.org>
3593 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
3596 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
3598 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
3599 loops. Remove now unneeded calls to gimple_switch_set_label() that
3600 just set removed labels to NULL_TREE.
3602 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
3604 * tree-ssanames.c (set_range_info_raw): Abstract from ...
3605 (set_range_info): ...here. Only call set_range_info_raw if domain
3607 (set_nonzero_bits): Call set_range_info_raw.
3608 * tree-ssanames.h (set_range_info_raw): New.
3610 2017-06-30 Jakub Jelinek <jakub@redhat.com>
3613 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
3614 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
3615 of nonimmediate_operand and <store_mask_constraint> instead of m
3616 for the input operand. For V8FI iterator, always split if input
3617 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
3618 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
3619 <store_mask_predicate> instead of register_operand and
3620 <store_mask_constraint> instead of v for the input operand. Make
3621 sure both operands aren't MEMs for if not <mask_applied>.
3623 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
3625 * lto-wrapper.c (copy_file) Close both file descriptors before
3628 2017-06-30 Martin Liska <mliska@suse.cz>
3631 * multiple_target.c (create_dispatcher_calls): Make ifunc
3632 also for function that don't have calls or are not referenced.
3634 2017-06-30 Richard Biener <rguenther@suse.de>
3636 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
3637 analyze the first scalar stmt. Move vector type computation
3638 for the BB case here from ...
3639 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
3640 live operation processing in the SLP case properly.
3642 2017-06-30 Richard Biener <rguenther@suse.de>
3644 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
3646 2017-06-30 Martin Liska <mliska@suse.cz>
3649 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
3650 before BUILT_IN_UNWIND_RESUME when ASAN is used.
3652 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
3654 * doc/invoke.texi (AArch64): Add missing options and remove redundant
3657 2017-06-30 Richard Biener <rguenther@suse.de>
3659 PR tree-optimization/81249
3660 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
3661 condition reduction result to original scalar type.
3663 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3665 * profile-count.h (enum profile_quality): Fix typos and whitespace
3668 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3670 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
3671 type for branch probabilities.
3673 2017-06-29 Julian Brown <julian@codesourcery.com>
3674 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3676 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
3677 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
3678 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
3679 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
3681 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3683 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
3684 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
3685 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
3686 CC usage from generic code to here.
3687 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
3688 CC usage into the target macros.
3690 2017-06-29 Maya Rashish <coypu@sdf.org>
3692 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
3695 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3697 * arm/arm-builtins.c: Include profile-count.h
3698 * except.c (sjlj_emit_function_enter): Use
3699 profile_probability::unlikely.
3701 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3703 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
3704 and tocrel_offset be pointer args rather than implicitly using
3706 (legitimate_constant_pool_address_p, rs6000_emit_move,
3707 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
3708 tocrel_offset and use in toc_relative_expr_p call.
3709 (print_operand, print_operand_address): Use static tocrel_base_oac
3710 and tocrel_offset_oac.
3711 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
3714 2017-06-29 Maya Rashish <coypu@sdf.org>
3716 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
3718 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
3720 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
3721 objects, take into account only the alignment of 'op0' and 'mode1' if
3724 2017-06-29 Steve Ellcey <sellcey@cavium.com>
3726 * ccmp.c (ccmp_tree_comparison_p): New function.
3727 (ccmp_candidate_p): Update to use above function.
3728 (get_compare_parts): New function.
3729 (expand_ccmp_next): Update to use new functions.
3730 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
3732 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
3733 take mode as argument.
3734 * ccmp.h (expand_ccmp_expr): Add mode as argument.
3735 * expr.c (expand_expr_real_1): Pass mode as argument.
3737 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
3739 * combine.c (combine_instructions): Print insns to dump_file, together
3742 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3744 * asan.c (asan_emit_stack_protection): Update.
3745 (create_cond_insert_point): Update.
3746 * auto-profile.c (afdo_propagate_circuit): Update.
3747 * basic-block.h (struct edge_def): Turn probability to
3748 profile_probability.
3749 (EDGE_FREQUENCY): Update.
3750 * bb-reorder.c (find_traces_1_round): Update.
3751 (better_edge_p): Update.
3752 (sanitize_hot_paths): Update.
3753 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
3754 (make_single_succ_edge): Update.
3755 (check_bb_profile): Update.
3756 (dump_edge_info): Update.
3757 (update_bb_profile_for_threading): Update.
3758 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
3759 probabilitycount to 0.
3760 * cfgbuild.c (compute_outgoing_frequencies): Update.
3761 * cfgcleanup.c (try_forward_edges): Update.
3762 (outgoing_edges_match): Update.
3763 (try_crossjump_to_edge): Update.
3764 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
3765 (expand_gimple_tailcall): Update.
3766 (construct_init_block): Use make_single_succ_edge.
3767 (construct_exit_block): Use make_single_succ_edge.
3768 * cfghooks.c (verify_flow_info): Update.
3769 (redirect_edge_succ_nodup): Update.
3770 (split_edge): Update.
3771 (account_profile_record): Update.
3772 * cfgloopanal.c (single_likely_exit): Update.
3773 * cfgloopmanip.c (scale_loop_profile): Update.
3774 (set_zero_probability): Remove.
3775 (duplicate_loop_to_header_edge): Update.
3776 * cfgloopmanip.h (loop_version): Update prototype.
3777 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3778 (force_nonfallthru_and_redirect): Update.
3779 (update_br_prob_note): Update.
3780 (rtl_verify_edges): Update.
3781 (purge_dead_edges): Update.
3782 (rtl_lv_add_condition_to_bb): Update.
3783 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
3784 * cgraphunit.c (init_lowered_empty_function): Update.
3785 (cgraph_node::expand_thunk): Update.
3786 * cilk-common.c: Include profile-count.h
3787 * dojump.c (inv): Remove.
3788 (jumpifnot): Update.
3789 (jumpifnot_1): Update.
3790 (do_jump_1): Update.
3792 (do_jump_by_parts_greater_rtx): Update.
3793 (do_compare_rtx_and_jump): Update.
3794 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
3795 do_jump_1. do_compare_rtx_and_jump): Update prototype.
3796 * dwarf2cfi.c: Include profile-count.h
3797 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
3798 (sjlj_emit_dispatch_table): Likewise.
3799 * explow.c: Include profile-count.h
3800 * expmed.c (emit_store_flag_force): Update.
3801 (do_cmp_and_jump): Update.
3802 * expr.c (compare_by_pieces_d::generate): Update.
3803 (compare_by_pieces_d::finish_mode): Update.
3804 (emit_block_move_via_loop): Update.
3805 (store_expr_with_bounds): Update.
3806 (store_constructor): Update.
3807 (expand_expr_real_2): Update.
3808 (expand_expr_real_1): Update.
3809 * expr.h (try_casesi, try_tablejump): Update prototypes.
3810 * gimple-pretty-print.c (dump_probability): Update.
3811 (dump_profile): New.
3812 (dump_gimple_label): Update.
3813 (dump_gimple_bb_header): Update.
3814 * graph.c (draw_cfg_node_succ_edges): Update.
3815 * hsa-gen.c (convert_switch_statements): Update.
3816 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
3817 (find_if_case_1): Update.
3818 (find_if_case_2): Update.
3819 * internal-fn.c (expand_arith_overflow_result_store): Update.
3820 (expand_addsub_overflow): Update.
3821 (expand_neg_overflow): Update.
3822 (expand_mul_overflow): Update.
3823 (expand_vector_ubsan_overflow): Update.
3824 * ipa-cp.c (good_cloning_opportunity_p): Update.
3825 * ipa-split.c (split_function): Use make_single_succ_edge.
3826 * ipa-utils.c (ipa_merge_profiles): Update.
3827 * loop-doloop.c (add_test): Update.
3828 (doloop_modify): Update.
3829 * loop-unroll.c (compare_and_jump_seq): Update.
3830 (unroll_loop_runtime_iterations): Update.
3831 * lra-constraints.c (lra_inheritance): Update.
3832 * lto-streamer-in.c (input_cfg): Update.
3833 * lto-streamer-out.c (output_cfg): Update.
3834 * mcf.c (adjust_cfg_counts): Update.
3835 * modulo-sched.c (sms_schedule): Update.
3836 * omp-expand.c (expand_omp_for_init_counts): Update.
3837 (extract_omp_for_update_vars): Update.
3838 (expand_omp_ordered_sink): Update.
3839 (expand_omp_for_ordered_loops): Update.
3840 (expand_omp_for_generic): Update.
3841 (expand_omp_for_static_nochunk): Update.
3842 (expand_omp_for_static_chunk): Update.
3843 (expand_cilk_for): Update.
3844 (expand_omp_simd): Update.
3845 (expand_omp_taskloop_for_outer): Update.
3846 (expand_omp_taskloop_for_inner): Update.
3847 * omp-simd-clone.c (simd_clone_adjust): Update.
3848 * optabs.c (expand_doubleword_shift): Update.
3849 (expand_abs): Update.
3850 (emit_cmp_and_jump_insn_1): Update.
3851 (expand_compare_and_swap_loop): Update.
3852 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
3853 * predict.c (predictable_edge_p): Update.
3854 (edge_probability_reliable_p): Update.
3855 (set_even_probabilities): Update.
3856 (combine_predictions_for_insn): Update.
3857 (combine_predictions_for_bb): Update.
3858 (propagate_freq): Update.
3859 (estimate_bb_frequencies): Update.
3860 (force_edge_cold): Update.
3861 * profile-count.c (profile_count::dump): Add missing space into dump.
3862 (profile_count::debug): Add newline.
3863 (profile_count::differs_from_p): Explicitly convert to unsigned.
3864 (profile_count::stream_in): Update.
3865 (profile_probability::dump): New member function.
3866 (profile_probability::debug): New member function.
3867 (profile_probability::differs_from_p): New member function.
3868 (profile_probability::differs_lot_from_p): New member function.
3869 (profile_probability::stream_in): New member function.
3870 (profile_probability::stream_out): New member function.
3871 * profile-count.h (profile_count_quality): Rename to ...
3872 (profile_quality): ... this one.
3873 (profile_probability): New.
3874 (profile_count): Update.
3875 * profile.c (compute_branch_probabilities): Update.
3876 * recog.c (peep2_attempt): Update.
3877 * sched-ebb.c (schedule_ebbs): Update.
3878 * sched-rgn.c (find_single_block_region): Update.
3879 (compute_dom_prob_ps): Update.
3880 (schedule_region): Update.
3881 * sel-sched-ir.c (compute_succs_info): Update.
3882 * stmt.c (struct case_node): Update.
3883 (do_jump_if_equal): Update.
3884 (get_outgoing_edge_probs): Update.
3885 (conditional_probability): Update.
3886 (emit_case_dispatch_table): Update.
3887 (expand_case): Update.
3888 (expand_sjlj_dispatch_table): Update.
3889 (emit_case_nodes): Update.
3890 * targhooks.c: Update.
3891 * tracer.c (better_p): Update.
3892 (find_best_successor): Update.
3893 * trans-mem.c (expand_transaction): Update.
3894 * tree-call-cdce.c: Update.
3895 * tree-cfg.c (gimple_split_edge): Upate.
3896 (move_sese_region_to_fn): Upate.
3897 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
3898 * tree-eh.c (lower_resx): Upate.
3899 (cleanup_empty_eh_move_lp): Upate.
3900 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3901 * tree-inline.c (copy_edges_for_bb): Update.
3902 (copy_cfg_body): Update.
3903 * tree-parloops.c (gen_parallel_loop): Update.
3904 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
3905 (gimple_gen_time_profiler): Update.
3906 * tree-ssa-dce.c (remove_dead_stmt): Update.
3907 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3908 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3909 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3910 (unloop_loops): Update.
3911 (try_peel_loop): Update.
3912 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3913 * tree-ssa-loop-split.c (connect_loops): Update.
3914 (split_loop): Update.
3915 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3916 (hoist_guard): Update.
3917 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3918 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3919 (value_replacement): Update.
3920 * tree-ssa-reassoc.c (branch_fixup): Update.
3921 * tree-ssa-tail-merge.c (replace_block_by): Update.
3922 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3923 (create_edge_and_update_destination_phis): Update.
3924 (compute_path_counts): Update.
3925 (recompute_probabilities): Update.
3926 (update_joiner_offpath_counts): Update.
3927 (freqs_to_counts_path): Update.
3928 (duplicate_thread_path): Update.
3929 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3930 (struct switch_conv_info): Update.
3931 (gen_inbound_check): Update.
3932 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3933 (vect_do_peeling): Update.
3934 (vect_loop_versioning): Update.
3935 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3936 (optimize_mask_stores): Update.
3937 * ubsan.c (ubsan_expand_null_ifn): Update.
3938 * value-prof.c (gimple_divmod_fixed_value): Update.
3939 (gimple_divmod_fixed_value_transform): Update.
3940 (gimple_mod_pow2): Update.
3941 (gimple_mod_pow2_value_transform): Update.
3942 (gimple_mod_subtract): Update.
3943 (gimple_mod_subtract_transform): Update.
3944 (gimple_ic): Update.
3945 (gimple_stringop_fixed_value): Update.
3946 (gimple_stringops_transform): Update.
3947 * value-prof.h: Update.
3949 2017-06-29 Carl Love <cel@us.ibm.com>
3951 * config/rs6000/rs6000-c.c: Add support for built-in functions
3952 vector signed int vec_signed (vector float);
3953 vector signed long long vec_signed (vector double);
3954 vector signed int vec_signed2 (vector double, vector double);
3955 vector signed int vec_signede (vector double);
3956 vector signed int vec_signedo (vector double);
3957 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
3958 instruction generator.
3959 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3960 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
3961 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
3963 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
3964 vunsignede_v2df): Add define_expands.
3965 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
3966 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
3967 VEC_UNSIGNEDO): Add definitions.
3968 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3969 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
3970 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
3971 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
3972 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
3973 * config/rs6000/altivec.h (vec_signed, vec_signed2,
3974 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
3975 vec_unsignede, vec_unsignedo): Add builtin defines.
3976 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
3978 * doc/extend.texi: Update the built-in documentation file for the
3979 new built-in functions.
3981 2017-06-29 Richard Biener <rguenther@suse.de>
3983 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
3984 reduction chains to LOOP_VINFO_REDUCTIONS.
3985 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
3986 SLP reductions after processing reduction chains.
3988 2017-06-29 Nathan Sidwell <nathan@acm.org>
3990 * builtins.c (fold_builtin_FUNCTION): Use
3991 lang_hooks.decl_printable_name.
3993 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
3996 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
3997 with only one label.
3998 * stmt.c (expand_case): Assert NCASES is greater than one.
4000 2017-06-29 Richard Biener <rguenther@suse.de>
4002 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
4004 (group_case_labels): Likewise.
4005 (find_taken_edge): Push sanity checking on val to workers...
4006 (find_taken_edge_cond_expr): ... here
4007 (find_taken_edge_switch_expr): ... and here, handle cases
4008 with just a default label.
4009 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
4010 (group_case_labels): Likewise.
4011 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
4012 group_case_labels does anything cleanup the CFG again.
4014 2017-06-29 Bin Cheng <bin.cheng@arm.com>
4016 PR tree-optimization/81196
4017 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
4018 exit condition comparing two IVs.
4020 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
4022 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
4023 profile to the dummy entry at the end of the list of architectures.
4024 * config/arm/arm-cpu-cdata.h: Regenerated.
4026 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4027 Michael Collison <michael.collison@arm.com>
4030 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
4032 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
4033 (*aarch64_reg_<mode>3_minus_mask): New pattern.
4034 (*aarch64_<optab>_reg_di3_mask2): New pattern.
4035 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
4036 of shift when the shift amount is masked with constant equal to
4037 the size of the mode.
4038 * config/aarch64/predicates.md (subreg_lowpart_operator): New
4041 2017-06-29 Martin Liska <mliska@suse.cz>
4043 * config/i386/i386.opt: Change range from [1,5] to [0,5].
4045 2017-06-29 Yury Gribov <tetra2005@gmail.com>
4048 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
4051 (edge_growth_cache_entry::edge_growth_cache_entry): New
4053 (reset_edge_growth_cache): Update to use constructor.
4055 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4057 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
4058 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
4059 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
4061 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
4063 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
4064 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
4066 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
4068 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
4069 (*-linux-uclibc*): Add t-uclibc tmake_file.
4070 * config/t-musl: New.
4071 * config/t-uclibc: New.
4073 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
4075 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
4077 (gen_comm_data): Emit architectural setting of arch_prof.
4078 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
4080 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
4081 (armv8-m.base, armv8-m.main): Likewise.
4082 * arm-protos.h (arm_build_target): Add profile field.
4083 (arch_option): Likewise.
4084 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
4086 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
4087 arm_active_target.profile.
4089 2017-06-28 Richard Biener <rguenther@suse.de>
4092 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
4093 TYPE_OVERFLOW_WRAPS.
4094 * match.pd (negate_expr_p): Likewise.
4095 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
4096 fold_build2, not fold_binary.
4098 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
4100 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
4101 Convert memory address to Pmode.
4102 (aarch64_print_operand): Assert MEM operands are always Pmode.
4104 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
4107 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
4108 Remove redundant if.
4109 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
4110 * config/arm/aarch-common-protos.h
4111 (aarch_forward_to_shift_is_not_shifted_re): Remove.
4112 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
4114 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
4117 * multiple_target.c (create_dispatcher_calls): Set the default
4118 clone to be static, not public.
4120 2017-06-28 Richard Biener <rguenther@suse.de>
4122 * tree-vect-loop.c (vectorizable_reduction): Move special
4123 cond reduction IV var creation ...
4124 (vect_create_epilog_for_reduction): ... here. Remove induction_index
4125 parameter. Use STMT_VINFO_VECTYPE.
4126 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
4129 2017-06-28 Martin Liska <mliska@suse.cz>
4132 * ipa-visibility.c (non_local_p): Handle visibility.
4134 2017-06-28 Martin Liska <mliska@suse.cz>
4137 * common.opt: Add IntegerRange to various options.
4138 * opt-functions.awk (integer_range_info): New function.
4139 * optc-gen.awk: Add integer_range_info to cl_options struct.
4140 * opts-common.c (decode_cmdline_option): Handle
4141 CL_ERR_INT_RANGE_ARG.
4142 (cmdline_handle_error): Likewise.
4143 * opts.c (print_filtered_help): Show valid interval in
4144 when --help is provided.
4145 * opts.h (struct cl_option): Add range_min and range_max fields.
4146 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
4148 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
4150 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
4151 (x * C EQ/NE y * C): New transformation.
4153 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
4155 * genmultilib (combination_space): Accept '+' in option names.
4157 2017-06-28 Martin Liska <mliska@suse.cz>
4160 * asan.c (instrument_derefs): Bail out inner references
4161 that are hard register variables.
4163 2017-06-28 Jakub Jelinek <jakub@redhat.com>
4166 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
4167 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
4169 2017-06-28 Richard Biener <rguenther@suse.de>
4171 * tree-vectorizer.h (vect_get_vec_defs): Remove.
4172 (vect_get_slp_defs): Adjust.
4173 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
4175 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
4177 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4178 get_initial_defs_for_reduction instead of vect_get_vec_defs.
4179 (vectorizable_reduction): Adjust.
4180 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
4182 (vect_get_slp_defs): Likewise.
4183 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
4184 (vectorizable_bswap): Adjust.
4185 (vectorizable_call): Likewise.
4186 (vectorizable_conversion): Likewise.
4187 (vectorizable_assignment): Likewise.
4188 (vectorizable_shift): Likewise.
4189 (vectorizable_operation): Likewise.
4190 (vectorizable_store): Likewise.
4191 (vectorizable_condition): Likewise.
4192 (vectorizable_comparison): Likewise.
4194 2017-06-28 Michael Collison <michael.collison@arm.com>
4197 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
4199 (arm_gen_movmemqi): Removed unused variable 'i'.
4200 Convert 'for' loop into 'while' loop.
4201 (arm_expand_prologue): Remove last unnecessary set of insn.
4202 (thumb_pop): Remove unused variable 'pushed_words'.
4203 (thumb_exit): Remove last unnecessary set of regs_to_pop.
4205 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4207 * config/s390/predicates.md: Use s390_rel_address_ok_p.
4208 * config/s390/s390-protos.h: Add prototype of
4209 s390_rel_address_ok_p.
4210 * config/s390/s390.c (s390_got_symbol): New function.
4211 (s390_rel_address_ok_p): New function.
4212 (legitimize_pic_address): Use s390_rel_address_ok_p.
4213 (s390_load_got): Use s390_got_symbol.
4214 (s390_option_override): Issue error if
4215 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
4216 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
4218 * config/s390/s390.opt: New option mpic-data-is-text-relative.
4220 2017-06-27 Andrew Pinski <apinski@cavium.com>
4222 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
4223 (X * copysign (1.0, X)): New pattern.
4224 (X * copysign (1.0, -X)): New pattern.
4225 (copysign (-1.0, CST)): New pattern.
4227 2017-06-27 Joseph Myers <joseph@codesourcery.com>
4229 * genmultilib (combination_space): Remove variable.
4230 Validate reuse rules against regular expression for any sequence
4231 of multilib options in any order.
4233 2017-06-27 Michael Collison <michael.collison@arm.com>
4235 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
4236 call aarch64_split_simd_combine.
4237 * (aarch64_combine_internal<mode>): Delete pattern.
4238 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
4239 Allow register and subreg operands.
4241 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4243 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
4244 specific need, just fallback on defaults.
4245 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
4247 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4248 Olivier Hainque <hainque@adacore.com>
4250 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
4252 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
4253 targets. Pick a default if no particular attempt applied.
4254 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
4257 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4259 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
4260 (x86_64-wrs-vxworks7): Likewise.
4262 2017-06-27 Marek Polacek <polacek@redhat.com>
4265 * ubsan.c (instrument_null): Check get_base_address's result for null.
4267 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
4269 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
4271 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
4273 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
4274 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
4275 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
4276 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
4278 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
4279 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
4280 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
4281 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
4282 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
4283 BUILT_IN_FEUPDATEENV): New builtins.
4284 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
4285 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
4286 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
4287 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
4289 (builtin_structptr_types): Adjust size.
4290 * tree.c (builtin_structptr_types): Add four entries.
4292 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4293 Olivier Hainque <hainque@adacore.com>
4295 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
4296 (TLS_SYM): New local macro, forcing reference to __tls__ on
4297 link command lines for VxWorks 7 RTPs, triggering initialization
4299 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
4300 OS features TLS support, true for RTPs on VxWorks 7.
4301 * config/vxworks.c (vxworks_override_options): Setup emutls
4304 2017-06-27 Jakub Jelinek <jakub@redhat.com>
4306 * predict.c (test_prediction_value_range): Use -1U instead of -1
4307 to avoid narrowing conversion warning.
4308 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
4309 to avoid narrowing conversion warning.
4310 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
4312 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
4314 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4316 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
4317 64bit configurations.
4318 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
4319 (SIZE_TYPE): Likewise.
4320 * config/vxworks.c (vxworks_emutls_var_fields): Use
4321 long_unsigned_type_node instead of unsigned_type_node as the offset
4322 field type, which is "pointer" mode in emutls.c.
4324 2017-06-27 Jakub Jelinek <jakub@redhat.com>
4327 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
4330 * gimple-fold.c (replace_call_with_call_and_fold): Handle
4331 gimple_vuse copying separately from gimple_vdef copying.
4333 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4335 * value-prof.c (free_hist): Remove call to memset and the enclosing if
4338 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
4339 Olivier Hainque <hainque@adacore.com>
4341 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
4342 for all vxworks7 targets.
4343 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
4344 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
4345 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
4346 variations for VX6/VX7 and 32/64bits later on in ...
4347 (VXWORKS_LIB_SPEC): Leverage new macros.
4348 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
4349 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
4351 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
4353 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
4354 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
4356 2017-06-26 Carl Love <cel@us.ibm.com>
4358 * config/rs6000/rs6000-c.c: Add support for built-in functions
4359 vector bool char vec_reve (vector bool char);
4360 vector signed char vec_reve (vector signed char);
4361 vector unsigned char vec_reve (vector unsigned char);
4362 vector bool int vec_reve (vector bool int);
4363 vector signed int vec_reve (vector signed int);
4364 vector unsigned int vec_reve (vector unsigned int);
4365 vector bool long long vec_reve (vector bool long long);
4366 vector signed long long vec_reve (vector signed long long);
4367 vector unsigned long long vec_reve (vector unsigned long long);
4368 vector bool short vec_reve (vector bool short);
4369 vector signed short vec_reve (vector signed short);
4370 vector double vec_reve (vector double);
4371 vector float vec_reve (vector float);
4372 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
4373 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
4374 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
4375 (altivec_vreve): New pattern.
4376 * config/rs6000/altivec.h (vec_reve): New define.
4377 * doc/extend.texi (vec_rev): Update the built-in documentation file
4378 for the new built-in functions.
4380 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4382 PR tree-optimization/71815
4383 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
4385 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
4387 (slsr_process_phi): Likewise.
4388 (replace_uncond_cands_and_profitable_phis): Don't replace a
4389 multiply candidate with a stride of 1 (copy or cast).
4390 (phi_incr_cost): Call uses_consumed_by_stmt rather than
4392 (lowest_cost_path): Likewise.
4393 (total_savings): Likewise.
4395 2017-06-26 Richard Biener <rguenther@suse.de>
4398 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
4399 Use def_builtin_pure for all gather builtins.
4401 2017-06-26 Richard Biener <rguenther@suse.de>
4403 PR tree-optimization/81203
4404 * tree-tailcall.c (find_tail_calls): Do not move stmts into
4407 2017-06-26 Marek Polacek <polacek@redhat.com>
4410 * doc/invoke.texi: Document -Wmultistatement-macros.
4412 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
4414 * doc/sourcebuild.texi (ARM-specific attributes): Document new
4415 arm_neon_ok_no_float_abi effective target.
4417 2017-06-26 Richard Biener <rguenther@suse.de>
4419 PR tree-optimization/80928
4420 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
4421 (copy_bbs): Set BB_DUPLICATED flag early.
4422 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
4424 (execute_on_shrinking_pred): Likewise.
4425 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
4426 BB_DUPLICATED blocks.
4427 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
4428 iterate over all PHIs considering removal of *gsi.
4430 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
4432 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
4435 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4437 * config/rs6000/rs6000-string.c: (expand_block_clear,
4438 do_load_for_compare, select_block_compare_mode,
4439 compute_current_alignment, expand_block_compare,
4440 expand_strncmp_align_check, expand_strn_compare,
4441 expand_block_move, rs6000_output_load_multiple)
4442 Move functions related to string/block move/compare
4444 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
4445 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
4446 for this function which is now used in two files.
4447 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
4448 * config.gcc: Add rs6000-string.o to extra_objs for
4449 targets powerpc*-*-* and rs6000*-*-*.
4451 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4454 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
4455 32-bit, since indexed is not valid for DImode.
4456 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
4457 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
4458 (define_peephole2 for Altivec d-form load): Add 32-bit support.
4459 (define_peephole2 for Altivec d-form store): Likewise.
4462 * multiple_target.c (create_dispatcher_calls): Only create the
4463 dispatcher call if the function is the default clone of a
4466 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
4469 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
4470 a call, force the call to not be a tail call.
4472 2017-06-23 Jeff Law <law@redhat.com>
4474 * doc/contrib.texi: Add entry for Steven Pemberton's work on
4477 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
4479 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
4480 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
4481 handling for early expansion of vector shifts (sl,sr,sra,rl).
4482 (builtin_function_type): Add vector shift right instructions
4483 to the unsigned argument list.
4485 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
4487 rtl-optimizatoin/79286
4488 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
4489 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
4490 trap. PIC register plus a const unspec without offset can never trap.
4492 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
4494 * tree.h (builtin_structptr_type): New type.
4495 (builtin_structptr_types): Declare new array.
4496 * tree.c (builtin_structptr_types): New array.
4497 (free_lang_data, build_common_tree_nodes): Use it.
4499 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
4502 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
4505 2017-06-22 Matt Turner <mattst88@gmail.com>
4507 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
4508 Lake models to skylake case. Assume skylake for unknown
4509 models with clflushopt.
4511 2017-06-22 Jeff Law <law@redhat.com>
4513 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
4514 frame sizes that do not satisfy aarch64_uimm12_shift.
4516 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
4518 * profile-count.h (apply_probability,
4519 apply_scale, probability_in): Fix checks for zero.
4521 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4523 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
4524 * doc/cppdiropts.texi (-I @var{dir}): Document it.
4526 2016-06-22 Richard Biener <rguenther@suse.de>
4528 * tree-vect-loop.c (vect_model_reduction_cost): Handle
4529 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
4530 REDUC_MAX_EXPR support.
4531 (vectorizable_reduction): Likewise.
4532 (vect_create_epilog_for_reduction): Likewise.
4534 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
4536 * match.pd (A / (1 << B) -> A >> B): New.
4537 * generic-match-head.c: Include optabs-tree.h.
4538 * gimple-match-head.c: Likewise.
4539 * optabs-tree.h (target_supports_op_p): New.
4540 * optabs-tree.c (target_supports_op_p): New.
4542 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4544 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
4545 $gcc_cv_ld --help output.
4546 (gcc_cv_ld_demangle): Likewise.
4547 (gcc_cv_ld_eh_frame_hdr): Likewise.
4548 (gcc_cv_ld_pie): Likewise.
4549 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
4550 (gcc_cv_ld_buildid): Likewise.
4551 (gcc_cv_ld_sysroot): Likewise.
4552 (ld_bndplt_support): Likewise.
4553 (ld_pushpopstate_support): Likewise.
4554 * configure: Regenerate.
4555 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
4557 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4560 * config/i386/sse.md (round<mode>2): Renumber match_dup and
4561 operands indexes to avoid gap between operands and match_dups.
4563 2017-06-21 Andrew Pinski <apinski@cavium.com>
4565 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
4566 Increment Arith_shift and Arith_shift_reg by 1.
4567 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
4569 * config/aarch64/aarch64.c (thunderx_tunings): Enable
4570 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
4571 (aarch64_strip_extend): Add new argument and test for it.
4572 (aarch64_cheap_mult_shift_p): New function.
4573 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
4574 add a cost if it is true.
4575 Update calls to aarch64_strip_extend.
4576 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
4578 2017-06-21 Andrew Pinski <apinski@cavium.com>
4580 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
4582 (thunderxt88): Likewise.
4583 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
4584 (thunderx_prefetch_tune): New variable.
4585 (thunderx2t99_prefetch_tune): Update for the correct values.
4586 (thunderxt88_tunings): New variable.
4587 (thunderx_tunings): Use thunderx_prefetch_tune instead of
4588 generic_prefetch_tune.
4589 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
4591 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4593 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
4594 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
4595 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
4596 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
4597 (aarch64_atomic_cas<mode>, GPI): Likewise.
4599 2017-06-21 Martin Liska <mliska@suse.cz>
4601 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
4602 statements on cold and hot labels.
4603 * predict.c (tree_estimate_probability_bb): Remove the
4604 prediction from this place.
4606 2017-06-21 Martin Liska <mliska@suse.cz>
4608 PR tree-optimization/79489
4609 * gimplify.c (maybe_add_early_return_predict_stmt): New
4611 (gimplify_return_expr): Call the function.
4612 * predict.c (tree_estimate_probability_bb): Remove handling
4614 * predict.def: Update comment about early return predictor.
4615 * gimple-predict.h (is_gimple_predict): New function.
4616 * predict.def: Change default value of early return to 66.
4617 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
4619 * passes.def: Put pass_strip_predict_hints to the beginning of
4622 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
4624 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
4625 FUNCTION_DECL declarations.
4626 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
4628 (dwaf2out_decl): Likewise.
4629 * godump.c (go_early_global_decl): Skip call to the real debug hook
4630 for FUNCTION_DECL declarations.
4631 * passes.c (rest_of_decl_compilation): Skip call to the
4632 early_global_decl debug hook for FUNCTION_DECL declarations, unless
4633 -fdump-go-spec is passed.
4635 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4637 * config/i386/i386.c (struct builtin_isa): New field pure_p.
4638 Reorder for compactness.
4639 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
4640 (def_builtin_pure, def_builtin_pure2): New functions.
4641 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
4643 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4645 * match.pd (nop_convert): New predicate.
4646 ((A +- CST1) +- CST2): Allow some NOP conversions.
4648 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4651 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
4652 with ctors/dtors if GOVD_SHARED is set.
4654 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4656 * config/aarch64/aarch64.md (movti_aarch64):
4657 Emit mov rather than orr.
4658 (movtf_aarch64): Likewise.
4659 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
4660 Emit mov rather than orr.
4662 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4664 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
4665 Swap alternatives, make integer dup more expensive.
4667 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4669 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
4670 Return true for non-tls symbols.
4672 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
4674 * config/aarch64/aarch64-cores.def (cortex-a55): New.
4675 (cortex-a75): Likewise.
4676 (cortex-a75.cortex-a55): Likewise.
4677 * config/aarch64/aarch64-tune.md: Regenerate.
4678 * doc/invoke.texi (-mtune): Document new values for -mtune.
4680 2017-06-21 Tom de Vries <tom@codesourcery.com>
4682 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
4684 (Effective-Target Keywords, Other attributes): Suggest using
4685 dg-add-options stack_size feature to get stack limit in stack_size
4686 effective target documentation.
4688 2017-06-21 Julian Brown <julian@codesourcery.com>
4689 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4691 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
4692 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
4693 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
4695 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
4696 attribute type list for neon_multiply.
4697 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
4698 attribute type list for neon_multiply.
4699 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
4700 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
4701 attribute type list for neon_multiply.
4702 * config/arm/types.md (crypto_pmull): Add.
4703 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
4704 attribute type list.
4706 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
4708 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
4711 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4713 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
4714 to make sure not to dereference a NULL cost_classes_ptr pointer.
4716 2017-06-20 Carl Love <cel@us.ibm.com>
4718 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4719 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4720 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4721 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4722 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4723 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4724 VMULOSW): New enum "unspec" values.
4725 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
4726 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
4727 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4728 altivec_vmulosw): New patterns.
4729 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4730 VMULOSW): Add definitions.
4732 2017-06-20 Julia Koval <julia.koval@intel.com>
4734 * config/i386/i386.c: Fix rounding expand for new pattern.
4735 * config/i386/subst.md: Fix pattern (parallel -> unspec).
4737 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4739 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
4740 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
4742 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4744 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
4747 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4749 * config/aarch64/aarch64-cores.def: Rearrange to sort by
4750 architecture, then by implementer ID.
4751 * config/aarch64/aarch64-tune.md: Regenerate.
4753 2017-06-20 Richard Biener <rguenther@suse.de>
4756 * fold-const.c (split_tree): Fold to type before negating.
4758 2017-06-20 David Malcolm <dmalcolm@redhat.com>
4760 * diagnostic-show-locus.c
4761 (selftest::test_fixit_deletion_affecting_newline): New function.
4762 (selftest::diagnostic_show_locus_c_tests): Call it.
4764 2017-06-20 Andreas Schwab <schwab@suse.de>
4767 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
4770 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
4772 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
4773 __ARM_FEATURE_COPROC according to support.
4775 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4777 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
4778 Rewritten to avoid overflow for > 32-bit pointers.
4781 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
4782 by removing enum keyword.
4783 (ubsan_type_descriptor): Likewise. Formatting fix.
4786 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
4787 splitter): Require TARGET_SSE2 in the condition.
4789 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
4792 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
4793 for doing vector set of SFmode on ISA 3.0.
4794 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
4795 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
4797 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
4798 SFmode value into a V4SF variable that was extracted from another
4799 V4SF variable without converting the element to double precision
4800 and back to single precision vector format.
4801 (vsx_insert_extract_v4sf_p9_2): Likewise.
4803 2017-06-19 Jakub Jelinek <jakub@redhat.com>
4805 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
4806 in UWHI to avoid undefined overflow.
4809 * ubsan.h (enum ubsan_encode_value_phase): New.
4810 (ubsan_encode_value): Change second argument to
4811 enum ubsan_encode_value_phase with default value of
4812 UBSAN_ENCODE_VALUE_GENERIC.
4813 * ubsan.c (ubsan_encode_value): Change second argument to
4814 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
4815 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
4816 create_tmp_var_raw instead of create_tmp_var and use a
4818 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
4819 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
4820 ubsan_encode_value callers.
4823 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
4824 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
4825 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
4827 2017-06-19 Richard Biener <rguenther@suse.de>
4830 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
4831 estimates if we changed anything.
4833 2017-06-19 Richard Biener <rguenther@suse.de>
4835 PR tree-optimization/80887
4836 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
4837 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
4838 simplified lookups, then reset mprts_hook.
4839 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
4841 (try_to_simplify): Likewise.
4843 2017-06-19 Martin Liska <mliska@suse.cz>
4846 * gimplify.c (gimplify_switch_expr):
4847 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
4849 2017-06-19 Martin Liska <mliska@suse.cz>
4851 * doc/install.texi: Document that PGO runs in 4 stages.
4853 2017-06-19 Martin Liska <mliska@suse.cz>
4856 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
4857 to dispatcher function name.
4858 * multiple_target.c (replace_function_decl): New function.
4859 (create_dispatcher_calls): Redirect both edges and references.
4861 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
4863 * profile-count.c (profile_count::dump): Dump quality.
4864 (profile_count::differs_from_p): Update for unsigned val.
4865 * profile-count.h (profile_count_quality): New enum.
4866 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
4868 2017-06-19 Richard Biener <rguenther@suse.de>
4870 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
4871 struct function as arg.
4872 (estimate_numbers_of_iterations): Export overload with loop arg.
4873 (free_numbers_of_iterations_estimates_loop): Use an overload of
4874 free_numbers_of_iterations_estimates instead.
4875 * tree-cfg.c (remove_bb): Adjust.
4876 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
4877 * tree-parloops.c (gen_parallel_loop): Likewise.
4878 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
4880 (tree_unroll_loops_completely): Likewise.
4881 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
4882 Use an overload instead and export.
4883 (estimated_loop_iterations): Adjust.
4884 (max_loop_iterations): Likewise.
4885 (likely_max_loop_iterations): Likewise.
4886 (estimate_numbers_of_iterations): Take struct function as arg
4888 (loop_exits_before_overflow): Adjust.
4889 (free_numbers_of_iterations_estimates_loop): Use an overload.
4890 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
4891 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
4893 2017-06-19 Richard Biener <rguenther@suse.de>
4896 * ipa-prop.c (find_constructor_constant_at_offset): Handle
4897 RANGE_EXPR conservatively.
4899 2017-06-16 Carl Love <cel@us.ibm.com>
4901 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4902 definitions for vec_float, vec_float2, vec_floato,
4903 vec_floate built-ins.
4904 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
4905 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
4907 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
4908 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
4909 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
4910 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
4911 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
4912 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
4913 vec_floato): Add builtin defines.
4914 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
4915 Update the built-in documentation file for the new built-in
4918 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4920 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
4921 (mthumb): Mark as the negative of -marm.
4923 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4925 * doc/invoke.texi (ARM Options, -mcpu): Document supported
4927 (ARM Options, -mtune): Document that this accepts the same
4928 extension options as -mcpu.
4929 (ARM Options, -mfpu): Document addition of -mfpu=auto.
4931 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4933 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
4934 permitted extensions.
4936 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4938 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
4939 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
4940 (armv8-m.main): Add option +nodsp.
4941 * config/arm/arm-cpu-cdata.h: Regenerated.
4943 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4945 * config/arm/t-fuchsia: New file.
4946 * config.gcc (arm*-*-fuchsia*): Use it.
4948 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4950 * config/arm/t-symbian: Rewrite for new option infrastructure.
4952 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4954 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
4955 (MULTILIB_REQUIRED): Likewise.
4957 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4959 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
4960 (MULTILIB_RESUE): Likewise.
4961 (MULTILIB_MATCHES): Likewise.
4962 (MULTLIB_REQUIRED): Likewise.
4964 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4966 * config/arm/t-rtems: Rewrite for new option framework.
4968 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4970 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
4971 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
4972 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
4973 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
4974 * config/arm/t-multilib: ... here.
4975 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
4976 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
4977 armv7-a and armv8*-a when A-profile libraries have not been built.
4978 * config/arm/t-rmprofile: Rewrite.
4980 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4982 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
4983 with a backslash. Remove the backslash after substituting unescaped
4985 * doc/fragments.texi (MULTILIB_REUSE): Document it.
4987 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4989 * config.gcc: (arm*-*-*): When building a-profile libraries, force
4990 the driver to pass through the default setting of -mfloat-abi.
4991 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
4993 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
4994 (all_feat_combs): New rule.
4995 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
4997 * config/arm/t-aprofile: Rewrite.
4999 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5001 * config/arm/arm.h (FPUTYPE_AUTO): Define.
5002 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
5003 fpu is not specified by the user/command-line.
5004 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
5005 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
5006 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
5007 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
5008 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
5009 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
5011 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5013 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
5014 * config/arm/t-arm-elf: Rewritten.
5016 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5018 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
5019 have some floating-point instructions.
5020 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
5021 (TARGET_MAYBE_HARD_FLOAT): New macro.
5022 * config/arm/arm-builtins.c (arm_init_builtins): Use
5023 TARGET_MAYBE_HARD_FLOAT.
5024 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
5026 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5028 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
5029 (configargs.h): Include it.
5030 (arm_print_hint_for_fpu_option): New function.
5031 (arm_parse_fpu_option): New function.
5032 (candidate_extension): New class.
5033 (arm_canon_for_multilib): New function.
5034 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
5035 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
5036 (ARCH_CANONICAL_SPECS): New macro.
5037 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
5039 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5041 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
5042 are set after handling multilib fragments. Set target_cpu_default2
5045 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5047 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
5049 (arm*-*-*): Set target_cpu_default2 to a quoted string.
5050 * config/arm/parsecpu.awk (check_cpu): Validate any extension
5052 (check_arch): Likewise.
5053 * config/arm/arm.c (arm_configure_build_target): Handle
5054 TARGET_CPU_DEFAULT being a string constant. Scan any feature
5055 options in the default.
5057 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5059 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
5060 when an option is an alias of another.
5061 * config/arm/parsecpu.awk (optalias): New parser token.
5062 (gen_comm_data): Mark non-alias options as such. Emit entries
5063 for extension aliases.
5064 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
5065 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
5066 (armv6kz, armv6zk, armv6t2): Likewise.
5067 (armv7): Make vfpv3-d16 an alias.
5068 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
5070 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
5071 Sort in canonical order.
5072 (armv8-a): Sort in canonical order.
5073 (armv8.1-a, armv8.2-a): Likewise.
5074 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
5076 (cortex-a9): Sort in canonical order.
5077 * config/arm/arm.c (selftests.h): Include it.
5078 (arm_test_cpu_arch_data): New function.
5079 (arm_run_self_tests): New function.
5080 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
5081 (targetm): Move declaration to the end of the file.
5082 * arm-cpu-cdata.h: Regenerated.
5084 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5086 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
5087 call to target_mode_check describing the type of option passed.
5088 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
5089 (arm_target_thumb_only): Use arm_parse_arch_option_name or
5090 arm_parse_cpu_option_name to match parameters against list of
5092 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
5093 arm_arch_core_flags data structure.
5094 * config/arm/arm-cpu_cdata.h: Regenerated.
5096 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5098 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
5100 (arm_print_hint_for_cpu_option): Likewise.
5101 (arm_print_hint_for_arch_option): Likewise.
5102 (arm_parse_cpu_option_name): Likewise.
5103 (arm_parse_arch_option_name): Likewise.
5104 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
5105 of entries in the all_fpus list.
5106 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
5107 (arm_parse_cpu_option_name): Declare.
5108 (arm_parse_arch_option_name): Declare.
5109 (arm_parse_option_features): Declare.
5110 (arm_intialize_isa): Declare.
5111 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
5113 (gen_comm_data): ... here. Make definitions non-static.
5114 * config/arm/arm-cpu-data.h: Regenerated.
5115 * config/arm/arm-cpu-cdata.h: Regenerated.
5117 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5119 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
5120 (cpu_arch_extension): New structure.
5121 (cpu_arch_option, arch_option, cpu_option): New structures.
5122 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
5124 (gen_data): Generate new format data tables.
5125 * config/arm/arm.c (cpu_tune): New structure.
5126 (cpu_option, processors): Delete.
5127 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
5128 (arm_print_hint_for_cpu_option): ... this and ...
5129 (arm_print_hint_for_arch_option): ... this.
5130 (arm_parse_arch_cpu_name): Delete. Replace with ...
5131 (arm_parse_cpu_option_name): ... this and ...
5132 (arm_parse_arch_option_name): ... this.
5133 (arm_unrecognized_feature): Change type of target parameter to
5135 (arm_parse_arch_cpu_features): Delete. Replace with ...
5136 (arm_parse_option_features): ... this.
5137 (arm_configure_build_target): Rework to use new configuration data
5139 (arm_print_tune_info): Rework for new configuration data tables.
5140 * config/arm/arm-cpu-data.h: Regenerated.
5141 * config/arm/arm-cpu.h: Regenerated.
5143 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5145 * Makefile.in (OBJS): Move sbitmap.o from here ...
5146 (OBJS-libcommon): ... to here.
5148 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5150 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
5151 (ISA_ALL_CRYPTO): New macro.
5152 (ISA_ALL_SIMD): New macro
5153 (ISA_ALL_FP): New macro.
5154 * config/arm/arm.c (fpu_bitlist): Update initializer.
5155 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
5157 (arm9e): Add fpu. Add option for nofp
5158 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
5159 (arm926ej-s, arm1026ej-s): Likewise.
5160 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
5161 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
5162 neon-fp16, neon-vfpv4, nofp and nosimd.
5163 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
5164 (cortex-a8): Add fpu. Add option for nofp.
5165 (cortex-a9): Add fpu. Add options for nosimd and nofp.
5166 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
5167 (cortex-r4f): Add fpu.
5168 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
5169 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
5171 (cortex-r8): Likewise.
5172 (cortex-m4): Add fpu. Add option for nofp.
5173 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
5174 (cortex-a17.cortex-a7): Likewise.
5175 (cortex-a32): Add fpu. Add options for crypto and nofp.
5176 (cortex-a35, cortex-a53): Likewise.
5177 (cortex-a57): Add fpu. Add option for crypto.
5178 (cortex-a72, cortex-a73): Likewise.
5179 (exynos-m1): Likewise.
5180 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
5181 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
5182 (cortex-m33): Add fpu. Add option for nofp.
5183 * config/arm/arm-cpu-cdata.h: Regenerated
5184 * config/arm/arm-cpu-data.h: Regenerated.
5186 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5188 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
5189 (armv5te, armv5tej): Likewise.
5190 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
5191 (armv7): Add options fp and vfpv3-d16.
5192 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
5193 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
5195 (armv7ve): Likewise.
5196 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
5197 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
5198 (armv8-a): Add nocrypto option.
5199 (armv8.1-a, armv8.2-a): Likewise.
5200 (armv8-m.main): add options fp, fp.dp and nofp.
5202 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5204 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
5206 (armv8-a+crc): Delete.
5207 (armv8.1-a): Add options simd, crypto and nofp.
5208 (armv8.2-a): Add options fp16, simd, crypto and nofp.
5209 (armv8.2-a+fp16): Delete.
5210 (armv8-m.main): Add option dsp.
5211 (armv8-m.main+dsp): Delete.
5212 (cortex-a8): Add fpu. Add nofp option.
5213 (cortex-a9): Add fpu. Add nofp and nosimd options.
5214 * config/arm/parsecpu.awk (gen_data): Generate option tables and
5215 link to main cpu and architecture data structures.
5216 (gen_comm_data): Only put isa attributes from the main architecture
5218 (option): New statement for architecture and CPU entries.
5219 * arm.c (struct cpu_option): New structure.
5220 (struct processors): Add entry for options.
5221 (arm_unrecognized_feature): New function.
5222 (arm_parse_arch_cpu_name): Ignore any characters after the first
5224 (arm_parse_arch_cpu_feature): New function.
5225 (arm_configure_build_target): Separate out any CPU and architecture
5226 features and parse separately. Don't error out if -mfpu=auto is
5227 used with only an architecture string.
5228 (arm_print_asm_arch_directives): New function.
5229 (arm_file_start): Call it.
5230 * config/arm/arm-cpu-cdata.h: Regenerated.
5231 * config/arm/arm-cpu-data.h: Likewise.
5232 * config/arm/arm-tables.opt: Likewise.
5234 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5236 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
5237 assembler when it is not -mfpu=auto.
5239 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5241 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
5242 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
5243 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
5244 (ASM_CPU_SPEC): Rewrite.
5245 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
5246 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
5247 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
5248 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
5249 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
5250 copied string is NUL-terminated. Also strip any characters prefixed
5252 (arm_rewrite_selected_arch): New function.
5253 (arm_rewrite_march): New function.
5255 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5257 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
5258 (x_arm_cpu_string, x_arm_tune_string): Likewise.
5259 (march, mcpu, mtune): Convert to string-based options.
5260 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
5261 (arm_parse_arch_cpu_name): New function.
5262 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
5263 identify selected architecture or CPU.
5264 (arm_option_save): New function.
5265 (TARGET_OPTION_SAVE): Redefine.
5266 (arm_option_restore): Restore string options.
5267 (arm_option_print): Print string options.
5269 2017-06-16 Martin Sebor <msebor@redhat.com>
5271 PR tree-optimization/80933
5272 PR tree-optimization/80934
5273 * builtins.c (fold_builtin_3): Do not handle bcmp here.
5274 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
5275 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
5276 (gimple_fold_builtin): Call them.
5278 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5280 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
5281 as unlikely; update profile.
5283 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5285 * predict.c (force_edge_cold): Handle declaring edges impossible
5288 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5290 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
5292 (try_unroll_loop_completely): Fix reporting.
5294 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5296 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
5298 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
5301 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
5302 if given a non-constant argument for an intrinsic which requires a
5305 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5307 * profile.c (compare_freqs): New function.
5308 (branch_prob): Sort edge list.
5309 (find_spanning_tree): Assume that the list is priority sorted.
5311 2017-06-16 Richard Biener <rguenther@suse.de>
5313 PR tree-optimization/81090
5314 * passes.def (pass_record_bounds): Remove.
5315 * tree-pass.h (make_pass_record_bounds): Likewise.
5316 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
5317 make_pass_record_bounds): Likewise.
5318 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
5319 not free niter estimates at the beginning but at the end.
5320 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
5322 2017-06-16 Richard Biener <rguenther@suse.de>
5324 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
5325 initializer to workaround ICE in host GCC 4.8.
5327 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5329 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
5331 (clone_inlined_nodes): Update.
5333 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5335 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
5336 prefetch settings, and enable prefetching by default at -O3.
5338 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5340 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5341 Set flag_prefetch_loop_arrays according to tuning data.
5343 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5345 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
5347 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
5348 [Unrelated to main purpose of the patch] Place the pointer field last
5349 to enable type checking errors when tune structure are wrongly merged.
5350 * config/aarch64/aarch64.c (generic_prefetch_tune,)
5351 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
5352 (thunderx2t99_prefetch_tune): New tune constants.
5353 (tune_params *_tunings): Update all tunings (no functional change).
5354 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
5355 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
5356 from tunings structures.
5358 2017-06-16 Jakub Jelinek <jakub@redhat.com>
5361 * ubsan.c (instrument_null): Add T argument, use it instead
5362 of computing it based on IS_LHS.
5363 (instrument_object_size): Likewise.
5364 (pass_ubsan::execute): Adjust instrument_null and
5365 instrument_object_size callers to pass gimple_get_lhs or
5366 gimple_assign_rhs1 result to it. Use instrument_null instead of
5367 calling get_base_address and instrument_mem_ref. Handle
5368 aggregate call arguments for object-size sanitization.
5370 2017-06-16 Yury Gribov <tetra2005@gmail.com>
5372 PR tree-optimization/81089
5373 * tree-vrp.c (is_masked_range_test): Validate operands of
5376 2017-06-15 Martin Sebor <msebor@redhat.com>
5379 * dumpfile.c (dump_register): Avoid calling memset to initialize
5380 a class with a default ctor.
5381 * gcc.c (struct compiler): Remove const qualification.
5382 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
5383 * hash-table.h: Ditto.
5384 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
5386 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
5387 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
5389 * params.h (struct param_info): Make struct members non-const.
5390 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
5391 with default initialization.
5392 * vec.h (vec_copy_construct, vec_default_construct): New helper
5394 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
5395 with vec_copy_construct.
5396 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
5397 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
5398 * doc/invoke.texi (-Wclass-memaccess): Document.
5400 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5402 * emit-rtl.h (is_leaf): Update comment about local
5405 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5408 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
5409 for a variable to have a section before checking if the section has a
5411 Set section to.persistent if persistent attribute is set.
5412 Warn if .persistent attribute is used on an automatic variable.
5414 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
5416 PR rtl-optimization/80474
5417 * reorg.c (update_block): Do not ignore instructions in a delay slot.
5419 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
5421 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
5424 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
5426 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
5427 (casesi): Emit bounds checking as RTL.
5428 (casesi_internal_mips16_<mode>): Remove bounds checking.
5430 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
5432 * config/xtensa/xtensa.c (xtensa_option_override): Append
5433 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
5434 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
5435 xtensa_doloop_hooks): Define unconditionally.
5436 (xtensa_reorg_loops): Only call reorg_loops in the presence of
5438 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
5439 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
5440 for it in xtensa_option_override.
5441 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
5442 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
5444 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
5446 * doc/cppopts.texi: Document '-' special value to -MF.
5448 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
5450 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
5451 (cortex_a53_fconst): Likewise.
5452 (cortex_a53_fpmul): Likewise.
5453 (cortex_a53_f_load_64): Likewise.
5454 (cortex_a53_f_load_many): Likewise.
5455 (cortex_a53_advsimd_alu): Likewise.
5456 (cortex_a53_advsimd_alu_q): Likewise.
5457 (cortex_a53_advsimd_mul): Likewise.
5458 (cortex_a53_advsimd_mul_q): Likewise.
5459 (fpmac bypass): Add new bypass for fpmac-fpmac case.
5460 Add missing fmul, r2f_cvt and fconst cases.
5462 2017-06-14 Richard Biener <rguenther@suse.de>
5465 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
5467 (fold_binary_loc): When associating do not treat pre-existing
5468 TREE_OVERFLOW on literal constants as a reason to allow
5469 TREE_OVERFLOW on associated literal constants.
5471 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5473 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
5474 (MASK_FEATURES): New macro.
5475 * config/sparc/sparc.c (sparc_option_override): Remove the special
5476 handling of -mfpu and generalize it to all MASK_FEATURES switches.
5478 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5480 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
5481 a division of 0 if non-call exceptions are enabled.
5483 2017-06-14 Andrew Pinski <apinski@cavium.com>
5484 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5487 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
5488 Improve vector initialization code gen for only variable case.
5490 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5492 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
5494 2017-06-14 Richard Biener <rguenther@suse.de>
5496 PR tree-optimization/81083
5497 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
5500 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5502 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
5503 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
5504 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
5505 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
5506 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
5507 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
5509 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5511 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
5512 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
5514 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5516 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
5518 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5520 * config/rs6000/t-rtems: Don't handle SPE.
5522 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5524 * config/rs6000/t-linux: Don't handle SPE.
5526 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5528 * config/rs6000/eabispe.h: Delete file.
5530 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5532 * config/rs6000/t-spe: Delete file.
5534 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5536 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
5537 (rs6000_legitimate_offset_address_p): Return false for anything in
5538 V2SImode or V2SFmode.
5540 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5542 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
5543 except V2SF and V2SI. Rearrange the vector modes, and add comments.
5544 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
5546 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
5547 (rs6000_legitimate_offset_address_p): Ditto.
5548 (rs6000_emit_move): Ditto.
5549 (rs6000_init_builtins): Remove V4HI_type_node.
5551 2017-06-13 Martin Liska <mliska@suse.cz>
5554 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
5555 (gate_asan): Likewise.
5556 * asan.h (asan_no_sanitize_address_p): Remove the function.
5557 (sanitize_flags_p): New function.
5558 * builtins.def: Fix coding style.
5559 * common.opt: Use renamed enum value.
5560 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
5561 * doc/extend.texi: Document no_sanitize attribute.
5562 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
5563 to SANITIZE_UNDEFINED_NONDEFAULT.
5564 * gcc.c (sanitize_spec_function): Use the renamed enum value.
5565 * gimple-fold.c (optimize_atomic_compare_exchange_p):
5566 Use sanitize_flags_p.
5567 * gimplify.c (gimplify_function_tree): Likewise.
5568 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
5569 * opts.c (parse_no_sanitize_attribute): New function.
5570 (common_handle_option): Use renamed enum value.
5571 * opts.h (parse_no_sanitize_attribute): Declare.
5572 * tree.c (sanitize_flags_p): New function.
5573 * tree.h: Declared here.
5574 * tsan.c: Use sanitize_flags_p.
5575 * ubsan.c (ubsan_expand_null_ifn): Likewise.
5576 (instrument_mem_ref): Likewise.
5577 (instrument_bool_enum_load): Likewise.
5578 (do_ubsan_in_current_function): Remove the function.
5579 (pass_ubsan::execute): Use sanitize_flags_p.
5580 * ubsan.h: Remove do_ubsan_in_current_function
5581 * tree-cfg.c (print_no_sanitize_attr_value): New function.
5582 (dump_function_to_file): Use it here.
5584 2017-06-13 Martin Jambor <mjambor@suse.cz>
5586 PR tree-optimization/80803
5587 PR tree-optimization/81063
5588 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
5589 (propagate_subaccesses_across_link): Enqueue subtree whenever
5590 necessary instead of relying on the caller.
5592 2017-06-13 Martin Jambor <mjambor@suse.cz>
5594 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
5595 that have a first_link.
5596 (sort_and_splice_var_accesses): Do not check first_link before
5598 (subtree_mark_written_and_enqueue): Likewise.
5599 (propagate_all_subaccesses): Likewise and do not stop at first
5600 parent with a first_link.
5602 2017-06-13 Martin Jambor <mjambor@suse.cz>
5604 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
5607 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5609 * match.pd: New pattern.
5611 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5613 * tree-vrp.c (is_masked_range_test): New function.
5614 (register_edge_assert_for): Determine ranges for
5617 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5619 PR tree-optimization/67328
5620 * fold-const.c (maskable_range_p): New function.
5621 (build_range_check): Generate bittests if possible.
5623 2017-06-13 Martin Liska <mliska@suse.cz>
5625 * gimple-pretty-print.c (dump_probability): Add new argument.
5626 (dump_edge_probability): Dump both probability and count.
5627 (dump_gimple_label): Likewise.
5628 (dump_gimple_bb_header): Likewise.
5630 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
5633 * config/avr/avr-devices.c: Fix indentation.
5634 * config/avr/gen-avr-mmcu-specs.c: Dito.
5636 2017-06-13 Richard Biener <rguenther@suse.de>
5638 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
5639 instead get vector type from stmt_info.
5640 (vectorizable_reduction): Adjust. Remove dead code.
5642 2017-06-13 Richard Biener <rguenther@suse.de>
5645 * fold-const.c (extract_muldiv_1): Remove bogus distribution
5646 case of C * (x * C2 + C3).
5647 (fold_addr_of_array_ref_difference): Properly fold index difference.
5649 2017-06-12 David S. Miller <davem@davemloft.net>
5652 * config/sparc/sparc.md (return expander): Emit frame blockage if
5653 function uses alloca.
5655 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
5657 * combine.c (make_field_assignment): Check len rather than the mode
5658 precision when calling force_to_mode.
5660 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
5662 Support multilibs and devices that see flash in RAM address range.
5665 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
5666 (avr_mcu_t) <flash_pm_offset>: New field.
5667 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
5668 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
5669 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
5670 (AVR_TINY_PM_OFFSET): Remove macro.
5671 * config/avr/avr.opt (-mshort-calls): New option.
5672 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
5673 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
5674 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
5675 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
5676 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
5677 instead of avr_arch->have_jmp_call.
5678 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
5679 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
5680 avr_arch->flash_pm_offset to define.
5681 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
5682 new field flash_pm_offset. Add entry for avrxmega3.
5683 (avr_texinfo): Add entry for avrxmega3.
5684 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
5685 attiny212, attiny214,
5686 attiny412, attiny414, attiny416, attiny417,
5687 attiny814, attiny816, attiny817,
5688 attiny1614, attiny1616, attiny1617,
5689 attiny3214, attiny3216, attiny3217.
5690 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
5691 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
5692 (avr_print_operand_address) [AVR_TINY]: Same.
5693 (avr_asm_init_sections) <readonly_data_section>: Only patch
5694 callback if avr_arch->flash_pm_offset = 0.
5695 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
5696 for rodata if avr_arch->flash_pm_offset != 0.
5697 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
5698 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
5699 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
5700 (m_options): Append opt_rcall.
5701 (m_dirnames): Append dir_rcall.
5702 * config/avr/t-multilib: Regenerate.
5704 * configure.ac [target=avr]: Check whether avrxmega3 default
5705 linker description file works as needed.
5706 * configure: Regenerate.
5707 * doc/avr-mmcu.texi: Regenerate.
5708 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
5709 <__AVR_ARCH__>: Document avrxmega3 and 103.
5710 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
5711 <__AVR_SHORT_CALLS__>: Document it.
5712 <__AVR_PM_BASE_ADDRESS__>: Document it.
5713 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
5714 (AVR Variable Attributes) <progmem>: Document this is
5715 not needed for avrxmega3.
5716 (AVR Named Address Spaces) <__flash>: Dito.
5718 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
5720 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
5722 2017-06-12 Doug Rupp <rupp@adacore.com>
5724 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
5725 Append vxworks-stdint.h to the tm_file list.
5726 * config/vxworks-stdint.h: New file.
5728 2017-06-12 Martin Liska <mliska@suse.cz>
5730 PR tree-optimization/81041
5731 * tree-profile.c (gimple_gen_ic_func_profiler):
5732 Create an extra BB in profile-generate
5733 (gimple_gen_time_profiler): Likewise.
5735 2017-06-12 Jakub Jelinek <jakub@redhat.com>
5737 PR tree-optimization/81003
5738 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
5739 (update_range_test): Use it instead of force_gimple_operand_gsi.
5741 2017-06-12 Richard Biener <rguenther@suse.de>
5743 PR tree-optimization/81053
5744 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
5745 with backedge value not defined in loop. Simplify def stmt
5748 2017-06-11 Tom de Vries <tom@codesourcery.com>
5751 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
5753 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
5754 nvptx_cannot_force_const_mem.
5756 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5758 * opts.c (finish_options): Move test for flag_split_stack after
5759 it has been initialized.
5761 2017-06-11 Jason Merrill <jason@redhat.com>
5763 * tree.h (id_equal): New.
5764 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
5765 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
5766 instead of strcmp of IDENTIFIER_POINTER.
5768 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5770 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
5771 (mark_all_inlined_calls_cdtor): Fix formating.
5772 (inline_transform): Rescale profile before inlining.
5774 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5776 * cgraph.h (cgraph_edge::clone): Update prototype.
5777 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
5778 (cgraph_node::create_clone): Update.
5779 (cgraph_node::create_version_clone): Update.
5780 * tree-inline.c (copy_bb): Update.
5781 (expand_call_inline): Update.
5783 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5785 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
5786 factored out from ...
5787 (rs6000_emit_prologue): ... here.
5789 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5791 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
5792 factored out from ...
5793 (rs6000_emit_prologue): ... here.
5795 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5797 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
5799 (handle_missing_profiles): Fix computation of tp_first_run.
5800 (counts_to_freqs): Do not touch freqs when count is 0.
5802 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5804 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
5807 2017-06-10 Tom de Vries <tom@codesourcery.com>
5809 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
5810 attributes): Document signal effective target.
5812 2017-06-10 Tom de Vries <tom@codesourcery.com>
5814 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
5815 Document effective target stack_size.
5817 2017-06-09 David Malcolm <dmalcolm@redhat.com>
5819 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
5820 to the edit_context if they can be auto-applied.
5822 2017-06-9 Ian Lance Taylor <iant@golang.org>
5824 * opts.c (finish_options): If -fsplit-stack, disable implicit
5825 -forder-blocks-and-partition.
5826 * doc/invoke.texi (Optimize Options): Document that when using
5827 -fsplit-stack -forder-blocks-and-partition is not implicitly
5830 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5832 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
5833 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
5834 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
5835 * builtins.def (abort, trap, unreachable): Declare cold.
5836 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
5837 * tree-core.h (ECF_COLD): New.
5838 * tree.c (set_call_expr_flags): Handle ECF_COLD.
5839 (build_common_builtin_nodes): Mark unreachable and abort as cold.
5841 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5843 * predict.c (unlikely_executed_stmt_p): Cleanup.
5845 2017-06-09 Richard Biener <rguenther@suse.de>
5847 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
5848 model if the ref is always written to.
5850 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5852 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
5854 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5856 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
5859 2017-06-09 Tom de Vries <tom@codesourcery.com>
5862 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
5863 "target cannot support label values" when encountering LABEL_REF.
5865 2017-06-09 Martin Liska <mliska@suse.cz>
5867 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
5868 (gimple_gen_ic_func_profiler): Emit direct comparison
5869 of __gcov_indirect_call_callee with NULL.
5870 (gimple_gen_time_profiler): Change probability from
5871 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
5873 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5875 * profile.c (edge_gcov_counts): Turn to pointer.
5876 (compute_branch_probabilities, compute_branch_probabilities): Update.
5877 (branch_prob): Do not clear edge_gcov_count.
5878 * profile.h (edge_gcov_counts): Turn to pointer.
5879 (edge_gcov_count): Update.
5881 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5883 * gimple.h (gimple_check_failed): Mark cold.
5885 2017-06-09 Richard Biener <rguenther@suse.de>
5887 PR tree-optimization/66623
5888 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
5889 refactor check_reduction into two parts, properly computing
5890 whether we have to check reduction validity for outer loop
5893 2017-06-09 Richard Biener <rguenther@suse.de>
5895 PR tree-optimization/79483
5896 * graphite-scop-detection.c (order): New global.
5897 (get_order): Compute bb to order mapping that satisfies code
5898 generation constraints.
5899 (cmp_pbbs): New helper.
5900 (build_scops): Start domwalk at entry block, sort generated
5903 2017-06-09 Richard Biener <rguenther@suse.de>
5906 * ipa-polymorphic-call.c
5907 (ipa_polymorphic_call_context::restrict_to_inner_class):
5908 Skip FIELD_DECLs with error_mark_node type.
5909 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
5912 2017-06-09 Martin Liska <mliska@suse.cz>
5914 * predict.c (struct branch_predictor): New struct.
5915 (test_prediction_value_range): New test.
5916 (predict_c_tests): New function.
5917 * selftest-run-tests.c (selftest::run_tests): Run the function.
5918 * selftest.h: Declare new tests.
5920 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
5923 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
5924 gen_add3_insn did not fail.
5925 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
5926 r0, construct that number in a temporary reg and add that reg to r0.
5927 If asked to put the result in r0 as well, fail.
5929 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
5931 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5932 for early expansion of vec_eqv.
5934 2017-06-08 Jakub Jelinek <jakub@redhat.com>
5937 * ubsan.c (instrument_null): Avoid pointless code temporary.
5938 (pass_ubsan::execute): Instrument aggregate arguments of calls.
5940 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
5944 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
5947 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
5950 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5952 * predict.c (unlikely_executed_edge_p): Move ahead.
5953 (probably_never_executed_edge_p): Use it.
5955 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5958 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
5959 gimple_call_builtin_p call.
5961 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5963 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
5964 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
5965 rtl_check_failed_type2, rtl_check_failed_code1,
5966 rtl_check_failed_code2, rtl_check_failed_code_mode,
5967 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
5968 rtvec_check_failed_bounds, rtl_check_failed_flag,
5969 _fatal_insn_not_found, _fatal_insn): Likewise.
5970 * tree.h (tree_contains_struct_check_failed,
5971 tree_check_failed, tree_not_check_failed,
5972 tree_class_check_failed, tree_range_check_failed,
5973 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
5974 tree_vec_elt_check_failed, phi_node_elt_check_failed,
5975 tree_operand_check_failed, omp_clause_check_failed,
5976 omp_clause_operand_check_failed, omp_clause_range_check_failed):
5979 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5981 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
5982 flag_branch_probabilities.
5983 * ipa-inline.c (edge_badness): Likewise.
5984 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
5985 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
5986 * predict.c (maybe_hot_frequency_p): Likewise.
5987 (probably_never_executed): Likewise.
5988 * sched-ebb.c (schedule_ebbs): Likewise.
5989 * sched-rgn.c (find_single_block_region): Likewise.
5990 * tracer.c (tail_duplicate): Likewise.
5992 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5994 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
5995 longer requires x_flag_profile_use.
5997 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5999 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
6000 instead of flag_reorder_blocks_and_partition.
6001 * dbxout.c (dbxout_function_end): Likewise.
6002 * dwarf2out.c (gen_subprogram_die): Likewise.
6003 * haifa-sched.c (sched_create_recovery_edges): Likewise.
6004 * hw-doloop.c (reorg_loops): Likewise.
6005 * varasm.c (assemble_start_function,
6006 assemble_end_function): Likewise.
6007 (decide_function_section): Do not check for
6008 flag_reorder_blocks_and_partition.
6010 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
6012 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
6014 (chkp_get_hard_register_fake_addr_expr): Ditto.
6015 (chkp_build_addr_expr): Add check for hard reg case.
6016 (chkp_parse_array_and_component_ref): Ditto.
6017 (chkp_find_bounds_1): Ditto.
6018 (chkp_process_stmt): Don't generate bounds store for
6021 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6023 * predict.c (maybe_hot_bb_p): Do not check profile status.
6024 (maybe_hot_edge_p): Likewise.
6025 (probably_never_executed): Check for zero counts even if profile
6027 (unlikely_executed_edge_p): New function.
6028 (unlikely_executed_stmt_p): New function.
6029 (unlikely_executed_bb_p): New function.
6030 (set_even_probabilities): Use unlikely predicates.
6031 (combine_predictions_for_bb): Likewise.
6032 (predict_paths_for_bb): Likewise.
6033 (predict_paths_leading_to_edge): Likewise.
6034 (determine_unlikely_bbs): New function.
6035 (estimate_bb_frequencies): Use it.
6036 (compute_function_frequency): Use zero counts even if profile is
6038 * profile-count.h: Fix typo.
6040 2017-08-08 Julia Koval <julia.koval@intel.com>
6042 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
6043 _mm512_mask_cvtsepi16_storeu_epi8,
6044 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
6045 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
6046 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
6047 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
6048 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
6049 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
6050 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
6051 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
6052 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
6053 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
6054 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
6055 __builtin_ia32_pmovuswb256mem_mask,
6056 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
6057 __builtin_ia32_pmovwb512mem_mask): New builtins.
6059 2017-08-08 Julia Koval <julia.koval@intel.com>
6061 PR target/73350,80862
6062 * config/i386/subst.md (round): Fix round pattern.
6063 * config/i386/i386.c (ix86_erase_embedded_rounding):
6064 Fix erasing rounding for the fixed pattern.
6066 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6068 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
6070 2017-06-08 Martin Liska <mliska@suse.cz>
6072 PR gcov-profile/80911
6073 * gcov.c (block_info::block_info): New constructor.
6075 2017-06-07 Carl Love <cel@us.ibm.com>
6077 * config/rs6000/rs6000-c: The return type of the following
6078 built-in functions was implemented as int not long long. Fix sign
6079 of return value for the unsigned version of vec_mulo and vec_mule.
6080 vector unsigned long long vec_bperm (vector unsigned long long,
6081 vector unsigned char)
6082 vector signed long long vec_mule (vector signed int,
6084 vector unsigned long long vec_mule (vector unsigned int,
6085 vector unsigned int)
6086 vector signed long long vec_mulo (vector signed int,
6088 vector unsigned long long vec_mulo (vector unsigned int,
6089 vector unsigned int)
6090 * doc/extend.texi: Fix the documentation for the built-in
6093 2017-06-07 Carl Love <cel@us.ibm.com>
6096 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
6099 2017-06-07 Carl Love <cel@us.ibm.com>
6101 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
6102 support, Generate doublehv for signed int/float for BE case only.
6104 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
6106 * doc/invoke.texi (mcx16): Rewrite.
6108 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6110 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
6111 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
6112 *mov<mode>_softfloat, and an anonymous splitter): Use
6113 nonimmediate_operand instead of rs6000_nonimmediate_operand.
6115 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6117 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
6119 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
6120 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
6121 (rs6000_debug_reg_global): Adjust.
6122 (rs6000_init_hard_regno_mode_ok): Adjust.
6123 (rs6000_dbx_register_number): Adjust.
6124 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
6125 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
6126 Remove SPE_ACC and SPEFSCR.
6127 (REG_ALLOC_ORDER): Ditto.
6128 (FRAME_POINTER_REGNUM): Change to 111.
6129 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
6130 (REG_CLASS_NAMES): Ditto.
6131 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
6132 (REGISTER_NAMES): Ditto.
6133 (ADDITIONAL_REG_NAMES): Ditto.
6134 (rs6000_reg_names): Ditto.
6135 * config/rs6000/rs6000.md: Renumber some register number
6138 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6140 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
6142 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
6143 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
6145 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
6146 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
6147 Delete the SPE high registers.
6148 (REG_ALLOC_ORDER): Ditto.
6149 (enum reg_class): Remove SPE_HIGH_REGS.
6150 (REG_CLASS_NAMES): Ditto.
6151 (REG_CLASS_CONTENTS): Delete the SPE high registers.
6152 (REGISTER_NAMES): Ditto.
6153 (rs6000_reg_names): Ditto.
6154 * doc/tm.texi.in: Remove SPE as example.
6155 * doc/tm.texi: Regenerate.
6157 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6159 * config/rs6000/8540.md (ppc8540_brinc): Delete.
6160 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
6161 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
6162 * config/rs6000/rs6000.md (type): Remove "brinc".
6164 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6166 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
6167 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
6168 * config/rs6000/linuxspe.h: Delete file.
6169 * config/rs6000/rs6000.md: Don't include spe.md.
6170 * config/rs6000/spe.h: Delete file.
6171 * config/rs6000/spe.md: Delete file.
6172 * config/rs6000/t-rs6000: Remove spe.md.
6174 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6176 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
6177 (reg_or_none500mem_operand): Delete.
6178 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
6179 instead of reg_or_none500mem_operand.
6181 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6183 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
6184 handling of SPE flags.
6185 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
6187 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6189 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
6191 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
6192 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
6193 paired_divv2sf3): Similar.
6194 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
6195 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
6196 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
6198 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
6199 Rename the paired_* instruction patterns.
6200 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
6202 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
6203 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
6204 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
6205 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
6206 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
6208 (struct machine_function): Delete field spe_insn_chain_scanned_p.
6209 (spe_func_has_64bit_regs_p): Delete.
6210 (spe_expand_predicate_builtin): Delete.
6211 (spe_expand_evsel_builtin): Delete.
6212 (TARGET_DWARF_REGISTER_SPAN): Do not define.
6213 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
6214 (invalid_e500_subreg): Delete.
6215 (rs6000_legitimize_address): Always force_reg op2 as well, for
6216 paired single memory accesses.
6217 (rs6000_member_type_forces_blk): Delete.
6218 (rs6000_spe_function_arg): Delete.
6219 (rs6000_expand_unop_builtin): Delete SPE handling.
6220 (rs6000_expand_binop_builtin): Ditto.
6221 (spe_expand_stv_builtin): Delete.
6222 (bdesc_2arg_spe): Delete.
6223 (spe_expand_builtin): Delete.
6224 (spe_expand_predicate_builtin): Delete.
6225 (spe_expand_evsel_builtin): Delete.
6226 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
6227 (spe_init_builtins): Delete.
6228 (spe_func_has_64bit_regs_p): Delete.
6229 (savres_routine_name): Delete "info" parameter. Adjust callers.
6230 (rs6000_emit_stack_reset): Ditto.
6231 (rs6000_dwarf_register_span): Delete.
6232 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
6233 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
6234 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
6236 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
6238 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
6239 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
6240 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
6241 mulv2sf3, divv2sf3): Delete expanders.
6243 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6245 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
6247 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6249 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
6250 * config/rs6000/rs6000.c: Ditto.
6252 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6254 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
6255 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
6256 comparison_operator.
6258 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6260 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
6261 * config/rs6000/rs6000.opt: Ditto.
6262 * config/rs6000/t-rtems: Ditto.
6264 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6266 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
6267 TARGET_E500_SINGLE by 0, simplify.
6268 * config/rs6000/rs6000.c: Ditto.
6269 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
6270 (spe_build_register_parallel): Delete.
6271 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
6272 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
6273 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
6274 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
6275 (E500_CONVERT): Delete.
6276 * config/rs6000/spe.md: Remove many patterns and all define_constants.
6278 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6280 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
6281 * config/rs6000/dfp.md: Ditto.
6282 (negdd2, *negdd2_fpr): Merge.
6283 (absdd2, *absdd2_fpr): Merge.
6284 (negtd2, *negtd2_fpr): Merge.
6285 (abstd2, *abstd2_fpr): Merge.
6286 * config/rs6000/e500.h: Delete file.
6287 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
6288 TARGET_FPRS by 1 and simplify.
6289 * config/rs6000/rs6000-c.c: Ditto.
6290 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
6292 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
6294 * config/rs6000/rs6000.md: Ditto.
6295 (floatdidf2, *floatdidf2_fpr): Merge.
6296 (move_from_CR_gt_bit): Delete.
6297 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
6298 (E500_CR_IOR_COMPARE): Delete.
6299 (All patterns that require !TARGET_FPRS): Delete.
6300 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
6302 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6304 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
6306 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6308 * graphds.c (add_edge): Intitialize edge's attached data.
6309 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
6310 pointer parameter. Call pointed function on each edge during
6311 graph traversing. Skip traversing the edge when the function
6313 (graphds_dfs, graphds_scc): Ditto.
6314 (for_each_edge): New parameter. Pass the new parameter to callback
6316 * graphds.h (skip_edge_callback): New function pointer type.
6317 (graphds_dfs, graphds_scc): New function pointer parameter.
6318 (graphds_edge_callback, for_each_edge): New parameter.
6320 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6322 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
6323 out code checking if runtime alias check is possible to below ...
6324 Call the new function.
6325 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
6326 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
6328 2017-06-07 Marek Polacek <polacek@redhat.com>
6331 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
6332 TYPE_OVERFLOW_WRAPS check.
6334 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6336 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
6337 if versioning is required.
6338 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
6339 peeling with the check for versioning.
6341 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6343 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
6344 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
6345 Set true to new parameter if new ssa variable is defined.
6346 (vect_gen_vector_loop_niters): Refactor. Set range information
6347 for the new vector loop bound variable.
6348 (vect_do_peeling): Ditto.
6350 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6352 * tree-affine.c (ssa.h): Include header file.
6353 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
6354 has wrapping overflow behavior.
6356 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6358 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
6360 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6362 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
6363 (tree_to_aff_combination): ... here.
6365 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6367 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
6368 reg_pressure model function.
6369 (ivopts_global_cost_for_size): Delete.
6370 (determine_set_costs, iv_ca_recount_cost): Call new model function
6371 ivopts_estimate_reg_pressure.
6373 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6375 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
6376 expensive than udiv. Remove floating point cases from mod.
6378 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6380 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
6383 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6385 * config/aarch64/aarch64.md
6386 (copysignsf3): Fix mask generation.
6388 2017-06-07 Jakub Jelinek <jakub@redhat.com>
6390 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
6392 (class dump_manager): Add register_dumps method.
6393 * dumpfile.c: Include langhooks.h.
6394 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
6395 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
6396 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
6397 (dump_manager::dump_register): Start with 512 entries instead of 32.
6398 (dump_manager::register_dumps): New method.
6399 * toplev.c (general_init): Instead of invoking register_dumps
6400 langhook, invoke register_dumps method on the dump manager.
6401 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
6404 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
6406 * doc/md.texi: Clarify the restrictions on a define_insn condition.
6407 Say that # requires an associated define_split to exist, and that
6408 the define_split must be suitable for use after register allocation.
6410 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6412 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
6413 (compute_outgoing_frequencies): Also initialize zero counts.
6414 (find_many_sub_basic_blocks): Do not produce uninitialized profile
6415 around loops; preserve more of profile when nothing changes.
6417 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
6419 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
6421 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
6422 * config/arm/arm-cpu-cdata.h: Regenerate.
6423 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
6424 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
6425 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
6426 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
6427 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
6429 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
6431 * config/arm/t-rmprofile: Likewise.
6432 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
6434 2017-06-06 David S. Miller <davem@davemloft.net>
6437 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
6438 blockage if function uses alloca.
6440 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
6442 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
6443 New "uid" fields to hold pretty-print IDs of group and ref.
6444 Memory references are now identified as <group_id>:<ref_id>
6445 instead of using [random] addresses.
6446 (dump_mem_details): Simplify, no functional change.
6447 (dump_mem_ref): Simplify and make output more concise.
6448 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
6449 (find_or_create_group): Initialize group uid.
6450 (record_ref): Initialize ref uid. Improve debug output.
6451 (prune_group_by_reuse, should_issue_prefetch_p,)
6452 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
6453 (mark_nontemporal_store, determine_loop_nest_reuse):
6454 Improve debug output.
6456 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
6458 * dbgcnt.def (prefetch): New debug counter.
6459 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
6460 (schedule_prefetches): Stop issueing prefetches if debug counter
6463 2017-06-06 Tom de Vries <tom@codesourcery.com>
6465 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
6466 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
6468 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6470 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
6471 Use aarch64_reg_or_zero predicate for operand 4.
6472 (aarch64_compare_and_swap<mode> define_insn_and_split):
6473 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
6474 (aarch64_store_exclusive<mode>): Likewise for operand 2.
6476 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
6478 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
6479 (arm_compute_save_core_reg_mask): This.
6480 (thumb1_compute_save_reg_mask): Rename into ...
6481 (thumb1_compute_save_core_reg_mask): This.
6482 (arm_compute_save_reg0_reg12_mask): Adapt comment.
6483 (arm_compute_frame_layout): Likewise.
6485 2017-06-06 Richard Biener <rguenther@suse.de>
6487 PR tree-optimization/80974
6488 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
6489 keep or clear leaders SSA info.
6491 2017-06-06 Tom de Vries <tom@codesourcery.com>
6493 * config/nvptx/nvptx.c (split_mode_p): New function.
6494 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
6496 2017-06-06 Tom de Vries <tom@codesourcery.com>
6498 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
6500 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6503 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
6506 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6508 * shrink-wrap.c (handle_simple_exit): Update profile.
6509 (try_shrink_wrapping): Upate profile.
6511 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6513 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
6514 (tree_guess_outgoing_edge_probabilities): New.
6515 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
6516 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
6518 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6520 * ipa-split.c (split_function): Initialize return bb profile.
6522 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6524 * profile.c (compute_branch_probabilities): Also initialize
6527 2017-06-06 Richard Biener <rguenther@suse.de>
6529 PR tree-optimization/80928
6530 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
6531 (vect_analyze_loop_operations): Properly guard analysis for
6533 (vect_transform_loop): Likewise.
6534 (vect_analyze_loop_2): Also reset SLP type on PHIs.
6535 (vect_model_induction_cost): Do not cost for pure SLP.
6536 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
6537 of induction in inner loop vectorization.
6538 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
6539 (vect_get_and_check_slp_defs): Handle vect_induction_def.
6540 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
6542 (vect_analyze_slp_cost_1): Cost induction.
6543 (vect_detect_hybrid_slp_stmts): Handle PHIs.
6544 (vect_get_slp_vect_defs): Likewise.
6545 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
6546 (vect_transform_stmt): Handle SLP reductions.
6547 * tree-vectorizer.h (vectorizable_induction): Adjust.
6549 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6551 * config/rs6000/rs6000.c (make_resolver_func): Update
6552 init_lowered_empty_function call.
6554 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
6556 * doc/invoke.texi: Document the -fprofile-abs-path option.
6557 * common.opt (fprofile-abs-path): New option.
6558 * gcov-io.h (gcov_write_filename): Declare.
6559 * gcov-io.c (gcov_write_filename): New function.
6560 * coverage.c (coverage_begin_function): Use gcov_write_filename.
6561 * profile.c (output_location): Likewise.
6563 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6565 * shring-wrap.c: Revert accidental commit.
6567 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
6569 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
6571 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6573 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
6575 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
6576 profile in callgraph edge.
6577 * profile-count.h (apply_probability): If THIS is 0, then result is 0
6578 (apply_scale): Likewise.
6579 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
6580 Also scale profile when inlining function with zero profile.
6581 (initialize_cfun): Update exit block profile even when it is zero.
6582 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
6583 when profile is read.
6585 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6587 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
6588 (CLONE_*): New constants to define the processors we can generate
6589 code for with the target_clone attribute.
6590 (rs6000_clone_map): New array to identify which clone processors
6591 the current program is running on.
6592 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
6593 target_clone attribute.
6594 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
6595 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
6596 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
6597 (cpu_expand_builtin): Add support for target_clone attribute.
6598 (rs6000_valid_attribute_p): Allow "default" attribute.
6599 (get_decl_name): New debug function to simplify printing the
6600 current function name in debugging statements.
6601 (rs6000_clone_priority): New functions to support the target_clone
6602 attribute, and be able to generate code to switch between ISA 2.05
6603 through ISA 3.0 (power6 through power9).
6604 (rs6000_compare_version_priority): Likewise.
6605 (rs6000_get_function_versions_dispatcher): Likewise.
6606 (make_resolver_func): Likewise.
6607 (add_condition_to_bb): Likewise.
6608 (dispatch_function_versions): Likewise.
6609 (rs6000_generate_version_dispatcher_body): Likewise.
6610 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
6611 (fusion_gpr_load_p): Fix a spacing issue.
6612 * doc/extend.texi (Common Function Attributes): Document that the
6613 PowerPC supports the target_clone attribute.
6615 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
6617 * config/arm/arm.h: explain F symbol found in description of ARM
6618 register allocation in its legend.
6620 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6622 * config/mips/frame-header-opt.c: Include profile-count.h.
6623 * config/riscv/riscv.c: Include profile-count.h
6625 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6627 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
6629 (sm_set_flag_if_changed): Add bbs field.
6630 (execute_sm_if_changed_flag_set): Pass BBS.
6631 (execute_sm): Update.
6633 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6635 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6638 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6640 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
6641 (peephole2): New peephole2 to emit the above.
6642 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
6644 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6646 * config/aarch64/aarch64.c (define_peephole2 above
6647 *sub_<shift>_<mode>): New peephole.
6649 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
6651 * config/i386/i386.c (make_resolver_func): Update.
6652 * Makefile.in: Add profile-count.h and profile-count.o
6653 * auto-profile.c (afdo_indirect_call): Update to new API.
6654 (afdo_set_bb_count): Update.
6655 (afdo_propagate_edge): Update.
6656 (afdo_propagate_circuit): Update.
6657 (afdo_calculate_branch_prob): Update.
6658 (afdo_annotate_cfg): Update.
6659 * basic-block.h: Include profile-count.h
6660 (struct edge_def): Turn count to profile_count.
6661 (struct basic_block_def): Likewie.
6662 (REG_BR_PROB_BASE): Move to profile-count.h
6663 (RDIV): Move to profile-count.h
6664 * bb-reorder.c (max_entry_count): Turn to profile_count.
6665 (find_traces): Update.
6666 (rotate_loop):Update.
6667 (connect_traces):Update.
6668 (sanitize_hot_paths):Update.
6669 * bt-load.c (migrate_btr_defs): Update.
6670 * cfg.c (RDIV): Remove.
6671 (init_flow): Use alloc_block.
6672 (alloc_block): Uninitialize count.
6673 (unchecked_make_edge): Uninitialize count.
6674 (check_bb_profile): Update.
6675 (dump_edge_info): Update.
6676 (dump_bb_info): Update.
6677 (update_bb_profile_for_threading): Update.
6678 (scale_bbs_frequencies_int): Update.
6679 (scale_bbs_frequencies_gcov_type): Update.
6680 (scale_bbs_frequencies_profile_count): New.
6681 * cfg.h (update_bb_profile_for_threading): Update.
6682 (scale_bbs_frequencies_profile_count): Declare.
6683 * cfgbuild.c (compute_outgoing_frequencies): Update.
6684 (find_many_sub_basic_blocks): Update.
6685 * cfgcleanup.c (try_forward_edges): Update.
6686 (try_crossjump_to_edge): Update.
6687 * cfgexpand.c (expand_gimple_tailcall): Update.
6688 (construct_exit_block): Update.
6689 * cfghooks.c (verify_flow_info): Update.
6690 (dump_bb_for_graph): Update.
6691 (split_edge): Update.
6692 (make_forwarder_block): Update.
6693 (duplicate_block): Update.
6694 (account_profile_record): Update.
6695 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
6696 (get_estimated_loop_iterations): Update.
6697 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
6698 (single_likely_exit): Update.
6699 * cfgloopmanip.c (scale_loop_profile): Update.
6701 (set_zero_probability): Update.
6702 (lv_adjust_loop_entry_edge): Update.
6703 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
6704 (purge_dead_edges): Update.
6705 (rtl_account_profile_record): Update.
6706 * cgraph.c (cgraph_node::create): Uninitialize count.
6707 (symbol_table::create_edge): Uninitialize count.
6708 (cgraph_update_edges_for_call_stmt_node): Update.
6709 (cgraph_edge::dump_edge_flags): Update.
6710 (cgraph_node::dump): Update.
6711 (cgraph_edge::maybe_hot_p): Update.
6712 * cgraph.h: Include profile-count.h
6713 (create_clone), create_edge, create_indirect_edge): Update.
6714 (cgraph_node): Turn count to profile_count.
6715 (cgraph_edge0: Likewise.
6716 (make_speculative, clone): Update.
6717 (create_edge): Update.
6718 (init_lowered_empty_function): Update.
6719 * cgraphclones.c (cgraph_edge::clone): Update.
6720 (duplicate_thunk_for_node): Update.
6721 (cgraph_node::create_clone): Update.
6722 * cgraphunit.c (cgraph_node::analyze): Update.
6723 (cgraph_node::expand_thunk): Update.
6724 * final.c (dump_basic_block_info): Update.
6725 * gimple-streamer-in.c (input_bb): Update.
6726 * gimple-streamer-out.c (output_bb): Update.
6727 * graphite.c (print_global_statistics): Update.
6728 (print_graphite_scop_statistics): Update.
6729 * hsa-brig.c: Include basic-block.h.
6730 * hsa-dump.c: Include basic-block.h.
6731 * hsa-gen.c (T sum_slice): Update.
6732 (convert_switch_statements):Update.
6733 * hsa-regalloc.c: Include basic-block.h.
6734 * ipa-chkp.c (chkp_produce_thunks): Update.
6735 * ipa-cp.c (struct caller_statistics): Update.
6736 (init_caller_stats): Update.
6737 (gather_caller_stats): Update.
6738 (ipcp_cloning_candidate_p): Update.
6739 (good_cloning_opportunity_p): Update.
6740 (get_info_about_necessary_edges): Update.
6741 (dump_profile_updates): Update.
6742 (update_profiling_info): Update.
6743 (update_specialized_profile): Update.
6744 (perhaps_add_new_callers): Update.
6745 (decide_about_value): Update.
6746 (ipa_cp_c_finalize): Update.
6747 * ipa-devirt.c (struct odr_type_warn_count): Update.
6748 (struct decl_warn_count): Update.
6749 (struct final_warning_record): Update.
6750 (possible_polymorphic_call_targets): Update.
6751 (ipa_devirt): Update.
6752 * ipa-fnsummary.c (redirect_to_unreachable): Update.
6753 * ipa-icf.c (sem_function::merge): Update.
6754 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
6755 * ipa-inline.c (compute_uninlined_call_time): Update.
6756 (compute_inlined_call_time): Update.
6757 (want_inline_small_function_p): Update.
6758 (want_inline_self_recursive_call_p): Update.
6759 (edge_badness): Update.
6760 (lookup_recursive_calls): Update.
6761 (recursive_inlining): Update.
6762 (inline_small_functions): Update.
6763 (dump_overall_stats): Update.
6764 (dump_inline_stats): Update.
6765 * ipa-profile.c (ipa_profile_generate_summary): Update.
6766 (ipa_propagate_frequency): Update.
6767 (ipa_profile): Update.
6768 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
6769 * ipa-utils.c (ipa_merge_profiles): Update.
6770 * loop-doloop.c (doloop_modify): Update.
6771 * loop-unroll.c (report_unroll): Update.
6772 (unroll_loop_runtime_iterations): Update.
6773 * lto-cgraph.c (lto_output_edge): Update.
6774 (lto_output_node): Update.
6775 (input_node): Update.
6776 (input_edge): Update.
6777 (merge_profile_summaries): Update.
6778 * lto-streamer-in.c (input_cfg): Update.
6779 * lto-streamer-out.c (output_cfg): Update.
6780 * mcf.c (create_fixup_graph): Update.
6781 (adjust_cfg_counts): Update.
6782 (sum_edge_counts): Update.
6783 * modulo-sched.c (sms_schedule): Update.
6784 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
6785 * predict.c (maybe_hot_count_p): Update.
6786 (probably_never_executed): Update.
6787 (dump_prediction): Update.
6788 (combine_predictions_for_bb): Update.
6789 (propagate_freq): Update.
6790 (handle_missing_profiles): Update.
6791 (counts_to_freqs): Update.
6792 (rebuild_frequencies): Update.
6793 (force_edge_cold): Update.
6794 * predict.h: Include profile-count.h
6795 (maybe_hot_count_p, counts_to_freqs): UPdate.
6796 * print-rtl-function.c: Do not include cfg.h
6797 * print-rtl.c: Include basic-block.h
6798 * profile-count.c: New file.
6799 * profile-count.h: New file.
6800 * profile.c (is_edge_inconsistent): Update.
6801 (correct_negative_edge_counts): Update.
6802 (is_inconsistent): Update.
6803 (set_bb_counts): Update.
6804 (read_profile_edge_counts): Update.
6805 (compute_frequency_overlap): Update.
6806 (compute_branch_probabilities): Update; Initialize and deinitialize
6808 (branch_prob): Update.
6809 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
6810 (edge_gcov_count): New.
6811 (bb_gcov_count): New.
6812 * shrink-wrap.c (try_shrink_wrapping): Update.
6813 * tracer.c (better_p): Update.
6814 * trans-mem.c (expand_transaction): Update.
6815 (ipa_tm_insert_irr_call): Update.
6816 (ipa_tm_insert_gettmclone_call): Update.
6817 * tree-call-cdce.c: Update.
6818 * tree-cfg.c (gimple_duplicate_sese_region): Update.
6819 (gimple_duplicate_sese_tail): Update.
6820 (gimple_account_profile_record): Update.
6821 (execute_fixup_cfg): Update.
6822 * tree-inline.c (copy_bb): Update.
6823 (copy_edges_for_bb): Update.
6824 (initialize_cfun): Update.
6825 (freqs_to_counts): Update.
6826 (copy_cfg_body): Update.
6827 (expand_call_inline): Update.
6828 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
6829 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
6830 (try_unroll_loop_completely): Update.
6831 (try_peel_loop): Update.
6832 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6833 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
6834 * tree-ssa-loop-split.c (connect_loops): Update.
6835 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
6836 * tree-ssa-reassoc.c (branch_fixup): Update.
6837 * tree-ssa-tail-merge.c (replace_block_by): Update.
6838 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
6839 (compute_path_counts): Update.
6840 (update_profile): Update.
6841 (recompute_probabilities): Update.
6842 (update_joiner_offpath_counts): Update.
6843 (estimated_freqs_path): Update.
6844 (freqs_to_counts_path): Update.
6845 (clear_counts_path): Update.
6846 (ssa_fix_duplicate_block_edges): Update.
6847 (duplicate_thread_path): Update.
6848 * tree-switch-conversion.c (case_bit_test_cmp): Update.
6849 (struct switch_conv_info): Update.
6850 * tree-tailcall.c (decrease_profile): Update.
6851 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
6852 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6853 * value-prof.c (check_counter): Update.
6854 (gimple_divmod_fixed_value): Update.
6855 (gimple_mod_pow2): Update.
6856 (gimple_mod_subtract): Update.
6857 (gimple_ic_transform): Update.
6858 (gimple_stringop_fixed_value): Update.
6859 * value-prof.h (gimple_ic): Update.
6861 2017-06-02 Carl Love <cel@us.ibm.com>
6863 * config/rs6000/rs6000-c: Add support for built-in functions
6864 vector double vec_doublee (vector signed int);
6865 vector double vec_doublee (vector unsigned int);
6866 vector double vec_doublee (vector float);
6867 vector double vec_doubleh (vector signed int);
6868 vector double vec_doubleh (vector unsigned int);
6869 vector double vec_doubleh (vector float);
6870 vector double vec_doublel (vector signed int);
6871 vector double vec_doublel (vector unsigned int);
6872 vector double vec_doublel (vector float);
6873 vector double vec_doubleo (vector signed int);
6874 vector double vec_doubleo (vector unsigned int);
6875 vector double vec_doubleo (vector float);.
6876 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
6877 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
6879 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
6880 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
6881 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
6883 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
6884 vec_doublel, vec_doubleh.
6885 * doc/extend.texi: Update the built-in documentation file for the
6886 new built-in functions.
6888 2017-06-02 David Malcolm <dmalcolm@redhat.com>
6891 * ipa-inline-analysis.c (free_growth_caches): Set
6892 edge_removal_hook_holder to NULL after removing it.
6894 2017-06-02 Sudakshina Das <sudi.das@arm.com>
6896 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
6897 comparision with zero.
6899 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6900 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6901 for early expansion of vec_min and vec_max builtins.
6902 (builtin_function_type): Add min/max unsigned variants to those
6903 identified as having unsigned arguments.
6905 2017-06-02 Olivier Hainque <hainque@adacore.com>
6907 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
6909 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6911 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
6912 Use VALL_F16 iterator rather than VALL.
6914 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6916 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
6917 Emit CBNZ inside loop when doing a strong exchange and comparing
6918 against zero. Generate the CC flags after the loop.
6920 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
6922 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
6923 (dl_section_ref): New.
6924 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
6925 On AIX, append an expression to subtract the size of the
6926 section length to dl_section_ref.
6928 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6930 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6931 for early expansion of vector absolute builtins.
6933 2017-06-02 Richard Biener <rguenther@suse.de>
6935 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
6936 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
6938 2017-06-02 Richard Biener <rguenther@suse.de>
6940 PR tree-optimization/80948
6941 * tree-tailcall.c (find_tail_calls): Track stmts to move in
6944 2017-06-02 Richard Biener <rguenther@suse.de>
6946 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
6948 * tree-vect-stmts.c (process_use): Do not mark backedge defs
6949 for inductions as relevant.
6951 2017-06-02 Richard Biener <rguenther@suse.de>
6953 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
6954 (vectorizable_induction): ... this. Remove dead code.
6956 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
6958 * builtins. (expand_builtin_alloca): Remove second parameter and
6959 infer its value from the first parameter instead.
6960 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
6962 2017-06-02 Jakub Jelinek <jakub@redhat.com>
6964 PR rtl-optimization/80903
6965 * loop-doloop.c (add_test): Unshare sequence.
6967 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
6969 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
6971 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
6973 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
6975 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
6976 xlogue_layout::get_instance, logue_layout::xlogue_layout,
6977 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
6978 (xlogue_layout::get_stub_rtx): Make static.
6979 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
6980 (xlogue_layout::compute_stub_managed_regs): Rename to...
6981 (xlogue_layout::count_stub_managed_regs): ...this.
6982 (xlogue_layout::is_stub_managed_reg): New function.
6983 (xlogue_layout::m_stub_names): Rename to...
6984 (xlogue_layout::s_stub_names): ...this, make static.
6985 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
6986 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
6987 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
6988 xlogue_layout::s_stub_names): Instantiate statics.
6989 (stub_managed_regs): Remove.
6990 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
6991 (disable_call_ms2sysv_xlogues): Rename to...
6992 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
6993 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
6995 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
6996 change after reload_completed.
6997 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
6999 (ix86_expand_prologue): Likewise.
7000 (ix86_expand_epilogue): Likewise.
7001 (ix86_expand_split_stack_prologue): Likewise.
7002 (ix86_compute_frame_layout): Remove frame parameter ...
7003 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
7004 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
7006 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
7007 (ix86_frame): Move from here ...
7008 * config/i386/i386.h (ix86_frame): ... to here.
7009 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
7010 complete ix86_frame data structure instead. Remove some_ld_name.
7012 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
7014 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
7015 symbols that hold a DECL_VALUE_EXPR.
7017 2017-06-01 Martin Jambor <mjambor@suse.cz>
7019 PR tree-optimization/80898
7020 * tree-sra.c (process_subtree_disqualification): Removed.
7021 (disqualify_candidate): Do not acll
7022 process_subtree_disqualification.
7023 (subtree_mark_written_and_enqueue): New function.
7024 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
7025 RHS has been disqualified and re-queue LHS if necessary. Apart
7026 from that, ignore disqualified RHS.
7028 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7030 * config/s390/s390.c (s390_emit_epilogue): Disable early return
7031 address fetch for z10 or later.
7033 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7035 * config/arc/arc.md (tst_movb): Add guard when splitting.
7037 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7039 * config/arc/arc.c (arc_can_eliminate): Test against
7040 arc_frame_pointer_needed.
7042 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7044 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
7045 to prevent store reordering.
7046 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
7047 (type): Add block type.
7048 (stack_tie): Define special instruction to be used in
7051 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7053 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
7054 constraint. It is not valid for the pattern.
7055 (noncommutative_binary_comparison): Likewise.
7057 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7059 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
7062 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7064 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
7065 be used by the reg-alloc.
7067 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7069 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
7070 reg-alloc when having mul64 or mul32x16 instructions.
7071 (mulsidi3): Likewise.
7072 (umulsidi3): Likewise.
7073 (mulsi32x16): New pattern.
7074 (mulsi64): Likewise.
7075 (mulsidi64): Likewise.
7076 (umulsidi64): Likewise.
7077 (MUL32x16_REG): Define.
7078 (mul64_600): Use MUL32x16_REG.
7079 (mac64_600): Likewise.
7080 (umul64_600): Likewise.
7081 (umac64_600): Likewise.
7083 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7085 * config/arc/arc.md (mulsi3_700): Make it commutative.
7087 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
7089 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
7091 (*sign_extendsidi2_insn): Likewise for movstosw.
7093 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
7095 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
7096 the type of the input discriminant value. Convert the
7097 discriminant value of signedness vary.
7099 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
7101 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
7102 Add to -Wall section.
7104 2017-06-01 Richard Biener <rguenther@suse.de>
7107 * fold-const.c (fold_plusminus_mult_expr): If the factored
7108 factor may be zero use a wrapping type for the inner operation.
7109 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
7110 and handle moved defs.
7111 (process_assignment): Properly guard the unary op case. Return a
7112 tri-state indicating that moving the stmt before the call may allow
7113 to continue. Pass through to_move.
7114 (find_tail_calls): Handle moving unrelated defs before
7117 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
7120 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
7121 splitter result in the canonical way.
7123 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
7125 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
7126 also for 32bit target. Update insn attributes.
7127 (zero-extendsidi2 splitter): Allow all registers for operand 1.
7129 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
7131 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
7132 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
7133 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
7134 (_mm_maskz_min_ss): New intrinsics.
7136 2017-05-31 Martin Liska <mliska@suse.cz>
7138 * tree-vect-loop.c (vect_create_epilog_for_reduction):
7139 Change comment style to one we normally use.
7140 (vectorizable_reduction): Likewise.
7141 (vectorizable_induction): Likewise.
7142 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7143 (vectorizable_call): Likewise.
7144 (vectorizable_simd_clone_call): Likewise.
7145 (vectorizable_conversion): Likewise.
7146 (vectorizable_assignment): Likewise.
7147 (vectorizable_shift): Likewise.
7148 (vectorizable_operation): Likewise.
7149 (vectorizable_store): Likewise.
7150 (vectorizable_load): Likewise.
7151 * tree-vectorizer.h: Likewise.
7153 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
7155 * passes.c (emergency_dump_function): New.
7156 * tree-pass.h (emergency_dump_function): Declare.
7157 * plugin.c (plugins_internal_error_function): Remove.
7158 * plugin.h (plugins_internal_error_function): Remove declaration.
7159 * toplev.c (internal_error_function): New static function. Use it...
7160 (general_init): ...here.
7162 2017-05-31 Graham Markall <graham.markall@embecosm.com>
7164 * config/arc/arc.c (arc_print_operand): Handle constant operands.
7165 (arc_rtx_costs): Add costs for new patterns.
7166 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
7167 * config/arc/predicates.md: Add _1_2_3_operand predicate.
7169 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
7171 * tree-ssa-strlen.c (get_next_strinfo): New function.
7172 (get_stridx_plus_constant): Use it.
7173 (zero_length_string): Likewise.
7174 (adjust_related_strinfos): Likewise.
7175 (adjust_last_stmt): Likewise.
7177 2017-05-31 Richard Biener <rguenther@suse.de>
7180 * config/i386/i386.c (ix86_expand_builtin): Remove assert
7181 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
7183 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
7185 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
7186 loop_vinfo argument and use of dependence distance vectors.
7187 Check instead whether the two references differ only in their
7188 initial value and assume that they have the same alignment if the
7189 difference is a multiple of the vector alignment.
7190 (vect_analyze_data_refs_alignment): Update call accordingly.
7192 2017-05-31 Martin Liska <mliska@suse.cz>
7195 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
7197 2017-05-31 Bin Cheng <bin.cheng@arm.com>
7199 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
7200 (create_intersect_range_checks): Move from ...
7201 * tree-data-ref.c (create_intersect_range_checks_index)
7202 (create_intersect_range_checks): ... to here.
7203 (create_runtime_alias_checks): New function factored from ...
7204 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
7205 here. Call above function.
7206 * tree-data-ref.h (create_runtime_alias_checks): New function.
7208 2017-05-31 Bin Cheng <bin.cheng@arm.com>
7210 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
7211 segment length for dr_b and compute it in wide_int.
7213 2017-05-31 Richard Biener <rguenther@suse.de>
7215 PR tree-optimization/80906
7216 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
7217 and pass through iv_map.
7218 (copy_bb_and_scalar_dependences): Adjust.
7219 (translate_pending_phi_nodes): Likewise.
7220 (copy_loop_close_phi_args): Handle code-generating IVs instead
7223 2017-05-30 David Malcolm <dmalcolm@redhat.com>
7225 * diagnostic-color.c (color_dict): Add "type-diff".
7226 (parse_gcc_colors): Update comment.
7227 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
7228 -fdiagnostics-show-template-tree and -fno-elide-type.
7229 (GCC_COLORS): Add type-diff to example.
7231 (-fdiagnostics-show-template-tree): New.
7232 (-fno-elide-type): New.
7233 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
7234 the pp_format_decoder callback. Call any m_format_postprocessor's
7236 (pretty_printer::pretty_printer): Initialize
7237 m_format_postprocessor.
7238 (pretty_printer::~pretty_printer): Delete any
7239 m_format_postprocessor.
7240 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
7241 (class format_postprocessor): New class.
7242 (struct pretty_printer::format_decoder): Document the new parameters.
7243 (struct pretty_printer::m_format_postprocessor): New field.
7244 * tree-diagnostic.c (default_tree_printer): Update for new
7245 bool and const char ** params.
7246 * tree-diagnostic.h (default_tree_printer): Likewise.
7248 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
7250 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
7251 (lwa_operand): Delete rs6000_gen_cell_microcode test.
7252 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
7253 rs6000_gen_cell_microcode code.
7254 (rs6000_final_prescan_insn): Delete.
7255 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
7256 "warn-cell-microcode" entries.
7257 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
7258 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
7259 throughout. Change cc_reg_not_micro_cr0_operand to
7260 cc_reg_not_cr0_operand throughout.
7261 (*extendhi<mode>2_noload): Delete.
7262 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
7263 (mwarn-cell-microcode): Delete.
7264 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
7265 -mgen-cell-microcode and -mwarn-cell-microcode.
7267 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
7270 * config/i386/constraints.md (Yd): New constraint.
7272 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
7273 and (?Yd, r) alternatives. Update insn attributes.
7274 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
7275 and (?*Yd, r) alternatives. Update insn attributes.
7276 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
7278 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
7280 * gimplify.c (gimplify_modify_expr): Don't create a
7281 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
7284 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
7286 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
7288 2017-05-30 Richard Biener <rguenther@suse.de>
7290 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
7291 and reduc_def fields.
7292 (STMT_VINFO_REDUC_TYPE): New define.
7293 (STMT_VINFO_REDUC_DEF): Likewise.
7294 (vect_force_simple_reduction): Adjust prototype.
7295 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
7296 (vect_is_simple_reduction): Remove check_reduction argument.
7297 (vect_force_simple_reduction): Adjust and set
7298 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
7299 (vectorizable_reduction): Do not re-do reduction analysis
7300 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
7301 * tree-parloops.c (gather_scalar_reductions): Adjust.
7303 2017-05-30 Richard Biener <rguenther@suse.de>
7306 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
7309 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
7311 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
7312 Introduce unknown_misalignment parameter and remove vf.
7313 (vect_peeling_hash_get_lowest_cost):
7314 Pass unknown_misalignment parameter.
7315 (vect_enhance_data_refs_alignment):
7316 Fix unsupportable data ref treatment.
7318 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7320 * tree-vect-data-refs.c (vect_get_data_access_cost):
7321 Workaround for SLP handling.
7322 (vect_enhance_data_refs_alignment):
7323 Compute costs for doing no peeling at all, compare to the best
7324 peeling costs so far and avoid peeling if cheaper.
7326 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7328 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
7329 Return peeling info and set costs to zero for unlimited cost
7331 (vect_enhance_data_refs_alignment): Also inspect all datarefs
7332 with unknown misalignment. Compute and costs for unknown
7333 misalignment, compare them to the costs for known misalignment
7334 and choose the cheapest for peeling.
7336 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7338 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
7339 (vect_get_peeling_costs_all_drs): Create function.
7340 (vect_peeling_hash_get_lowest_cost):
7341 Use vect_get_peeling_costs_all_drs.
7342 (vect_peeling_supportable): Create function.
7343 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
7345 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7347 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
7348 DR_HAS_NEGATIVE_STEP.
7349 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
7350 (vect_enhance_data_refs_alignment): Use.
7351 (vect_duplicate_ssa_name_ptr_info): Use.
7352 * tree-vectorizer.h (dr_misalignment): Use.
7353 (known_alignment_for_access_p): Use.
7355 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7358 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
7360 (has_section_name): New function.
7362 2017-05-30 Martin Liska <mliska@suse.cz>
7365 * auto-profile.c (get_function_decl_from_block): Fix
7368 2017-05-30 Richard Biener <rguenther@suse.de>
7371 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
7373 2017-05-30 Martin Liska <mliska@suse.cz>
7375 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
7376 * dumpfile.h (struct dump_file_info): Remove ctors.
7378 2017-05-30 Martin Liska <mliska@suse.cz>
7380 * predict.def: Fix GNU coding style.
7382 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
7384 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
7385 Mark 'to' argument with ATTRIBUTE_UNUSED.
7387 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
7389 * config/xtensa/xtensa.c (xtensa_emit_call): Use
7390 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
7391 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
7394 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
7396 * doc/install.texi (Options specification): Restore entry of
7397 --enable-sjlj-exceptions.
7399 2017-05-27 Michael Eager <eager@eagercon.com>
7402 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
7404 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
7406 * config/microblaze/microblaze.h
7407 (FIXED_REGISTERS): Update in macro.
7408 (CALL_USED_REGISTERS): Update in macro.
7410 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
7412 * doc/install.texi: Add links to macOS binary distributions.
7414 2017-05-27 Jakub Jelinek <jakub@redhat.com>
7418 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7420 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
7422 2017-05-26 Martin Liska <mliska@suse.cz>
7424 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
7426 2017-05-26 Martin Liska <mliska@suse.cz>
7428 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
7429 always leading ';; '.
7430 (dump_bb_info): Likewise.
7431 (brief_dump_cfg): Likewise.
7432 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
7433 * dumpfile.c: Remove usage of TDF_VERBOSE.
7434 * dumpfile.h (enum dump_kind): Likewise.
7435 (dump_gimple_bb_header): Do not use TDF_COMMENT.
7436 * print-tree.c (debug_verbose): Remove.
7437 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
7438 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
7439 * tree-diagnostic.c (default_tree_printer): Replace
7440 TDF_DIAGNOSTIC with TDF_SLIM.
7442 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7444 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
7445 in parameter loop, rather than loop_vinfo.
7446 (create_intersect_range_checks): Ditto.
7447 (vect_create_cond_for_alias_checks): Update call to above functions.
7449 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7451 PR tree-optimization/80815
7452 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
7453 for merging runtime alias checks. Handle negative DR_STEPs.
7455 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7457 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
7459 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
7460 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
7461 out code pruning runtime alias checks.
7462 * tree-data-ref.c (prune_runtime_alias_test_list): New function
7463 factored out from above.
7464 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
7466 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
7468 (prune_runtime_alias_test_list): New decalaration.
7470 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7472 * tree-vect-data-refs.c (compare_tree): Rename and move ...
7473 * tree-data-ref.c (data_ref_compare_tree): ... to here.
7474 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
7475 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
7476 (operator==, comp_dr_with_seg_len_pair): Ditto.
7477 (vect_prune_runtime_alias_test_list): Ditto.
7479 2017-05-26 Martin Liska <mliska@suse.cz>
7482 * params.def: Bound partial-inlining-entry-probability param.
7484 2017-05-26 Marek Polacek <polacek@redhat.com>
7487 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
7490 2017-05-26 Richard Biener <rguenther@suse.de>
7492 PR tree-optimization/80842
7493 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
7496 2017-05-26 Richard Biener <rguenther@suse.de>
7498 PR tree-optimization/80844
7499 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
7501 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
7503 * doc/md.texi (Machine Constraints): Update x86 family
7504 machine constraints section to match 'config/i386/constraints.md'.
7506 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
7508 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
7510 2017-05-25 Nathan Sidwell <nathan@acm.org>
7512 * doc/invoke.texi (--enable-languages): Update documentation.
7514 2017-05-25 Martin Liska <mliska@suse.cz>
7516 * dumpfile.c: Add TDF_FOLDING.
7517 * dumpfile.h (enum dump_kind): Likewise.
7518 * genmatch.c (dt_simplify::gen_1): Use it.
7520 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7522 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
7524 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7526 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
7527 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
7529 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7531 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
7532 * match.pd (X == C): Rewrite it here.
7533 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
7534 with_certain_nonzero_bits2): New predicates.
7535 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
7537 2017-05-24 Nathan Sidwell <nathan@acm.org>
7539 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
7542 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
7545 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
7547 * config/powerpcspe: New port. Files are copied from the rs6000
7548 port, with "rs6000" in filenames replaced by "powerpcspe".
7550 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
7552 PR rtl-optimization/80754
7553 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
7555 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
7557 * config/sparc/sparc.md (length): Return the correct value for -mflat
7558 sibcalls to match output_sibcall.
7560 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
7564 * config/rs6000/rs6000.c (struct machine_function): Add new field
7566 (rs6000_get_separate_components): Init that field, use it.
7567 (rs6000_components_for_bb): Use the field.
7569 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
7571 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
7573 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
7576 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
7578 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7581 * config/s390/s390.c (s390_check_qrst_address): Check incoming
7582 address against address_operand predicate.
7583 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
7585 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
7587 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
7588 parameters passed indirectly.
7590 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
7592 * config/i386/i386.md (*movdi_internal): Remove SSE4
7593 alternative 18 (?r, *v). Update insn attributes.
7594 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
7595 Update insn attributes.
7596 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
7597 Update insn attributes.
7598 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
7599 alternative 1 (r, v). Remove isa attribute.
7600 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
7601 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
7602 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
7604 2017-05-23 Tom de Vries <tom@codesourcery.com>
7606 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
7609 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7611 * cgraphunit.c (symbol_table::process_new_functions): Update.
7612 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
7613 (inline_generate_summary): Rename to ...
7614 (ipa_fn_summary_generate): ... this one.
7615 (inline_read_summary): Rename to ...
7616 (ipa_fn_summary_read): ... this one.
7617 (inline_write_summary): Rename to ...
7618 (ipa_fn_summary_write): ... this one.
7619 (inline_free_summary): Rename to ...
7620 (ipa_free_fn_summary): ... this one.
7621 (pass_data_local_fn_summary, pass_local_fn_summary,
7622 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
7623 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
7624 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
7625 make_pass_ipa_fn_summary): New.
7626 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
7627 inline_write_summary, inline_free_summary): Remove.
7628 (ipa_free_fn_summary) : New.
7629 * ipa-inline.c (ipa_inline): Update.
7630 (pass_ipa_inline): Do not generate summaries.
7631 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
7633 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
7634 and add pass_ipa_fn_summary.
7635 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
7637 (make_pass_inline_parameters): Remove.
7639 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
7641 * omp-low.c (struct omp_context): Remove "default_kind" member.
7644 * omp-offload.c (execute_oacc_device_lower): Remove the
7645 parallelism dimensions function attributes for unparallelized
7646 OpenACC kernels constructs.
7648 2017-05-23 Martin Liska <mliska@suse.cz>
7650 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
7652 (cgraph_edge::make_speculative): Likewise.
7653 (cgraph_edge::resolve_speculation): Likewise.
7654 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
7655 (cgraph_node::dump): Likewise.
7656 * cgraph.h: Likewise.
7657 * cgraphunit.c (analyze_functions): Likewise.
7658 (symbol_table::compile): Likewise.
7659 * ipa-cp.c (print_all_lattices): Likewise.
7660 (determine_versionability): Likewise.
7661 (initialize_node_lattices): Likewise.
7662 (ipcp_verify_propagated_values): Likewise.
7663 (estimate_local_effects): Likewise.
7664 (update_profiling_info): Likewise.
7665 (create_specialized_node): Likewise.
7666 (perhaps_add_new_callers): Likewise.
7667 (decide_about_value): Likewise.
7668 (decide_whether_version_node): Likewise.
7669 (identify_dead_nodes): Likewise.
7670 (ipcp_store_bits_results): Likewise.
7671 * ipa-devirt.c (dump_targets): Likewise.
7672 (ipa_devirt): Likewise.
7673 * ipa-icf.c (sem_item::dump): Likewise.
7674 (sem_function::equals): Likewise.
7675 (sem_variable::equals): Likewise.
7676 (sem_item_optimizer::read_section): Likewise.
7677 (sem_item_optimizer::execute): Likewise.
7678 (congruence_class::dump): Likewise.
7679 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
7680 (dump_inline_summary): Likewise.
7681 (estimate_node_size_and_time): Likewise.
7682 (inline_analyze_function): Likewise.
7683 * ipa-inline-transform.c (inline_call): Likewise.
7684 * ipa-inline.c (report_inline_failed_reason): Likewise.
7685 (want_early_inline_function_p): Likewise.
7686 (edge_badness): Likewise.
7687 (update_edge_key): Likewise.
7688 (inline_small_functions): Likewise.
7689 * ipa-profile.c (ipa_profile): Likewise.
7690 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
7691 (ipa_make_edge_direct_to_target): Likewise.
7692 (remove_described_reference): Likewise.
7693 (ipa_impossible_devirt_target): Likewise.
7694 (propagate_controlled_uses): Likewise.
7695 (ipa_print_node_params): Likewise.
7696 (ipcp_transform_function): Likewise.
7697 * ipa-pure-const.c (pure_const_read_summary): Likewise.
7698 (propagate_pure_const): Likewise.
7699 * ipa-reference.c (generate_summary): Likewise.
7700 (read_write_all_from_decl): Likewise.
7701 (propagate): Likewise.
7702 (ipa_reference_read_optimization_summary): Likewise.
7703 * ipa-utils.c (ipa_merge_profiles): Likewise.
7704 * ipa.c (walk_polymorphic_call_targets): Likewise.
7705 (symbol_table::remove_unreachable_nodes): Likewise.
7706 (ipa_single_use): Likewise.
7707 * passes.c (execute_todo): Likewise.
7708 * predict.c (drop_profile): Likewise.
7709 * symtab.c (symtab_node::get_dump_name): New function.
7710 (symtab_node::dump_name): Likewise.
7711 (symtab_node::dump_asm_name): Likewise.
7712 (symtab_node::dump_references): Likewise.
7713 (symtab_node::dump_referring): Likewise.
7714 (symtab_node::dump_base): Likewise.
7715 (symtab_node::debug_symtab): Likewise.
7716 * tree-sra.c (convert_callers_for_node): Likewise.
7717 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
7718 * value-prof.c (init_node_map): Likewise.
7720 2017-05-23 Martin Liska <mliska@suse.cz>
7722 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
7723 and symtab_node::debug_symtab to symbol_table::debug.
7724 * cgraphunit.c (analyze_functions): Use the renamed function.
7725 (symbol_table::compile): Likewise.
7726 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
7727 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
7728 * passes.c (execute_todo): Likewise.
7729 * symtab.c (symbol_table::dump): New function.
7730 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
7732 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7734 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
7735 that nonconst implies exec.
7737 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7739 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7740 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7741 (inline_edge_summary_vec): Turn into ...
7742 (ipa_call_summaries): ... this one.
7743 (redirect_to_unreachable, edge_set_predicate,
7744 evaluate_properties_for_edge, inline_summary_alloc,
7745 reset_ipa_call_summary, reset_inline_summary,
7746 inline_summary_t::duplicate): Update.
7747 (inline_edge_duplication_hook): Turn to ...
7748 (ipa_call_summary_t::duplicate): ... this one.
7749 (inline_edge_removal_hook): Turn to ...
7750 (ipa_call_summary_t::remove): ... this one.
7751 (dump_inline_edge_summary): Turn to ...
7752 (dump_ipa_call_summary): ... this one.
7753 (estimate_function_body_sizes): Update.
7754 (inline_update_callee_summaries): Update.
7755 (remap_edge_change_prob): Update.
7756 (remap_edge_summaries): Update.
7757 (inline_merge_summary): Update.
7758 (do_estimate_edge_time): Update.
7759 (inline_generate_summary): Update.
7760 (inline_read_section): Update.
7761 (inline_read_summary): Update.
7762 (inline_free_summary): Update.
7763 * ipa-inline.c (can_inline_edge_p): Update.
7764 (compute_inlined_call_time): Update.
7765 (want_inline_small_function_p): Update.
7766 (edge_badness): Update.
7767 (early_inliner): Update.
7768 * ipa-inline.h (inline_edge_summary): Turn to ...
7769 (ipa_call_summary): ... this one.
7770 (ipa_call_summary_t): New class.
7771 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
7772 (ipa_call_summaries): New.
7773 (inline_edge_summary): Remove.
7774 (estimate_edge_growth): Update.
7775 * ipa-profile.c (ipa_propagate_frequency_1): Update.
7776 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7777 * ipa-split.c (execute_split_functions): Update.
7778 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
7780 2017-05-23 Tom de Vries <tom@codesourcery.com>
7782 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7783 attributes): Document rdrand effective target.
7785 2017-05-23 Tom de Vries <tom@codesourcery.com>
7787 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7788 attributes): Sort alphabetically.
7790 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
7792 * config/avr/genmultilib.awk: Use gsub instead of gensub.
7794 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
7797 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
7798 V2DF/V2DI splat into two separate patterns, one that handles
7799 registers, and the other that only handles memory. Drop support
7800 for splatting from a GPR on ISA 2.07 and then splitting the
7801 splat into direct move and splat.
7802 (vsx_splat_<mode>_reg): Likewise.
7803 (vsx_splat_<mode>_mem): Likewise.
7805 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
7807 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
7809 2017-05-22 Jakub Jelinek <jakub@redhat.com>
7812 * omp-low.c (finish_taskreg_remap): New function.
7813 (finish_taskreg_scan): If unit size of ctx->record_type
7814 is non-constant, unshare the size expression and replace
7815 decls in it with possible outer var refs.
7818 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
7819 GOVD_SHARED rather than GOVD_PRIVATE with it.
7820 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
7821 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
7824 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
7825 as last argument to build_outer_var_ref for pointer bases of array
7828 2017-05-19 Martin Sebor <msebor@redhat.com>
7830 * print-tree.c (print_node): Print DECL_READ_P flag.
7832 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7834 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
7835 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
7836 * cgraph.c: Likewise.
7837 * cgraphunit.c: Likewise.
7838 * gengtype.c: Likewise.
7839 * ipa-cp.c: Likewise.
7840 * ipa-devirt.c: Likewise.
7841 * ipa-icf.c: Likewise.
7842 * ipa-predicate.c: Likewise.
7843 * ipa-profile.c: Likewise.
7844 * ipa-prop.c: Likewise.
7845 * ipa-split.c: Likewise.
7847 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
7848 edge_predicate_pool, dump_inline_hints,
7849 inline_summary::account_size_time, redirect_to_unreachable,
7850 edge_set_predicate, set_hint_predicate,
7851 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
7852 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
7853 inline_summary_t::remove, remap_hint_predicate_after_duplication,
7854 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
7855 ipa_call_summary_t::remove, initialize_growth_caches,
7856 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
7857 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
7858 mark_modified, unmodified_parm_1, unmodified_parm,
7859 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
7860 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
7861 compute_bb_predicates, will_be_nonconstant_expr_predicate,
7862 will_be_nonconstant_predicate, record_modified_bb_info,
7863 get_minimal_bb, record_modified, param_change_prob,
7864 phi_result_unknown_predicate, predicate_for_phi_result,
7865 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
7866 estimate_function_body_sizes, compute_inline_parameters,
7867 compute_inline_parameters_for_curren, pass_data_inline_parameters,
7868 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
7869 inline_update_callee_summaries, remap_edge_change_prob,
7870 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
7871 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
7872 inline_analyze_function, inline_summary_t::insert,
7873 inline_generate_summary, read_ipa_call_summary, inline_read_section,
7874 inline_read_summary, write_ipa_call_summary, inline_write_summary,
7875 inline_free_summary): Move to ipa-fnsummary.h
7876 (predicate_t): Remove.
7877 * ipa-fnsummary.c: New file.
7878 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
7879 (enum inline_hints_vals, inline_hints, agg_position_info,
7880 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
7881 inline_summaries, ipa_call_summary, ipa_call_summary_t,
7882 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
7883 dump_inline_summary, dump_inline_hints, inline_generate_summary,
7884 inline_read_summary, inline_write_summary, inline_free_summary,
7885 inline_analyze_function, initialize_inline_failed,
7886 inline_merge_summary, inline_update_overall_summary,
7887 compute_inline_parameters): Move to ipa-fnsummary.h
7888 * ipa-fnsummary.h: New file.
7889 * ipa-inline-transform.h: Include ipa-inline.h.
7890 * ipa-inline.c: LIkewise.
7892 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7894 * ipa-inline.c (edge_badness): Use inlined_time instead of
7895 inline_summaries->get.
7897 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7899 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
7901 2017-05-22 Nathan Sidwell <nathan@acm.org>
7903 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
7904 (fdump-lang): Document 'raw' option.
7905 * dumpfile.h (TDI_tu): Delete.
7906 * dumpfile.c (dump_files): Remove translation-unit.
7907 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
7909 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
7911 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
7912 command option from $(AWK) call.
7913 * config/avr/genmultilib.awk: Simplify and rewrite so that it
7914 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
7915 [FORMAT]: Remove handling of variable.
7916 * config/avr/t-multilib: Regenerate.
7918 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7920 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
7922 (dump_inline_summary): Do not print self_time.
7923 (estimate_function_body_sizes): Do not set self_time.
7924 (compute_inline_parameters): Likewise.
7925 (inline_read_section, inline_write_summary): Do not stream self_time.
7926 * ipa-inline.h (inline_summary): Drop self_time.
7928 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7930 * ipa-inline-analysis.c (account_size_time): Rename to ...
7931 (inline_summary::account_size_time): ... this one.
7932 (reset_ipa_call_summary): Turn to ...
7933 (ipa_call_summary::reset): ... this one.
7934 (reset_inline_summary): Turn to ...
7935 (inline_summary::reset): ... this one.
7936 (inline_summary_t::remove): Update.
7937 (inline_summary_t::duplicate): Update.
7938 (ipa_call_summary_t::remove): Update.
7939 (dump_inline_summary): Update.
7940 (estimate_function_body_sizes): Update.
7941 (compute_inline_parameters): Update.
7942 (estimate_node_size_and_time): Update.
7943 (inline_merge_summary): Update.
7944 (inline_update_overall_summary): Update.
7945 (inline_read_section): Update.
7946 (inline_write_summary): Update.
7947 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
7948 add account_size_time and reset member functions.
7949 (ipa_call_summary): Add reset function.
7950 * ipa-predicate.h (predicate::operator &): Constify.
7952 2017-05-22 Richard Biener <rguenther@suse.de>
7954 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
7956 2017-05-19 Jason Merrill <jason@redhat.com>
7958 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
7960 2017-05-19 Marek Polacek <polacek@redhat.com>
7963 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
7964 TYPE_OVERFLOW_WRAPS checks.
7966 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
7968 * tree-core.h (enum omp_clause_default_kind): Add
7969 "OMP_CLAUSE_DEFAULT_PRESENT".
7970 * tree-pretty-print.c (dump_omp_clause): Handle it.
7971 * gimplify.c (enum gimplify_omp_var_data): Add
7972 "GOVD_MAP_FORCE_PRESENT".
7973 (gimplify_adjust_omp_clauses_1): Map it to
7974 "GOMP_MAP_FORCE_PRESENT".
7975 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
7977 * gimplify.c (oacc_default_clause): Clarify.
7979 2017-05-19 Nathan Sidwell <nathan@acm.org>
7981 LANG_HOOK_REGISTER_DUMPS
7982 * toplev.c (general_init): Call register dump lang hook.
7983 * doc/invoke.texi: Document -fdump-lang option family.
7984 * dumpfile.c (dump_files): Remove class dump here.
7985 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
7986 * dumpfile.h (tree_dump_index): Remove TDI_class.
7987 * langhooks-def.h (lhd_register_dumps): Declare.
7988 (LANG_HOOKS_REGISTER_DUMPS): Define.
7989 (LANG_HOOKS_INITIALIZER): Add it.
7990 * langhooks.c (lhd_register_dumps): Define.
7991 * langhooks.h (struct lang_hooks): Add register_dumps.
7993 2017-05-19 Nathan Sidwell <nathan@acm.org>
7995 * context.h (context::set_passes): New.
7996 * context.c (context::context): Do not create pass manager.
7997 * toplev.c (general_init): Create pass manager here.
7999 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
8001 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
8002 use this splitter if two add or or instructions would also work for
8003 the constant we want to generate.
8005 2017-05-19 Richard Biener <rguenther@suse.de>
8008 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
8009 predicate evaluation.
8011 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
8013 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
8015 * ipa-inline.c (want_inline_small_function_p): Do not cast to
8018 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
8020 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
8021 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
8022 (inline_edge_summary_vec): Turn into ...
8023 (ipa_call_summaries): ... this one.
8024 (redirect_to_unreachable, edge_set_predicate,
8025 evaluate_properties_for_edge, inline_summary_alloc,
8026 reset_ipa_call_summary, reset_inline_summary,
8027 inline_summary_t::duplicate): Update.
8028 (inline_edge_duplication_hook): Turn to ...
8029 (ipa_call_summary_t::duplicate): ... this one.
8030 (inline_edge_removal_hook): Turn to ...
8031 (ipa_call_summary_t::remove): ... this one.
8032 (dump_inline_edge_summary): Turn to ...
8033 (dump_ipa_call_summary): ... this one.
8034 (estimate_function_body_sizes): Update.
8035 (inline_update_callee_summaries): Update.
8036 (remap_edge_change_prob): Update.
8037 (remap_edge_summaries): Update.
8038 (inline_merge_summary): Update.
8039 (do_estimate_edge_time): Update.
8040 (inline_generate_summary): Update.
8041 (inline_read_section): Update.
8042 (inline_read_summary): Update.
8043 (inline_free_summary): Update.
8044 * ipa-inline.c (can_inline_edge_p): Update.
8045 (compute_inlined_call_time): Update.
8046 (want_inline_small_function_p): Update.
8047 (edge_badness): Update.
8048 (early_inliner): Update.
8049 * ipa-inline.h (inline_edge_summary): Turn to ...
8050 (ipa_call_summary): ... this one.
8051 (ipa_call_summary_t): New class.
8052 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
8053 (ipa_call_summaries): New.
8054 (inline_edge_summary): Remove.
8055 (estimate_edge_growth): Update.
8056 * ipa-profile.c (ipa_propagate_frequency_1): Update.
8057 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
8058 * ipa-split.c (execute_split_functions): Update.
8059 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
8061 2017-05-19 Richard Biener <rguenther@suse.de>
8064 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
8066 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
8068 * config/rs6000/rs6000.c (struct machine_function): Add field
8069 fpr_is_wrapped_separately.
8070 (rs6000_get_separate_components): Use 64 components. Handle the
8072 (rs6000_components_for_bb): Handle the FPR components.
8073 (rs6000_emit_prologue_components): Handle the FPR components.
8074 (rs6000_emit_epilogue_components): Handle the FPR components.
8075 (rs6000_set_handled_components): Handle the FPR components.
8076 (rs6000_emit_prologue): Don't output prologue code for those FPRs
8077 that are already separately shrink-wrapped.
8078 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
8079 that are already separately shrink-wrapped.
8081 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
8084 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
8087 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
8088 (define_peephole2 for Altivec d-form load): Add peepholes to catch
8089 cases where the register allocator uses a move and an offsettable
8090 memory operation to/from a FPR register on ISA 2.06/2.07.
8091 (define_peephole2 for Altivec d-form store): Likewise.
8093 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
8096 * config/i386/mmx.md (*mov<mode>_internal): Enable
8097 alternatives 11, 12, 13 and 14 also for 32bit targets.
8098 Remove alternatives 15, 16, 17 and 18.
8099 * config/i386/sse.md (vec_concatv2di): Change
8100 alternative (!x, *y) to (x, ?!*Yn).
8102 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
8104 * dumpfile.h (enum dump_kind): Remove stray comma.
8106 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
8108 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
8109 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
8110 predicate::num_conditions
8111 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
8112 (CHANGED): turn into predicate::changed.
8113 (agg_position_info): Move to ipa-predicate.h
8114 (add_condition, predicate::add_clause, predicate::operator &=,
8115 predicate::or_with, predicate::evaluate, predicate::probability,
8116 dump_condition, dump_clause, predicate::dump,
8117 predicate::remap_after_duplication, predicate::remap_after_inlining,
8118 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
8119 (evaluate_conditions_for_known_args): Update.
8120 (set_cond_stmt_execution_predicate): Update.
8121 * ipa-inline.h: Include ipa-predicate.h
8122 (condition, inline_param_summary, conditions, agg_position_info,
8123 predicate): Move to ipa-predicate.h
8124 * ipa-predicate.c: New file.
8125 * ipa-predicate.h: New file.
8127 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
8129 * final.c (leaf_function_p): Check we are not in a sequence.
8131 2017-05-18 Martin Liska <mliska@suse.cz>
8133 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
8134 * dumpfile.c (dump_register): Use new enum dump_kind.
8135 (get_dump_file_name): Likewise.
8136 (dump_enable_all): Likewise.
8137 (dump_switch_p_1): Likewise.
8138 (enable_rtl_dump_file): Remove usage of TDF_RTL.
8139 * dumpfile.h (enum dump_kind): New enum type.
8140 (struct dump_file_info): Create constructor and
8141 format fields and comments.
8142 * passes.c (pass_manager::register_one_dump_file):
8144 * statistics.c (statistics_early_init): Likewise.
8145 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
8146 TDF_TREE with TDF_SLIM.
8147 (gather_memory_references_ref): Likewise.
8149 2017-05-18 Martin Liska <mliska@suse.cz>
8151 * vec.h (struct vnull): Use it.
8153 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
8155 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
8156 (true_predicate, false_predicate, true_predicate_p,
8157 false_predicate_p): Remove.
8158 (single_cond_predicate, not_inlined_predicate): Turn to member function
8160 (add_condition): Update.
8161 (add_clause): Turn to...
8162 (predicate::add_clause): ... this one; update; allow passing NULL
8164 (and_predicates): Turn to ...
8165 (predicate::operator &=): ... this one.
8166 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
8167 (or_predicates): Turn to ...
8168 (predicate::or_with): ... this one.
8169 (evaluate_predicate): Turn to ...
8170 (predicate::evaluate): ... this one.
8171 (predicate_probability): Turn to ...
8172 (predicate::probability): ... this one.
8173 (dump_condition): Update.
8174 (dump_predicate): Turn to ...
8175 (predicate::dump): ... this one.
8176 (account_size_time): Update.
8177 (edge_set_predicate): Update.
8178 (set_hint_predicate): UPdate.
8179 (evaluate_conditions_for_known_args): Update.
8180 (evaluate_properties_for_edge): Update.
8181 (remap_predicate_after_duplication): Turn to...
8182 (predicate::remap_after_duplication): ... this one.
8183 (remap_hint_predicate_after_duplication): Update.
8184 (inline_summary_t::duplicate): UPdate.
8185 (dump_inline_edge_summary): Update.
8186 (dump_inline_summary): Update.
8187 (set_cond_stmt_execution_predicate): Update.
8188 (set_switch_stmt_execution_predicate): Update.
8189 (compute_bb_predicates): Update.
8190 (will_be_nonconstant_expr_predicate): Update.
8191 (will_be_nonconstant_predicate): Update.
8192 (phi_result_unknown_predicate): Update.
8193 (predicate_for_phi_result): Update.
8194 (array_index_predicate): Update.
8195 (estimate_function_body_sizes): Update.
8196 (estimate_node_size_and_time): Update.
8197 (estimate_ipcp_clone_size_and_time): Update.
8198 (remap_predicate): Rename to ...
8199 (predicate::remap_after_inlining): ... this one.
8200 (remap_hint_predicate): Update.
8201 (inline_merge_summary): Update.
8202 (inline_update_overall_summary): Update.
8203 (estimate_size_after_inlining): Update.
8204 (read_predicate): Rename to ...
8205 (predicate::stream_in): ... this one.
8206 (read_inline_edge_summary): Update.
8207 (write_predicate): Rename to ...
8208 (predicate::stream_out): ... this one.
8209 (write_inline_edge_summary): Update.
8210 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
8211 (clause_t): Turn to uint32_t
8212 (predicate): Turn to class; implement constructor and operators
8214 (size_time_entry): Update.
8215 (inline_summary): Update.
8216 (inline_edge_summary): Update.
8218 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
8220 * fold-const.c (fold_binary_loc): Move transformation...
8221 * match.pd (C - X CMP X): ... here.
8223 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
8225 * config/sparc/sparc.c (sparc_option_override): Set function
8226 alignment for -mcpu=niagara7 to 64 to match the I$ line.
8227 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
8229 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
8231 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
8233 2017-05-18 Marek Polacek <polacek@redhat.com>
8236 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
8237 (pass_ubsan::execute): Call gimple_assign_single_p instead of
8238 gimple_assign_load_p.
8240 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
8243 * real.c (do_compare): Give decimal_do_compare preference over
8244 comparing just the signs.
8246 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
8248 * doc/md.texi (Canonicalization of Instructions): Describe the
8249 canonical form of instructions that inherently set a condition
8252 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
8255 * tree-cfg.c: Move deletion of unreachable case statements to after
8256 the merging of consecutive case labels.
8258 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
8260 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
8261 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
8262 restoring of callee-saved registers.
8264 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
8266 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
8267 * config/visium/visium.c (single_set_and_flags): Likewise.
8268 * config/visium/visium.md (Substitutions): Likewise.
8270 2017-05-17 Martin Liska <mliska@suse.cz>
8272 * cfg.c: Introduce dump_flags_t type and
8273 use it instead of int type.
8275 * cfghooks.c: Likewise.
8276 * cfghooks.h (struct cfg_hooks): Likewise.
8277 * cfgrtl.c: Likewise.
8278 * cfgrtl.h: Likewise.
8279 * cgraph.c (cgraph_node::get_body): Likewise.
8280 * coretypes.h: Likewise.
8281 * domwalk.c: Likewise.
8282 * domwalk.h: Likewise.
8283 * dumpfile.c (struct dump_option_value_info): Likewise.
8284 (dump_enable_all): Likewise.
8285 (dump_switch_p_1): Likewise.
8286 (opt_info_switch_p): Likewise.
8287 * dumpfile.h (enum tree_dump_index): Likewise.
8288 (struct dump_file_info): Likewise.
8289 * genemit.c: Likewise.
8290 * generic-match-head.c: Likewise.
8291 * gengtype.c (open_base_files): Likewise.
8292 * gimple-pretty-print.c: Likewise.
8293 * gimple-pretty-print.h: Likewise.
8294 * graph.c (print_graph_cfg): Likewise.
8295 * graphite-scop-detection.c (dot_all_sese): Likewise.
8296 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
8297 * loop-unroll.c (report_unroll): Likewise.
8298 * passes.c (pass_manager::register_one_dump_file): Likewise.
8299 * print-tree.c: Likewise.
8300 * statistics.c: Likewise.
8301 * tree-cfg.c: Likewise.
8302 * tree-cfg.h: Likewise.
8303 * tree-dfa.c: Likewise.
8304 * tree-dfa.h: Likewise.
8305 * tree-dump.c (dump_function): Likewise.
8306 * tree-dump.h (struct dump_info): Likewise.
8307 * tree-pretty-print.c: Likewise.
8308 * tree-pretty-print.h: Likewise.
8309 * tree-ssa-live.c: Likewise.
8310 * tree-ssa-live.h: Likewise.
8311 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
8312 * tree-vect-loop.c: Likewise.
8313 * tree-vect-slp.c: Likewise.
8315 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
8316 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8318 PR tree-optimization/80457
8319 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
8320 of all arguments to a statement as scalar_to_vec operations.
8321 (vectorizable_call): Adjust call to vect_model_simple_cost for
8323 (vectorizable_conversion): Likewise.
8324 (vectorizable_assignment): Likewise.
8325 (vectorizable_shift): Likewise.
8326 (vectorizable_operation): Likewise.
8327 (vectorizable_comparison): Likewise.
8328 (vect_is_simple_cond): Record the def types for operands.
8329 (vectorizable_condition): Likewise, call vect_model_simple_cost.
8330 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
8331 for statement argument count.
8333 2017-05-16 Carl Love <cel@us.ibm.com>
8335 * config/rs6000/rs6000-c: Add support for built-in functions
8336 vector unsigned long long vec_bperm (vector unsigned long long,
8337 vector unsigned char)
8338 vector signed long long vec_mule (vector signed int,
8340 vector unsigned long long vec_mule (vector unsigned int,
8341 vector unsigned int)
8342 vector signed long long vec_mulo (vector signed int,
8344 vector unsigned long long vec_mulo (vector unsigned int,
8345 vector unsigned int)
8346 vector signed char vec_sldw (vector signed char,
8349 vector unsigned char vec_sldw (vector unsigned char,
8350 vector unsigned char,
8352 vector signed short vec_sldw (vector signed short,
8353 vector signed short,
8355 vector unsigned short vec_sldw (vector unsigned short,
8356 vector unsigned short,
8358 vector signed int vec_sldw (vector signed int,
8361 vector unsigned int vec_sldw (vector unsigned int,
8362 vector unsigned int,
8364 vector signed long long vec_sldw (vector signed long long,
8365 vector signed long long,
8367 vector unsigned long long vec_sldw (vector unsigned long long,
8368 vector unsigned long long,
8370 * config/rs6000/rs6000-c: Add support for built-in functions
8371 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
8372 * config/rs6000/altivec.h: Add defintion for vec_sldw.
8373 * doc/extend.texi: Update the built-in documentation for the
8374 new built-in functions.
8376 2017-05-16 Marek Polacek <polacek@redhat.com>
8380 * tree.c (save_expr): Don't fold the expression.
8382 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
8384 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
8385 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
8386 and (?*y,m). Update insn attributes.
8388 2017-05-16 Martin Liska <mliska@suse.cz>
8390 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
8391 flags argument of print_gimple_stmt, print_gimple_expr,
8392 print_generic_stmt and print_generic_expr.
8393 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
8394 * coretypes.h: Likewise.
8395 * except.c (dump_eh_tree): Likewise.
8396 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8397 * gimple-pretty-print.h: Likewise.
8398 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
8399 (backprop::push_to_worklist): Likewise.
8400 (backprop::pop_from_worklist): Likewise.
8401 (backprop::process_use): Likewise.
8402 (backprop::intersect_uses): Likewise.
8403 (note_replacement): Likewise.
8404 * gimple-ssa-store-merging.c
8405 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
8406 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
8407 (pass_store_merging::execute): Likewise.
8408 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
8409 (ssa_base_cand_dump_callback): Likewise.
8410 (dump_incr_vec): Likewise.
8411 (replace_refs): Likewise.
8412 (replace_mult_candidate): Likewise.
8413 (create_add_on_incoming_edge): Likewise.
8414 (create_phi_basis): Likewise.
8415 (insert_initializers): Likewise.
8416 (all_phi_incrs_profitable): Likewise.
8417 (introduce_cast_before_cand): Likewise.
8418 (replace_one_candidate): Likewise.
8419 * gimplify.c (gimplify_expr): Likewise.
8420 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
8421 (set_rename): Likewise.
8422 (rename_uses): Likewise.
8423 (copy_loop_phi_nodes): Likewise.
8424 (add_close_phis_to_merge_points): Likewise.
8425 (copy_loop_close_phi_args): Likewise.
8426 (copy_cond_phi_args): Likewise.
8427 (graphite_copy_stmts_from_block): Likewise.
8428 (translate_pending_phi_nodes): Likewise.
8429 * graphite-poly.c (print_pdr): Likewise.
8430 (dump_gbb_cases): Likewise.
8431 (dump_gbb_conditions): Likewise.
8432 (print_scop_params): Likewise.
8433 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
8434 (build_cross_bb_scalars_use): Likewise.
8435 (gather_bbs::before_dom_children): Likewise.
8436 * hsa-dump.c (dump_hsa_immed): Likewise.
8437 * ipa-cp.c (print_ipcp_constant_value): Likewise.
8438 (get_replacement_map): Likewise.
8439 * ipa-inline-analysis.c (dump_condition): Likewise.
8440 (estimate_function_body_sizes): Likewise.
8441 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
8442 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
8443 * ipa-prop.c (ipa_dump_param): Likewise.
8444 (ipa_print_node_jump_functions_for_edge): Likewise.
8445 (ipa_modify_call_arguments): Likewise.
8446 (ipa_modify_expr): Likewise.
8447 (ipa_dump_param_adjustments): Likewise.
8448 (ipa_dump_agg_replacement_values): Likewise.
8449 (ipcp_modif_dom_walker::before_dom_children): Likewise.
8450 * ipa-pure-const.c (check_stmt): Likewise.
8451 (pass_nothrow::execute): Likewise.
8452 * ipa-split.c (execute_split_functions): Likewise.
8453 * omp-offload.c (dump_oacc_loop_part): Likewise.
8454 (dump_oacc_loop): Likewise.
8455 * trans-mem.c (tm_log_emit): Likewise.
8456 (tm_memopt_accumulate_memops): Likewise.
8457 (dump_tm_memopt_set): Likewise.
8458 (dump_tm_memopt_transform): Likewise.
8459 * tree-cfg.c (gimple_verify_flow_info): Likewise.
8460 (print_loop): Likewise.
8461 * tree-chkp-opt.c (chkp_print_addr): Likewise.
8462 (chkp_gather_checks_info): Likewise.
8463 (chkp_get_check_result): Likewise.
8464 (chkp_remove_check_if_pass): Likewise.
8465 (chkp_use_outer_bounds_if_possible): Likewise.
8466 (chkp_reduce_bounds_lifetime): Likewise.
8467 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
8468 (chkp_mark_completed_bounds): Likewise.
8469 (chkp_register_incomplete_bounds): Likewise.
8470 (chkp_mark_invalid_bounds): Likewise.
8471 (chkp_maybe_copy_and_register_bounds): Likewise.
8472 (chkp_build_returned_bound): Likewise.
8473 (chkp_get_bound_for_parm): Likewise.
8474 (chkp_build_bndldx): Likewise.
8475 (chkp_get_bounds_by_definition): Likewise.
8476 (chkp_generate_extern_var_bounds): Likewise.
8477 (chkp_get_bounds_for_decl_addr): Likewise.
8478 * tree-chrec.c (chrec_apply): Likewise.
8479 * tree-data-ref.c (dump_data_reference): Likewise.
8480 (dump_subscript): Likewise.
8481 (dump_data_dependence_relation): Likewise.
8482 (analyze_overlapping_iterations): Likewise.
8483 * tree-inline.c (expand_call_inline): Likewise.
8484 (tree_function_versioning): Likewise.
8485 * tree-into-ssa.c (dump_defs_stack): Likewise.
8486 (dump_currdefs): Likewise.
8487 (dump_names_replaced_by): Likewise.
8488 (dump_update_ssa): Likewise.
8489 (update_ssa): Likewise.
8490 * tree-object-size.c (pass_object_sizes::execute): Likewise.
8491 * tree-parloops.c (build_new_reduction): Likewise.
8492 (try_create_reduction_list): Likewise.
8493 (ref_conflicts_with_region): Likewise.
8494 (oacc_entry_exit_ok_1): Likewise.
8495 (oacc_entry_exit_single_gang): Likewise.
8496 * tree-pretty-print.h: Likewise.
8497 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
8498 (get_scalar_evolution): Likewise.
8499 (add_to_evolution): Likewise.
8500 (get_loop_exit_condition): Likewise.
8501 (analyze_evolution_in_loop): Likewise.
8502 (analyze_initial_condition): Likewise.
8503 (analyze_scalar_evolution): Likewise.
8504 (instantiate_scev): Likewise.
8505 (number_of_latch_executions): Likewise.
8506 (gather_chrec_stats): Likewise.
8507 (final_value_replacement_loop): Likewise.
8508 (scev_const_prop): Likewise.
8509 * tree-sra.c (dump_access): Likewise.
8510 (disqualify_candidate): Likewise.
8511 (create_access): Likewise.
8513 (maybe_add_sra_candidate): Likewise.
8514 (create_access_replacement): Likewise.
8515 (analyze_access_subtree): Likewise.
8516 (analyze_all_variable_accesses): Likewise.
8517 (sra_modify_assign): Likewise.
8518 (initialize_constant_pool_replacements): Likewise.
8519 (find_param_candidates): Likewise.
8520 (decide_one_param_reduction): Likewise.
8521 (replace_removed_params_ssa_names): Likewise.
8522 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
8523 * tree-ssa-copy.c (dump_copy_of): Likewise.
8524 (copy_prop_visit_cond_stmt): Likewise.
8525 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
8526 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
8527 (record_equivalences_from_stmt): Likewise.
8528 * tree-ssa-dse.c (compute_trims): Likewise.
8529 (delete_dead_call): Likewise.
8530 (delete_dead_assignment): Likewise.
8531 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
8532 (forward_propagate_into_cond): Likewise.
8533 (pass_forwprop::execute): Likewise.
8534 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8535 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
8537 (move_computations_worker): Likewise.
8538 (execute_sm): Likewise.
8539 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
8540 (remove_exits_and_undefined_stmts): Likewise.
8541 (remove_redundant_iv_tests): Likewise.
8542 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
8543 (adjust_iv_update_pos): Likewise.
8544 * tree-ssa-math-opts.c (bswap_replace): Likewise.
8545 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
8546 (value_replacement): Likewise.
8547 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
8548 * tree-ssa-pre.c (print_pre_expr): Likewise.
8549 (get_representative_for): Likewise.
8550 (create_expression_by_pieces): Likewise.
8551 (insert_into_preds_of_block): Likewise.
8552 (eliminate_insert): Likewise.
8553 (eliminate_dom_walker::before_dom_children): Likewise.
8554 (eliminate): Likewise.
8555 (remove_dead_inserted_code): Likewise.
8556 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
8557 * tree-ssa-reassoc.c (get_rank): Likewise.
8558 (eliminate_duplicate_pair): Likewise.
8559 (eliminate_plus_minus_pair): Likewise.
8560 (eliminate_not_pairs): Likewise.
8561 (undistribute_ops_list): Likewise.
8562 (eliminate_redundant_comparison): Likewise.
8563 (update_range_test): Likewise.
8564 (optimize_range_tests_var_bound): Likewise.
8565 (optimize_vec_cond_expr): Likewise.
8566 (rewrite_expr_tree): Likewise.
8567 (rewrite_expr_tree_parallel): Likewise.
8568 (linearize_expr): Likewise.
8569 (break_up_subtract): Likewise.
8570 (linearize_expr_tree): Likewise.
8571 (attempt_builtin_powi): Likewise.
8572 (attempt_builtin_copysign): Likewise.
8573 (transform_stmt_to_copy): Likewise.
8574 (transform_stmt_to_multiply): Likewise.
8575 (dump_ops_vector): Likewise.
8576 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
8577 (print_scc): Likewise.
8578 (set_ssa_val_to): Likewise.
8579 (visit_reference_op_store): Likewise.
8580 (visit_use): Likewise.
8581 (sccvn_dom_walker::before_dom_children): Likewise.
8582 (run_scc_vn): Likewise.
8583 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
8585 (expr_hash_elt::print): Likewise.
8586 (const_and_copies::pop_to_marker): Likewise.
8587 (const_and_copies::record_const_or_copy_raw): Likewise.
8588 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
8589 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
8590 (dump_predicates): Likewise.
8591 (find_uninit_use): Likewise.
8592 (warn_uninitialized_phi): Likewise.
8593 (pass_late_warn_uninitialized::execute): Likewise.
8594 * tree-ssa.c (verify_vssa): Likewise.
8595 (verify_ssa): Likewise.
8596 (maybe_optimize_var): Likewise.
8597 * tree-vrp.c (dump_value_range): Likewise.
8598 (dump_all_value_ranges): Likewise.
8599 (dump_asserts_for): Likewise.
8600 (register_edge_assert_for_2): Likewise.
8601 (vrp_visit_cond_stmt): Likewise.
8602 (vrp_visit_switch_stmt): Likewise.
8603 (vrp_visit_stmt): Likewise.
8604 (vrp_visit_phi_node): Likewise.
8605 (simplify_cond_using_ranges_1): Likewise.
8606 (fold_predicate_in): Likewise.
8607 (evrp_dom_walker::before_dom_children): Likewise.
8608 (evrp_dom_walker::push_value_range): Likewise.
8609 (evrp_dom_walker::pop_value_range): Likewise.
8610 (execute_early_vrp): Likewise.
8612 2017-05-16 Richard Biener <rguenther@suse.de>
8614 * dwarf2out.c (loc_list_from_tree_1): Do not create
8615 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
8617 2017-05-16 Richard Biener <rguenther@suse.de>
8619 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
8621 (note_variable_value_in_expr): If we resolved the decl ref
8622 do not push to the stack.
8624 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
8626 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
8627 operations in fast-math mode.
8628 (vaddq_f16): Likewise.
8629 (vmul_f16): Likewise.
8630 (vmulq_f16): Likewise.
8631 (vsub_f16): Likewise.
8632 (vsubq_f16): Likewise.
8633 * config/arm/neon.md (add<mode>3): New.
8635 (fma:<VH:mode>3): New. Also remove outdated comment.
8638 2017-05-16 Martin Liska <mliska@suse.cz>
8642 * ipa-devirt.c (warn_types_mismatch): Fix typo.
8643 (odr_types_equivalent_p): Likewise.
8645 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
8647 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
8649 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
8652 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
8653 non-interunit SSE move alternatives with '?'.
8654 (zero-extendsidi peephole2): New peephole to skip intermediate
8655 general register in SSE zero-extend sequence.
8657 2017-05-15 Jeff Law <law@redhat.com>
8659 * reorg.c (relax_delay_slots): Create a new variable to hold
8660 the temporary target rather than clobbering TARGET_LABEL.
8662 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
8663 missing argument to extract_bit_field call.
8664 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
8666 2017-05-15 Martin Liska <mliska@suse.cz>
8669 * gcc.c (process_command): Do not allow empty argument of -o option.
8671 2017-05-15 Renlin Li <renlin.li@arm.com>
8673 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
8674 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
8675 * config/aarch64/constraints.md (Usf): Add long call check.
8676 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
8677 (call_value): Likewise.
8678 (sibcall): Likewise.
8679 (sibcall_value): Likewise.
8681 (call_value_insn): New.
8682 (sibcall_insn): Update rtx pattern.
8683 (sibcall_value_insn): Likewise.
8684 (call_internal): Remove.
8685 (call_value_internal): Likewise.
8686 (sibcall_internal): Likewise.
8687 (sibcall_value_internal): Likewise.
8688 (call_reg): Likewise.
8689 (call_symbol): Likewise.
8690 (call_value_reg): Likewise.
8691 (call_value_symbol): Likewise.
8693 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
8696 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
8698 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
8700 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
8701 compatible with CCGOCmode and with CCZmode.
8703 2017-05-14 Martin Sebor <msebor@redhat.com>
8706 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
8707 (gimple_fold_builtin_snprintf): Same.
8708 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
8709 (gimple_fold_builtin_snprintf): Same.
8710 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
8712 (is_call_safe): New function.
8713 (try_substitute_return_value): Call it.
8714 (try_simplify_call): New function.
8715 (pass_sprintf_length::handle_gimple_call): Call it.
8717 2017-05-14 Martin Sebor <msebor@redhat.com>
8720 * builtins.c (expand_builtin_stpncpy): Simplify.
8722 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
8724 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
8725 * config/i386/i386.h
8726 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
8727 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
8728 (struct machine_function): Add new members call_ms2sysv,
8729 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
8730 (struct machine_frame_state): New fields sp_realigned and
8731 sp_realigned_offset.
8732 * config/i386/i386.c
8733 (enum xlogue_stub): New enum.
8734 (enum xlogue_stub_sets): New enum.
8735 (class xlogue_layout): New class.
8736 (struct ix86_frame): New fields stack_realign_allocate_offset,
8737 stack_realign_offset and outlined_save_offset. Modify comments to
8738 detail stack layout when using out-of-line stubs.
8739 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
8740 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
8741 -mcall-ms2sysv-xlogues.
8742 (stub_managed_regs): New static variable.
8743 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
8744 registers managed by out-of-line stub.
8745 (disable_call_ms2sysv_xlogues): New function.
8746 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
8747 m->call_ms2sysv when appropriate and compute frame layout for
8749 (sp_valid_at, fp_valid_at): New inline functions.
8750 (choose_basereg): New function.
8751 (choose_baseaddr): Add align parameter, use choose_basereg and modify
8753 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
8754 Use align parameter of choose_baseaddr to generated aligned SSE movs
8756 (pro_epilogue_adjust_stack): Modify to track
8757 machine_frame_state::sp_realigned.
8758 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
8759 (ix86_nsaved_sseregs): Likewise.
8760 (ix86_emit_save_regs): Likewise.
8761 (ix86_emit_save_regs_using_mov): Likewise.
8762 (ix86_emit_save_sse_regs_using_mov): Likewise.
8763 (get_scratch_register_on_entry): Likewise.
8764 (gen_frame_set): New function.
8765 (gen_frame_load): Likewise.
8766 (gen_frame_store): Likewise.
8767 (emit_outlined_ms2sysv_save): Likewise.
8768 (emit_outlined_ms2sysv_restore): Likewise.
8769 (ix86_expand_prologue): Modify stack re-alignment code and call
8770 emit_outlined_ms2sysv_save when appropriate.
8771 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
8772 parameter rtx_insn *insn, which allows the function to be used to only
8774 (ix86_expand_epilogue): Modify validity checks of frame and stack
8775 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
8776 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
8777 * config/i386/predicates.md
8778 (save_multiple): New predicate.
8779 (restore_multiple): Likewise.
8780 * config/i386/sse.md
8781 (save_multiple<mode>): New pattern.
8782 (save_multiple_realign<mode>): Likewise.
8783 (restore_multiple<mode>): Likewise.
8784 (restore_multiple_and_return<mode>): Likewise.
8785 (restore_multiple_leave_return<mode>): Likewise.
8786 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
8788 2017-05-14 Julia Koval <julia.koval@intel.com>
8790 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
8791 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
8792 (__builtin_ia32_xsetbv): New builtins.
8793 * config/i386/i386.c (ix86_expand_special_args_builtin):
8795 (ix86_expand_builtin): Special expand for new intrinsics.
8796 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
8797 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
8798 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
8800 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8802 * cfganal.c (inverted_post_order_compute): Change argument type
8804 * cfganal.h (inverted_post_order_compute): Adjust prototype.
8805 * df-core.c (rest_of_handle_df_initialize): Adjust.
8806 (rest_of_handle_df_finish): Likewise.
8807 (df_analyze_1): Likewise.
8808 (df_analyze): Likewise.
8809 (loop_inverted_post_order_compute): Change argument to be a vec *.
8810 (df_analyze_loop): Adjust.
8811 (df_get_n_blocks): Likewise.
8812 (df_get_postorder): Likewise.
8813 * df.h (struct df_d): Change field to be a vec.
8814 * lcm.c (compute_laterin): Adjust.
8815 (compute_available): Likewise.
8816 * lra-lives.c (lra_create_live_ranges_1): Likewise.
8817 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
8818 * tree-ssa-pre.c (compute_antic): Likewise.
8820 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8822 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
8823 (depth_first_search::depth_first_search): Change structure init
8824 function to this constructor.
8825 (depth_first_search::add_bb): Rename function to this member.
8826 (depth_first_search::execute): Likewise.
8827 (flow_dfs_compute_reverse_finish): Adjust.
8829 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8831 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
8832 (longest_simple_path): Likewise.
8833 * shrink-wrap.c (spread_components): Likewise.
8834 (disqualify_problematic_components): Likewise.
8835 (emit_common_heads_for_components): Likewise.
8836 (emit_common_tails_for_components): Likewise.
8837 (insert_prologue_epilogue_for_components): Likewise.
8839 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8841 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
8844 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8846 * df-core.c (df_set_blocks): Start using auto_bitmap.
8847 (df_compact_blocks): Likewise.
8848 * df-problems.c (df_rd_confluence_n): Likewise.
8849 * df-scan.c (df_insn_rescan_all): Likewise.
8850 (df_process_deferred_rescans): Likewise.
8851 (df_update_entry_block_defs): Likewise.
8852 (df_update_exit_block_uses): Likewise.
8853 (df_entry_block_bitmap_verify): Likewise.
8854 (df_exit_block_bitmap_verify): Likewise.
8855 (df_scan_verify): Likewise.
8856 * lra-constraints.c (lra_constraints): Likewise.
8857 (undo_optional_reloads): Likewise.
8858 (lra_undo_inheritance): Likewise.
8859 * lra-remat.c (calculate_gen_cands): Likewise.
8860 (do_remat): Likewise.
8861 * lra-spills.c (assign_spill_hard_regs): Likewise.
8862 (spill_pseudos): Likewise.
8863 * tree-ssa-pre.c (bitmap_set_and): Likewise.
8864 (bitmap_set_subtract_values): Likewise.
8866 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8868 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
8869 management with auto_bitmap.
8870 (fix_inter_tick): Likewise.
8871 (fix_recovery_deps): Likewise.
8872 * ira.c (add_store_equivs): Likewise.
8873 (find_moveable_pseudos): Likewise.
8874 (split_live_ranges_for_shrink_wrap): Likewise.
8875 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
8876 (rtx_reuse_manager::seen_def_p): Likewise.
8877 (rtx_reuse_manager::set_seen_def): Likewise.
8878 * print-rtl.h (class rtx_reuse_manager): Likewise.
8880 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8882 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
8884 (migrate_btr_def): Likewise.
8885 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
8886 * df-core.c (loop_post_order_compute): Likewise.
8887 (loop_inverted_post_order_compute): Likewise.
8888 * hsa-common.h: Likewise.
8889 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
8890 * init-regs.c (initialize_uninitialized_regs): Likewise.
8891 * ipa-inline.c (resolve_noninline_speculation): Likewise.
8892 (inline_small_functions): Likewise.
8893 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
8894 * ira.c (combine_and_move_insns): Likewise.
8895 (build_insn_chain): Likewise.
8896 * loop-invariant.c (find_invariants): Likewise.
8897 * lower-subreg.c (propagate_pseudo_copies): Likewise.
8898 * predict.c (tree_predict_by_opcode): Likewise.
8899 (predict_paths_leading_to): Likewise.
8900 (predict_paths_leading_to_edge): Likewise.
8901 (estimate_loops_at_level): Likewise.
8902 (estimate_loops): Likewise.
8903 * shrink-wrap.c (try_shrink_wrapping): Likewise.
8904 (spread_components): Likewise.
8905 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
8906 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
8907 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
8908 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
8909 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
8910 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
8911 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
8912 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
8913 (mark_threaded_blocks): Likewise.
8914 (thread_through_all_blocks): Likewise.
8915 * tree-ssa.c (verify_ssa): Likewise.
8916 (execute_update_addresses_taken): Likewise.
8917 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
8919 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8921 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
8923 (post_order_compute): Likewise.
8924 (inverted_post_order_compute): Likewise.
8925 (pre_and_rev_post_order_compute_fn): Likewise.
8927 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8929 * genrecog.c (int_set::int_set): Explicitly construct our
8930 auto_vec base class.
8931 * vec.h (auto_vec::auto_vec): New constructor.
8933 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8935 * bitmap.h (class auto_bitmap): New constructor taking
8936 bitmap_obstack * argument.
8938 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8940 * bitmap.h (class auto_bitmap): Change type of m_bits to
8941 bitmap_head, and adjust ctor / dtor and member operators.
8943 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
8945 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
8946 when returned register mode doesn't match original mode.
8948 2017-05-12 Jeff Law <law@redhat.com>
8949 Jakub Jelinek <jakub@redhat.com>
8951 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
8952 we look for cc setter after the compare-elim changes.
8953 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
8954 within the vector to match what compare-elim now expects.
8955 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
8956 (xorsi3_flags, one_cmplsi2_flags): Likewise.
8958 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
8959 after the compare-elim changes.
8960 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
8961 the vector to match what compare-elim now expects.
8962 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
8963 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
8964 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
8965 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
8966 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
8968 * config/visium/visium.c (single_set_and_flags): Fix where
8969 we look for cc setter after the compare-elim changes.
8970 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
8971 with the vector to match what compare-elim now expects.
8972 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
8973 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
8974 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
8975 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
8976 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
8977 (neg<mode>2_insn_set_overflow): Likewise.
8979 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
8982 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
8983 maybe_expand_insn call, set ops[0].target. If still set after call,
8984 set alt_rtl. Add extra arg to recursive calls.
8985 (extract_bit_field): Add alt_rtl argument. Pass to
8987 * expmed.h (extract_bit_field): Fix prototype.
8988 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
8989 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
8990 to extract_bit_field_calls.
8991 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
8992 Pass alt_rtl to extract_bit_field calls.
8993 * calls.c (store_unaligned_arguments_into_psuedos)
8994 load_register_parameters): Pass extra NULL to extract_bit_field calls.
8995 * optabs.c (maybe_legitimize_operand): Clear op->target when call
8997 * optabs.h (struct expand_operand): Add target bitfield.
8999 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
9001 * compare-elim.c (try_eliminate_compare): Canonicalize
9002 operation with embedded compare to
9003 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
9004 (set (reg) (operation)].
9006 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
9008 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
9011 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
9012 cost of adding a carry flag for ADC instruction.
9013 [case MINUS]: Ignore the cost of subtracting a carry flag
9014 for SBB instruction.
9016 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
9018 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
9020 * config/rs6000/bmiintrin.h: New file.
9021 * config/rs6000/bmi2intrin.h: New file.
9022 * config/rs6000/x86intrin.h: New file.
9024 2017-05-12 Jeff Law <law@redhat.com>
9026 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
9029 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
9032 * tree-cfg.c: Remove cfg edges of unreachable case statements.
9034 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9036 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
9037 early expansion of vector divide builtins.
9038 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
9039 builtins identified as having unsigned arguments.
9041 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9043 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
9044 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
9045 expansion of vector logical operations (and, andc, or, xor,
9048 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9050 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
9051 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
9053 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9055 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
9056 early GIMPLE expansion of vector multiplies.
9058 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9060 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
9061 TARGET_HAVE_MOVT conditional.
9062 (movt splitter): Likewise.
9064 2017-05-12 Richard Biener <rguenther@suse.de>
9066 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
9067 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9068 Fold all stmts not inplace.
9070 2017-05-12 Richard Biener <rguenther@suse.de>
9072 PR tree-optimization/80713
9073 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
9074 inserted_exprs bit for not removed stmts.
9076 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
9079 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
9080 parallelized" attribute for parallelized OpenACC kernels.
9081 * omp-offload.c (execute_oacc_device_lower): Use it.
9083 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
9084 Set "oacc kernels" attribute.
9085 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
9086 parameter. Adjust all users.
9087 (oacc_fn_attrib_kernels_p): Remove function.
9088 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
9089 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
9090 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
9091 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
9092 assert "oacc kernels" attribute is set.
9094 2017-05-11 Carl Love <cel@us.ibm.com>
9096 * config/rs6000/rs6000-c: Add support for built-in functions
9097 vector unsigned char vec_popcnt (vector signed char)
9098 vector unsigned char vec_popcnt (vector unsigned char)
9099 vector unsigned short vec_popcnt (vector signed short)
9100 vector unsigned short vec_popcnt (vector unsigned short)
9101 vector unsigned int vec_popcnt (vector signed int)
9102 vector unsigned int vec_popcnt (vector unsigned int)
9103 vector unsigned long long vec_popcnt (vector signed long long)
9104 vector unsigned long long vec_popcnt (vector unsigned long long)
9105 vector signed long long vec_slo (vector signed long long,
9107 vector signed long long vec_slo (vector signed long long,
9108 vector unsigned char)
9109 vector unsigned long long vec_slo (vector unsigned long long,
9111 vector unsigned long long vec_slo (vector unsigned long long,
9112 vector unsigned char)
9113 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
9114 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
9115 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
9116 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
9117 * doc/extend.texi: Update the built-in documentation file for the
9118 new built-in functions.
9120 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
9122 * attribs.h (sorted_attr_string): Move machine independent
9123 functions for target clone support from the i386 port to common
9124 code. Rename ix86_function_versions to common_function_versions.
9125 Rename make_name to make_unique_name.
9126 (common_function_versions): Likewise.
9127 (make_unique_name): Likewise.
9128 (make_dispatcher_decl): Likewise.
9129 (is_function_default_version): Likewise.
9130 * attribs.c (attr_strcmp): Likewise.
9131 (sorted_attr_string): Likewise.
9132 (common_function_versions): Likewise.
9133 (make_unique_name): Likewise.
9134 (make_dispatcher_decl): Likewise.
9135 (is_function_default_version): Likewise.
9136 * config/i386/i386.c (attr_strcmp): Likewise.
9137 (sorted_attr_string): Likewise.
9138 (ix86_function_versions): Likewise.
9139 (make_name): Likewise.
9140 (make_dispatcher_decl): Likewise.
9141 (is_function_default_version): Likewise.
9142 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
9144 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9147 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
9148 Account for direct move costs for vec_construct of integer
9151 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
9154 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
9155 (UNSPEC_STX_ATOMIC): Ditto.
9156 (loaddi_via_sse): New insn.
9157 (storedi_via_sse): Ditto.
9158 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
9159 Update corresponding peephole2 patterns.
9160 (atomic_storedi_fpu): Ditto.
9162 2017-05-11 Julia Koval <julia.koval@intel.com>
9164 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
9165 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
9167 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
9168 (__builtin_ia32_rsqrt14ss_mask): New builtins.
9169 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
9171 2017-05-11 Nathan Sidwell <nathan@acm.org>
9173 * graphite-poly.c: Include dumpfile.h.
9175 * dumpfle.h (dump_function): Declare here ...
9176 * tree-dump.h (dump_function): ... not here.
9177 * dumpfile.c: #include tree-cfg.h.
9178 (dump_function): Move here from ...
9179 * tree-dump.c (dump_function): ... here.
9180 * gimplify.c: #include splay-tree.h, not tree-dump.h.
9181 * graphite-poly.c: Don't include tree-dump.h.
9182 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
9183 * print-tree.c: Likewise.
9184 * stor-layout.c: Likewise.
9185 * tree-nested.c: Likewise.
9187 * dumpfile.c (dump_start): Use TDF_FLAGS.
9188 (dump_enable_all): Fix TDF_KIND check thinko.
9190 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
9192 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9193 array entries to represent two legal parameterizations of the
9194 overloaded __builtin_cmpb function, as represented by the
9195 P6_OV_BUILTIN_CMPB constant.
9196 (altivec_resolve_overloaded_builtin): Add special case handling
9197 for the __builtin_cmpb function, as represented by the
9198 P6_OV_BUILTIN_CMPB constant.
9199 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
9200 (BU_P6_64BIT_2): New macro.
9201 (BU_P6_OVERLOAD_2): New macro
9202 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
9203 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
9204 (CMPB): Add overload support to represent both 32-bit and 64-bit
9205 compare-bytes function.
9206 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
9207 support for TARGET_CMPB.
9208 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
9209 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
9210 documentation of the __builtin_cmpb overloaded built-in function.
9212 2017-05-11 Richard Biener <rguenther@suse.de>
9214 PR tree-optimization/80705
9215 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
9216 bases are not vectorizable.
9218 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9220 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
9221 when counting register pressure.
9223 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9225 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
9226 (iv_ca_more_deps): Renamed to ...
9227 (iv_ca_compare_deps): ... this.
9228 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
9230 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9232 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
9234 (determine_group_iv_costs): ... here.
9235 (find_inv_vars_cb): Record inv var if it's not recorded before.
9237 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9239 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
9240 (get_shiftadd_cost): Ditto.
9242 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9244 * tree-ssa-address.c: Include header file.
9245 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
9247 (add_to_parts): Refactor.
9248 (addr_to_parts): New parameter. Update use of move_hint_to_base.
9249 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
9252 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9254 PR tree-optimization/53090
9255 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
9257 (extract_cond_operands): Detect condition with IV on both sides
9258 and return COMP_IV_EXPR_2.
9259 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
9260 (rewrite_use_compare): Simplify by removing call to function
9261 extract_cond_operands.
9263 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9265 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
9266 (extract_cond_operands): Detect condition comparing against non-
9267 invariant bound and return appropriate enum value.
9268 (find_interesting_uses_cond): Update use of extract_cond_operands.
9269 Handle its return value accordingly.
9270 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
9272 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9274 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
9275 nonlinear iv_use computation in loop invariant sensitive way.
9277 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9279 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
9280 (find_iv_candidates): Call relate_compare_use_with_all_cands.
9282 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9284 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
9285 (dump_cand): Support iv_cand.inv_exprs.
9286 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
9288 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
9291 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9293 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
9295 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
9296 as local function. Include necessary header files.
9297 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
9299 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9301 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
9303 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9305 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
9306 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
9307 RSHIFT_EXPR and BIT_NOT_EXPR.
9309 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9311 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
9312 (adjust_setup_cost): New parameter supporting round up adjustment.
9313 (struct address_cost_data): Delete.
9314 (force_expr_to_var_cost): Don't bound cost with spill_cost.
9315 (split_address_cost, ptr_difference_cost): Delete.
9316 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
9317 (struct ainc_cost_data): New struct.
9318 (get_address_cost_ainc): New function.
9319 (get_address_cost, get_computation_cost): Reimplement.
9320 (determine_group_iv_cost_address): Record inv_expr for all uses of
9322 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
9323 (iv_ca_has_deps): Reimplemented to ...
9324 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
9326 (iv_ca_extend): Call iv_ca_more_deps.
9328 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9330 * tree-ssa-address.c (struct mem_address): Move to header file.
9331 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
9332 * tree-ssa-address.h (struct mem_address): Move from C file.
9333 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
9335 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9337 * tree-affine.h (aff_combination_type): New interface.
9338 (aff_combination_zero_p): Remove static.
9339 (aff_combination_const_p): New interface.
9340 (aff_combination_singleton_var_p): New interfaces.
9342 2017-05-11 Richard Biener <rguenther@suse.de>
9344 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9345 Skip unreachable blocks and destinations.
9346 (eliminate): Move stmt removal and fixup ...
9347 (fini_eliminate): ... here. Skip inserted exprs.
9348 (pass_pre::execute): Move fini_pre after fini_eliminate.
9349 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
9350 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
9351 PRE to get rid of dead code that has invalid SSA form and
9352 split critical edges again.
9354 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9356 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
9358 2017-05-11 Richard Biener <rguenther@suse.de>
9360 * passes.c (execute_function_todo): Verify loops if they are
9361 said to be up-to-date.
9362 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
9363 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
9365 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
9368 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
9369 handle calling assemble_external ourself.
9372 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
9373 modes with zero size. Enhance comment.
9375 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9377 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
9378 built-ins for vec_xl and vec_xst with short and char pointer
9381 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
9383 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
9384 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
9385 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
9386 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
9387 (_mm_maskz_min_round_ss): New intrinsics.
9388 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
9389 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
9390 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
9391 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
9392 (__builtin_ia32_minss_mask_round): New builtins.
9393 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9394 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9395 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
9397 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
9398 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
9400 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
9403 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
9405 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
9406 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
9407 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
9408 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
9409 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
9410 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
9411 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
9412 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9413 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
9414 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
9415 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
9416 (__builtin_ia32_mulss_mask_round): New builtins.
9417 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9418 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9419 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
9421 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
9422 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
9424 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
9427 2017-05-10 Julia Koval <julia.koval@intel.com>
9429 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
9430 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
9431 (_mm256_setr_m128i): New intrinsics.
9433 2017-05-10 Julia Koval <julia.koval@intel.com>
9435 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
9436 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
9437 (_mm_maskz_rcp14_ss): New intrinsics.
9438 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
9439 (__builtin_ia32_rcp14ss_mask): New builtins.
9440 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
9442 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
9444 PR tree-optimization/51513
9445 * tree-cfg.c (gimple_seq_unreachable_p): New function.
9446 (assert_unreachable_fallthru_edge_p): Use it.
9447 (group_case_labels_stmt): Likewise.
9448 * tree-cfg.h: Prototype it.
9449 * stmt.c: Include cfghooks.h and tree-cfg.h.
9450 (emit_case_dispatch_table) <gap_label>: New local variable.
9451 Use it to fill dispatch table gaps.
9452 Test for default_label before updating probabilities.
9453 (expand_case) <default_label>: Remove unneeded initialization.
9454 Test for unreachable default case statement and remove its edge.
9455 Set default_label accordingly.
9456 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
9458 2017-05-10 Carl Love <cel@us.ibm.com>
9460 * config/rs6000/rs6000-c: Add support for built-in functions
9461 vector signed char vec_neg (vector signed char)
9462 vector signed short int vec_neg (vector short int)
9463 vector signed int vec_neg (vector signed int)
9464 vector signed long long vec_neg (vector signed long long)
9465 vector float vec_neg (vector float)
9466 vector double vec_neg (vector double)
9467 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
9469 * config/rs6000/altivec.h: Add define for vec_neg
9470 * doc/extend.texi: Update the built-in documentation for the
9471 new built-in functions.
9473 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9475 PR tree-optimization/77644
9476 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
9478 2017-05-10 Nathan Sidwell <nathan@acm.org>
9480 * dumpfile.h (TDI_lang_all): New.
9481 (TDF_KIND): New. Renumber others
9482 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
9484 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
9486 (get_dump_file_name): Adjust suffix generation.
9487 (dump_enable_all): Use TDF_KIND.
9488 * doc/invoke.texi (-fdump-lang-all): Document.
9490 * dumpfile.h: Tabify.
9492 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
9495 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
9496 Move member access before delete.
9498 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
9500 * tree-inline.c (expand_call_inline): Split block at stmt
9503 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9506 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
9507 are now unused after splitting mov{sf,sd}_hardfloat.
9508 (f32_lr2): Likewise.
9510 (f32_lm2): Likewise.
9512 (f32_li2): Likewise.
9515 (f32_sr2): Likewise.
9517 (f32_sm2): Likewise.
9519 (f32_si2): Likewise.
9522 (f32_vsx): Likewise.
9524 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
9525 For movsf, order stores so the VSX stores occur before the GPR
9526 store which encourages the register allocator to use a traditional
9527 FPR instead of a GPR. For movsd, order the stores so that the GPR
9528 store comes before the VSX stores to allow the power6 to work.
9529 This is due to the power6 not having a 32-bit integer store
9530 instruction from a FPR.
9531 (movsf_hardfloat): Likewise.
9532 (movsd_hardfloat): Likewise.
9534 2017-05-09 Martin Sebor <msebor@redhat.com>
9536 PR translation/80280
9537 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
9540 PR translation/80280
9541 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
9542 data member added in r247778.
9543 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
9545 2017-05-09 Nathan Sidwell <nathan@acm.org>
9547 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
9549 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
9552 2017-05-09 Marek Polacek <polacek@redhat.com>
9554 * doc/invoke.texi: Fix typo.
9556 2017-05-09 Richard Biener <rguenther@suse.de>
9558 * tree-vrp.c (vrp_val_is_max): Adjust comment.
9559 (vrp_val_is_min): Likewise.
9560 (set_value_range_to_value): Likewise.
9561 (set_value_range_to_nonnegative): Likewise.
9562 (gimple_assign_nonzero_p): Likewise.
9563 (gimple_stmt_nonzero_p): Likewise.
9564 (vrp_int_const_binop): Likewise. Remove unreachable case.
9565 (adjust_range_with_scev): Adjust comments.
9566 (compare_range_with_value): Likewise.
9567 (extract_range_from_phi_node): Likewise.
9568 (test_for_singularity): Likewise.
9570 2017-05-09 Richard Biener <rguenther@suse.de>
9572 * tree-vrp.c (get_single_symbol): Add assert that we don't
9573 get overflowed constants as invariant part.
9574 (compare_values_warnv): Add comment before the TREE_NO_WARNING
9575 checks. Use wi::cmp instead of recursing for integer constants.
9576 (compare_values): Just ignore whether we assumed undefined
9577 overflow instead of failing the compare.
9578 (extract_range_for_var_from_comparison_expr): Add comment before the
9579 TREE_NO_WARNING sets.
9580 (test_for_singularity): Likewise.
9581 (extract_range_from_comparison): Do not disable optimization
9582 when we assumed undefined overflow.
9583 (extract_range_basic): Remove init of unused var.
9585 2017-05-09 Richard Biener <rguenther@suse.de>
9587 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
9588 (extract_range_from_multiplicative_op_1): Adjust.
9589 (extract_range_from_binary_expr_1): Use int_const_binop.
9591 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
9594 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
9595 rs6000_store_data_bypass_p in seven define_bypass directives and
9596 in several comments.
9597 * config/rs6000/rs6000-protos.h: Add prototype for
9598 rs6000_store_data_bypass_p function.
9599 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
9600 function implements slightly different (rs6000-specific) semantics
9601 than store_data_bypass_p, returning false rather than aborting
9602 with assertion error when arguments do not satisfy the
9603 requirements of store data bypass.
9604 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
9605 rs6000_store_data_bypass_p.
9607 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
9609 * config/xtensa/xtensa-protos.h
9610 (xtensa_initial_elimination_offset): New declaration.
9611 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
9612 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
9613 macro definition, add case for FRAME_POINTER_REGNUM when
9614 FRAME_GROWS_DOWNWARD.
9615 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
9616 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
9617 xtensa_initial_elimination_offset.
9619 2017-05-08 Nathan Sidwell <nathan@acm.org>
9621 * doc/invoke.texi: Alphabetize -fdump options.
9623 2017-05-08 Martin Sebor <msebor@redhat.com>
9625 PR translation/80280
9626 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
9628 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
9630 * target.def (compute_frame_layout): New optional target hook.
9631 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
9632 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
9633 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
9635 * reload1.c (verify_initial_elim_offsets): Likewise.
9636 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
9637 (use_simple_return_p): Call arm_compute_frame_layout if needed.
9638 (arm_get_frame_offsets): Split up into this ...
9639 (arm_compute_frame_layout): ... and this function.
9641 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
9643 * config/aarch64/constraints.md (Usa): New constraint.
9644 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
9646 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
9648 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
9649 with_multilib_list after it has been checked.
9651 2017-05-08 Richard Biener <rguenther@suse.de>
9653 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
9654 (bitmap_set_subtract_values): Likewise.
9656 2017-05-08 Richard Biener <rguenther@suse.de>
9658 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
9659 (gimple_assign_nonzero): ... this and remove strict_overflow_p
9661 (gimple_stmt_nonzero_warnv_p): Rename to ...
9662 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
9664 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
9665 (extract_range_basic): Adjust, do not disable propagation on
9666 strict overflow sensitive simplification.
9667 (vrp_visit_cond_stmt): Likewise.
9669 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
9671 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
9672 body size unconditionally.
9674 2017-05-07 Jeff Law <law@redhat.com>
9677 2017-05-06 Jeff Law <law@redhat.com>
9678 PR tree-optimization/78496
9679 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9682 PR tree-optimization/78496
9683 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9684 (simplify_stmt_using_ranges): Call it.
9685 (vrp_dom_walker::before_dom_children): Extract equivalences
9686 from an ASSERT_EXPR with an equality comparison against a
9689 2017-05-06 Jeff Law <law@redhat.com>
9691 PR tree-optimization/78496
9692 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9695 PR tree-optimization/78496
9696 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9697 (simplify_stmt_using_ranges): Call it.
9698 (vrp_dom_walker::before_dom_children): Extract equivalences
9699 from an ASSERT_EXPR with an equality comparison against a
9702 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9704 * lra-constraints.c (lra_copy_reg_equiv): New function.
9705 (split_reg): Use it to copy equivalence information from the
9706 original register to the spill register.
9708 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9710 PR rtl-optimization/75964
9711 * simplify-rtx.c (simplify_const_relational_operation): Remove
9712 invalid handling of comparisons of integer ABS.
9714 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
9716 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
9718 (init_regs): Remove declaration.
9719 (function_arg_advance_32): Initialize error_p as boolean variable.
9721 2017-05-05 Nathan Sidwell <nathan@acm.org>
9723 * store-motion.c (remove_reachable_equiv_notes): Reformat long
9724 lines. Use for (;;).
9726 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9728 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
9729 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
9730 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
9731 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
9732 VF=2 that require versioning.
9734 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9736 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
9739 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9741 * diagnostic.h (diagnostic_override_option_index): Convert from
9742 macro to inline function.
9744 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9746 * diagnostic.c (last_module_changed_p): New function.
9747 (set_last_module): New function.
9748 (diagnostic_report_current_module): Convert macro usage to
9749 the above functions.
9750 * diagnostic.h (diagnostic_context::last_module): Strengthen
9751 from const line_map * to const line_map_ordinary *.
9752 (diagnostic_last_module_changed): Delete macro.
9753 (diagnostic_set_last_module): Delete macro.
9755 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9757 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
9758 with diagnostic_report_diagnostic.
9759 (diagnostic_n_impl_richloc): Likewise.
9760 * diagnostic.h (report_diagnostic): Delete macro.
9761 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
9762 with diagnostic_report_diagnostic.
9763 * substring-locations.c (format_warning_va): Likewise.
9765 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9767 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
9768 save/restor of format_spec. Move option-printing code to...
9769 (print_option_information): ...this new function, and
9770 reimplement by simply printing to the pretty_printer,
9771 rather than appending to the format string.
9773 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9775 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
9776 handling logic into...
9777 (update_effective_level_from_pragmas): ...this new function.
9779 2017-05-04 Andrew Waterman <andrew@sifive.com>
9781 * config/riscv/riscv.opt (mstrict-align): New option.
9782 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
9783 (SLOW_UNALIGNED_ACCESS): Define.
9784 (riscv_slow_unaligned_access): Declare.
9785 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
9787 (riscv_slow_unaligned_access): New variable.
9788 (rocket_tune_info): Set slow_unaligned_access to true.
9789 (optimize_size_tune_info): Set slow_unaligned_access to false.
9790 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
9791 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
9792 (riscv_option_override): Set riscv_slow_unaligned_access.
9793 * doc/invoke.texi: Add -mstrict-align to RISC-V.
9795 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
9797 * config/riscv/riscv.md: Unify indentation.
9799 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
9804 * config/rs6000/rs6000.md (u code attribute): Add FIX and
9806 (extendsi<mode>2): Add support for doing sign extension via
9807 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
9808 don't have ISA 3.0 instructions.
9809 (extendsi<mode>2 splitter): Likewise.
9810 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
9811 generate the normal insns since SImode can now go in vector
9812 registers. Disallow the special UNSPECs needed for previous
9813 machines to hide SImode being used. Add new insns
9814 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
9815 (fix_trunc<mode>si2_stfiwx): Likewise.
9816 (fix_trunc<mode>si2_internal): Likewise.
9817 (fixuns_trunc<mode>si2): Likewise.
9818 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9819 (fctiw<u>z_<mode>_smallint): Likewise.
9820 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
9821 of floating point to 32-bit integer from doing a direct move to
9822 the GPR registers to do a store.
9823 (fctiwz_<mode>): Break long line.
9825 2017-05-05 Bin Cheng <bin.cheng@arm.com>
9827 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
9828 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
9829 (addr_list, addr_offset_valid_p): New.
9830 (split_address_groups): Check offset validity with above function.
9831 (gt-tree-ssa-loop-ivopts.h): Include header file.
9833 2017-05-05 Nathan Sidwell <nathan@acm.org>
9835 * config.gcc (arm*-*-*): Add missing 'fi'.
9837 2017-05-05 Steve Ellcey <sellcey@cavium.com>
9839 * doc/invoke.texi (-fopt-info): Explicitly say order of options
9840 included in -fopt-info does not matter.
9841 * doc/optinfo.texi (-fopt-info): Fix description of default
9842 behavour. Explicitly say order of options included in -fopt-info
9845 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
9847 * config.gcc: Allow combinations of aprofile and rmprofile values for
9848 --with-multilib-list.
9849 * config/arm/t-multilib: New file.
9850 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
9851 variables. Remove setting of ISA and floating-point ABI in
9852 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
9853 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
9854 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
9855 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
9857 * config/arm/t-rmprofile: Likewise except for the matches changes.
9858 * doc/install.texi (--with-multilib-list): Document the combination of
9859 aprofile and rmprofile values and warn about pitfalls in doing that.
9861 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9863 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
9864 (movdi_aarch64): Likewise.
9866 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9868 PR tree-optimization/80632
9869 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
9871 (build_arrays): Initialize it for virtual phis.
9872 (fix_phi_nodes): Use it for virtual phis.
9874 PR tree-optimization/80558
9875 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
9876 [x, y] op z into [x op, y op z] for op & or | if conditions
9879 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
9880 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9883 * config/arm/arm.md (use_literal_pool): Remove.
9884 (64-bit immediate split): No longer takes cost into consideration
9885 if arm_disable_literal_pool is enabled.
9886 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
9887 used when arm_disable_literal_pool is enabled.
9888 (arm_max_const_double_inline_cost): Remove use of
9889 arm_disable_literal_pool.
9890 (push_minipool_fix): Add assert.
9891 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
9892 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
9893 (no_literal_pool_sf_immediate): New.
9895 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9897 PR tree-optimization/80613
9898 * tree-ssa-dce.c (propagate_necessity): Remove cases for
9899 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
9901 2017-05-05 Richard Biener <rguenther@suse.de>
9903 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
9905 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
9907 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
9908 of this flag from insn conditions due to removal from r247495.
9910 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9912 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
9914 (arm_early_store_addr_dep_ptr): Likewise.
9915 * config/arm/aarch-common-protos.h
9916 (arm_early_load_addr_dep_ptr): Add prototype.
9917 (arm_early_store_addr_dep_ptr): Likewise.
9918 * config/arm/cortex-a53.md: Add new bypasses.
9920 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9922 * tree.c (next_type_uid): Change type to unsigned.
9923 (type_hash_canon): Decrement back next_type_uid if
9924 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
9925 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
9928 2017-05-04 Martin Sebor <msebor@redhat.com>
9930 * builtins.c: Fix a trivial typo in a comment.
9933 * builtins.c (check_sizes): Adjust to handle reading past the end.
9934 Avoid printing excessive upper bound of ranges. Use %E to print
9935 tree nodes instead of converting them to %wu.
9936 (expand_builtin_memchr): New function.
9937 (compute_dest_size): Rename...
9938 (compute_objsize): ...to this.
9939 (expand_builtin_memcpy): Adjust.
9940 (expand_builtin_mempcpy): Adjust.
9941 (expand_builtin_strcat): Adjust.
9942 (expand_builtin_strcpy): Adjust.
9943 (check_strncat_sizes): Adjust.
9944 (expand_builtin_strncat): Adjust.
9945 (expand_builtin_strncpy): Adjust and simplify.
9946 (expand_builtin_memset): Adjust.
9947 (expand_builtin_bzero): Adjust.
9948 (expand_builtin_memcmp): Adjust.
9949 (expand_builtin): Handle memcmp.
9950 (maybe_emit_chk_warning): Check strncat just once.
9952 2017-05-04 Martin Sebor <msebor@redhat.com>
9954 PR preprocessor/79214
9957 * builtins.c (check_sizes): Add inlining context and issue
9958 warnings even when -Wno-system-headers is set.
9959 (check_strncat_sizes): Same.
9960 (expand_builtin_strncat): Same.
9961 (expand_builtin_memmove): New function.
9962 (expand_builtin_stpncpy): Same.
9963 (expand_builtin): Handle memmove and stpncpy.
9965 2017-05-04 Bin Cheng <bin.cheng@arm.com>
9967 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
9968 which is not used any more.
9970 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9972 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
9974 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9976 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
9977 (cortexa53_tunings): Likewise.
9978 (cortexa57_tunings): Likewise.
9979 (cortexa72_tunings): Likewise.
9980 (cortexa73_tunings): Likewise.
9982 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9984 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
9985 Set loop alignment to 8.
9987 2017-05-04 Martin Sebor <msebor@redhat.com>
9989 PR translation/80280
9990 * builtins.c (expand_builtin_object_size): Add missing quoting to
9991 %D and like directives.
9992 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
9993 (hsa_type_for_tree_type): Same.
9994 (verify_function_arguments): Same.
9995 * symtab.c (symbol_table::change_decl_assembler_name): Same.
9996 * varasm.c (get_section): Same.
9999 2017-05-04 Martin Sebor <msebor@redhat.com>
10001 PR translation/80280
10002 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
10004 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
10006 * config/aarch64/aarch64.c (generic_addrcost_table):
10007 Change HI/TI mode setting.
10009 2017-05-04 Martin Jambor <mjambor@suse.cz>
10011 PR tree-optimization/80622
10012 * tree-sra.c (comes_initialized_p): New function.
10013 (build_accesses_from_assign): Only set write lazily when
10014 comes_initialized_p is false.
10015 (analyze_access_subtree): Use comes_initialized_p.
10016 (propagate_subaccesses_across_link): Assert !comes_initialized_p
10017 instead of testing for PARM_DECL.
10019 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10021 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
10022 constraint on operand 0 to allow more general addressing modes.
10023 Adjust output template.
10024 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
10026 * config/aarch64/aarch64-protos.h
10027 (aarch64_address_valid_for_prefetch_p): Declare prototype.
10028 * config/aarch64/constraints.md (Dp): New address constraint.
10029 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
10032 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
10034 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
10035 update use of estimate_ipcp_clone_size_and_time.
10036 (estimate_local_effects): Update use of
10037 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
10038 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
10039 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
10040 Return nonspecialized time.
10042 2017-05-04 Richard Biener <rguenther@suse.de>
10044 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
10045 for the last VUSE which def dominates the PHI. Directly call
10047 (get_continuation_for_phi_1): Remove.
10049 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
10051 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
10052 to explain the use of truncating division. Cap the number of
10053 iterations to the maximum given by nb_iterations_upper_bound,
10056 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
10058 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
10059 * configure: Regenerate.
10060 * config.in: Regenerate.
10061 * config/i386/driver-mingw32.c: new file.
10062 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
10063 * config.host: Link driver-mingw32.o on MinGW host.
10064 * doc/install.texi: Document new --enable-mingw-wildcard configure
10067 2017-05-04 Marek Polacek <polacek@redhat.com>
10069 PR tree-optimization/80612
10070 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
10072 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10073 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
10075 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
10076 (movt splitter): Likewise.
10077 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
10078 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
10079 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
10080 block for Thumb-1 with MOVT.
10081 (thumb2_legitimate_address_p): Move code block ...
10082 (can_avoid_literal_pool_for_label_p): ... into this new function.
10083 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
10085 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
10086 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
10087 "M-profile targets with the MOVT instruction".
10089 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10091 * config/arm/arm-builtins.c (arm_init_builtins): Rename
10092 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
10093 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
10095 2017-05-04 Martin Liska <mliska@suse.cz>
10097 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
10098 variable cond_code.
10100 2017-05-04 Richard Biener <rguenther@suse.de>
10102 * tree.c (array_at_struct_end_p): Handle arrays at struct
10103 end with flexarrays more conservatively. Refactor and treat
10104 arrays of arrays or aggregates more strict. Fix
10105 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
10106 * tree.c (array_at_struct_end_p): Adjust prototype.
10107 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
10108 * gimple-fold.c (get_range_strlen): Likewise.
10109 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
10111 2017-05-04 Richard Biener <rguenther@suse.de>
10113 PR tree-optimization/31130
10114 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
10116 (supports_overflow_infinity): Likewise.
10117 (is_negative_overflow_infinity): Likewise.
10118 (is_positive_overflow_infinity): Likewise.
10119 (is_overflow_infinity): Likewise.
10120 (stmt_overflow_infinity): Likewise.
10121 (overflow_infinity_range_p): Likewise.
10122 (usable_range_p): Remove as always returning true.
10123 (make_overflow_infinity): Remove.
10124 (negative_overflow_infinity): Likewise.
10125 (positive_overflow_infinity): Likewise.
10126 (avoid_overflow_infinity): Likewise.
10127 (set_value_range): Adjust accordingly.
10128 (set_value_range_to_nonnegative): Likewise, remove now unused
10129 overflow_infinity arg.
10130 (vrp_operand_equal_p): Adjust.
10131 (update_value_range): Likewise.
10132 (range_int_cst_singleton_p): Likewise.
10133 (operand_less_p): Likewise.
10134 (compare_values_warnv): Likewise.
10135 (extract_range_for_var_from_comparison_expr): Likewise.
10136 (vrp_int_const_binop): Likewise.
10137 (zero_nonzero_bits_from_vr): Likewise.
10138 (extract_range_from_multiplicative_op_1): Likewise.
10139 (extract_range_from_binary_expr_1): Likewise.
10140 (extract_range_from_unary_expr): Likewise.
10141 (extract_range_from_comparison): Likewise.
10142 (extract_range_basic): Likewise.
10143 (adjust_range_with_scev): Likewise.
10144 (compare_ranges): Likewise.
10145 (compare_range_with_value): Likewise.
10146 (dump_value_range): Likewise.
10147 (test_for_singularity): Likewise, remove strict_overflow_p parameter
10149 (simplify_cond_using_ranges): Adjust.
10151 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
10153 * brig-builtins.def: Added a builtin for class_f64.
10154 * builtin-types.def: Added a builtin type needed by class_f64.
10156 2017-05-03 Jason Merrill <jason@redhat.com>
10158 * timevar.def: Add TV_CONSTEXPR.
10160 2017-05-03 David Malcolm <dmalcolm@redhat.com>
10162 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
10164 2017-05-03 Martin Jambor <mjambor@suse.cz>
10166 * ipa-prop.c (ipa_update_after_lto_read): Removed.
10167 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
10168 * ipa-cp.c (ipcp_propagate_stage): Do not call
10169 ipa_update_after_lto_read.
10170 * ipa-inline.c (ipa_inline): Likewise.
10172 2017-05-03 Martin Jambor <mjambor@suse.cz>
10174 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
10175 tag. Added a default constructor and a destructor.
10176 (ipa_edge_args_sum_t): New class;
10177 (ipa_edge_args_sum): Declare.
10178 (ipa_edge_args_vector): Remove declaration.
10179 (IPA_EDGE_REF): Use ipa_edge_args_sum.
10180 (ipa_free_edge_args_substructures): Remove declaration.
10181 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
10182 (ipa_edge_args_info_available_for_edge_p): Likewise.
10183 * ipa-prop.c (ipa_edge_args_vector): Removed.
10184 (edge_removal_hook_holder): Likewise.
10185 (edge_duplication_hook_holder): Likewise.
10186 (ipa_edge_args_sum): New variable.
10187 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
10188 ipa_edge_args_vector.
10189 (ipa_free_edge_args_substructures): Likewise.
10190 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
10191 ipa_edge_args_vector.
10192 (ipa_edge_removal_hook): Turned into method
10193 ipa_edge_args_sum_t::remove.
10194 (ipa_edge_duplication_hook): Turned into method
10195 ipa_edge_args_sum_t::duplicate.
10196 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
10197 registering edge hooks.
10198 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
10199 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
10200 ipa_edge_args_sum instead of ipa_edge_args_vector.
10201 * ipa-profile.c (ipa_profile): Likewise.
10203 2017-05-03 Martin Jambor <mjambor@suse.cz>
10205 * symbol-summary.h (function_summary): New method exists.
10206 (function_summary::symtab_removal): Deallocate through release.
10207 (call_summary): New class.
10208 (gt_ggc_mx): New overload.
10209 (gt_pch_nx): Likewise.
10210 (gt_pch_nx): Likewise.
10212 2017-05-03 Jeff Law <law@redhat.com>
10214 PR tree-optimization/78496
10215 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
10216 from simplify_cond_using_ranges. Split off code to walk
10217 backwards through casts into ...
10218 (simplify_cond_using_ranges_2): New function.
10219 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
10220 (execute_vrp): After identifying jump threads, call
10221 simplify_cond_using_ranges_2.
10223 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
10226 * ipa-inline.h (inline_summary): Add ctor.
10227 (create_ggc): Do not use ggc_cleared_alloc.
10229 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
10230 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10232 * gcc.c (handle_braces): Support escaping in switch matching
10234 * doc/invoke.texi (Spec Files): Document it.
10235 Remove superfluous @code markup in items.
10237 2017-05-03 David Malcolm <dmalcolm@redhat.com>
10239 * diagnostic-show-locus.c (struct column_range): New struct.
10240 (get_affected_columns): New function.
10241 (get_printed_columns): New function.
10242 (struct correction): New struct.
10243 (correction::ensure_capacity): New function.
10244 (correction::ensure_terminated): New function.
10245 (struct line_corrections): New struct.
10246 (line_corrections::~line_corrections): New dtor.
10247 (line_corrections::add_hint): New function.
10248 (layout::print_trailing_fixits): Reimplement in terms of the new
10250 (selftest::test_overlapped_fixit_printing): New function.
10251 (selftest::diagnostic_show_locus_c_tests): Call it.
10253 2017-05-03 Nathan Sidwell <nathan@acm.org>
10255 Canonicalize canonical type hashing
10256 * tree.h (type_hash_canon_hash): Declare.
10257 * tree.c (type_hash_list, attribute_hash_list): Move into
10258 type_hash_canon_hash.
10259 (build_type_attribute_qual_variant): Break out hash code calc into
10260 type_hash_canon_hash.
10261 (type_hash_canon_hash): New. Generic type hash computation.
10262 (build_range_type_1, build_array_type_1, build_function_type,
10263 build_method_type_directly, build_offset_type, build_complex_type,
10264 make_vector_type): Call it.
10266 2017-05-03 Richard Biener <rguenther@suse.de>
10268 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
10269 When all DRs have unknown misaligned do not always peel
10270 when there is a store but apply the same costing model as if
10271 there were only loads.
10273 2017-05-03 Richard Biener <rguenther@suse.de>
10276 PR tree-optimization/80492
10277 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
10278 compare_base_decls returning dont-know properly.
10280 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
10282 * config/arm/iterators.md (CCSI): New mode iterator.
10283 (arch): New mode attribute.
10284 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
10285 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
10286 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
10287 code iterator for success result mode.
10288 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
10289 the corresponding new insn generators.
10291 2017-05-03 Bin Cheng <bin.cheng@arm.com>
10294 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10295 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
10297 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
10299 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
10300 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
10301 (DDR_A): Wrap DDR argument in brackets.
10302 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
10303 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
10304 (DDR_REVERSED_P): Likewise.
10306 2017-05-03 Jakub Jelinek <jakub@redhat.com>
10308 PR tree-optimization/79472
10309 * tree-switch-conversion.c (struct switch_conv_info): Add
10310 contiguous_range and default_case_nonstandard fields.
10311 (collect_switch_conv_info): Compute contiguous_range and
10312 default_case_nonstandard fields, don't clear final_bb if
10313 contiguous_range and only the default case doesn't have the required
10315 (check_all_empty_except_final): Set default_case_nonstandard instead
10316 of failing if contiguous_range and the default case doesn't have empty
10318 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
10319 and only the default case doesn't have the required constants. Skip
10321 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
10322 if default_case_nonstandard.
10323 (build_constructors): Build constant 1 just once. Assert that default
10324 values aren't inserted in between cases if contiguous_range. Skip
10326 (build_arrays): Skip virtual phis.
10327 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
10328 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
10329 Handle virtual phis.
10330 (gen_inbound_check): Handle default_case_nonstandard case.
10331 (process_switch): Adjust check_final_bb caller. Call
10332 gather_default_values with the first non-default case instead of
10333 default case if default_case_nonstandard.
10335 2017-05-02 Nathan Sidwell <nathan@acm.org>
10337 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
10338 check. Fix formatting.
10340 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
10342 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
10343 errors when comparing specialized and unspecialized times.
10345 2017-05-02 David Malcolm <dmalcolm@redhat.com>
10347 * diagnostic-show-locus.c
10348 (layout::should_print_annotation_line_p): Make private.
10349 (layout::print_annotation_line): Make private.
10350 (layout::annotation_line_showed_range_p): Make private.
10351 (layout::show_ruler): Make private.
10352 (layout::print_source_line): Make private. Pass in line and
10353 line_width, rather than calling location_get_source_line. Drop
10355 (layout::print_leading_fixits): New method.
10356 (layout::print_any_fixits): Rename to...
10357 (layout::print_trailing_fixits): ...this, and make private.
10358 Don't print newline fixits.
10359 (diagnostic_show_locus): Move logic for printing one row into...
10360 (layout::print_line): ...this new function. Move the
10361 location_get_source_line call and error-handling from
10362 print_source_line to here. Call print_leading_fixits, and rename
10363 print_any_fixits to print_trailing_fixits.
10364 (selftest::test_fixit_insert_containing_newline): Update now that
10365 newlines are partially supported.
10366 (selftest::test_fixit_insert_containing_newline_2): New test.
10367 (selftest::test_fixit_replace_containing_newline): Update comments.
10368 (selftest::diagnostic_show_locus_c_tests): Call the new test.
10369 * edit-context.c (class added_line): New class.
10370 (class edited_line): Describe newline handling in comment.
10371 (edited_line::actually_edited_p): New method.
10372 (edited_line::print_content): Delete redundant decl.
10373 (edited_line::m_predecessors): New field.
10374 (edited_file::print_content): Call edited_line::print_content.
10375 (edited_file::print_diff): Update to support newlines.
10376 (edited_file::print_diff_hunk): Likewise.
10377 (edited_file::print_run_of_changed_lines): New function.
10378 (edited_file::print_diff_line): Convert to...
10379 (print_diff_line): ...this.
10380 (edited_file::get_effective_line_count): New function.
10381 (edited_line::edited_line): Initialize new field m_predecessors.
10382 (edited_line::~edited_line): Clean up m_predecessors.
10383 (edited_line::apply_fixit): Handle newlines.
10384 (edited_line::get_effective_line_count): New function.
10385 (edited_line::print_content): New function.
10386 (edited_line::print_diff_lines): New function.
10387 (selftest::test_applying_fixits_insert_containing_newline): New
10389 (selftest::test_applying_fixits_replace_containing_newline): New
10391 (selftest::insert_line): New function.
10392 (selftest::test_applying_fixits_multiple_lines): Add example of
10394 (selftest::edit_context_c_tests): Call the new tests.
10396 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10398 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
10399 parameter cand. Update dump information.
10400 (get_computation_cost): Update uses.
10402 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10404 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
10405 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
10406 (get_computation_at, rewrite_use_address): Update use of
10407 get_computation_aff.
10409 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10411 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
10412 (get_computation): Delete.
10413 (get_computation_cost): Implement like get_computation_cost_at.
10414 Use get_computation_at.
10415 (get_computation_cost_at): Delete.
10416 (rewrite_use_nonlinear_expr): Use get_computation_at.
10417 (rewrite_use_compare, remove_unused_ivs): Ditto.
10419 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10421 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
10423 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10425 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
10426 (ivopts_global_cost_for_size): Rename parameter and update uses.
10427 (iv_ca_recount_cost): Update uses.
10428 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
10429 candidates seperately in n_invs and n_cands.
10430 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
10432 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10434 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
10435 (find_inv_vars_cb): New.
10436 (find_depends): Renamed to ...
10437 (find_inv_vars): ... this.
10438 (add_candidate_1, force_var_cost): Call find_inv_vars.
10439 (split_address_cost, determine_group_iv_cost_cond): Ditto.
10441 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10443 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
10444 inv_vars. Add inv_exprs.
10445 (struct iv_cand): Rename depends_on to inv_vars.
10446 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
10447 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
10448 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
10449 (dump_cand): Dump inv_vars.
10450 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
10451 (record_invariant, find_depends, add_candidate_1): Ditto.
10452 (set_group_iv_cost, force_var_cost): Ditto.
10453 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
10454 (get_computation_cost_at, get_computation_cost): Ditto.
10455 (determine_group_iv_cost_generic): Ditto.
10456 (determine_group_iv_cost_address): Ditto.
10457 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
10458 (determine_group_iv_costs): Ditto.
10459 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
10460 (iv_ca_set_remove_invariants): Renamed to ...
10461 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
10462 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
10463 (iv_ca_set_add_invariants): Renamed to ...
10464 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
10465 (iv_ca_set_cp): Use iv_ca_set_add_invs.
10466 (iv_ca_has_deps): Support inv_vars and inv_exprs.
10467 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
10468 (create_new_ivs): Remove useless dump.
10470 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10472 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
10474 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
10475 (iv_ca_set_no_cp, create_new_iv): Ditto.
10477 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10479 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
10481 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10483 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
10484 function tree_check2.
10486 2017-05-02 Martin Liska <mliska@suse.cz>
10488 * doc/gcov.texi: Add missing preposition.
10489 * gcov.c (function_info::function_info): Properly fill up
10490 all member variables.
10492 2017-05-02 Tamar Christina <tamar.christina@arm.com>
10494 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
10496 2017-05-02 Tamar Christina <tamar.christina@arm.com>
10498 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
10500 2017-05-02 Martin Liska <mliska@suse.cz>
10503 * lto-streamer-in.c (lto_read_tree_1): Remove
10504 LTO_STREAMER_DEBUG.
10505 * lto-streamer.c (struct tree_hash_entry): Likewise.
10506 (struct tree_entry_hasher): Likewise.
10507 (tree_entry_hasher::hash): Likewise.
10508 (tree_entry_hasher::equal): Likewise.
10509 (lto_streamer_init): Likewise.
10510 (lto_orig_address_map): Likewise.
10511 (lto_orig_address_get): Likewise.
10512 (lto_orig_address_remove): Likewise.
10513 * lto-streamer.h: Likewise.
10514 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
10515 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
10517 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
10519 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
10520 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
10521 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
10522 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
10523 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
10524 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
10525 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
10526 (mm_maskz_sub_ss): New intrinsics.
10527 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
10528 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
10529 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
10530 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
10531 (__builtin_ia32_subss_mask_round): New builtins.
10532 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
10533 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
10534 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
10536 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
10537 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
10539 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
10542 2017-05-02 Martin Jambor <mjambor@suse.cz>
10544 PR tree-optimization/78687
10545 * tree-sra.c (access): New field parent.
10546 (process_subtree_disqualification): New function.
10547 (disqualify_candidate): Call it.
10548 (build_accesses_from_assign): Reset write flag if creating an
10550 (build_access_subtree): Fill in parent field and also prpagate
10551 down grp_write flag.
10552 (create_artificial_child_access): New parameter set_grp_write, set
10553 grp_write to its value.
10554 (propagate_subaccesses_across_link): Also propagate grp_write flag
10556 (propagate_all_subaccesses): Push the closest parent back to work
10557 queue if add_access_to_work_queue returned true.
10559 2017-05-02 Richard Biener <rguenther@suse.de>
10561 * common.opt (fstrict-overflow): Alias negative to fwrapv.
10562 * doc/invoke.texi (fstrict-overflow): Remove all traces of
10563 -fstrict-overflow documentation.
10564 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
10565 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
10566 flag_strict_overflow.
10567 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
10568 * lto-opts.c (lto_write_options): Do not stream it.
10569 * lto-wrapper.c (merge_and_complain): Do not handle it.
10570 * opts.c (default_options_table): Do not set -fstrict-overflow.
10571 (finish_options): Likewise do not clear it when sanitizing.
10572 * simplify-rtx.c (simplify_const_relational_operation): Do not
10573 test flag_strict_overflow.
10575 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10577 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
10578 using enabled attribute.
10579 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
10580 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
10581 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
10582 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
10583 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
10584 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
10585 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
10586 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
10587 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
10588 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
10590 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10592 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
10594 2017-05-02 Richard Biener <rguenther@suse.de>
10596 PR tree-optimization/80591
10598 2017-04-10 Richard Biener <rguenther@suse.de>
10600 * tree-ssa-structalias.c (find_func_aliases): Properly handle
10603 2017-05-02 Richard Biener <rguenther@suse.de>
10605 PR tree-optimization/80549
10606 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
10607 (cleanup_tree_cfg_noloop): Create forwarders to known loop
10608 headers if they do not have a preheader.
10610 2017-05-02 Martin Liska <mliska@suse.cz>
10613 * common.opt: Fix typo.
10614 * doc/invoke.texi: Likewise.
10616 2017-05-01 Jan Beulich <jbeulich@suse.com>
10618 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
10619 swapping, add (x,x,m,x,n) alternative.
10621 2017-05-01 Nathan Sidwell <nathan@acm.org>
10623 * calls.c (combine_pending_stack_adjustment_and_call): Remove
10624 unnecessary unadjusted_alignment check.
10626 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10629 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
10631 * gimplify.c (gimplify_cilk_detach): New function.
10632 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
10633 * tree-core.h: Document EXPR_CILK_SPAWN.
10634 * tree.h (EXPR_CILK_SPAWN): Define.
10636 2017-05-01 David Malcolm <dmalcolm@redhat.com>
10638 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
10639 to use new fixit_hint representation, using the "replace" logic.
10640 (get_line_span_for_fixit_hint): Likewise.
10641 (layout::print_any_fixits): Likewise.
10642 (selftest::test_one_liner_many_fixits): Rename to...
10643 (selftest::test_one_liner_many_fixits_1): ...this, and update
10644 comment and expected output to reflect that the multiple fix-it
10645 hints are now consolidated into one insertion.
10646 (selftest::test_one_liner_many_fixits_2): New test.
10647 (selftest::test_diagnostic_show_locus_one_liner): Update for
10649 (selftest::test_fixit_consolidation): Update for fix-it API
10651 * diagnostic.c (print_parseable_fixits): Likewise.
10652 * edit-context.c (edited_line::m_line_events): Convert from
10653 auto_vec <line_event *> to auto_vec <line_event>.
10654 (class line_event): Convert from abstract base class to a concrete
10655 class, taking over the role of replace_event.
10656 (class insert_event): Delete.
10657 (class replace_event): Rename to class line_event. Convert to
10659 (edit_context::add_fixits): Reimplement.
10660 (edit_context::apply_insert): Delete.
10661 (edit_context::apply_replace): Rename to...
10662 (edit_context::apply_fixit): ...this. Convert to half-open range.
10663 (edited_file::apply_insert): Delete.
10664 (edited_file::apply_replace): Rename to...
10665 (edited_file::apply_fixit): ...this.
10666 (edited_line::~edited_line): Drop deletion of events.
10667 (edited_line::apply_insert): Delete.
10668 (edited_line::apply_replace): Rename to...
10669 (edited_line::apply_fixit): ...this. Convert to half-open range.
10670 Update for change to type of m_line_events.
10671 * edit-context.h (edit_context::apply_insert): Delete.
10672 (edit_context::apply_replace): Rename to...
10673 (edit_context::apply_fixit): ...this.
10675 2017-05-01 Martin Sebor <msebor@redhat.com>
10677 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
10680 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
10683 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
10684 __get_cpuid_max returns 0.
10685 (__get_cpuid_count): Ditto.
10687 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
10689 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
10690 replacement expression is another instance of one of its arguments.
10692 2017-05-01 Jakub Jelinek <jakub@redhat.com>
10695 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
10696 check for stack push/pop autoinc.
10697 * config/i386/i386.c (ix86_agi_dependent): Return false
10698 if the only reason why modified_in_p returned true is that
10699 addr is SP based and set_insn is a push or pop.
10701 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10703 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
10706 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10709 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
10710 (account_size_time): Use two predicates - exec_pred and
10712 (evaluate_conditions_for_known_args): Compute both clause and
10714 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
10715 (inline_summary_t::duplicate): Update.
10716 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
10718 (compute_inline_parameters): Likewise.
10719 (estimate_edge_size_and_time): Update caluclation of time.
10720 (estimate_node_size_and_time): Compute both time and nonspecialized
10722 (estimate_ipcp_clone_size_and_time): Update.
10723 (inline_merge_summary): Update.
10724 (do_estimate_edge_time): Update.
10725 (do_estimate_edge_size): Update.
10726 (do_estimate_edge_hints): Update.
10727 (inline_read_section, inline_write_summary): Stream both new predicates.
10728 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
10730 (compute_inlined_call_time): Cleanup.
10731 (big_speedup_p): Update.
10732 (edge_badness): Update.
10733 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
10734 (size_time_entry): Replace predicate by exec_predicate and
10735 nonconst_predicate.
10736 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
10737 (estimate_edge_time): Return also nonspec_time.
10738 (reset_edge_growth_cache): Update.
10740 2017-04-29 Jakub Jelinek <jakub@redhat.com>
10742 PR rtl-optimization/80491
10743 * ifcvt.c (noce_process_if_block): When looking for x setter
10744 with missing else_bb, don't check only the insn right before
10745 cond_earliest, but look for the last insn that x is modified in
10746 within the same bb.
10748 PR rtl-optimization/80491
10749 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
10751 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
10753 PR tree-optimization/80487
10754 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
10756 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10758 PR tree-optimization/79697
10759 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
10760 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
10761 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
10763 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
10764 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
10766 2017-04-28 Martin Sebor <msebor@redhat.com>
10768 PR tree-optimization/80523
10769 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
10770 (init_target_to_host_charmap, target_to_host, target_strtol10): New
10772 (maybe_warn, format_directive, parse_directive): Use new functions.
10773 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
10775 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
10777 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
10779 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
10781 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
10782 target_header_dir): Set correctly.
10783 * configure: Regenerated.
10784 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
10785 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
10786 instead of SYSTEM_HEADER_DIR.
10788 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
10790 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
10791 (estimate_local_effects): Likewise.
10792 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
10793 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
10794 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
10795 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
10796 do_estimate_edge_time, estimate_edge_time): Likewise.
10797 * ipa-inline-analysis.c (estimate_node_size_and_time,
10798 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
10799 (estimate_time_after_inlining): Remove.
10801 2017-04-28 Martin Liska <mliska@suse.cz>
10803 * doc/gcov.texi: Enhance documentation of gcov.
10805 2017-04-28 Martin Liska <mliska@suse.cz>
10807 * doc/gcov.texi: Sort options in alphabetic order.
10808 * doc/gcov-dump.texi: Likewise.
10809 * doc/gcov-tool.texi: Likewise.
10810 * gcov.c (print_usage): Likewise.
10811 * gcov-dump.c (print_usage): Likewise.
10812 * gcov-tool.c (print_merge_usage_message): Likewise.
10813 (print_rewrite_usage_message): Likewise.
10814 (print_overlap_usage_message): Likewise.
10816 2017-04-28 Martin Liska <mliska@suse.cz>
10818 PR gcov-profile/53915
10819 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
10821 2017-04-28 Martin Liska <mliska@suse.cz>
10823 PR gcov-profile/79891
10824 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
10825 is marked by compiler as living on a line.
10826 (get_cycles_count): Remove usage of the union.
10827 (output_intermediate_file): Likewise.
10828 (find_source): Fix GNU coding style.
10829 (accumulate_line_counts): Remove old non-all block mode.
10830 (output_lines): Remove usage of the union.
10831 * profile.c (output_location): Include all BBs, even if
10832 belonging to a same line (and file) as a previous BB.
10834 2017-04-28 Martin Liska <mliska@suse.cz>
10836 * gcov.c (process_args): Handle new argument 'w'.
10837 (read_graph_file): Assign ID to BBs.
10838 (output_branch_count): Display BB # if verbose flag is set.
10839 (output_lines): Likewise for arcs.
10840 (print_usage): Add '--verbose' option help.
10841 * doc/gcov.texi: Document --verbose (-w) option.
10843 2017-04-28 Martin Liska <mliska@suse.cz>
10845 * gcov.c (struct block_location_info): New struct.
10846 (process_file): Fill up the new structure.
10847 (read_graph_file): Replace usage of encoding by the newly added
10849 (add_line_counts): Likewise.
10850 (accumulate_line_counts): Remove usage of the union.
10851 (function_info::function_info): New function.
10852 (function_info::~function_info): Likewise.
10853 (process_file): Call delete instead of release_function.
10854 (release_function): Release the function.
10855 (release_structures): Call delete instead of release_function.
10856 (solve_flow_graph): Replace usage of num_blocks.
10857 (find_exception_blocks): Likewise.
10858 (output_lines): Fix GNU coding style.
10860 2017-04-28 Martin Liska <mliska@suse.cz>
10863 * coverage.c (coverage_remove_note_file): New function.
10864 * coverage.h: Declare the function.
10865 * toplev.c (finalize): Clean if an error has been seen.
10867 2017-04-28 Martin Liska <mliska@suse.cz>
10869 PR gcov-profile/80031
10870 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
10871 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
10872 * gcov.c (read_graph_file): Read just number of blocks.
10873 * profile.c (branch_prob): Do not stream 0 flags per a basic
10876 2017-04-28 Martin Liska <mliska@suse.cz>
10878 * gcov-dump.c (tag_*): Add new argument to declarations.
10879 (dump_gcov_file): Likewise.
10880 (tag_blocks): Add and use new argument depth.
10881 (tag_arcs): Likewise.
10882 (tag_lines): Likewise.
10883 (tag_counters): Likewise.
10884 (tag_summary): Likewise.
10885 (dump_working_sets): Use depth to do a proper indentation.
10887 2017-04-28 Jakub Jelinek <jakub@redhat.com>
10890 * cgraph.h (symtab_node::debug_symtab): No longer inline.
10891 * symtab.c (symtab_node::debug_symtab): Move definition here.
10893 2017-04-28 Richard Biener <rguenther@suse.de>
10895 * lto-streamer.h (LTO_major_version): Bump to 7.
10897 2017-04-28 Richard Biener <rguenther@suse.de>
10899 * tree-vrp.c (assert_info): New struct.
10900 (add_assert_info): New helper.
10901 (register_edge_assert_for_2): Refactor to add asserts to a vector
10903 (register_edge_assert_for_1): Likewise.
10904 (register_edge_assert_for): Likewise.
10905 (finish_register_edge_assert_for): New helper actually registering
10906 asserts where live on edge.
10907 (find_conditional_asserts): Adjust.
10908 (find_switch_asserts): Likewise.
10909 (evrp_dom_walker::try_find_new_range): Generalize.
10910 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
10912 2017-04-27 Marek Polacek <polacek@redhat.com>
10915 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
10916 arg10 and arg11 to itype.
10918 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
10920 * doc/extend.texi (Object Size Checking): Improve grammar.
10922 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
10925 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
10926 that the logic for permitting reciprocal estimates matches that
10929 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10932 * tree.c (type_cache_hasher::equal): Only compare
10933 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
10934 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
10935 non-aggregate element types.
10936 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
10937 about the flag on ARRAY_TYPEs in the comment, formatting fix.
10939 2017-04-27 Richard Biener <rguenther@suse.de>
10941 PR middle-end/80533
10942 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
10943 stripping ARRAY_REFs from MEM_EXPR make sure we're not
10944 keeping a reference to a trailing array.
10946 2017-04-27 Richard Biener <rguenther@suse.de>
10948 PR middle-end/80539
10949 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
10950 being in loop-closed SSA form conservatively.
10951 (chrec_fold_multiply_poly_poly): Likewise.
10953 2017-04-27 Tamar Christina <tamar.christina@arm.com>
10955 PR middle-end/79665
10956 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
10957 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
10959 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10962 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
10963 (aarch64_function_arg_alignment): Return unsigned int again, but still
10964 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
10965 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
10966 Don't emit -Wpsabi note.
10967 (aarch64_function_arg_boundary): Likewise.
10968 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
10971 2017-04-26 Nathan Sidwell <nathan@acm.org>
10973 * tree.h (crc32_unsigned_n): Declare.
10974 (crc32_unsigned, crc32_unsigned): Make inline.
10975 * tree.c (crc32_unsigned_bits): Replace with ...
10976 (crc32_unsigned_n): ... this.
10977 (crc32_unsigned, crc32_byte): Remove.
10978 (crc32_string): Remove unnecessary braces.
10980 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
10982 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
10983 * ipa-inline-analysis.c (MAX_TIME): Remove.
10984 (account_size_time): Use sreal for time.
10985 (dump_inline_summary): Update.
10986 (estimate_function_body_sizes): Update.
10987 (estimate_edge_size_and_time): Update.
10988 (estimate_calls_size_and_time): Update.
10989 (estimate_node_size_and_time): Update.
10990 (inline_merge_summary): Update.
10991 (inline_update_overall_summary): Update.
10992 (estimate_time_after_inlining): Update.
10993 (inline_read_section): Update.
10994 (inline_write_summary): Update.
10995 * ipa-inline.c (compute_uninlined_call_time): Update.
10996 (compute_inlined_call_time): Update.
10997 (recursive_inlining): Update.
10998 (inline_small_functions): Update.
10999 (dump_overall_stats): Update.
11000 * ipa-inline.h: Include sreal.h.
11001 (size_time_entry): Turn time to sreal.
11002 (inline_summary): Turn self_time nad time to sreal.
11004 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
11006 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
11008 (sreal::stream_out, sreal::stream_in): New.
11009 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
11011 2017-04-25 Jakub Jelinek <jakub@redhat.com>
11013 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
11016 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
11019 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
11021 (dimode_scalar_chain::compute_convert_gain): Ditto.
11022 (dimode_scalar_chain::make_vector_copies): Ditto.
11023 (dimode_scalar_chain::convert_reg): Ditto.
11024 (dimode_scalar_chain::convert_insn): Ditto.
11025 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
11026 (VI248_AVX512BW_1): New mode iterator.
11027 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
11028 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
11031 2017-04-25 Martin Sebor <msebor@redhat.com>
11033 PR tree-optimization/80497
11034 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
11035 constants are representable in HOST_WIDE_INT.
11036 (parse_directive): Ditto.
11038 2017-04-25 Martin Sebor <msebor@redhat.com>
11041 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
11042 (new_zero_array): Adjust signature.
11043 (dom_info::dom_init): Used unsigned rather that size_t.
11044 (dom_info::dom_info): Same.
11046 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11047 Jakub Jelinek <jakub@redhat.com>
11050 * config/arm/arm.c: Include gimple.h.
11051 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
11052 returns negative, increment ncrn only if it returned positive.
11053 (arm_needs_doubleword_align): Return int instead of bool,
11054 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
11055 members, but if there is any such non-FIELD_DECL
11056 > PARM_BOUNDARY aligned decl, return -1 instead of false.
11057 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
11058 returns negative, increment nregs only if it returned positive.
11059 (arm_setup_incoming_varargs): Likewise.
11060 (arm_function_arg_boundary): Emit -Wpsabi note if
11061 arm_needs_doubleword_align returns negative, return
11062 DOUBLEWORD_ALIGNMENT only if it returned positive.
11064 2017-04-25 Marek Polacek <polacek@redhat.com>
11067 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
11068 first argument to type.
11070 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
11073 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
11074 type checks to test for compatibility instead of equality.
11076 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11077 Jakub Jelinek <jakub@redhat.com>
11080 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
11082 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
11083 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
11084 the alignment computation, but return their maximum in warn_alignment.
11085 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
11086 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
11088 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
11089 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
11092 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11094 * config/arc/simdext.md (dmpyh): Fix typo.
11096 2017-04-25 Richard Biener <rguenther@suse.de>
11098 PR tree-optimization/80492
11099 * alias.c (compare_base_decls): Handle registers with asm
11100 specification conservatively.
11101 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
11102 compare_base_decls returning dont-know properly.
11104 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11106 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
11107 (legitimate_offset_address_p): New function.
11108 (arc_legitimate_address_p): Use above function.
11110 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11112 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
11114 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11116 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
11117 ACCH registers whenever they are available.
11119 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11121 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
11122 double regs fix when not used.
11124 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11126 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
11128 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
11129 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
11131 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11133 * config/arc/arc.c (arc_output_addsi): Check for h-register class
11134 when emitting short ADD instructions.
11136 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11138 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
11140 (cmpsi_cc_c_insn): Likewise.
11141 (cbranchsi4_scratch): Compute proper instruction length using
11142 compact_hreg_operand.
11143 * config/arc/predicates.md (compact_hreg_operand): New predicate.
11145 2017-04-25 Richard Biener <rguenther@suse.de>
11147 PR middle-end/80509
11148 * passes.c (pass_manager::pass_manager): Initialize
11149 m_name_to_pass_map.
11151 2017-04-25 Richard Biener <rguenther@suse.de>
11153 PR tree-optimization/79201
11154 * tree-ssa-sink.c (statement_sink_location): Handle calls.
11156 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11159 * config/s390/vector.md: Split MEM->GPR vector moves for
11160 non-s_operand addresses.
11162 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11165 * config/s390/predicates.md (reload_const_wide_int_operand): New
11167 * config/s390/s390.md ("movti"): Remove d/P alternative.
11168 ("movti_bigconst"): New pattern definition.
11170 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11173 * s390-protos.h (s390_expand_cs_hqi): Removed.
11174 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
11175 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
11176 modes as well as CCZ1mode and CCZmode.
11177 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
11178 signature of s390_emit_compare_and_swap.
11179 (s390_expand_cs_hqi): Likewise, make static.
11180 (s390_expand_cs_tdsi): Generate an explicit compare before trying
11181 compare-and-swap, in some cases.
11182 (s390_expand_cs): Wrapper function.
11183 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
11185 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
11186 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
11187 patterns for small and large integers. Forbid symref memory operands.
11188 Move expander to s390.c. Require cc register.
11189 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
11190 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
11191 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
11192 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
11193 symref memory operands. Remove CC mode and call s390_match_ccmode
11195 ("atomic_exchange<mode>"): Allow and implement all integer modes.
11197 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11199 * config/s390/s390.md (define_peephole2): New peephole to help
11200 combining the load-and-test pattern with volatile memory.
11202 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11204 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
11205 with CCZmode for TARGET_Z196.
11207 2017-04-25 Jakub Jelinek <jakub@redhat.com>
11209 PR rtl-optimization/80501
11210 * combine.c (make_compound_operation_int): Set subreg_code to SET
11211 even for AND with mask of the sign bit of mode.
11213 PR rtl-optimization/80500
11214 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
11215 sum's initial value.
11217 2017-04-25 Julian Brown <julian@codesourcery.com>
11218 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11220 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
11222 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
11224 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
11226 2017-04-25 Julian Brown <julian@codesourcery.com>
11227 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11229 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
11230 (thunderx2t99_sha): New Reservation.
11232 2017-04-25 Julian Brown <julian@codesourcery.com>
11233 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11235 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
11236 type for 1-element load.
11238 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
11240 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
11242 2017-04-24 Martin Jambor <mjambor@suse.cz>
11244 PR tree-optimization/80293
11245 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
11246 char arrays not totally scalarizable if it is false.
11247 (analyze_all_variable_accesses): Pass correct value in the new
11248 parameter. Add a statistics counter.
11250 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
11252 PR middle-end/79931
11253 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
11255 2017-04-24 Richard Biener <rguenther@suse.de>
11257 PR tree-optimization/80494
11258 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
11259 out for complex types.
11261 2017-04-24 Richard Biener <rguenther@suse.de>
11263 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
11264 * tree-ssa-sccvn.c (print_scc): Print SCC size.
11265 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
11266 (DFS): Adjust and never fail.
11267 (sccvn_dom_walker::fail): Remove.
11268 (sccvn_dom_walker::before_dom_children): Adjust.
11269 (run_scc_vn): Likewise and never fail.
11270 * tree-ssa-pre.c (pass_pre::execute): Adjust.
11271 (pass_fre::execute): Likewise.
11273 2017-04-24 Richard Biener <rguenther@suse.de>
11275 PR tree-optimization/79725
11276 * tree-ssa-sink.c (statement_sink_location): Return whether
11277 failure reason was zero uses. Move that check later.
11278 (sink_code_in_bb): Deal with zero uses by removing the stmt
11281 2017-04-24 Richard Biener <rguenther@suse.de>
11284 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
11285 pointer-based references.
11287 2017-04-24 Richard Biener <rguenther@suse.de>
11290 * pass_manager.h (pass_manager::operator new): Remove.
11291 (pass_manager::operator delete): Likewise.
11292 * passes.c (pass_manager::operator new): Remove.
11293 (pass_manager::operator delete): Likewise.
11294 (pass_manager::pass_manager): Zero individual pass members.
11296 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
11299 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
11300 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
11301 Check "XEXP (src, 1)" operand here.
11302 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
11303 Check "XEXP (src, 1)" operand here.
11304 (dimode_scalar_chain::make_vector_copies): Detect count register
11305 of a shift instruction. Zero extend count register from QImode
11306 to DImode to satisfy vector shift pattern count operand predicate.
11307 Substitute vector shift count operand with a DImode copy.
11308 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
11311 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
11313 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
11314 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
11315 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
11316 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
11317 (UNSPEC_NOREX_MEM): Remove definition.
11319 2017-04-21 Richard Biener <rguenther@suse.de>
11321 PR tree-optimization/79547
11322 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
11323 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
11324 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
11325 without any constraints.
11327 2017-04-21 Richard Biener <rguenther@suse.de>
11329 PR tree-optimization/78847
11330 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
11332 2017-04-21 Richard Biener <rguenther@suse.de>
11334 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
11335 (build_distinct_type_copy): Likewise.
11336 (build_variant_type_copy): Likewise.
11337 * tree.c (build_qualified_type): Pass down mem-stat info.
11338 (build_distinct_type_copy): Likewise.
11339 (build_variant_type_copy): Likewise.
11341 2017-04-21 Richard Biener <rguenther@suse.de>
11343 PR tree-optimization/80237
11344 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
11346 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
11347 for a simplified result.
11349 2016-04-21 Richard Biener <rguenther@suse.de>
11351 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
11352 sth as strict as a simple_iv but a chrec without symbols and an
11353 operand defined in the loop we are peeling (and not some subloop).
11354 (propagate_constants_for_unrolling): Propagate all constants.
11356 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
11359 * config/i386/i386.c (print_reg): Remove assert for disalowed
11360 regno values, call output_operand_lossage instead.
11362 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
11365 * config/i386/constraints.md (Yc): New register constraint.
11366 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
11367 Use Yc constraint for alternative 2 of operand 0. Remove
11368 preferred_for_speed attribute.
11370 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
11372 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
11373 lastprivate clauses in SIMT case.
11375 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
11377 * doc/invoke.texi (-Wextra-semi): Document new warning option.
11379 2017-04-20 Richard Biener <rguenther@suse.de>
11381 PR tree-optimization/57796
11382 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
11383 as N scalar stores.
11384 (vect_model_load_cost): Cost gathers as N scalar loads.
11386 2017-04-20 Richard Biener <rguenther@suse.de>
11388 * ggc-page.c (ggc_allocated_p): Rename to ...
11389 (safe_lookup_page_table_entry): ... this and return the lookup
11391 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
11393 2017-04-20 Richard Biener <rguenther@suse.de>
11395 PR tree-optimization/80453
11396 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
11397 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
11398 from the conditions.
11399 (vn_phi_eq): Pass them down.
11400 (vn_phi_lookup): Record them.
11401 (vn_phi_insert): Likewise.
11403 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
11405 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
11406 uninitialized variable warning to avoid buffer overrun.
11408 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
11411 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
11412 is suppressed for '{ 0 }' in C.
11414 2017-04-20 Jakub Jelinek <jakub@redhat.com>
11416 * BASE-VER: Set to 8.0.0.
11418 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
11420 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
11421 priority .init_array and .fini_array section with SECTION_NOTYPE
11424 2017-04-20 Jakub Jelinek <jakub@redhat.com>
11426 PR middle-end/80423
11427 * tree.h (build_array_type): Add typeless_storage default argument.
11428 * tree.c (type_cache_hasher::equal): Also compare
11429 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
11430 (build_array_type): Add typeless_storage argument, set
11431 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
11433 (build_nonshared_array_type): Adjust build_array_type_1 caller.
11434 (build_array_type): Likewise. Add typeless_storage argument.
11436 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
11437 Jakub Jelinek <jakub@redhat.com>
11439 PR tree-optimization/80426
11440 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
11441 operation on symbolic operands, also compute the overflow for the
11442 invariant part when the operation degenerates into a negation.
11444 2017-04-19 Jakub Jelinek <jakub@redhat.com>
11447 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
11448 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
11451 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
11453 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
11456 * config/avr/avr.c (tree.h): Include it.
11457 (cgraph.h): Include it.
11458 (avr_encode_section_info): Don't warn for uninitialized progmem
11459 variable if it's just an alias.
11461 2017-04-19 Richard Biener <rguenther@suse.de>
11464 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
11465 when needed by AutoPGO.
11467 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
11470 * doc/lto.texi: Remove an extra 'that'.
11472 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
11474 PR rtl-optimization/80429
11475 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
11476 are only used in debug insns.
11478 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
11479 Vladimir Makarov <vmakarov@redhat.com>
11481 * config/sparc/predicates.md (input_operand): Add comment. Return
11482 true for any memory operand when LRA is in progress.
11483 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
11485 2017-04-18 Jeff Law <law@redhat.com>
11488 * mips.md ({return,simple_return}_internal): Do not overwrite
11491 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11493 PR tree-optimization/80443
11494 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
11495 instead of adding 1, subtract -1 and similarly instead of subtracting
11498 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
11500 PR rtl-optimization/80357
11501 * haifa-sched.c (tmp_bitmap): New variable.
11502 (model_recompute): Handle duplicate use records.
11503 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
11504 (free_global_sched_pressure_data): Free it.
11506 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
11509 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
11510 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
11511 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
11512 instead of SYSTEM_HEADER_DIR.
11514 2017-04-18 Jeff Law <law@redhat.com>
11516 PR middle-end/80422
11517 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
11518 predecessors after walking up the insn chain.
11520 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11523 * dwarf2out.c (modified_type_die): Try harder not to emit internal
11524 sizetype type into debug info.
11526 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
11529 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
11530 unneeded test for TARGET_UPPER_REGS_SF.
11531 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
11533 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11536 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
11537 instead of gsi_after_labels.
11539 2017-04-18 Jeff Law <law@redhat.com>
11541 * regcprop.c (maybe_mode_change): Avoid creating copies of the
11545 2017-04-13 Jeff Law <law@redhat.com>
11546 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11547 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11549 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
11552 * config/avr/avr.c (intl.h): Include it.
11553 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
11555 2017-04-18 Martin Liska <mliska@suse.cz>
11557 PR gcov-profile/78783
11558 * gcov-tool.c (gcov_output_files): Validate that destination
11559 file is either removed by the tool or by a user.
11561 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
11562 Guy Benyei <guybe@mellanox.com>
11564 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
11565 block, and do not negate it, the stored id is already negative.
11567 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
11569 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
11571 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
11574 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
11575 masks of options that should be turned off if the VSX vector
11576 options are turned off.
11577 (OTHER_P8_VECTOR_MASKS): Likewise.
11578 (OTHER_VSX_VECTOR_MASKS): Likewise.
11579 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
11580 rs6000_disable_incompatible_switches to validate no type switches
11582 (rs6000_incompatible_switch): New function to disallow turning on
11583 other vector options if -mno-vsx, -mno-power8-vector, or
11584 -mno-power9-vector are specified.
11586 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11588 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
11590 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11592 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
11593 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
11594 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
11595 (ARG_POINTER_CFA_OFFSET): Likewise.
11597 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11599 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
11600 conditions to take advantage of various optimizations.
11602 2017-04-13 Jeff Law <law@redhat.com>
11604 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11605 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11606 (zero_extendsidi2_dext): Likewise.
11608 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11611 * fold-const.c (fold_ternary_loc): Revert
11612 use op0 instead of fold_convert_loc (loc, type, arg0) part of
11615 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
11617 PR rtl-optimization/80343
11618 * lra-remat.c (update_scratch_ops): Assign original hard reg to
11619 new scratch pseudo.
11621 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
11624 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
11625 to ubsan_encode_value.
11627 2017-04-13 Jeff Law <law@redhat.com>
11629 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
11630 appearing in DEBUG_INSNs.
11632 2017-04-13 Martin Liska <mliska@suse.cz>
11634 PR gcov-profile/80413
11635 * gcov-io.c (gcov_write_string): Copy to buffer just when
11636 allocated size is greater than zero.
11638 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11641 * dwarf2out.c (decls_for_scope): Ignore declarations of
11642 current_function_decl in BLOCK_NONLOCALIZED_VARS.
11644 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
11647 * ipa-visibility.c (non_local_p): Fix typos.
11648 (localize_node): When localizing symbol in same comdat group,
11649 dissolve the group only when we know external symbols are going
11651 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
11653 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11655 PR tree-optimization/79390
11656 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
11657 order does not result in usable sequence, retry with reversed operand
11663 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
11664 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
11665 op0 instead of fold_convert_loc (loc, type, arg0).
11667 2017-04-12 Jeff Law <law@redhat.com>
11669 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
11670 has a delay slot in the generated code.
11672 * config/cris/cris.md (cris_preferred_reload_class): Return
11673 GENNONACR_REGS rather than GENERAL_REGS.
11675 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11678 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
11679 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
11680 signedness of the result type.
11682 2017-04-12 Richard Biener <rguenther@suse.de>
11683 Jeff Law <law@redhat.com>
11685 PR tree-optimization/80359
11686 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
11687 trim stores to TARGET_MEM_REFs.
11689 2017-04-12 Richard Biener <rguenther@suse.de>
11691 PR tree-optimization/79390
11692 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
11693 threading case even more.
11695 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
11698 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
11699 for quad_address_p for TImode, instead of just not indexed_address.
11701 2017-04-12 Richard Biener <rguenther@suse.de>
11702 Bernd Edlinger <bernd.edlinger@hotmail.de>
11704 PR middle-end/79671
11705 * alias.c (component_uses_parent_alias_set_from): Handle
11706 TYPE_TYPELESS_STORAGE.
11707 (get_alias_set): Likewise.
11708 * tree-core.h (tree_type_common): Add typeless_storage flag.
11709 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
11710 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
11711 for types containing members with TYPE_TYPELESS_STORAGE.
11712 (place_field): Likewise.
11713 (layout_type): Likewise for ARRAY_TYPE.
11714 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
11715 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
11716 TYPE_TYPELESS_STORAGE.
11717 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
11719 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11722 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
11723 first argument to type.
11725 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11729 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
11730 CONST0_RTX (mode) rather than const0_rtx where appropriate.
11731 (rs6000_expand_binop_builtin): Likewise.
11732 (rs6000_expand_ternop_builtin): Likewise; also add missing
11733 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
11734 vshasigma built-ins.
11735 * doc/extend.texi: Document that vec_xxpermdi's third argument
11736 must be a constant.
11738 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
11740 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
11741 Use shift_const cost parameter when calculating gain of STV shifts.
11743 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
11745 PR rtl-optimization/70478
11746 * lra-constraints.c (process_alt_operands): Check memory for
11747 disfavoring memory insn operand.
11749 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11751 PR middle-end/80100
11752 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
11753 left shift in unsigned HOST_WIDE_INT type.
11755 PR rtl-optimization/80385
11756 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
11757 (not (neg X)) into (plus X -1) for complex or non-integral modes.
11760 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
11761 if they have any depend clauses.
11763 2017-04-11 Martin Liska <mliska@suse.cz>
11766 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
11767 * ipa-split.c (split_function): Create a local comdat symbol
11768 if caller is in a comdat group.
11770 2017-04-11 Martin Liska <mliska@suse.cz>
11773 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
11776 2017-04-11 Martin Sebor <msebor@redhat.com>
11778 PR middle-end/80364
11779 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
11780 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
11782 (directive::set_width, directive::set_precision, format_character):
11784 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
11787 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
11790 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
11791 conflict, set target->arch_name instead of target->cpu_name.
11793 2017-04-11 Richard Biener <rguenther@suse.de>
11795 PR tree-optimization/80374
11796 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
11797 build_zero_cst, remove fold_convertible_p check again.
11799 2017-04-11 Martin Liska <mliska@suse.cz>
11802 * ubsan.c (instrument_object_size): Do not instrument register
11805 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11808 * config/i386/i386-builtin-types.def
11809 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
11810 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
11811 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
11812 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
11813 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
11814 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
11815 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
11816 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
11817 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
11818 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
11819 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
11820 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
11821 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
11822 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
11823 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
11824 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
11825 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
11826 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
11827 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
11828 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
11829 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
11830 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
11831 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
11832 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
11833 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
11834 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
11835 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
11836 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
11837 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
11838 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
11839 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
11840 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
11841 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
11842 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
11843 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
11844 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
11845 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
11846 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
11847 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
11848 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
11849 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
11850 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
11851 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
11852 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
11853 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
11854 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
11856 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
11857 flag to second_arg_count, handle 4 argument function type _COUNT
11858 aliases, handle second_arg_count on second argument rather than last.
11860 2017-04-10 Jeff Law <law@redhat.com>
11862 PR tree-optimization/80374
11863 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
11864 record anything if we can not convert integer_zero_node to the
11867 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
11870 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11871 Enhance special handling given to the TARGET_P9_MINMAX option in
11872 relation to certain other options.
11874 2017-04-10 Bin Cheng <bin.cheng@arm.com>
11876 PR tree-optimization/80153
11877 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
11878 remove POINTER_PLUS_EXPR's base part directly, rather than through
11881 2017-04-10 Richard Biener <rguenther@suse.de>
11882 Bin Cheng <bin.cheng@arm.com>
11884 PR tree-optimization/80153
11885 * tree-affine.c (aff_combination_to_tree): Get base pointer from
11886 the first element of pointer type aff_tree. Build result expr in
11888 (add_elt_to_tree): Convert to type unconditionally. Remove other
11889 fold_convert calls.
11890 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
11891 (rewrite_use_nonlinear_expr): Check invariant using iv information.
11893 2017-04-10 Richard Biener <rguenther@suse.de>
11895 * tree-ssa-structalias.c (find_func_aliases): Properly handle
11898 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
11900 PR rtl-optimization/70478
11901 * lra-constraints.c (curr_small_class_check): New.
11902 (update_and_check_small_class_inputs): New.
11903 (process_alt_operands): Update curr_small_class_check. Disfavor
11904 alternative insn memory operands. Check available regs for small
11907 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
11910 * config/mips/mips.opt (-mvirt): Update description.
11911 * doc/invoke.texi (-mvirt): Likewise.
11913 2017-04-10 Richard Biener <rguenther@suse.de>
11915 PR middle-end/80362
11916 * fold-const.c (fold_binary_loc): Look at unstripped ops when
11917 looking for NEGATE_EXPR in -A / -B to A / B folding.
11919 2017-04-10 Martin Liska <mliska@suse.cz>
11921 PR gcov-profile/80224
11922 * gcov.c (print_usage): Fix usage string.
11923 (get_gcov_intermediate_filename): Remove.
11924 (output_gcov_file): Use both for normal and intermediate format.
11925 (generate_results): Do not initialize special file for
11926 intermediate format.
11928 2017-04-10 Richard Biener <rguenther@suse.de>
11930 PR tree-optimization/80304
11931 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
11934 2017-04-10 Nathan Sidwell <nathan@acm.org>
11937 * config/rs6000/rs6000.c (rs6000_vector_type): New.
11938 (rs6000_init_builtins): Use it.
11940 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11942 * config/arm/arm.md (<mrc>): Add mode to SET source.
11943 (<mrrc>): Likewise.
11945 2017-04-10 Richard Biener <rguenther@suse.de>
11947 PR middle-end/80344
11948 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
11950 2017-04-10 Jakub Jelinek <jakub@redhat.com>
11953 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
11954 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
11955 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
11956 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
11957 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
11958 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
11959 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
11960 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
11961 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
11962 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
11963 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
11964 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
11965 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
11966 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
11967 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
11968 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
11969 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
11970 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
11971 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
11972 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
11973 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
11974 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
11975 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
11977 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
11979 PR rtl-optimization/70478
11980 * lra-constraints.c: Reverse the last patch.
11982 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
11984 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
11985 Add comment for WCHAR_T.
11987 2017-04-08 Martin Liska <mliska@suse.cz>
11990 2017-04-07 Martin Liska <mliska@suse.cz>
11993 * ipa-split.c (split_function): Add function part to a same comdat
11996 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11999 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
12001 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
12003 * rs6000/rs6000.c (vec_load_pendulum): Rename...
12004 (vec_pairing): ...to this.
12005 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
12006 (rs6000_sched_init): Adjust for name change.
12007 (struct rs6000_sched_context): Likewise.
12008 (rs6000_init_sched_context): Likewise.
12009 (rs6000_set_sched_context): Likewise.
12011 2017-04-07 Jakub Jelinek <jakub@redhat.com>
12017 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
12019 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
12020 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
12021 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
12023 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
12025 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
12027 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
12029 PR rtl-optimization/70703
12030 * ira-color.c (update_conflict_hard_regno_costs): Use
12031 int64_t instead of HOST_WIDE_INT.
12033 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
12035 PR rtl-optimization/70478
12036 * lra-constraints.c (process_alt_operands): Disfavor alternative
12037 insn memory operands.
12039 2017-04-07 Jeff Law <law@redhat.com>
12041 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
12042 CALL and NOTE_INSN_CALL_ARG_LOCATION.
12044 2017-04-07 Martin Liska <mliska@suse.cz>
12047 * config/aarch64/aarch64.c (aarch64_process_target_attr):
12048 Show error message instead of an ICE.
12050 2017-04-07 Martin Liska <mliska@suse.cz>
12053 * ipa-split.c (split_function): Add function part to a same comdat
12056 2017-04-07 Richard Biener <rguenther@suse.de>
12058 PR middle-end/80341
12059 * tree.c (get_unwidened): Also handle ! for_type case for
12061 * convert.c (do_narrow): Split out from ...
12062 (convert_to_integer_1): ... here. Do not pass final truncation
12063 type to get_unwidened for TRUNC_DIV_EXPR.
12065 2017-04-07 Richard Biener <rguenther@suse.de>
12067 * tree-affine.c (wide_int_ext_for_comb): Take type rather
12069 (aff_combination_const): Adjust.
12070 (aff_combination_scale): Likewise.
12071 (aff_combination_add_elt): Likewise.
12072 (aff_combination_add_cst): Likewise.
12073 (aff_combination_convert): Likewise.
12074 (add_elt_to_tree): Likewise. Remove unused argument.
12075 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
12077 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
12079 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
12081 * config/arm/arm.c (arm_default_short_enums): Use
12082 ARM_DEFAULT_SHORT_ENUMS.
12083 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
12085 2017-04-06 Jakub Jelinek <jakub@redhat.com>
12088 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
12089 members with redundant out-of-class redeclaration.
12091 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
12094 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
12095 * config/i386/i386.md (*zero_extendsidi2):
12096 Add (?*x,*x) and (?*v,*v) alternatives.
12098 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
12101 * config/i386/i386.c (ix86_expand_builtin)
12102 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
12103 mode from insn data. Convert operands to insn operand mode.
12104 Copy operands that don't satisfy insn predicate to a register.
12106 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
12108 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
12111 2017-04-06 Richard Biener <rguenther@suse.de>
12113 PR tree-optimization/80334
12114 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
12115 preserve alignment of accesses.
12117 2017-04-06 Richard Biener <rguenther@suse.de>
12119 PR tree-optimization/80262
12120 * tree-sra.c (build_ref_for_offset): Preserve address-space
12122 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
12123 Drop useless address-space information on MEM_REF offsets.
12125 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
12127 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
12129 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
12131 PR rtl-optimization/70703
12132 * ira-color.c (update_conflict_hard_regno_costs): Use
12133 HOST_WIDE_INT instead of long.
12135 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
12138 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
12139 not defined for x86_64 target. Add -mmmx target option when __SSE2__
12141 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
12142 for x86_64 target. Handle -m3dnowa option.
12144 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
12146 PR rtl-optimization/70703
12147 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
12148 (update_conflict_hard_regno_costs): Use long instead of unsigned
12149 arithmetic for cost calculation.
12151 2017-04-05 Jakub Jelinek <jakub@redhat.com>
12152 Bernd Edlinger <bernd.edlinger@hotmail.de>
12155 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
12158 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
12161 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
12162 ptr_mode with Pmode throughout.
12163 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
12164 into probe_stack_range and use DImode.
12166 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
12169 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
12170 call_eh_return is true.
12172 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12174 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
12175 Initialize last_match_fntype_index.
12177 2017-04-05 Jakub Jelinek <jakub@redhat.com>
12180 * tree-nvr.c: Include internal-fn.h.
12181 (pass_return_slot::execute): Ignore internal calls without
12184 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12185 Richard Biener <rguenther@suse.de>
12188 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
12189 captures used multiple times, except for the last use.
12190 * generic-match-head.c: Include gimplify.h.
12192 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12194 PR tree-optimization/79390
12195 * target.h (struct noce_if_info): Declare.
12196 * targhooks.h (default_noce_conversion_profitable_p): Declare.
12197 * target.def (noce_conversion_profitable_p): New target hook.
12198 * ifcvt.h (struct noce_if_info): New type, moved from ...
12199 * ifcvt.c (struct noce_if_info): ... here.
12200 (noce_conversion_profitable_p): Renamed to ...
12201 (default_noce_conversion_profitable_p): ... this. No longer
12203 (noce_try_store_flag_constants, noce_try_addcc,
12204 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
12205 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
12206 instead of noce_conversion_profitable_p.
12207 * config/i386/i386.c: Include ifcvt.h.
12208 (ix86_option_override_internal): Don't override
12209 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
12210 (ix86_noce_conversion_profitable_p): New function.
12211 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
12212 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
12213 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
12214 * doc/tm.texi: Regenerated.
12216 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12218 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
12221 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
12224 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
12225 instructions for small multiply cores.
12227 2017-04-04 Jeff Law <law@redhat.com>
12229 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
12231 (mips_expand_vec_perm_const): Initialize elements in orig_perm
12232 that are not set by the loop over the elements.
12234 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12237 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
12238 int mode, convert_modes it to mode as unsigned, otherwise use
12239 lowpart_subreg to mode rather than SImode.
12240 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
12241 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
12242 Use DImode instead of SImode for the shift count operand.
12243 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
12246 2017-04-04 Richard Biener <rguenther@suse.de>
12248 PR middle-end/80281
12249 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
12250 arithmetic done for the negate or the plus. Simplify.
12251 (A - (-B) -> A + B): Likewise.
12252 * fold-const.c (split_tree): Make sure to not negate pointers.
12254 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
12256 PR rtl-optimization/60818
12257 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
12258 a compare of comparisons with the thing compared if this results
12259 in a different machine mode.
12261 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
12263 * alias.c (base_alias_check): Fix typo in comment.
12264 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
12265 * cgraphunit.c (symbol_table::compile): Likewise.
12266 * collect2.c (maybe_run_lto_and_relink): Likewise.
12267 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
12268 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
12269 * config/avr/avr.c (avr_map_op_t): Likewise.
12270 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
12271 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
12272 * config/epiphany/epiphany.md (movcc): Likewise.
12273 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
12274 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
12276 * config/mips/mips.c (mips_save_restore_reg): Likewise.
12277 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
12278 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
12279 * config/sh/sh.c (sh_rtx_costs): Likewise.
12280 * fold-const.c (fold_truth_andor): Likewise.
12281 * genautomata.c (collapse_flag): Likewise.
12282 * gengtype.h (struct type::u::s): Likewise.
12283 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
12284 * input.c (FORMAT_AMOUNT): Likewise.
12285 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
12286 (known_aggs_to_agg_replacement_list): Likewise.
12287 * ipa-inline-analysis.c: Likewise.
12288 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
12289 * ipa-polymorphic-call.c
12290 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
12291 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
12292 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
12294 * modulo-sched.c (apply_reg_moves): Likewise.
12295 * omp-expand.c (build_omp_regions_1): Likewise.
12296 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
12297 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
12298 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
12299 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
12300 * value-prof.c: Likewise.
12301 * var-tracking.c (val_reset): Likewise.
12303 2017-04-03 Richard Biener <rguenther@suse.de>
12305 PR tree-optimization/80275
12306 * fold-const.c (split_address_to_core_and_offset): Handle
12309 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
12311 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
12312 descriptors is at least equal to that of functions.
12314 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
12316 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
12318 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
12321 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
12322 (mov<IMOD4:mode>): New expander.
12323 (*mov<IMOD4:mode>_internal): New insn and split pattern.
12325 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
12327 PR rtl-optimization/79405
12328 * fwprop.c (propagations_left): New variable.
12329 (forward_propagate_into): Decrement it.
12330 (fwprop_init): Initialize it.
12331 (fw_prop): If the variable has reached zero, stop propagating.
12332 (fwprop_addr): Ditto.
12334 2017-03-31 Jakub Jelinek <jakub@redhat.com>
12337 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
12338 a FUNCTION_DECL, pass it as decl instead of origin to
12341 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
12343 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
12346 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
12349 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
12350 TARGET_VSX_SMALL_INTEGER.
12352 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12354 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
12355 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
12357 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
12359 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
12360 extraction from odd-numbered MSA register.
12362 2017-03-31 Jakub Jelinek <jakub@redhat.com>
12364 PR middle-end/80173
12365 * expmed.c (store_bit_field_1): Don't attempt to create
12366 a word subreg out of hard registers wider than word if they
12367 have HARD_REGNO_NREGS of 1 for their mode.
12369 PR middle-end/80163
12370 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
12371 conversions to integer types wider than word and pointer.
12374 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
12375 (rtx_equal_for_cselib_p): Pass 0 to it.
12376 * cselib.c (cselib_hasher::equal): Likewise.
12377 (rtx_equal_for_cselib_1): Add depth argument. If depth
12378 is 128, don't look up VALUE locs and punt. Increment
12379 depth in recursive calls when walking VALUE locs.
12381 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
12383 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
12384 (make_gcov_file_name): Use the canonical path name for generating
12386 (read_line): Fix handling of files with ascii null bytes.
12388 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
12390 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
12391 to initialise a vector register instead
12392 of using a const_int.
12394 2017-03-30 Jakub Jelinek <jakub@redhat.com>
12396 PR translation/80189
12397 * gimplify.c (omp_default_clause): Use %qs instead of %s in
12398 diagnostic messages.
12400 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
12403 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
12404 (dfp_diex_<mode>): Update mode of operand 1.
12405 * doc/extend.texi (dxex, dxexq): Document change to return type.
12406 (diex, diexq): Document change to argument type.
12408 2017-03-30 Martin Jambor <mjambor@suse.cz>
12411 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
12412 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
12413 it reflects the signature changes performed at the callee side.
12414 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
12415 to cgraph_build_function_type_skip_args.
12416 (build_function_decl_skip_args): Adjust call to the above function.
12418 2017-03-30 Jakub Jelinek <jakub@redhat.com>
12421 * config/i386/sse.md
12422 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
12423 register as dest whenever it is a MEM not rtx_equal_p to the
12424 corresponding dup operand, and when forcing into reg move the
12425 reg into the memory afterwards.
12426 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
12427 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
12428 for the force_reg mode.
12429 (avx512vl_vextractf128<mode>): Use register as dest either
12430 always when a MEM, or when it is a MEM not rtx_equal_p to the
12431 corresponding dup operand, or even not when it is a CONST_VECTOR
12432 depending on the mode and lo vs. hi.
12433 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
12435 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
12436 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
12437 Likewise. Require that operands[2] is even.
12438 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
12439 Remove extraneous parens. Require that operands[2] is a multiple
12441 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
12442 operands[0] is a MEM if <mask_applied>, the predicates/constraints
12443 disallow memory then.
12445 2017-03-30 Richard Biener <rguenther@suse.de>
12447 PR tree-optimization/77498
12448 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
12449 to non-constants over backedges.
12451 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
12453 PR rtl-optimization/80233
12454 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
12455 as last_combined_insn. Do not test for BARRIER_P separately.
12457 2017-03-29 Andreas Schwab <schwab@suse.de>
12460 * calls.c (prepare_call_address): Convert funexp to Pmode before
12461 copying to temp reg.
12463 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12465 PR tree-optimization/80158
12466 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
12467 Handle possible future case of more than one alternate
12469 (replace_rhs_if_not_dup): Likewise.
12470 (replace_one_candidate): Likewise.
12472 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
12474 PR rtl-optimization/80193
12475 * ira.c (ira): Do not check allocation for LRA.
12477 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
12479 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
12480 (nvptx_output_simt_exit): Declare.
12481 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
12482 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
12483 (init_softstack_frame): Move initialization of crtl->is_leaf to...
12484 (nvptx_declare_function_name): ...here. Emit declaration of local
12485 memory space buffer for omp_simt_enter insn.
12486 (nvptx_output_unisimt_switch): New.
12487 (nvptx_output_softstack_switch): New.
12488 (nvptx_output_simt_enter): New.
12489 (nvptx_output_simt_exit): New.
12490 * config/nvptx/nvptx.h (struct machine_function): New fields
12491 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
12492 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
12493 (UNSPECV_SIMT_EXIT): Ditto.
12494 (omp_simt_enter_insn): New insn.
12495 (omp_simt_enter): New expansion.
12496 (omp_simt_exit): New insn.
12497 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
12499 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
12500 (expand_GOMP_SIMT_ENTER_ALLOC): New.
12501 (expand_GOMP_SIMT_EXIT): New.
12502 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
12503 (GOMP_SIMT_ENTER_ALLOC): Ditto.
12504 (GOMP_SIMT_EXIT): Ditto.
12505 * target-insns.def (omp_simt_enter): New insn.
12506 (omp_simt_exit): Ditto.
12507 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
12509 (lower_rec_simd_input_clauses): Implement SIMT privatization.
12510 (lower_rec_input_clauses): Likewise.
12511 (lower_lastprivate_clauses): Handle SIMT privatization.
12513 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
12514 (ompdevlow_adjust_simt_enter): New.
12515 (find_simtpriv_var_op): New.
12516 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
12517 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
12519 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
12520 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
12521 (copy_decl_for_dup_finish): Ditto.
12523 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
12525 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
12528 * config/i386/i386.c (ix86_option_override_internal): Always
12529 allow -mpreferred-stack-boundary=3 for 64-bit targets.
12531 2017-03-28 Bin Cheng <bin.cheng@arm.com>
12533 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
12535 2017-03-28 Bin Cheng <bin.cheng@arm.com>
12537 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
12538 mark new edge's irreducible flag accordign to it.
12539 (vect_do_peeling): Check loop preheader edge's irreducible flag
12540 and pass it to function slpeel_add_loop_guard.
12542 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
12544 PR tree-optimization/80218
12545 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
12546 Update block frequencies and counts.
12548 2017-03-28 Richard Biener <rguenther@suse.de>
12550 PR tree-optimization/78644
12551 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
12552 of a simplification result we may not use it at all.
12554 2017-03-28 Richard Biener <rguenther@suse.de>
12557 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
12558 without arguments, generate default definition of a SSA name.
12560 2017-03-28 Richard Biener <rguenther@suse.de>
12562 PR middle-end/80222
12563 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
12564 TYPE_REF_CAN_ALIAS_ALL references.
12565 * fold-const.c (fold_indirect_ref_1): Likewise.
12567 2017-03-28 Martin Liska <mliska@suse.cz>
12570 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
12571 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
12573 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12574 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
12576 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
12577 (EXTRA_SPECS): Define.
12578 (SUBTARGET_EXTRA_SPECS): Likewise.
12579 (SUBTARGET_CPP_SPEC): Likewise.
12580 * config/arc/elf.h (EXTRA_SPECS): Renamed to
12581 SUBTARGET_EXTRA_SPECS.
12582 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
12584 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12586 * config/arc/simdext.md (vst64_insn): Update pattern.
12587 (vld32wh_insn): Likewise.
12588 (vld32wl_insn): Likewise.
12589 (vld64_insn): Likewise.
12590 (vld32_insn): Likewise.
12592 2017-03-28 Marek Polacek <polacek@redhat.com>
12595 * fold-const.c (fold_comparison): Use protected_set_expr_location
12596 instead of SET_EXPR_LOCATION.
12598 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
12600 * tree.c (add_expr): Avoid name lookup warning.
12602 2017-03-27 Jeff Law <law@redhat.com>
12604 PR tree-optimization/80216
12605 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
12606 function name. Limit recursion depth.
12607 (record_temporary_equivalences): Corresponding changes.
12609 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
12611 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
12614 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12617 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
12619 * cfgcleanup.c (reg_note_cfa_p): New array.
12620 (insns_have_identical_cfa_notes): New function.
12621 (old_insns_match_p): Don't cross-jump in between /f
12622 and non-/f instructions. If both i1 and i2 are frame related,
12623 verify all CFA notes, their order and content.
12625 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
12628 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
12629 HImode and SImode with zero extend to DImode to one insn.
12630 (bswap<mode>2_extenddi): Likewise.
12631 (bswapsi2_extenddi): Likewise.
12632 (bswaphi2_extendsi): Likewise.
12633 (bswaphi2): Combine bswap HImode and SImode into one insn.
12634 Separate memory insns from swapping register.
12635 (bswapsi2): Likewise.
12636 (bswap<mode>2): Likewise.
12637 (bswaphi2_internal): Delete, no longer used.
12638 (bswapsi2_internal): Likewise.
12639 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
12640 store, and gpr<-gpr swap insns.
12641 (bswap<mode>2_store): Likewise.
12642 (bswaphi2_reg): Register only splitter, combine with the splitter.
12643 (bswaphi2 splitter): Likewise.
12644 (bswapsi2_reg): Likewise.
12645 (bswapsi2 splitter): Likewise.
12646 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
12647 the insns into load, store, and register/register insns.
12648 (bswapdi2_ldbrx): Likewise.
12649 (bswapdi2_load): Likewise.
12650 (bswapdi2_store): Likewise.
12651 (bswapdi2_reg): Likewise.
12653 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
12655 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
12656 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
12658 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12661 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
12663 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
12664 special handling for target option conflicts between dform
12665 options (-mpower9-dform, -mpower9-dform-vector,
12666 -mpower9-dform-scalar) and -mno-direct-move.
12668 2017-03-27 Richard Biener <rguenther@suse.de>
12670 PR tree-optimization/80181
12671 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
12673 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12675 * config/arc/predicates.md (move_double_src_operand): Replace the
12676 call to move_double_src_operand with a call to address_operand.
12678 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12680 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
12681 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
12682 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
12684 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12686 * config/arc/predicates.md (long_immediate_loadstore_operand):
12687 Consider scaled addresses cases.
12689 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12691 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
12692 restored when in interrupt.
12693 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
12694 doesn't have delay slot.
12696 2017-03-27 Richard Biener <rguenther@suse.de>
12699 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
12700 inlined thunk clones.
12702 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12705 * asan.c (instrument_derefs): Copy over last operand from
12706 original COMPONENT_REF to the new COMPONENT_REF with
12707 DECL_BIT_FIELD_REPRESENTATIVE.
12708 * ubsan.c (instrument_object_size): Likewise.
12710 2017-03-27 Richard Biener <rguenther@suse.de>
12712 PR tree-optimization/80170
12713 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
12714 sure DR/SCEV didnt fold in constants we do not see when looking
12715 at the reference base alignment.
12717 2017-03-27 Richard Biener <rguenther@suse.de>
12719 PR middle-end/80171
12720 * gimple-fold.c (fold_ctor_reference): Properly guard against
12721 NULL return value from canonicalize_constructor_val.
12723 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
12726 * config/i386/i386.c (ix86_expand_builtin)
12727 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
12728 flags reg setting and flags reg using instructions.
12729 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
12730 clobbering instructions to zero extend op2.
12732 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
12734 * doc/install.texi (Configuration) <--with-aix-soname>:
12735 Update link to AIX ld.
12737 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
12739 PR rtl-optimization/80160
12740 PR rtl-optimization/80159
12741 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
12742 reg_alternate_class into account.
12744 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
12747 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
12748 to consider in curr_insn_transform.
12750 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12752 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
12753 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
12754 and emit_mode_inner.
12756 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12758 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
12759 argument to the overloaded builtin variants. Use the new flag to
12760 deprecate certain builtin variants.
12761 * config/s390/s390-builtin-types.def: Add new builtin types.
12762 * config/s390/s390-builtins.h: Support new flags field for
12763 overloaded builtins.
12764 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
12765 (s390_macro_to_expand): Enable vector float data type.
12766 (s390_cpu_cpp_builtins_internal): Indicate support of the new
12767 builtins by incrementing the __VEC__ version number.
12768 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
12770 (s390_resolve_overloaded_builtin): Emit error messages depending
12771 on the builtin flags.
12772 * config/s390/s390.c (s390_expand_builtin): Support additional
12773 flags argument. Change error message to match the messages
12774 emitted in s390-c.c.
12775 * config/s390/s390.md: New UNSPEC_* constants.
12776 (op_type): Add new instruction types.
12777 * config/s390/vecintrin.h: Add new builtins and test data class
12779 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
12780 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
12781 (VEC_INEXACT, VEC_NOINEXACT): New constants.
12782 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
12783 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
12784 ("vec_mergel<mode>"): V_HW -> VEC_HW.
12786 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
12787 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
12788 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
12789 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
12791 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
12792 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
12793 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
12794 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
12796 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
12797 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
12798 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
12799 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
12800 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
12801 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
12802 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
12804 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
12805 ("vec_scatter_element<V_HW_4:mode>_DI")
12806 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
12807 ("vec_fpint<mode>", "vflls")
12808 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
12809 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
12810 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
12811 ("*vec_cmphe<mode>_cc"): ... these.
12813 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
12814 mode constant instead of magic value.
12816 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12818 * config/s390/s390.c (s390_expand_vec_compare): Support other
12819 vector floating point modes than just V2DF.
12820 (s390_expand_vcond): Likewise.
12821 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
12822 (s390_cannot_change_mode_class): Prevent mode changes between TF
12823 and V1TF in vector registers.
12824 * config/s390/s390.md (DF, SF): New mode attributes.
12825 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
12826 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
12827 SFmode support for VRs.
12828 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
12830 (VFT, VF_HW): New mode iterators.
12831 (vw, sdx): New mode attributes.
12832 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
12833 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
12834 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
12835 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
12836 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
12837 also the new vector floating point modes. Renaming to ...
12839 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
12840 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
12841 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
12842 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
12843 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
12844 ("vec_unordered<mode>"): ... these.
12846 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
12847 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
12848 ("*vec_extendv2df"): New insn definitions.
12850 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12852 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
12853 ("mulditi3_2", "*muldi3_sign"): New patterns.
12854 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
12855 rename the pattern definition.
12857 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12859 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
12861 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
12863 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12865 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
12866 instruction if possible.
12867 * config/s390/vector.md (vec_halfnumelts): New mode
12869 ("*vec_vllezlf<mode>"): New pattern.
12871 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12873 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
12874 ("popcountv4si2", "popcountv2di2"): Rename to ...
12875 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
12876 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
12878 ("popcount<mode>2_vxe"): New pattern.
12880 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12882 * common/config/s390/s390-common.c (processor_flags_table): Add
12884 * config.gcc: Add arch12.
12885 * config/s390/driver-native.c (s390_host_detect_local_cpu):
12886 Default to arch12 for unknown CPU model numbers.
12887 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
12888 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
12889 PROCESSOR_max sanity check.
12890 * config/s390/s390-opts.h (enum processor_type): Add
12892 * config/s390/s390.c (processor_table): Add arch12.
12893 (s390_expand_builtin): Add check for B_VXE flag.
12894 (s390_issue_rate): Add PROCESSOR_ARCH12.
12895 (s390_get_sched_attrmask): Likewise.
12896 (s390_get_unit_mask): Likewise.
12897 (s390_sched_score): Enable z13 scheduling for arch12.
12898 (s390_sched_reorder): Likewise.
12899 (s390_sched_variable_issue): Likewise.
12900 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
12902 (s390_tune_attr): Use z13 scheduling also for arch12.
12903 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
12904 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
12905 (TARGET_VXE_P): New macros.
12906 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
12907 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
12908 * config/s390/s390.opt: Add arch12 as processor_type.
12910 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12912 * config/s390/s390.md
12913 ("fixuns_truncdddi2", "fixuns_trunctddi2")
12914 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
12915 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
12917 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
12918 Rename expanders to ...
12920 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
12921 ("fixuns_truncdddi2_emu"): ... these.
12923 ("fixuns_trunc<mode>si2_emu"): New expander.
12925 ("*fixuns_truncdfdi2_z13"): Rename to ...
12926 ("*fixuns_truncdfdi2_vx"): ... this.
12928 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12930 * config/s390/2964.md: Remove the single element vector compare
12931 instructions which are no longer used.
12932 * config/s390/s390.c (s390_select_ccmode): Remove handling of
12934 (s390_canonicalize_comparison): Remove handling of DFmode
12936 (s390_expand_vec_compare_scalar): Remove function.
12937 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
12938 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
12940 ("*cmp<mode>_ccs"): Add wfcdb instruction.
12942 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12944 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
12946 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
12947 will anyway by matched by mov<mode>_64dfp.
12949 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12951 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
12952 vlef/vstef. Add missing operand to vleif.
12954 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12956 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
12957 pair for all vector types with 64 bit elements.
12958 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
12959 * config/s390/vector.md (V_HW_64): ... here.
12960 (V_128_NOSINGLE): New mode iterator.
12961 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
12962 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
12963 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
12964 ("*vec_load_pairv2di"): Change to ...
12965 ("*vec_load_pair<mode>"): ... this one.
12967 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12969 * config/s390/constraints.md: Add comments.
12970 (jKK): Reject element sizes > 8 bytes.
12971 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
12973 * config/s390/s390.md: Add the s_operand checks formerly in
12974 s390_split_ok_p to various splitters where they are still
12976 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
12977 for 128 bit vectors. Plus two splitters.
12979 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12981 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
12984 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12987 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
12988 error if the boundary argument is not constant.
12990 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12992 PR rtl-optimization/80112
12993 * loop-doloop.c (doloop_condition_get): Don't check condition
12994 if cmp isn't SET with IF_THEN_ELSE src.
12996 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12998 PR tree-optimization/80158
12999 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
13000 replacing a candidate statement, also replace it for the
13001 candidate's alternate interpretation.
13002 (replace_rhs_if_not_dup): Likewise.
13003 (replace_one_candidate): Likewise.
13005 2017-03-24 Richard Biener <rguenther@suse.de>
13007 PR tree-optimization/80167
13008 * graphite-isl-ast-to-gimple.c
13009 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
13011 (translate_isl_ast_to_gimple::get_rename): Likewise.
13013 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
13015 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
13016 handling of certain combinations of target options, including the
13017 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
13018 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
13020 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13023 * config/arm/arm.md (*load_multiple): Add reload_completed to
13024 matching condition.
13026 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13027 Richard Biener <rguenth@suse.de>
13029 PR tree-optimization/79908
13030 PR tree-optimization/80136
13031 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
13032 been cast away, gimplify_and_add suffices.
13034 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
13036 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
13038 2017-03-23 Richard Biener <rguenther@suse.de>
13040 PR tree-optimization/80032
13041 * gimplify.c (gimple_push_cleanup): Forced unconditional
13042 cleanups still have to go to the conditional_cleanups
13045 2017-03-22 Jakub Jelinek <jakub@redhat.com>
13047 PR tree-optimization/80072
13048 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
13050 (next_operand_entry_id): Change type to unsigned int.
13051 (sort_by_operand_rank): Make sure to return the right return value
13052 even if unsigned fields are bigger than INT_MAX.
13053 (struct oecount): Change cnt and id type to unsigned int.
13054 (oecount_hasher::equal): Formatting fix.
13055 (oecount_cmp): Make sure to return the right return value
13056 even if unsigned fields are bigger than INT_MAX.
13057 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
13060 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
13061 TREE_READONLY on result if writing it more than once.
13064 * doc/invoke.texi (-fsanitize=thread): Document that with
13065 -fnon-call-exceptions atomics are not able to throw
13069 * tsan.c: Include tree-eh.h.
13070 (instrument_builtin_call): Call maybe_clean_eh_stmt or
13071 maybe_clean_or_replace_eh_stmt where needed.
13072 (instrument_memory_accesses): Add cfg_changed argument.
13073 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
13074 if it returned true.
13075 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
13077 PR rtl-optimization/63191
13078 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
13079 wrapper function, moved the whole old content into ...
13080 (ix86_delegitimize_address_1): ... this. New inline function.
13081 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
13082 true as last argument instead of ix86_delegitimize_address.
13084 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
13086 * config/aarch64/aarch64.c (generic_branch_cost): Copy
13087 cortexa57_branch_cost.
13089 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
13091 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
13093 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13096 * doc/md.texi (Constraints): Document wA constraint.
13097 * config/rs6000/constraints.md (wA): New.
13098 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
13099 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
13100 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
13101 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
13103 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
13106 * gimplify.c (is_oacc_declared): New function.
13107 (oacc_default_clause): Use it to set default flags for acc declared
13108 variables inside parallel regions.
13109 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
13110 declared variables.
13111 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
13112 declare attribute to any decl as necessary.
13114 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
13117 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
13118 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
13119 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
13120 (arm_arch_lpae): This.
13121 * config/arm/arm.c (arm_arch7ve): Rename into ...
13122 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
13123 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
13126 2017-03-22 Martin Liska <mliska@suse.cz>
13129 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
13130 error message instead of an ICE.
13132 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13134 * doc/extend.texi (6.11 Additional Floating Types): Revise.
13136 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
13138 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
13140 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
13143 2017-03-21 Martin Sebor <msebor@redhat.com>
13145 * doc/extend.texi: Use "cannot" instead of "can't."
13146 * doc/hostconfig.texi: Same.
13147 * doc/install.texi: Same.
13148 * doc/invoke.texi: Same.
13149 * doc/loop.texi: Same.
13150 * doc/md.texi: Same.
13151 * doc/objc.texi: Same.
13152 * doc/rtl.texi: Same.
13153 * doc/tm.texi: Same.
13154 * doc/tm.texi.in: Same.
13155 * doc/trouble.texi: Same.
13157 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
13160 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
13161 (collect_checksum_attributes): Set it.
13162 (die_checksum_ordered): Use it.
13164 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13166 PR tree-optimization/79908
13167 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
13168 change: For a VA_ARG whose LHS has been cast away, use
13169 force_gimple_operand to construct the side effects.
13171 2017-03-21 David Malcolm <dmalcolm@redhat.com>
13173 PR translation/80001
13174 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
13175 more amenable to translation.
13176 (oacc_loop_auto_partitions): Likewise.
13178 2017-03-21 Marek Polacek <polacek@redhat.com>
13179 Martin Sebor <msebor@redhat.com>
13181 PR tree-optimization/80109
13182 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
13183 on INTEGRAL_TYPE_P.
13185 2017-03-21 Jakub Jelinek <jakub@redhat.com>
13186 Segher Boessenkool <segher@kernel.crashing.org>
13189 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
13190 check reg_used_between_p between insn and one of succ or succ2
13191 depending on if succ is artificial insn not inserted into insn
13194 2017-03-21 Martin Liska <mliska@suse.cz>
13196 PR gcov-profile/80081
13197 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
13198 * doc/gcc.texi: Include gcov-dump stuff.
13199 * doc/gcov-dump.texi: New file.
13201 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
13203 PR rtl-optimization/79150
13204 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
13205 conditional jump, if the jump is the last insn of the loop.
13207 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13208 Richard Biener <rguenth@suse.de>
13210 PR tree-optimization/79908
13211 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
13212 been cast away, use force_gimple_operand to construct the side
13215 2017-03-21 Martin Liska <mliska@suse.cz>
13217 PR libfortran/79956
13218 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
13221 2017-03-21 Brad Spengler <spender@grsecurity.net>
13224 * plugin.c (htab_hash_plugin): New function.
13225 (add_new_plugin): Use it and adjust.
13226 (parse_plugin_arg_opt): Adjust.
13227 (init_one_plugin): Likewise.
13229 2017-03-21 Richard Biener <rguenther@suse.de>
13231 PR tree-optimization/80032
13232 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
13233 if set force the cleanup to happen unconditionally.
13234 (gimplify_target_expr): Push inserted clobbers with force_uncond
13235 to avoid them being removed by control-dependent DCE.
13237 2017-03-21 Richard Biener <rguenther@suse.de>
13239 PR tree-optimization/80122
13240 * tree-inline.c (copy_bb): Do not expans va-arg packs or
13241 va_arg_pack_len when the inlined call stmt requires pack
13243 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
13245 2017-03-21 Jakub Jelinek <jakub@redhat.com>
13248 * tsan.c (instrument_builtin_call): If the memory model argument
13249 is not a constant, assume it is valid.
13252 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
13255 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
13257 PR rtl-optimization/79910
13258 * combine.c (can_combine_p): Do not allow combining an I0 or I1
13259 if its dest is used by an insn before I2 (other than the combined
13260 insns themselves, which are properly handled already).
13262 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
13265 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
13267 * combine.c (record_used_regs): New static function.
13268 (try_combine): Handle situations where there is an additional
13269 instruction between I2 and I3 which needs to have a LOG_LINK
13273 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
13275 * combine.c (try_combine): Delete redundant i1 test. Call
13276 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
13278 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13281 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
13282 alternatives 13/14.
13284 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13286 PR tree-optimization/80054
13287 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
13288 the optimization if a PHI or any of its arguments is not dominated
13289 by the candidate's basis. Use gphi* rather than gimple* as
13291 (replace_profitable_candidates): Clean up a gimple* variable that
13292 should be a gphi* variable.
13294 2017-03-20 Martin Sebor <msebor@redhat.com>
13297 * doc/extend.texi (attribute constructor): Document present limitation.
13299 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
13302 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
13303 __POWER9_VECTOR__ #ifdef control, change template definition to
13304 use Power9-specific built-in function.
13305 (vec_any_eq): Likewise.
13306 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
13307 to control outcomes from this test.
13308 (vector_ae_<mode>p): For VEC_F modes, likewise.
13310 2017-03-20 Ian Lance Taylor <iant@google.com>
13312 * config/i386/i386.c (ix86_function_regparm): Save an extra
13313 register for -fsplit-stack with DECL_STATIC_CHAIN.
13315 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
13318 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
13319 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
13321 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
13323 * config/riscv/riscv.c (riscv_print_operand): Use "fence
13325 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
13328 2017-03-20 Marek Polacek <polacek@redhat.com>
13331 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
13333 2017-03-20 Richard Biener <rguenther@suse.de>
13335 PR tree-optimization/80113
13336 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
13337 allocate extra SSA name for PHI def.
13338 (add_close_phis_to_outer_loops): Likewise.
13339 (add_close_phis_to_merge_points): Likewise.
13340 (copy_loop_close_phi_args): Likewise.
13341 (copy_cond_phi_nodes): Likewise.
13343 2017-03-20 Martin Liska <mliska@suse.cz>
13345 PR middle-end/79753
13346 * tree-chkp.c (chkp_build_returned_bound): Do not build
13347 returned bounds for a LHS that's not a BOUNDED_P type.
13349 2017-03-20 Martin Liska <mliska@suse.cz>
13353 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
13354 COMPLEX_CST and VECTOR_CST.
13356 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13359 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
13360 target operand. A new splitter adds the clobber statement in case
13361 the target operand is dead anyway.
13363 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
13365 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
13366 to age-old versions of binutils and glibc.
13368 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
13370 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
13372 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
13374 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
13376 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
13378 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
13379 requirement for binutils 2.13.
13381 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
13383 * combine.c (try_combine): Delete redundant i1 test. Call
13384 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
13386 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
13388 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
13389 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
13391 <riscv64-*-elf>: Re-arrange section
13392 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
13393 <riscv32-*-linux>: Likewise.
13394 <riscv64-*-elf>: Likewise
13395 <riscv64-*-linux>: Likewise.
13397 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
13400 * aarch64.opt(verbose-cost-dump): Fix typo.
13402 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
13405 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
13406 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
13408 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
13410 * reload.c (find_reloads): When reloading a nonoffsettable address,
13411 use RELOAD_OTHER for it and its address reloads.
13413 PR rtl-optimization/79910
13414 * combine.c (record_used_regs): New static function.
13415 (try_combine): Handle situations where there is an additional
13416 instruction between I2 and I3 which needs to have a LOG_LINK
13419 2017-03-17 Jeff Law <law@redhat.com>
13421 PR tree-optimization/71437
13422 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
13423 conditional in the hash table first.
13424 (vrp_dom_walker::before_dom_children): Extract condition from
13425 ASSERT_EXPR. Record condition, its inverion and any implied
13426 conditions as well.
13428 2017-03-17 Marek Polacek <polacek@redhat.com>
13429 Markus Trippelsdorf <markus@trippelsdorf.de>
13431 PR tree-optimization/80079
13432 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
13435 2017-03-17 Richard Biener <rguenther@suse.de>
13437 PR middle-end/80075
13438 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
13439 Properly verify the LHS before the RHS possibly claims to be
13441 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
13444 2017-03-17 Martin Jambor <mjambor@suse.cz>
13446 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
13447 (List of -O2 options): Likewise.
13448 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
13451 2017-03-17 Tom de Vries <tom@codesourcery.com>
13453 * gcov-dump.c (print_usage): Print bug_report_url.
13455 2017-03-17 Richard Biener <rguenther@suse.de>
13457 PR middle-end/80050
13458 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
13459 (parser::peek): Likewise.
13461 2017-03-17 Richard Biener <rguenther@suse.de>
13463 PR tree-optimization/80048
13464 * sese.c (free_sese_info): Properly release rename_map and
13465 copied_bb_map elements.
13467 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
13469 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
13470 Add linked-list forward and backlinks. Insert on
13471 construction, remove on destruction.
13472 (class pass_store_merging): Add m_stores_head field.
13473 (pass_store_merging::terminate_and_process_all_chains):
13474 Iterate over m_stores_head list.
13475 (pass_store_merging::terminate_all_aliasing_chains):
13477 (pass_store_merging::execute): Check for debug stmts first.
13478 Push new chains onto the m_stores_head stack.
13480 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
13483 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
13484 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
13485 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
13487 2017-03-16 Jeff Law <law@redhat.com>
13489 PR tree-optimization/71437
13490 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
13491 member function. Implementation moved into after_dom_children
13492 member function and into the threader's thread_outgoing_edges
13494 (dom_opt_dom_walker::after_dom_children): Simplify by moving
13495 some code into new thread_outgoing_edges.
13496 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
13497 definition. Simplify marker handling (do it here). Assume we always
13498 have the available expression and the const/copies tables.
13499 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
13501 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
13502 * tree-vrp.c (equiv_stack): No longer file scoped.
13503 (vrp_dom_walker): New class.
13504 (vrp_dom_walker::before_dom_children): New member function.
13505 (vrp_dom_walker::after_dom_children): Likewise.
13506 (identify_jump_threads): Setup domwalker. Use it rather than
13507 walking edges in a random order by hand. Simplify setup/finalization.
13508 (finalize_jump_threads): Remove.
13509 (vrp_finalize): Do not call identify_jump_threads here.
13510 (execute_vrp): Do it here instead and call thread_through_all_blocks
13513 PR tree-optimization/71437
13514 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
13516 (simplify_stmt_for_jump_threading): Add basic_block argument. All
13518 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
13519 (dom_opt_dom_walker::thread_across_edge): Remove
13520 handle_dominating_asserts argument. All callers changed.
13521 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
13522 changes. Remove calls to lhs_of_dominating_assert. Other
13523 uses of handle_dominating_asserts turn into unconditional code
13524 (simplify_control_stmt_condition_1): Likewise.
13525 (simplify_control_stmt_condition): Likewise.
13526 (thread_through_normal_block, thread_across_edge): Likewise.
13527 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
13528 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
13529 object if it is not an SSA_NAME.
13530 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
13531 before calling into the VRP specific simplifiers.
13532 (identify_jump_threads): Remove handle_dominating_asserts
13535 2017-03-16 Jakub Jelinek <jakub@redhat.com>
13538 * tree-diagnostic.c (default_tree_printer): No longer static.
13539 * tree-diagnostic.h (default_tree_printer): New prototype.
13541 2017-03-16 Tamar Christina <tamar.christina@arm.com>
13543 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
13544 Change ins into fmov.
13546 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13548 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
13549 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
13550 Use h_con constraint for operand 1.
13551 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
13552 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
13554 2017-03-15 Jeff Law <law@redhat.com>
13556 PR tree-optimization/71437
13557 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
13558 (record_temporary_equivalences): Use it.
13560 PR tree-optimization/71437
13561 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
13562 tree-ssa-scopedtables.
13563 (lookup_avail_expr, build_and_record_new_cond): Likewise.
13564 (record_conditions, record_cond, vuse_eq): Likewise.
13565 (record_edge_info): Adjust to API tweak of record_conditions.
13566 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
13567 (record_temporary_equivalences, optimize_stmt): Likewise.
13568 (eliminate_redundant_computations): Likewise.
13569 (record_equivalences_from_stmt): Likewise.
13570 * tree-ssa-scopedtables.c: Include options.h and params.h.
13571 (vuse_eq): New function, moved from tree-ssa-dom.c
13572 (build_and_record_new_cond): Likewise.
13573 (record_conditions): Likewise. Accept vector of conditions rather
13574 than edge_equivalence structure for first argument.
13575 for the first argument.
13576 (avail_exprs_stack::lookup_avail_expr): New member function, moved
13577 from tree-ssa-dom.c.
13578 (avail_exprs_stack::record_cond): Likewise.
13579 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
13580 from tree-ssa-dom.c.
13581 (avail_exprs_stack): Add new member functions lookup_avail_expr
13583 (record_conditions): Declare.
13585 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
13588 * lra-constraints.c (process_alt_operands): Increase reject for
13589 reloading an input/output operand.
13591 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
13594 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
13595 insns to convert from signed/unsigned char/short to IEEE 128-bit
13597 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
13599 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
13602 * config/i386/i386.c (ix86_vector_duplicate_value): Create
13603 subreg of inner mode for values already in registers.
13605 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
13607 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
13608 iteration reg is used after the loop.
13610 2017-03-14 Martin Sebor <msebor@redhat.com>
13612 PR tree-optimization/79800
13613 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
13614 precision in negative-positive range.
13615 (format_floating): Call non-const overload with adjusted precision.
13617 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
13620 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
13623 2017-03-14 Martin Sebor <msebor@redhat.com>
13625 PR middle-end/80020
13626 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
13627 * builtins.def (aligned_alloc): Use it.
13630 * Makefile.in (GTFILES): Add calls.c.
13631 * calls.c: Include "gt-calls.h".
13633 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
13635 PR rtl-optimization/79728
13636 * regs.h (struct target_regs): New field
13637 x_contains_allocatable_regs_of_mode.
13638 (contains_allocatable_regs_of_mode): New macro.
13639 * reginfo.c (init_reg_sets_1): Initialize it, and change
13640 contains_reg_of_mode so it includes global regs as well.
13641 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
13642 rather than contains_regs_of_mode.
13644 2017-03-14 Martin Liska <mliska@suse.cz>
13646 * doc/invoke.texi: Document options that can't be combined with
13647 -fcheck-pointer-bounds.
13649 2017-03-14 Martin Liska <mliska@suse.cz>
13651 PR middle-end/79831
13652 * doc/invoke.texi (-Wchkp): Document the option.
13654 2017-03-14 Martin Liska <mliska@suse.cz>
13656 * Makefile.in: Install gcov-dump.
13658 2017-03-14 Martin Liska <mliska@suse.cz>
13660 * multiple_target.c (expand_target_clones): Bail out for
13661 an invalid attribute.
13663 2017-03-14 Richard Biener <rguenther@suse.de>
13665 * alias.c (struct alias_set_entry): Pack properly.
13666 * cfgloop.h (struct loop): Likewise.
13667 * cse.c (struct set): Likewise.
13668 * ipa-utils.c (struct searchc_env): Likewise.
13669 * loop-invariant.c (struct invariant): Likewise.
13670 * lra-remat.c (struct cand): Likewise.
13671 * recog.c (struct change_t): Likewise.
13672 * rtl.h (struct address_info): Likewise.
13673 * symbol-summary.h (function_summary): Likewise.
13674 * tree-loop-distribution.c (struct partition): Likewise.
13675 * tree-object-size.c (struct object_size_info): Likewise.
13676 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
13677 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
13678 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
13679 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
13680 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
13681 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
13682 (struct _stmt_vec_info): Likewise.
13684 2017-03-14 Martin Liska <mliska@suse.cz>
13687 * multiple_target.c (create_dispatcher_calls): Check that
13688 a target can create a function dispatcher.
13690 2017-03-14 Martin Liska <mliska@suse.cz>
13693 * multiple_target.c (expand_target_clones): Drop local.local
13694 flag for default implementation.
13696 2017-03-14 Richard Biener <rguenther@suse.de>
13698 PR tree-optimization/80030
13699 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
13701 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
13703 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
13704 gcc_fallthrough() instead of __attribute__((fallthrough));
13706 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13708 * doc/gcc.texi: Remove "up" link to (DIR).
13709 * doc/gccint.texi: Ditto.
13711 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13713 * doc/install.texi (Specific) <avr>: Remove reference to
13716 2017-03-13 Jeff Law <law@redhat.com>
13718 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
13719 attribute rather than comments.
13721 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
13722 match_scratch operand is highest.
13724 2017-03-13 Martin Liska <mliska@suse.cz>
13726 PR middle-end/78339
13727 * ipa-pure-const.c (warn_function_noreturn): If the declarations
13728 is a CHKP clone, use original declaration.
13730 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13732 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
13733 (arc_conditional_register_usage): Use a different allocation order
13734 when optimizing for size.
13735 * common/config/arc/arc-common.c (arc_option_optimization_table):
13736 Section anchors default on when optimizing for size.
13738 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13740 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
13742 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13744 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
13745 * config/arc/arc.md (cpu_facility): Add cd variant.
13746 (*movqi_insn): Add code density variant.
13747 (*movhi_insn): Likewise.
13748 (*movqi_insn): Likewise.
13749 (*addsi3_mixed): Likewise.
13750 (subsi3_insn): Likewise.
13752 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13754 * config/arc/arc.md (movsi_cond_exec): Update constraint.
13756 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13758 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
13759 expressions with MINUS and UNARY ops.
13761 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13764 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
13766 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
13767 between vec_select and vector argument.
13768 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
13769 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
13770 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
13771 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
13772 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
13773 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
13775 2017-03-13 Richard Biener <rguenther@suse.de>
13778 * params.def (vect-max-peeling-for-alignment): Fix typo.
13780 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13782 * doc/install.texi (Specific) <mips-*-*>: Remove description of
13783 issue that only occurred with binutils below 2.18.
13785 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13787 * doc/install.texi (Specific) <cris-axis-elf>: No longer
13788 refer to binutils 2.11/2.12 minimum.
13790 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13792 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
13793 ftp.kernel.org and simplify binutils requirement.
13795 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
13797 * doc/invoke.texi (Warning Options): Fix spelling of link-time
13799 (Optimize Options): Ditto. Also remove redundancy.
13801 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13803 PR translation/79848
13804 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
13806 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
13807 to G_ to avoid double translation.
13809 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13811 PR translation/79923
13812 * auto-profile.c (get_combined_location): Convert leading
13813 character of diagnostics to lower case and remove trailing period.
13814 (read_profile): Likewise for various diagnostics.
13815 * config/arm/arm.c (arm_option_override): Remove trailing period
13816 from various diagnostics.
13817 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
13818 (msp430_expand_delay_cycles): Likewise.
13820 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13823 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
13824 full command-line argument, rather than just "str".
13825 (aarch64_validate_march): Likewise.
13826 (aarch64_validate_mtune): Likewise.
13828 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
13830 PR rtl-optimization/78911
13831 * lra-assigns.c (must_not_spill_p): New function.
13832 (spill_for): Use it.
13834 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13836 PR tree-optimization/79981
13837 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
13838 ATOMIC_COMPARE_EXCHANGE ifn result.
13839 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
13840 IFN_ATOMIC_COMPARE_EXCHANGE.
13842 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13845 * opts.c (parse_sanitizer_options): Add missing question mark to
13846 "did you mean" message.
13848 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13850 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
13852 (VMULEUH_UNS): Likewise.
13853 (VMULOUB_UNS): Likewise.
13854 (VMULOUH_UNS): Likewise.
13855 * config/rs6000/rs6000.c (builtin_function_type): Remove
13856 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
13858 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13861 * read-rtl-function.c (function_reader::read_rtx_operand): Update
13862 x with result of extra_parsing_for_operand_code_0.
13863 (function_reader::extra_parsing_for_operand_code_0): Convert
13864 return type from void to rtx, returning x. When reading
13865 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
13866 larger size containing struct block_symbol.
13868 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
13870 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
13871 -mfloat128-hardware without -m64.
13873 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
13876 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
13877 entries to the case statement that marks unsigned arguments to
13878 overloaded functions.
13880 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
13882 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
13883 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
13885 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
13888 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
13889 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
13891 2017-03-10 Martin Liska <mliska@suse.cz>
13895 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
13896 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
13899 2017-03-10 Olivier Hainque <hainque@adacore.com>
13901 * tree-switch-conversion (array_value_type): Start by resetting
13902 candidate type to it's main variant.
13904 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13906 PR rtl-optimization/79909
13907 * combine.c (try_combine): Use simplify_replace_rtx on individual
13908 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
13909 of the whole CALL_INSN_FUNCTION_USAGE.
13911 PR tree-optimization/79972
13912 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
13913 get_range_info on SSA_NAMEs. Formatting fixes.
13915 2017-03-10 Richard Biener <rguenther@suse.de>
13916 Jakub Jelinek <jakub@redhat.com>
13918 PR tree-optimization/77975
13919 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
13920 edge to be constant.
13921 (get_val_for): For constant x return it. Formatting fix.
13922 (loop_niter_by_eval): Avoid pointless looping if the next iteration
13923 would use the same bases as the current one.
13925 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13927 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
13928 instead of vec_select for V1TImode.
13929 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
13931 (VSX_LE_128): Add V1TI to this mode iterator.
13932 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
13933 (*vsx_le_perm_store_<mode>): Likewise.
13934 (pre-reload splitter for VSX stores): Likewise.
13935 (post-reload splitter for VSX stores): Likewise.
13936 (*vsx_xxpermdi2_le_<mode>): Likewise.
13937 (*vsx_lxvd2x2_le_<mode>): Likewise.
13938 (*vsx_stxvd2x2_le_<mode>): Likewise.
13940 2017-03-09 Michael Eager <eager@eagercon.com>
13942 Correct failures with --enable-checking=yes,rtl.
13944 * config/microblaze/microblaze.c (microblaze_expand_shift):
13945 Replace GET_CODE test with CONST_INT_P and INTVAL test with
13946 test for const0_rtx.
13947 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
13948 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
13950 2017-03-09 Richard Biener <rguenther@suse.de>
13952 PR tree-optimization/79977
13953 * graphite-scop-detection.c (scop_detection::merge_sese):
13954 Handle the case of extra exits to blocks dominating the entry.
13956 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
13958 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13961 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
13963 PR rtl-optimization/79949
13964 * lra-constraints.c (process_alt_operands): Check memory when
13965 trying to predict a cycle. Print about the overall increase.
13967 2017-03-09 Richard Biener <rguenther@suse.de>
13969 PR middle-end/79971
13970 * gimple-expr.c (useless_type_conversion_p): Preserve
13971 TYPE_SATURATING for fixed-point types.
13973 2017-03-09 Richard Biener <rguenther@suse.de>
13976 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
13977 alignment of BLKmode params.
13979 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13982 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
13983 (VALL_NO_V2Q): Likewise.
13985 * config/aarch64/aarch64-simd.md
13986 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
13988 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
13989 VALL_NO_V2Q mode iterator.
13990 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
13992 2017-03-09 Martin Liska <mliska@suse.cz>
13994 PR tree-optimization/79631
13995 * tree-chkp-opt.c (chkp_is_constant_addr): Call
13996 tree_int_cst_sign_bit just for INTEGER constants.
13998 2017-03-09 Martin Liska <mliska@suse.cz>
14002 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
14005 2017-03-09 Marek Polacek <polacek@redhat.com>
14008 * tree.c (inchash::add_expr): Handle TREE_VEC.
14010 2017-03-09 Martin Liska <mliska@suse.cz>
14013 (chkp_narrow_size_and_offset): New function.
14014 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
14015 (void chkp_parse_bit_field_ref): New function.
14016 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
14017 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
14019 2017-03-09 Martin Liska <mliska@suse.cz>
14022 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
14023 (chkp_find_bounds_1): Remove gcc_unreachable.
14025 2017-03-09 Jakub Jelinek <jakub@redhat.com>
14028 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
14029 BUILT_IN_SYNC*, determine the access type from the size suffix and
14030 always build a MEM_REF with that type. Handle forgotten
14031 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
14034 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
14035 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
14036 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
14037 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
14038 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
14039 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
14040 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
14041 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
14042 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
14043 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
14044 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
14045 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
14046 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
14047 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
14048 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
14049 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
14050 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
14051 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
14052 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
14053 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
14054 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
14055 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
14056 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
14057 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
14058 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
14059 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
14060 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
14061 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
14062 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
14063 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
14064 definitions outside of __OPTIMIZE__ guarded section.
14067 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
14068 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
14069 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
14070 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
14073 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14075 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
14076 ("vfenez<mode>"): Add missing constraints.
14078 2017-03-08 Martin Sebor <msebor@redhat.com>
14081 * config/nds32/nds32.c (nds32_option_override):
14082 Fix misspelled diagnostic.
14084 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14087 * gimplify.c (gimplify_omp_for): Replace index var in outer
14088 taskloop statement with an artificial variable and add
14089 OMP_CLAUSE_PRIVATE clause for it.
14091 2017-03-08 Richard Biener <rguenther@suse.de>
14093 PR tree-optimization/79955
14094 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
14095 for accesses that are completely outside of the variable.
14097 2017-03-08 Andrew Haley <aph@redhat.com>
14099 PR tree-optimization/79943
14100 * tree-ssa-loop-split.c (compute_new_first_bound): When
14101 calculating the new upper bound, (END-BEG) should be added, not
14104 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14106 * config/avr/avr.md (setmemhi): Make sure match_dup
14107 operand number comes before match_scratch.
14109 2017-03-08 Richard Biener <rguenther@suse.de>
14111 PR tree-optimization/79920
14112 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
14113 with ncopies == 1 to ...
14114 (vect_transform_slp_perm_load): ... here. Properly compute
14115 all element loads by iterating VF times over the group. Do
14116 not handle ncopies (computed in a broken way) in
14117 vect_create_mask_and_perm.
14119 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14122 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
14123 is a uniform vector, use uniform_vector_p return value instead of
14124 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
14126 2017-03-07 Marek Polacek <polacek@redhat.com>
14128 PR middle-end/79809
14129 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
14130 (alloca_call_type): Likewise.
14132 2017-03-07 Martin Liska <mliska@suse.cz>
14134 * gcov.c (process_args): Put comment to correct location.
14136 2017-03-07 Martin Liska <mliska@suse.cz>
14138 PR middle-end/68270
14139 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
14140 Use array_at_struct_end_p instead of DECL_CHAIN (field).
14141 (chkp_narrow_bounds_for_field): Likewise.
14142 (chkp_parse_array_and_component_ref): Pass one more argument to
14145 2017-03-07 Richard Biener <rguenther@suse.de>
14147 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
14150 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
14152 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
14153 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
14155 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14158 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
14159 to end of description.
14160 (PARAM_MAX_STORES_TO_MERGE): Likewise.
14162 2017-03-07 Jakub Jelinek <jakub@redhat.com>
14164 PR rtl-optimization/79901
14165 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
14167 (*avx512f_<code><mode>3<mask_name>): ... this.
14168 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
14169 iterator instead of VI8_AVX2_AVX512BW.
14171 PR rtl-optimization/79901
14172 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
14173 min/max expander, expand it using expand_vec_cond_expr.
14176 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
14179 2017-03-06 Jakub Jelinek <jakub@redhat.com>
14182 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
14183 to void * for PCH reasons.
14184 * dwarf2out.c (output_loc_operands, output_die): Cast
14185 v.val_vec.array to unsigned char *.
14187 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
14190 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
14193 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
14195 PR rtl-optimization/79571
14196 * lra-constraints.c (process_alt_operands): Calculate static
14197 reject and subtract it from overall when only addresses will be
14200 2017-03-06 Julia Koval <julia.koval@intel.com>
14203 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
14204 incoming stack boundary to 128 for 64-bit targets.
14206 2017-03-06 Richard Biener <rguenther@suse.de>
14208 PR tree-optimization/79894
14209 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
14210 to NULL after folding it.
14212 2017-03-06 Richard Biener <rguenther@suse.de>
14214 PR tree-optimization/79824
14215 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
14216 check disabling peeling for gaps.
14218 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
14220 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
14221 attributes): Document gettimeofday.
14223 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
14225 * config/s390/s390.c (s390_option_override_internal): Set
14226 PARAM_MIN_VECT_LOOP_BOUND
14228 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
14230 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
14231 * config/s390/s390.md: Likewise.
14233 2017-03-06 Jakub Jelinek <jakub@redhat.com>
14236 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
14237 (<avx2_avx512>_perm<mode>): Rename to ...
14238 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
14240 (<avx512>_perm<mode>_mask): Rename to ...
14241 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
14243 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
14244 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
14245 instead of VI8F_256_512.
14246 (avx512f_perm<mode>): New define_expand.
14247 (avx512f_perm<mode>_mask): Likewise.
14248 (avx512f_perm<mode>_1<mask_name>): New define_insn.
14249 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
14251 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14253 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
14254 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
14256 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
14258 2017-03-06 Martin Liska <mliska@suse.cz>
14261 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
14262 when having a SSA NAME w/o VAR_DECL assigned to it.
14264 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14266 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
14267 msa_dpsub_<su>_d): Fix MODE for vec_select.
14269 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14271 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
14273 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
14275 2017-03-06 Richard Biener <rguenther@suse.de>
14277 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
14278 * plugin.c (register_plugin_info): Likewise.
14279 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
14281 2017-03-05 Jakub Jelinek <jakub@redhat.com>
14283 * config/i386/sse.md (sse_storehps, sse_storelps,
14284 avx_<castmode><avxsizesuffix>_<castmode>,
14285 avx512f_<castmode><avxsizesuffix>_<castmode>,
14286 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
14287 in condition that at least one operand is not a MEM.
14289 2017-03-03 Jakub Jelinek <jakub@redhat.com>
14291 PR middle-end/79805
14292 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
14293 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
14295 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
14296 gimple_call_nothrow_p flag based on whether original builtin can throw.
14297 If it can, emit following stmts on the fallthrough edge.
14298 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
14299 don't create new bb if inserting just debug stmts on the edge, try to
14300 insert them on the fallthru bb or just reset debug stmts.
14302 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
14305 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
14306 restore recog_data (including the operand rtxes inside it) around
14307 the call to get_insn_template.
14309 2017-03-03 Martin Sebor <msebor@redhat.com>
14311 PR tree-optimization/79699
14312 * context.c (context::~context): Free MPFR caches to avoid
14313 a memory leak on program exit.
14315 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14317 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
14318 Use wide_int::ulow () instead of .elt (0).
14320 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
14322 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
14323 (*pushxf): Limit oF constraint to 32bit targets and add oC
14324 constraint for 64bit targets.
14325 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
14326 (*pushdf): Change rmF constraint to rmC.
14328 2017-03-03 Martin Liska <mliska@suse.cz>
14330 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
14331 Remove unused variable.
14333 2017-03-03 Jakub Jelinek <jakub@redhat.com>
14336 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
14337 is a memory operand, increase num_memory.
14338 (ix86_expand_args_builtin): Likewise.
14340 2017-03-03 Jan Hubicka <jh@suse.cz>
14343 * ipa-devirt.c (maybe_record_node): Properly handle
14344 __cxa_pure_virtual visibility.
14346 2017-03-03 Martin Liska <mliska@suse.cz>
14348 PR tree-optimization/79803
14349 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
14351 (pass_loop_prefetch::execute): Disabled optimization if an
14352 assumption about L1 cache size is not met.
14354 2017-03-03 Martin Liska <mliska@suse.cz>
14356 PR rtl-optimization/79574
14357 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
14358 (hash_scan_set): Likewise.
14359 (dump_hash_table): Likewise.
14360 (hoist_code): Likewise.
14362 2017-03-03 Richard Biener <rguenther@suse.de>
14364 * fixed-value.c (fixed_from_string): Restore use of elt (1)
14365 in place of uhigh ().
14366 (fixed_convert_from_real): Likewise.
14368 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
14371 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
14373 2017-03-03 Richard Biener <rguenther@suse.de>
14375 PR middle-end/79818
14376 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
14377 TYPE_OVERFLOW_UNDEFINED check.
14379 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14381 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
14383 (vector_ae_<mode>_p): Likewise.
14384 (vector_nez_<mode>_p): Likewise.
14385 (vector_ne_v2di_p): Likewise.
14386 (vector_ae_v2di_p): Likewise.
14387 (vector_ne_<mode>_p): Likewise.
14388 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
14390 (vsx_tsqrt<mode>2_fe): Likewise.
14392 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
14395 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
14397 2017-03-02 Jakub Jelinek <jakub@redhat.com>
14399 PR rtl-optimization/79780
14400 * cprop.c (one_cprop_pass): When second and further conditional trap
14401 in a single basic block is turned into an unconditional trap, turn it
14402 into a deleted note to avoid RTL verification failures.
14404 2017-03-02 Richard Biener <rguenther@suse.de>
14406 * fold-const.c (const_binop): Use ulow () instead of elt (0).
14408 2017-03-02 Richard Biener <rguenther@suse.de>
14410 PR tree-optimization/79345
14412 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
14413 param and abort the walk, returning -1 if it is hit.
14414 (walk_aliased_vdefs): Take a limit param and pass it on.
14415 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
14416 defaulting to 0 and return a signed int.
14417 * tree-ssa-uninit.c (struct check_defs_data): New struct.
14418 (check_defs): New helper.
14419 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
14420 about uninitialized memory.
14421 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
14422 bogus uninitialized warning.
14423 (fixed_convert_from_real): Likewise.
14425 2017-03-02 Bin Cheng <bin.cheng@arm.com>
14427 PR tree-optimization/66768
14428 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
14429 iv_use if base object can't be determined.
14431 2017-03-02 Jakub Jelinek <jakub@redhat.com>
14433 PR tree-optimization/79345
14434 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
14435 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
14436 (get_pattern_stats): Initialize it.
14437 * genemit.c (gen_expand): Verify match_scratch numbers come after
14438 match_operand/match_dup numbers.
14439 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
14440 match_scratch numbers.
14441 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
14443 * config/s390/s390.md (trunctdsd2): Likewise.
14445 2017-03-02 Richard Biener <rguenther@suse.de>
14447 * wide-int.h (wide_int_storage::operator=): Implement in terms
14450 2017-03-02 Richard Biener <rguenther@suse.de>
14452 PR tree-optimization/79777
14453 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
14454 the to insert expression to sth existing.
14456 2017-03-01 Martin Sebor <msebor@redhat.com>
14458 PR middle-end/79692
14459 * gimple-ssa-sprintf.c
14460 (directive::known_width_and_precision): New function.
14461 (format_integer): Use it.
14462 (get_mpfr_format_length): Consider the full range of precision
14463 when computing %g output with the # flag. Set the likely byte
14464 count to 3 rather than 1 when precision is indeterminate.
14465 (format_floating): Correct the lower bound of precision.
14467 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14469 * doc/invoke.texi: Document default code model for 64-bit Linux.
14471 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14474 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
14475 udiv rather than div since input pattern is unsigned.
14477 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
14479 * config/i386/i386.c (print_reg): Warn for values of
14480 unsupported size in integer register.
14482 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
14485 * config/rs6000/predicates.md (current_file_function_operand): Do
14486 not allow self calls to be local if the function is replaceable.
14488 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
14491 * config/rs6000/altivec.h (vec_ctz and others): Change the
14492 preprocessor macro that controls conditional compilation from
14493 _ARCH_PWR9 to __POWER9_VECTOR__.
14494 (vec_all_ne): Change parameterization of __altivec_scalar_pred
14495 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
14496 control (instead of _ARCH_PWR9 control) so that template
14497 definition uses power9-specific function.
14498 (vec_any_eq): Likewise.
14499 (vec_all_ne): Change macro definition to use a power9-specific
14500 expansion under #ifdef __POWER9_VECTOR__ control (instead of
14501 _ARCH_PWR9 control).
14502 (vec_any_eq) Likewise.
14503 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
14504 expansion for CMPNEF to remove support for xvcmpnesp instruction.
14505 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
14506 support for xvcmpnedp instruction.
14507 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
14508 macro expansion so that Power9 implementation of vec_all_ne does
14509 not use the AltiVec predicate framework.
14510 (VCMPNEH_P): Likewise.
14511 (VCMPNEW_P): Likewise.
14512 (VCMPNED_P): Likewise.
14513 (VCMPNEFP_P): Likewise.
14514 (VCMPNEDP_P): Likewise.
14515 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
14516 implementation of vec_any_eq to not use AltiVec predicate
14518 (VCMPAEH_P): Likewise.
14519 (VCMPAEW_P): Likewise.
14520 (VCMPAED_P): Likewise.
14521 (VCMPAEFP_P): Likewise.
14522 (VCMPAEDP_P): Likewise.
14523 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
14524 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
14525 not use the AltiVec predicate framework.
14526 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
14527 of vec_any_eq to not use AltiVec predicate framework.
14528 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
14529 support for predefined __POWER9_VECTOR__ macro to indicate that
14530 Power9 instruction selection is enabled.
14531 (altivec_overloaded_builtins): Remove extraneous
14532 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
14533 function argument types RS6000_BTI_bool_V16QI and
14534 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
14535 entry for overloaded function argument types RS6000_BTI_bool_V4SI
14536 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
14537 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
14538 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
14539 Power9 for implementations of vec_cmpne. Change the signature for
14540 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
14541 (representing vec_all_ne) to remove the previously described first
14542 argument of type RS6000_BTI_INTSI, as this was an artifact of
14543 reliance on the AltiVec predicate framework, which is no longer
14544 used in the implementation of these functions. Add
14545 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
14546 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
14547 since, unlike the AltiVec predicate framework implementation, we
14548 do not share function descriptors between vec_alle and vec_anyeq.
14549 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
14550 set of modes that receive special treatment even when
14551 TARGET_P9_VECTOR is true. The special treatment emits code that
14552 does not depend on Power9 instructions.
14553 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
14554 define_expand to not rely on AltiVec predicate framework.
14555 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
14557 (vector_ne_v2di_p): Change this define_expand to not rely on
14558 AltiVec predicate framework.
14559 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
14561 (vector_ne_<mode>_p): Change this define_expand to not rely on
14562 AltiVec predicate framework.
14563 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
14565 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
14566 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
14567 define_insn pattern.
14568 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
14569 define_insn pattern because the xvcmpne<VSs>. instruction is not
14571 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
14572 instruction is not supported.
14574 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14576 * config/nvptx/nvptx.c: Include intl.h.
14578 2017-03-01 Martin Jambor <mjambor@suse.cz>
14581 * ipa-prop.h (ipa_bits): Removed field known.
14582 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
14583 to pointers. Adjusted their comments to warn about their sharing.
14584 (ipcp_transformation_summary): Change bits to a vector of pointers.
14585 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
14586 (ipa_get_ipa_bits_for_value): Declare.
14587 * tree-vrp.h (value_range): Mark as GTY((for_user)).
14588 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
14589 (ipa_bits_hash_table): Likewise.
14590 (ipa_vr_ggc_hash_traits): Likewise.
14591 (ipa_vr_hash_table): Likewise.
14592 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
14593 being pointers and vr_known being removed.
14594 (ipa_set_jf_unknown): Likewise.
14595 (ipa_get_ipa_bits_for_value): New function.
14596 (ipa_set_jfunc_bits): Likewise.
14597 (ipa_get_value_range): New overloaded functions.
14598 (ipa_set_jfunc_vr): Likewise.
14599 (ipa_compute_jump_functions_for_edge): Use the above functions to
14600 construct bits and vr parts of jump functions.
14601 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
14602 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14604 (ipcp_grow_transformations_if_necessary): Also allocate
14605 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14607 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
14608 them. Fix too long lines.
14609 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
14610 vr_known being removed.
14611 (ipa_read_jump_function): Use new setter functions to construct bits
14612 and vr parts of jump functions or set them to NULL.
14613 (write_ipcp_transformation_info): Adjust for bits being pointers.
14614 (read_ipcp_transformation_info): Likewise.
14615 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
14617 Include gt-ipa-prop.h.
14618 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
14620 (ipcp_store_bits_results): Likewise.
14621 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
14622 Do not write to existing jump functions but use a temporary instead.
14624 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14627 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
14628 attempt to use its first operand as BIT_FIELD_REF base.
14630 2017-03-01 Richard Biener <rguenther@suse.de>
14632 PR middle-end/79721
14633 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
14634 interpolating formula in wrapping arithmetic.
14635 (chrec_apply): Convert chrec_evaluate return value to wanted type.
14637 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14639 PR tree-optimization/79734
14640 * tree-vect-generic.c (expand_vector_condition): Optimize
14641 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
14642 Handle VEC_COND_EXPR where comparison has different inner width from
14643 type's inner width.
14645 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
14647 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
14648 markup, and similar issues. Remove @opindex entries for things
14649 that aren't options. Add missing -mmpy-option entries.
14651 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14653 PR tree-optimization/79737
14654 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
14655 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
14656 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
14657 instead of byte_size. Formatting fix.
14658 (shift_bytes_in_array_right): Formatting fix.
14660 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
14663 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
14664 condition on optimize for the leaf function test.
14666 2017-02-28 Martin Liska <mliska@suse.cz>
14669 * read-rtl-function.c (function_reader::handle_unknown_directive):
14670 Bail out when one uses -flto.
14672 2017-02-28 Martin Liska <mliska@suse.cz>
14674 * common.opt: Replace space with tabular for options of <number>
14676 * config/i386/i386.opt: Show <number> value for
14677 -mlarge-data-threshold.
14678 * opts.c (print_filtered_help): Do not display number in hexadecimal
14681 2017-02-28 Martin Liska <mliska@suse.cz>
14683 * common.opt: Fix --help=option -Q for options which are of
14686 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
14688 * config/i386/i386.c (print_reg): Error out for values
14689 of 8-bit size in invalid integer register.
14691 2017-02-28 Martin Sebor <msebor@redhat.com>
14693 PR tree-optimization/79691
14694 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
14696 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14699 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
14700 gcc_unreachable with output_operand_lossage.
14702 2017-02-28 Richard Biener <rguenther@suse.de>
14704 PR tree-optimization/79740
14705 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
14707 (visit_nary_op): Insert the nary into the hashtable if we
14708 pattern-matched sth.
14709 * tree-ssa-pre.c (eliminate_insert): Robustify.
14711 2017-02-28 Richard Biener <rguenther@suse.de>
14713 PR middle-end/79731
14714 * fold-const.c (decode_field_reference): Reject out-of-bound
14717 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14719 * config/i386/i386.c: Include intl.h.
14720 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
14721 instead of just cond ? "..." : "...".
14722 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
14723 * coverage.c (read_counts_file): Likewise.
14724 * omp-offload.c: Include intl.h.
14725 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
14726 of just cond ? "..." : "...".
14727 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
14728 of just cond ? "..." : "...".
14730 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
14733 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
14735 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
14736 'tune for' CPU name.
14737 * config/arm/arm-cpu-data.h: Regenerated.
14739 2017-02-28 Richard Biener <rguenther@suse.de>
14741 PR tree-optimization/79732
14742 * tree-inline.c (expand_call_inline): Do not shadow var.
14744 2017-02-28 Richard Biener <rguenther@suse.de>
14746 PR tree-optimization/79723
14747 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
14748 address-space properly.
14750 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
14752 * doc/optinfo.texi (Optimization groups): Fix option used for
14754 * doc/invoke.texi (-fopt-info): Document "omp".
14755 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
14756 (OPTGROUP_ALL): Add OPTGROUP_OMP.
14757 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
14758 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
14759 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
14761 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
14763 * dumpfile.c (optgroup_options): Instead of "openmp", associate
14764 OPTGROUP_OMP with "omp".
14766 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
14769 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
14770 for arithmetic shift of unsigned V2DI.
14772 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
14774 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
14775 arc/linux.h headers.
14776 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
14777 (LINK_SPEC): Likewise.
14778 (ARC_TLS_EXTRA_START_SPEC): Likewise.
14779 (EXTRA_SPECS): Likewise.
14780 (STARTFILE_SPEC): Likewise.
14781 (ENDFILE_SPEC): Likewise.
14782 (LIB_SPEC): Likewise.
14783 (TARGET_SDATA_DEFAULT): Likewise.
14784 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
14785 (MULTILIB_DEFAULTS): Likewise.
14786 (DWARF2_UNWIND_INFO): Likewise.
14787 * config/arc/big.h: New file.
14788 * config/arc/elf.h: Likewise.
14789 * config/arc/linux.h: Likewise.
14790 * config/arc/t-uClibc: Remove.
14792 2017-02-27 Bin Cheng <bin.cheng@arm.com>
14794 PR tree-optimization/77536
14795 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
14796 (tree_transform_and_unroll_loop): Use above function to compute the
14797 estimated niter of unrolled loop and use it when scaling profile.
14798 Also use count info rather than frequency if it's non-zero.
14799 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
14800 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
14801 (vect_transform_loop): Call above function.
14803 2017-02-27 Richard Biener <rguenther@suse.de>
14805 PR tree-optimization/45397
14806 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
14807 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
14808 (visit_nary_op): Add pattern matching for CSEing sign-changed
14809 or truncated operations with wider ones.
14811 2017-02-27 Richard Biener <rguenther@suse.de>
14813 PR tree-optimization/79690
14814 * tree-vect-stmts.c (vectorizable_store): Use vector type
14815 built from the DR with address-space.
14817 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
14819 * doc/invoke.texi (Optimize Options): Refine the description
14820 of asan-use-after-return.
14822 2017-02-25 Alan Modra <amodra@gmail.com>
14824 PR rtl-optimization/79584
14825 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
14826 base, not ad->base_term, the reg within base. Remove assertion
14827 that ad->base == ad->base_term. Replace gen_int_mode using
14828 bogus mode with const0_rtx.
14830 2017-02-25 Jakub Jelinek <jakub@redhat.com>
14832 PR middle-end/79396
14833 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
14834 FMA_EXPR like tcc_binary or tcc_unary.
14836 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
14839 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
14841 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
14842 (output_loc_operands): Handle DW_OP_call_ref and
14843 DW_OP_GNU_variable_value.
14844 (struct variable_value_struct): New type.
14845 (struct variable_value_hasher): Likewise.
14846 (variable_value_hash): New variable.
14847 (string_types): Remove.
14848 (copy_loc_descr): New function.
14849 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
14850 (prepend_loc_descr_to_each): New function.
14851 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
14852 instead of add_loc_descr_to_each if the first argument is single
14853 location list and the second has multiple.
14854 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
14855 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
14856 when looking for variable value which doesn't have other location info.
14857 (loc_list_from_tree): Formatting fix.
14858 (gen_array_type_die): Simplify DW_AT_string_length handling.
14859 (adjust_string_types): Remove.
14860 (gen_subprogram_die): Don't call adjust_string_types nor test/set
14861 string_types. Call resolve_variable_values.
14862 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
14863 (resolve_addr_in_expr): Likewise. Add A argument.
14864 (copy_deref_exprloc): Remove deref argument. Adjust for the
14865 original expression being DW_OP_GNU_variable_value with optionally
14866 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
14867 optionally after it.
14868 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
14869 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
14870 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
14871 (variable_value_hasher::hash, variable_value_hasher::equal): New
14873 (resolve_variable_value_in_expr, resolve_variable_value,
14874 resolve_variable_values, note_variable_value_in_expr,
14875 note_variable_value): New functions.
14876 (dwarf2out_early_finish): Call note_variable_value on all toplevel
14879 2017-02-24 Jakub Jelinek <jakub@redhat.com>
14882 * opts.h (handle_generated_option): Add GENERATED_P argument.
14883 * opts-common.c (handle_option): Adjust function comment.
14884 (handle_generated_option): Add GENERATED_P argument, pass it to
14886 (control_warning_option): Pass false to handle_generated_option
14888 * opts.c (maybe_default_option): Pass true to handle_generated_option
14890 * optc-gen.awk: Likewise.
14892 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14894 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
14895 a REG, look at the REG it is a SUBREG of.
14896 (splitter for cmpeqsi_t): Ditto.
14898 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14900 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
14901 the special USEs with the pattern of the insn, not the insn itself.
14903 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
14906 * doc/invoke.texi: Document -mload-store-pairs.
14908 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14909 Sandra Loosemore <sandra@codesourcery.com>
14911 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
14912 argument isn't a CONST_INT.
14913 (nios2_alternate_compare_const): Assert op is a CONST_INT.
14914 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
14915 (nios2_validate_compare): Bypass alternate compare logic if *op2
14916 is not a CONST_INT.
14917 (ldstwm_operation_p): Return false if first_base is not a REG or
14918 if first_offset is not a CONST_INT.
14920 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14922 * config/cris/cris.md: Use correct operand in a define_peephole2.
14924 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14926 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
14928 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14930 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
14931 this_insn if it is an INSN or JUMP_INSN.
14932 (force_offsettable): Look at base, not at addr.
14933 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
14934 on things that aren't necessarily CONST_INTs.
14936 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
14938 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
14939 -mfpmath=sse is the default also for x86-32 targets with SSE2
14940 instruction set when @option{-ffast-math} is enabled
14942 2017-02-24 Jeff Law <law@redhat.com>
14944 PR rtl-optimizatoin/79286
14945 * ira.c (update_equiv_regs): Drop may_trap_p exception to
14948 2017-02-24 Richard Biener <rguenther@suse.de>
14950 PR tree-optimization/79389
14951 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
14954 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
14956 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
14957 function comment to reflect reality.
14958 (loop_exits_before_overflow): Fix typo in function description.
14960 2017-02-24 Richard Biener <rguenther@suse.de>
14962 PR tree-optimization/79389
14963 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
14964 properly that a threading opportunity exists. Detect conditional
14965 copy/constant propagation opportunities.
14967 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
14969 * config/visium/visium.md (type): Add trap.
14970 (b): New mode attribute.
14971 (*btst): Rename into...
14972 (*btst<mode>): ...this and adjust.
14973 (*cbranchsi4_btst_insn): Rename into...
14974 (*cbranch<mode>4_btst_insn): ...this and adjust.
14975 (trap): New define_insn.
14977 2017-02-23 Jakub Jelinek <jakub@redhat.com>
14979 PR tree-optimization/79389
14980 * ifcvt.c (struct noce_if_info): Add rev_cond field.
14981 (noce_reversed_cond_code): New function.
14982 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
14983 reversed_comparison_code. Formatting fix.
14984 (noce_try_store_flag): Test rev_cond != NULL in addition to
14985 reversed_comparison_code.
14986 (noce_try_store_flag_constants): Likewise.
14987 (noce_try_store_flag_mask): Likewise.
14988 (noce_try_addcc): Use rev_cond if non-NULL instead of
14989 reversed_comparison_code.
14990 (noce_try_cmove_arith): Likewise. Formatting fixes.
14991 (noce_try_minmax, noce_try_abs): Clear rev_cond.
14992 (noce_find_if_block): Initialize rev_cond.
14993 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
14994 instead of false as last argument never attempt to reverse it
14997 2017-02-23 Bin Cheng <bin.cheng@arm.com>
14999 PR tree-optimization/79663
15000 * tree-predcom.c (combine_chains): Process refs in reverse order
15001 only for ZERO length chains, and add explaining comment.
15003 2017-02-23 Jeff Law <law@redhat.com>
15005 PR tree-optimization/79578
15006 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
15007 in call to operand_equal_p.
15009 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
15012 * config/i386/cpuid.h: Fix another undefined behavior.
15014 2017-02-23 Richard Biener <rguenther@suse.de>
15016 PR tree-optimization/79683
15017 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
15018 vector types for data-refs.
15020 2017-02-23 Martin Liska <mliska@suse.cz>
15022 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
15024 2017-02-23 Jakub Jelinek <jakub@redhat.com>
15026 PR middle-end/79665
15027 * internal-fn.c (get_range_pos_neg): Moved to ...
15028 * tree.c (get_range_pos_neg): ... here. No longer static.
15029 * tree.h (get_range_pos_neg): New prototype.
15030 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
15031 are known to be in between 0 and signed maximum inclusive, try to
15032 expand both unsigned and signed divmod and use the cheaper one from
15035 2017-02-22 Jeff Law <law@redhat.com>
15037 PR tree-optimization/79578
15038 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
15039 to compare base operands.
15041 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
15044 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
15045 gpc_reg_operand instead of fpr_reg_operand.
15047 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
15049 * config/mips/mips.c (mips_return_in_memory): Force FP
15050 vector types to be returned in memory for o32 ABI.
15052 2017-02-22 Jakub Jelinek <jakub@redhat.com>
15054 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
15055 instead of DW_TAG_member for static data member declarations and don't
15056 set no_linkage_name for static inline data members.
15057 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
15060 2017-02-22 Martin Liska <mliska@suse.cz>
15062 * doc/invoke.texi: Replace inequality signs with square brackets
15065 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15067 PR tree-optimization/68644
15068 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
15070 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
15073 * lra-constraints.c (simplify_operand_subreg): Handle
15074 WORD_REGISTER_OPERATIONS targets.
15076 2017-02-22 Jakub Jelinek <jakub@redhat.com>
15079 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
15080 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
15081 elimination by swapping fld*.
15083 2017-02-22 Richard Biener <rguenther@suse.de>
15085 PR tree-optimization/79673
15086 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
15087 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
15088 irrelevant address-space qualifiers and avoiding a
15089 ADDR_SPACE_CONVERT_EXPR from fold_convert.
15091 2017-02-22 Richard Biener <rguenther@suse.de>
15093 PR tree-optimization/79666
15094 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
15095 to not symbolically negate if that may introduce undefined
15098 2017-02-22 Martin Liska <mliska@suse.cz>
15101 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
15102 * data-streamer-out.c (streamer_write_gcov_count_stream):
15104 * value-prof.c (stream_out_histogram_value): Make assert more
15105 precise based on type of counter.
15107 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
15110 * config/i386/i386.md (standard_x87sse_constant_load splitter):
15111 Use nonimmediate_operand instead of memory_operand for operand 1.
15112 (float-extend standard_x87sse_constant_load splitter): Ditto.
15114 2017-02-21 Jeff Law <law@redhat.com>
15116 PR tree-optimization/79621
15117 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
15118 blocks with edges to themselves.
15120 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15123 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
15124 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
15125 Use gimple_call_builtin_p.
15128 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
15129 on temporarily removed DEBUG_INSNs.
15131 PR tree-optimization/79649
15132 * tree-loop-distribution.c (classify_partition): Give up on
15133 non-generic address space loads/stores.
15135 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
15137 * doc/loop.texi (Loop manipulation): Remove nonexistent
15138 tree_ssa_loop_version from the documentation.
15139 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
15141 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15144 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
15145 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
15146 * config/rs6000/rs6000.c: Include except.h.
15147 (rs6000_expand_split_stack_prologue): Call
15148 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
15150 2017-02-21 Martin Jambor <mjambor@suse.cz>
15153 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
15154 have been analyzed.
15156 2017-02-21 Martin Jambor <mjambor@suse.cz>
15158 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
15159 for backward compatibility only.
15160 * doc/invoke.texi (Option Summary): Remove all references to
15161 -fipa-cp-alignment.
15163 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
15167 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15169 * lra-constraints.c (curr_insn_transform): Handle
15170 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
15172 2017-02-21 Martin Liska <mliska@suse.cz>
15174 * config/i386/i386.opt: Replace -masm-dialect with -masm.
15176 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
15178 PR translation/79638
15179 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
15181 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
15184 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
15185 (arm_function_ok_for_sibcall): Return false for an indirect call by
15186 descriptor if all the argument registers are used.
15187 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
15188 alignment of the function.
15190 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15192 PR tree-optimization/61441
15193 * simplify-rtx.c (simplify_const_unary_operation): For
15194 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
15195 the sNaN unmodified.
15197 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
15199 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
15200 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
15201 instead of SYSTEM_HEADER_DIR.
15203 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
15204 Martin Liška <mliska@suse.cz>
15206 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
15207 Fix typos and grammar, use active voice, and clarify.
15209 2017-02-20 Marek Polacek <polacek@redhat.com>
15211 PR middle-end/79537
15212 * gimplify.c (gimplify_expr): Handle unused *&&L;.
15215 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
15217 2017-02-20 Jakub Jelinek <jakub@redhat.com>
15220 * config/i386/i386.c (ix86_expand_builtin): Handle
15221 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
15222 ix86_builtins_isa[fcode].isa as a requirement of those
15223 flags and any other flag in the bitmask.
15224 (ix86_init_mmx_sse_builtins): Use 0 instead of
15225 ~OPTION_MASK_ISA_64BIT as mask.
15226 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
15227 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
15228 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
15229 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
15231 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15234 * lra-constraints.c (split_reg): Check requested split mode
15235 is supported by the register.
15237 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15239 * lra-constraints.c (simplify_operand_subreg): Remove early
15242 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15245 * lra-constraints.c (curr_insn_transform): Tighten condition
15246 for converting SUBREG reloads from OP_OUT to OP_INOUT.
15248 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15251 * lra-constraints.c (curr_insn_transform): Handle
15252 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
15254 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
15257 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
15259 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
15261 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
15264 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
15267 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15269 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
15272 2017-02-18 Jakub Jelinek <jakub@redhat.com>
15274 * final.c (last_columnnum, override_columnnum): New variables.
15275 (final_start_function): Set last_columnnum, pass it to begin_prologue
15276 hook and pass 0 to dwarf2out_begin_prologue.
15277 (final_scan_insn): Update override_columnnum. Pass last_columnnum
15278 to source_line debug hook.
15279 (notice_source_line): Compute last_columnnum and for debug_column_info
15280 return true on column changes.
15281 * debug.h (struct gcc_debug_hooks): Add column argument to
15282 source_line and begin_prologue hooks.
15283 (debug_nothing_int_charstar_int_bool): Remove prototype.
15284 (debug_nothing_int_int_charstar,
15285 debug_nothing_int_int_charstar_int_bool): New prototypes.
15286 (dwarf2out_begin_prologue): Add column argument.
15287 * debug.c (do_nothing_debug_hooks): Adjust source_line and
15288 begin_prologue hooks.
15289 (debug_nothing_int_charstar_int_bool): Remove.
15290 (debug_nothing_int_int_charstar,
15291 debug_nothing_int_int_charstar_int_bool): New functions.
15292 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
15293 through to dwarf2out_source_line.
15294 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
15295 (dwarf2out_source_line): Add column argument, emit it if requested.
15296 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
15298 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
15299 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
15300 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
15301 through to dwarf2out_begin_prologue.
15302 (vmsdbgout_source_line): Add column argument, pass it through to
15303 dwarf2out_source_line.
15304 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
15305 dbxout_source_line caller.
15306 (dbxout_source_line): Add column argument.
15308 * common.opt (gno-column-info, gcolumn-info): New options.
15309 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
15310 (check_die): Also test for multiple DW_AT_decl_column attributes.
15311 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
15312 DW_AT_decl_column if requested.
15313 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
15315 (gen_variable_die): Likewise.
15316 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
15317 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
15320 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
15321 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
15322 (ix86_handle_option): Handle OPT_m3dnowa.
15323 * doc/invoke.texi (-m3dnowa): Document.
15324 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
15325 -m3dnowa instead of -m3dnow -march=athlon.
15328 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
15329 instead of gcc_assert for K, r and R code checks. Formatting fixes.
15331 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15334 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
15335 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
15336 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
15337 generator for vsx_xxpermdi_<mode>_be.
15338 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
15339 force big-endian semantics.
15340 (vsx_xxpermdi_<mode>_be): New define_expand with same
15341 implementation as previous version of vsx_xxpermdi_<mode>.
15343 2017-02-17 Jakub Jelinek <jakub@redhat.com>
15345 PR tree-optimization/79327
15346 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
15347 variable, its initialization and use.
15349 2017-02-17 Julia Koval <julia.koval@intel.com>
15351 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
15352 (OPTION_MASK_ISA_PKU_UNSET): New.
15353 (ix86_handle_option): Handle -mrdpid.
15354 * config/i386/cpuid.h (bit_RDPID): New.
15355 * config/i386/driver-i386.c (host_detect_local_cpu):
15356 Detect RDPID feature.
15357 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
15358 * config/i386/i386-c.c (ix86_target_macros_internal):
15360 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
15361 (ix86_valid_target_attribute_inner_p): Add "rdpid".
15362 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
15363 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
15364 * config/i386/i386.md (define_insn "rdpid"): New.
15365 * config/i386/i386.opt Add -mrdpid.
15366 * config/i386/immintrin.h (_rdpid_u32): New.
15368 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
15370 PR rtl-optimization/79541
15371 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
15372 instead of transforming it into USE.
15374 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
15376 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
15377 If HONOR_SNANS (SFmode) force the input to a register.
15378 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
15379 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
15380 an frsp or similar insn.
15382 2017-02-17 Martin Liska <mliska@suse.cz>
15384 PR rtl-optimization/79577
15385 * params.def (selsched-max-sched-times): Increase minimum to 1.
15387 2017-02-17 Martin Liska <mliska@suse.cz>
15389 PR rtl-optimization/79574
15390 * gcse.c (want_to_gcse_p): Prevent integer overflow.
15392 2017-02-17 Martin Liska <mliska@suse.cz>
15394 PR tree-optimization/79529
15395 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
15396 ssa_defined_default_def_p to handle cases which are implicitly
15398 * tree-ssa.c (ssa_defined_default_def_p): New function.
15399 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
15400 which are implicitly defined.
15401 * tree-ssa.h (ssa_defined_default_def_p): Declare.
15403 2017-02-17 Richard Biener <rguenther@suse.de>
15405 PR middle-end/79576
15406 * params.def (max-ssa-name-query-depth): Limit to 10.
15408 2017-02-17 Richard Biener <rguenther@suse.de>
15410 PR tree-optimization/79552
15411 * tree-ssa-structalias.c (visit_loadstore): Properly verify
15414 2017-02-17 Richard Biener <rguenther@suse.de>
15417 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
15419 2017-02-17 Marek Polacek <polacek@redhat.com>
15421 PR middle-end/79536
15422 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
15423 (fold_negate_expr): New wrapper.
15425 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
15427 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
15428 Correct terminology and de-emphasize pre-standard behavior.
15430 2017-02-16 Alan Modra <amodra@gmail.com>
15432 PR rtl-optimization/79286
15433 * ira.c (def_dominates_uses): New function.
15434 (update_equiv_regs): Don't create an equivalence for insns that
15435 may trap where the register def does not dominate the use.
15437 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
15439 PR rtl-optimization/78127
15440 * lra.c (lra): Call lra_eliminate before finish the loop after
15443 2017-02-16 Richard Biener <rguenther@suse.de>
15445 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
15447 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
15448 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
15449 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
15450 (isl_val_int_from_wi): New function.
15451 (extract_affine_gmp): Rename to ...
15452 (extract_affine_wi): ... this, take a widest_int.
15453 (extract_affine_int): Just wrap extract_affine_wi.
15454 (add_param_constraints): Use isl_val_int_from_wi.
15455 (add_loop_constraints): Likewise, and extract_affine_wi.
15457 2017-02-15 Jeff Law <law@redhat.com>
15459 PR middle-end/79521
15460 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
15461 ira_init_register_move_cost_if_necessary.
15463 2017-02-15 Martin Sebor <msebor@redhat.com>
15465 PR middle-end/32003
15466 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
15467 removed in a prior commit.
15469 2017-02-15 Bin Cheng <bin.cheng@arm.com>
15471 PR tree-optimization/79347
15472 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
15473 counters during peeling.
15475 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
15477 * Makefile.in (site.exp): Remove "set ISLVER".
15479 2017-02-15 Jakub Jelinek <jakub@redhat.com>
15482 * real.c (real_from_integer): Call real_convert even for decimal.
15484 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
15487 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
15489 2017-02-14 Andrew Pinski <apinski@cavium.com>
15491 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
15492 cores and change the partno/implementer to be correct.
15493 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
15494 the 'B" as the implementer.
15495 * config/aarch64/aarch64-tune.md: Regenerate.
15497 2017-02-14 Carl Love <cel@us.ibm.com>
15499 * config/rs6000/rs6000.c: Add case statement entry to make the
15500 xvcvuxdsp built-in argument unsigned.
15501 * config/rs6000/vsx.md: Fix the source and return operand types so they
15502 match the instruction definitions from the ISA document. Fix typo
15503 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
15506 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
15509 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
15510 member early_clobber_alts.
15511 * lra-lives.c (reg_early_clobber_p): New.
15512 (process_bb_lives): Use it.
15513 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
15514 (debug_operand_data): Initialize early_clobber_alts.
15515 (setup_operand_alternative): Set up early_clobber_alts.
15516 (collect_non_operand_hard_regs): Ditto. Pass early clobber
15517 alternatives to new_insn_reg.
15518 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
15520 (lra_update_insn_regno_info): Pass the new arg.
15522 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15524 PR middle-end/79505
15525 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
15526 (new_oacc_loop_raw): Don't clear already cleared fields.
15529 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
15530 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
15531 _mm512_prefetch_i64gather_ps): New inline functions and macros.
15533 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
15536 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
15538 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
15541 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
15542 the extra instruction to the right place to store 128-bit constant
15545 2017-02-14 Martin Sebor <msebor@redhat.com>
15547 PR middle-end/79448
15548 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
15549 warning for strings of unknown length.
15551 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
15553 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
15555 2017-02-14 Jeff Law <law@redhat.com>
15558 * ira-costs.c (scan_one_insn): Initialize register move costs
15559 for pseudos seen in USE/CLOBBER insns.
15561 PR tree-optimization/79095
15562 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
15563 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
15564 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
15565 if the operands are known to be not equal, then the resulting range
15567 (intersect_ranges): If the new range is ~[0,0] and the old range is
15568 wide, then prefer ~[0,0].
15569 * tree-vrp.c (overflow_comparison_p_1): New function.
15570 (overflow_comparison_p): New function.
15571 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
15572 if NAME is used in an overflow test.
15573 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
15574 overflow check that can be expressed as an equality test, then adjust
15575 ops to be that equality test.
15577 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15579 * config/s390/s390-builtin-types.def: Remove flags argument.
15580 * config/s390/s390.c (s390_init_builtins): Likewise.
15582 2017-02-14 Martin Liska <mliska@suse.cz>
15584 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
15585 vector. Fix trailing white spaces.
15587 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
15589 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
15592 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15594 PR rtl-optimization/68664
15595 * config/arm/arm.c (arm_sched_can_speculate_insn):
15596 New function. Declare prototype.
15597 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15599 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15601 PR rtl-optimization/68664
15602 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
15604 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15606 2017-02-14 Amit Pawar <amit.pawar@amd.com>
15608 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
15609 max skip bytes for function, loop and jump.
15611 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15613 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
15614 ABS_EXPR for gimple dump.
15616 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15619 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
15621 PR tree-optimization/79408
15622 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
15623 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
15624 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
15625 also if rhs1 is INTEGER_CST.
15627 2017-02-14 Richard Biener <rguenther@suse.de>
15629 PR middle-end/79432
15630 * tree-into-ssa.c (insert_phi_nodes): When the function can
15631 have abnormal edges rewrite SSA names with broken use-def
15632 dominance out of SSA and register them for PHI insertion.
15634 2017-02-13 Martin Sebor <msebor@redhat.com>
15636 PR middle-end/79496
15637 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
15638 clearing info.nowrite flag when snprintf size argument is a range.
15640 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15642 * cprop.c (cprop_jump): Add missing space in string literal.
15643 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
15644 (get_constraint_for_component_ref): Likewise.
15645 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
15646 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
15647 * lra-constraints.c (process_alt_operands): Likewise.
15648 * ipa-inline.c (inline_small_functions): Likewise.
15649 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
15650 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15651 * trans-mem.c (diagnose_tm_1_op): Likewise.
15652 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
15653 (grid_parallel_clauses_gridifiable): Likewise.
15655 * config/nvptx/mkoffload.c (process): Add space in between
15658 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
15659 "MOD4_SSE_REGS" and "ALL_REGS".
15661 * spellcheck.c (test_data): Add , in between "foo" and "food".
15663 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15666 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
15667 boundary crossing check and subsequent code generation agree.
15669 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15671 * config/aarch64/aarch64.c (has_memory_op): Delete.
15672 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
15675 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15677 PR rtl-optimization/79388
15678 PR rtl-optimization/79450
15679 * combine.c (distribute_notes): When removing TEM_INSN for which
15680 corresponding dest has last value recorded, invalidate that last
15683 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15685 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
15686 of explicit '@'. Add missing assembly comment marker on branch costs
15689 2017-02-13 Nathan Sidwell <nathan@acm.org>
15691 * gengtype-lex.l (<in_struct>): Add '/'.
15693 2017-02-13 Martin Liska <mliska@suse.cz>
15696 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
15698 2017-02-13 Richard Biener <rguenther@suse.de>
15700 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
15702 * configure: Re-generate.
15703 * config.in: Likewise.
15704 * graphite-dependences.c: Simplify as if
15705 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
15706 * graphite-isl-ast-to-gimple.c: Likewise.
15707 * graphite-optimize-isl.c: Likewise.
15708 * graphite-poly.c: Likewise.
15709 * graphite-sese-to-poly.c: Likewise.
15710 * graphite.h: Likewise.
15711 * toplev.c: Include isl/version.h and use isl_version () for
15712 printing the ISL version.
15713 * doc/install.texi: Update ISL requirement.
15715 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15717 * doc/standards.texi (Standards): Update reference to
15720 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15722 * doc/extend.texi (Named Address Spaces): sourceware.org now
15724 * doc/install.texi (Binaries): Ditto.
15727 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15729 * doc/cpp.texi: Replace "stringify"/"stringification" with C
15730 standard terminology "stringize"/"stringizing" throughout.
15731 * doc/cppinternals.texi: Likewise.
15733 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15735 * doc/extend.texi: Fix some spelling mistakes and typos.
15736 * doc/invoke.texi: Likewise.
15738 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15741 * params.def (inline-min-speedup) Change from 10 to 8.
15743 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15745 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
15748 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15751 * ipa-inline-analysis.c (get_minimal_bb): New function.
15752 (record_modified): Use it.
15753 (remap_edge_change_prob): Handle also ancestor functions.
15755 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
15757 * doc/contrib.texi (Contributors): Remove broken link into
15758 the Mauve CVS repository.
15760 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15762 PR middle-end/79454
15763 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
15764 result computation whenever lhs doesn't have vector mode, not
15765 just when it has BLKmode.
15767 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15769 * doc/makefile.texi (profiledbootstrap): Refer to the
15770 installation instructions only in textual form.
15772 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15775 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
15777 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15779 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
15780 (Specific): Update mingw-w64 reference.
15782 (Specific): Remove broken link to Renesas RX processor.
15784 2017-02-10 Richard Biener <rguenther@suse.de>
15786 * toplev.c (process_options): Do not mention obsolete graphite
15787 options when printing sorry message about missing graphite support.
15788 Mention -floop-nest-optimize.
15790 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
15792 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
15793 (vtst_p16): Likewise.
15794 (vtstq_p8): Likewise.
15795 (vtstq_p16): Likewise.
15797 (vtstq_p64): Likewise.
15798 * config/arm/arm_neon.h (vgetq_lane_p64): New.
15799 (vset_lane_p64): New.
15800 (vsetq_lane_p64): New.
15802 2017-02-10 Jakub Jelinek <jakub@redhat.com>
15804 PR tree-optimization/79411
15805 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
15806 stmt operands are SSA_NAMEs used in abnormal phis.
15807 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
15810 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15813 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
15816 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15818 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
15820 2017-02-09 Jakub Jelinek <jakub@redhat.com>
15822 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
15826 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
15827 not arbitrary TREE_CONSTANT.
15830 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
15831 "omp declare target link" attribute unless is_global_var.
15832 * omp-offload.c (find_link_var_op): Likewise.
15834 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
15835 Chung-Lin Tang <cltang@codesourcery.com>
15837 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
15839 (gimplify_adjust_omp_clauses): Don't delete TILE.
15840 (gimplify_omp_for): Deal with TILE.
15841 * internal-fn.c (expand_GOACC_TILE): New function.
15842 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
15844 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
15845 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
15847 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
15848 avoid DIV for outermost collapse var.
15849 (expand_oacc_for): Insert tile element loop as needed. Adjust.
15850 Remove out of date comments, fix whitespace.
15851 * omp-general.c (omp_extract_for_data): Deal with tiling.
15852 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
15853 adjust OLF_DIM_BASE value.
15854 (struct omp_for_data): Add tiling field.
15855 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
15856 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
15857 for auto loops. Remove default auto determining, moved to
15858 oacc_loop_fixed_partitions.
15859 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
15860 stmts, add e_mask field.
15861 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
15862 (oacc_thread_numbers): Use oacc_dim_call.
15863 (oacc_xform_tile): New.
15864 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
15865 (finish_oacc_loop): Adjust for ifns vector.
15866 (oacc_loop_discover_walk): Append loop abstraction sites to list,
15867 add case for GOACC_TILE fns.
15868 (oacc_loop_xform_loop): Delete.
15869 (oacc_loop_process): Iterate over call list directly, and add
15870 handling for GOACC_TILE fns.
15871 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
15873 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
15874 vector partitioning to outer loops. Assign 2 partitions to loops
15875 when available. Add TILE handling.
15876 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
15877 (execite_oacc_device_lower): Process GOACC_TILE fns,
15878 ignore unknown specs.
15879 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
15880 * tree.c (omp_clause_num_ops): Adjust TILE ops.
15881 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
15883 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
15885 * configure.ac (ACX_BUGURL): Update.
15886 * configure: Regenerate.
15888 2017-02-09 Richard Biener <rguenther@suse.de>
15890 PR tree-optimization/69823
15891 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
15892 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
15894 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15896 * config/arc/arc-c.def: Add __NPS400__ definition.
15897 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
15898 (TARGET_NPS400): Define.
15900 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15902 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
15904 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
15905 pointer, arch_info.
15906 (arc_cpu_types): Fill the arch_info field with a pointer into the
15907 arc_arch_types table.
15908 (arc_selected_cpu): Declare.
15909 * config/arc/arc.c (arc_selected_cpu): Make global.
15910 (arc_selected_arch): Delete.
15911 (arc_base_cpu): Delete.
15912 (arc_override_options): Remove references to deleted variables,
15913 update access to arch information.
15914 (ARC_OPT): Update access to arch information.
15915 (ARC_OPTX): Likewise.
15916 * config/arc/arc.h (arc_base_cpu): Remove declaration.
15917 (TARGET_ARC600): Update access to arch information.
15918 (TARGET_ARC601): Likewise.
15919 (TARGET_ARC700): Likewise.
15920 (TARGET_EM): Likewise.
15921 (TARGET_HS): Likewise.
15922 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
15925 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
15928 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
15929 condition/operands for integer GE/LE/GEU/LEU operations.
15931 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
15933 PR translation/79397
15934 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
15937 2017-02-08 Martin Jambor <mjambor@suse.cz>
15940 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
15941 whether allocation happened.
15942 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
15943 nothing was allocated.
15945 2017-02-08 Jakub Jelinek <jakub@redhat.com>
15947 PR tree-optimization/79408
15948 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
15949 constant, but SSA_NAME with a known integer range, use the minimum
15950 of that range instead of op1 to determine if modulo can be replaced
15951 with its first operand.
15953 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15955 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
15957 2017-02-08 Richard Biener <rguenther@suse.de>
15959 PR tree-optimization/71824
15960 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
15961 Check all loops contained in the merged region.
15963 2017-02-07 Andrew Pinski <apinski@cavium.com>
15965 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
15967 2017-02-07 Andrew Pinski <apinski@cavium.com>
15969 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
15970 (thunderxt88): Likewise.
15971 (thunderxt81): Disable LSE and change v8.1 to v8.
15972 (thunderxt83): Likewise.
15974 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15975 Richard Biener <rguenther@suse.de>
15977 PR middle-end/79399
15978 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
15979 type from int to size_t.
15980 * ira-costs.c (struct_costs_size): Change type from int to size_t.
15982 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15984 PR rtl-optimization/79386
15985 * cprop.c (bypass_conditional_jumps): Initialize
15986 bypass_last_basic_block already before splitting bbs after
15987 unconditional traps...
15988 (bypass_conditional_jumps): ... rather than here.
15991 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
15992 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
15993 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
15994 fix -masm=intel patterns.
15996 2017-02-07 Richard Biener <rguenther@suse.de>
15998 PR tree-optimization/79256
15999 PR middle-end/79278
16000 * builtins.c (get_object_alignment_2): Use min_align_of_type
16001 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
16002 and ADJUST_FIELD_ALIGN.
16004 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
16006 * doc/tm.texi: Regenerate.
16007 * stor-layout.c (layout_decl): Adjust.
16008 (update_alignment_for_field): Likewise.
16009 (place_field): Likewise.
16010 (min_align_of_type): Likewise.
16011 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
16012 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
16013 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
16014 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
16015 * config/frv/frv.c (frv_adjust_field_align): Likewise.
16016 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
16017 * config/i386/i386.c (x86_field_alignment): Likewise.
16018 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
16019 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
16020 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
16021 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
16022 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
16023 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
16027 2017-01-30 Richard Biener <rguenther@suse.de>
16029 PR tree-optimization/79256
16030 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
16031 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
16034 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
16036 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
16037 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
16038 builtins to SImode and emit a zero-extend, if necessary.
16040 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
16042 * docs/invoke.texi (RISC-V Options): Alphabetize.
16044 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
16046 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
16049 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
16051 * config/riscv/riscv.c: New file.
16052 * common/config/riscv/riscv-common.c: Likewise.
16053 * config.gcc: Likewise.
16054 * config/riscv/constraints.md: Likewise.
16055 * config/riscv/elf.h: Likewise.
16056 * config/riscv/generic.md: Likewise.
16057 * config/riscv/linux.h: Likewise.
16058 * config/riscv/multilib-generator: Likewise.
16059 * config/riscv/peephole.md: Likewise.
16060 * config/riscv/pic.md: Likewise.
16061 * config/riscv/predicates.md: Likewise.
16062 * config/riscv/riscv-builtins.c: Likewise.
16063 * config/riscv/riscv-c.c: Likewise.
16064 * config/riscv/riscv-ftypes.def: Likewise.
16065 * config/riscv/riscv-modes.def: Likewise.
16066 * config/riscv/riscv-opts.h: Likewise.
16067 * config/riscv/riscv-protos.h: Likewise.
16068 * config/riscv/riscv.h: Likewise.
16069 * config/riscv/riscv.md: Likewise.
16070 * config/riscv/riscv.opt: Likewise.
16071 * config/riscv/sync.md: Likewise.
16072 * config/riscv/t-elf-multilib: Likewise.
16073 * config/riscv/t-linux: Likewise.
16074 * config/riscv/t-linux-multilib: Likewise.
16075 * config/riscv/t-riscv: Likewise.
16076 * configure.ac: Likewise.
16077 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
16078 Waterman as RISC-V maintainers.
16079 * doc/install.texi: Add RISC-V entries.
16080 * doc/invoke.texi: Add RISC-V options section.
16081 * doc/md.texi: Add RISC-V constraints section.
16082 * configure: Regenerated.
16084 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
16087 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
16088 false values to be constant vectors with all 0 or all 1 bits set.
16089 (vcondu<mode><mode>): Likewise.
16090 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
16092 (fpmask_comparison_operator): Update comment.
16093 (vecint_comparison_operator): New predicate.
16094 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
16095 vector conditionals when the true and false values are constant
16096 vectors with all 0 bits or all 1 bits set.
16098 2017-02-06 Martin Sebor <msebor@redhat.com>
16100 PR tree-optimization/79376
16101 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
16103 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
16105 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
16106 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
16107 to simplify split condition.
16109 2017-02-06 Jakub Jelinek <jakub@redhat.com>
16111 * omp-expand.c (oxpand_omp_atomic_fetch_op,
16112 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
16115 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
16117 PR rtl-optimization/68664
16118 * target.def (can_speculate_insn): New hook.
16119 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
16120 * doc/tm.texi: Regenerate.
16121 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
16122 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
16123 (rs6000_sched_can_speculate_insn): New function.
16125 2017-02-06 Jakub Jelinek <jakub@redhat.com>
16127 PR tree-optimization/79284
16128 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
16129 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
16130 vectorizable_mask_load_store, vectorizable_operation,
16131 vect_is_simple_cond, get_same_sized_vectype): Use it instead
16132 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
16133 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
16134 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
16135 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
16136 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
16137 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
16138 is_gimple_assign (stmt). Replace another such test with
16139 is_gimple_assign (stmt).
16141 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
16144 * config/avr/avr.c (rtl-iter.h): Include it.
16145 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
16146 (avr_legitimate_combined_insn): ...and implementation.
16148 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
16150 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
16151 * config/s390/s390.c (s390_const_operand_ok)
16152 (s390_canonicalize_comparison, s390_extract_part)
16153 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
16154 (s390_contiguous_bitmask_p, s390_rtx_costs)
16155 (legitimize_pic_address): Likewise.
16156 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
16157 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
16158 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
16159 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
16160 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
16162 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
16164 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
16165 REGNO($0) == REGNO($1).
16167 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16169 * config/s390/linux.h(SIZE_TYPE): Add comment.
16171 2017-02-06 Julian Brown <julian@codesourcery.com>
16172 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16173 Virendra Pathak <virendra.pathak@broadcom.com>
16175 * config/aarch64/aarch64-cores.def: Change the scheduler
16177 * config/aarch64/aarch64.md: Include thunderx2t99.md.
16178 * config/aarch64/thunderx2t99.md: New file.
16180 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16182 * doc/standards.texi (Go Language): Update link to language
16185 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
16187 * tree-eh.c (lower_resx): Sanitize profile.
16188 (cleanup_empty_eh_move_lp): Likewise.
16190 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
16193 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
16195 * cfgloopmanip.h (loop_version): Update prototype.
16196 * modulo-sched.c (sms_schedule): Update call of loop_version.
16197 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
16198 * tree-parloops.c (gen_parallel_loop): Likewise.
16199 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
16200 * tree-ssa-loop-split.c (split_loop): Likewise.
16201 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
16202 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
16204 2017-02-05 Martin Liska <mliska@suse.cz>
16207 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
16209 (print_operand_address): Initialize a struct to zero.
16211 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16213 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
16214 garbage collector only in textual form.
16216 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16218 * doc/extend.texi (x86 specific memory model extensions for
16219 transactional memory): Simplify a phrase.
16221 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
16224 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
16225 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
16226 (atomic_storedi_1): Likewise.
16228 2017-02-04 Jakub Jelinek <jakub@redhat.com>
16230 PR tree-optimization/79338
16231 * tree-parloops.c (gather_scalar_reductions): Don't call
16232 vect_analyze_loop_form for loop->inner before destroying loop's
16235 2017-02-03 Martin Sebor <msebor@redhat.com>
16237 PR tree-optimization/79327
16238 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
16239 when precision has resulted in leading zeros.
16240 (format_integer): Adjust the likely counter to assume an unknown
16241 argument that may be zero is non-zero.
16243 2017-02-03 Jason Merrill <jason@redhat.com>
16246 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
16247 avoid copying non-taken branch.
16249 2017-02-03 Jakub Jelinek <jakub@redhat.com>
16251 PR tree-optimization/79340
16252 * tree-vect-loop.c (vectorizable_reduction): Release
16253 vec_defs elements after safe_splicing them into other vectors.
16256 PR tree-optimization/79327
16257 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
16258 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
16260 (format_integer): Use wide_int_to_tree instead of build_int_cst
16261 + to_?hwi. If argmin is NULL, just set argmin and argmax to
16262 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
16263 of shortest and longest sequence.
16265 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
16267 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
16268 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
16270 2017-02-03 Walter Lee <walt@tilera.com>
16273 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
16274 after initial stackframe link reg save.
16275 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
16277 2017-02-03 Jakub Jelinek <jakub@redhat.com>
16280 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
16281 wu for stxssp alternative.
16283 2017-02-03 Martin Sebor <msebor@redhat.com>
16285 PR tree-optimization/79352
16286 * gimple-fold.c (get_range_strlen): Add argument.
16287 (get_range_strlen): Change return type to bool.
16288 (get_maxval_strlen): Pass in a dummy argument.
16289 * gimple-fold.h (get_range_strlen): Change return type to bool.
16290 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
16291 * tree.h (array_at_struct_end_p): Add argument.
16292 * tree.c (array_at_struct_end_p): Handle it.
16294 2017-02-03 Martin Liska <mliska@suse.cz>
16297 * multiple_target.c (create_dispatcher_calls): Redirect edge
16298 from a caller of a dispatcher.
16299 (expand_target_clones): Make the clones local.
16300 (ipa_target_clone): Do both target clones and resolvers.
16301 (ipa_dispatcher_calls): Remove the pass.
16302 (pass_dispatcher_calls::gate): Likewise.
16303 (make_pass_dispatcher_calls): Likewise.
16304 * passes.def (pass_target_clone): Put as very first IPA early
16307 2017-02-03 Martin Liska <mliska@suse.cz>
16309 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
16310 in case of a function with ifunc attribute.
16312 2017-02-03 Martin Liska <mliska@suse.cz>
16314 * cgraph.c (cgraph_node::dump): Dump function version info.
16315 * symtab.c (symtab_node::dump_base): Add missing new line.
16317 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
16319 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
16320 (ifcombine_ifandif): Use it.
16322 2017-02-03 Martin Liska <mliska@suse.cz>
16324 * doc/invoke.texi: Document default value for
16325 use-after-scope-direct-emission-threshold.
16327 2017-02-03 Martin Liska <mliska@suse.cz>
16329 PR tree-optimization/79339
16330 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
16331 (format_floating): Likewise.
16333 2017-02-03 Martin Liska <mliska@suse.cz>
16336 * ipa-prop.c (ipa_node_params_t::insert): Remove current
16338 (ipa_node_params_t::remove): Likewise.
16339 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
16340 initialization from removed ipa_node_params_t::insert.
16341 (ipa_node_params::~ipa_node_params): Move from removed
16342 ipa_node_params_t::release.
16343 * symbol-summary.h (symbol_summary::m_released): New member.
16344 Do not release a summary twice. Do not allow to call finalizer
16345 for types of a summary that live in GGC memory.
16347 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16349 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
16352 2017-02-02 Martin Sebor <msebor@redhat.com>
16354 PR middle-end/79275
16355 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
16356 (format_string): Tighten up the range of output for non-constant
16357 strings and correct the expected range for wide non-constant strings.
16359 2017-02-02 Martin Sebor <msebor@redhat.com>
16361 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
16363 PR middle-end/32003
16364 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
16366 (-fdump-tree-@var): Add to index and document how to come up
16367 with pass-specific option and dump file names.
16368 (-fdump-passes): Clarify where to look for output.
16370 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
16372 PR middle-end/77445
16373 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
16374 statistics of the analyzed path; allow threading for speed when
16375 any of BBs along the path are optimized for speed.
16377 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
16379 PR middle-end/78468
16380 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
16381 settings of the virtual registers.
16384 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
16386 * explow.c (get_dynamic_stack_size): Take known alignment of stack
16387 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
16390 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16392 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
16393 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
16395 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16397 * config/s390/s390.md: Add missing comments with the expanded
16399 * config/s390/vector.md: Likewise.
16400 * config/s390/vx-builtins.md: Likewise.
16402 2017-02-02 Jakub Jelinek <jakub@redhat.com>
16405 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
16406 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
16407 conditions on a single line.
16409 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16411 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
16412 __S390_VX__ to __VX__.
16414 2017-02-01 Andrew Pinski <apinski@cavium.com>
16416 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
16417 stmt_info to record_stmt_cost.
16418 (vect_get_known_peeling_cost): Pass stmt_info if known to
16420 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
16421 cpu_vector_cost field into
16422 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
16423 field into vec_int_stmt_cost and vec_fp_stmt_cost.
16424 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
16425 splitting of scalar_stmt_cost and vec_stmt_cost.
16426 (thunderx_vector_cost): Likewise.
16427 (cortexa57_vector_cost): LIkewise.
16428 (exynosm1_vector_cost): Likewise.
16429 (xgene1_vector_cost): Likewise.
16430 (thunderx2t99_vector_cost): Improve after the splitting of the two
16432 (aarch64_builtin_vectorization_cost): Update for the splitting of
16433 scalar_stmt_cost and vec_stmt_cost.
16435 2017-02-01 Torvald Riegel <triegel@redhat.com>
16436 Richard Henderson <rth@redhat.com>
16438 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
16439 conditional on existance of a fast atomic load.
16440 * optabs-query.c (can_atomic_load_p): New function.
16441 * optabs-query.h (can_atomic_load_p): Declare it.
16442 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
16443 no fast atomic load is available for the particular size of access.
16444 (expand_atomic_compare_and_swap): Likewise.
16445 (expand_atomic_load): Likewise.
16446 (expand_atomic_store): Likewise.
16447 (expand_atomic_fetch_op): Likewise.
16448 * testsuite/lib/target-supports.exp
16449 (check_effective_target_sync_int_128): Remove x86 because it provides
16450 no fast atomic load.
16451 (check_effective_target_sync_int_128_runtime): Likewise.
16453 2017-02-01 Richard Biener <rguenther@suse.de>
16455 * graphite.c: Include tree-vectorizer.h for find_loop_location.
16456 (graphite_transform_loops): Provide opt-info for optimized nests.
16457 * tree-parloop.c (parallelize_loops): Provide opt-info for
16458 parallelized loops.
16460 2017-02-01 Richard Biener <rguenther@suse.de>
16462 PR middle-end/79315
16463 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
16464 was not set before.
16466 2017-02-01 Richard Biener <rguenther@suse.de>
16468 PR tree-optimization/71824
16469 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
16470 Verify the loops are valid in the merged SESE region.
16471 (scop_detection::can_represent_loop_1): Check analyzing the
16472 evolution of the number of iterations in the region succeeds.
16474 2017-01-31 Ian Lance Taylor <iant@golang.org>
16476 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
16477 REG_ARGS_SIZE note to 32-bit push insns and call insn.
16479 2017-01-31 David Malcolm <dmalcolm@redhat.com>
16481 PR preprocessor/79210
16482 * input.c (get_substring_ranges_for_loc): Replace line_width
16483 assertion with error-handling.
16485 2017-01-31 Richard Biener <rguenther@suse.de>
16487 PR tree-optimization/77318
16488 * graphite-sese-to-poly.c (extract_affine): Fix assert.
16489 (create_pw_aff_from_tree): Take loop parameter.
16490 (add_condition_to_pbb): Pass loop of the condition to
16491 create_pw_aff_from_tree.
16493 2017-01-31 Jakub Jelinek <jakub@redhat.com>
16495 * config/s390/s390.c (s390_asan_shadow_offset): New function.
16496 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
16498 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
16502 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
16504 (convert_int_to_float128): Likewise.
16505 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
16506 (convert_int_to_float128): Likewise.
16507 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
16508 (UNSPEC_IEEE128_CONVERT): Likewise.
16509 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
16510 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
16511 Use local variables for IBM extended format.
16512 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
16513 (fix_trunc<mode>si2_fprs): Likewise.
16514 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
16515 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
16516 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
16517 to know that we can now have integers of all sizes in vector
16519 (fix<uns>_<mode>di2_hw): Likewise.
16520 (float<uns>_<mode>si2_hw): Likewise.
16521 (fix_<mode>si2_hw): Likewise.
16522 (fixuns_<mode>si2_hw): Likewise.
16523 (float<uns>_<mode>di2_hw): Likewise.
16524 (float_<mode>di2_hw): Likewise.
16525 (float_<mode>si2_hw): Likewise.
16526 (floatuns_<mode>di2_hw): Likewise.
16527 (floatuns_<mode>si2_hw): Likewise.
16528 (xscvqp<su>wz_<mode>): Delete, no longer used.
16529 (xscvqp<su>dz_<mode>): Likewise.
16530 (xscv<su>dqp_<mode>): Likewise.
16531 (ieee128_mfvsrd_64bit): Likewise.
16532 (ieee128_mfvsrd_32bit): Likewise.
16533 (ieee128_mfvsrwz): Likewise.
16534 (ieee128_mtvsrw): Likewise.
16535 (ieee128_mtvsrd_64bit): Likewise.
16536 (ieee128_mtvsrd_32bit): Likewise.
16538 2017-01-31 Martin Liska <mliska@suse.cz>
16541 * ipa-prop.c (ipa_free_all_node_params): Call release method
16542 instead of ~sumbol_summary to not to trigger double times
16545 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
16547 PR tree-optimization/71691
16548 * bitmap.h (class auto_bitmap): New.
16549 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
16550 is_maybe_undefined instead of ssa_undefined_value_p.
16552 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16554 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
16555 __S390_ARCH_LEVEL__ to __ARCH__.
16557 2017-01-31 Jakub Jelinek <jakub@redhat.com>
16559 PR tree-optimization/79267
16560 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
16561 if should_remove_lhs_p is true.
16563 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
16566 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
16567 (add_alignment_attribute): New.
16568 (base_type_die): Add alignment attribute.
16569 (subrange_type_die): Likewise.
16570 (modified_type_die): Likewise.
16571 (gen_array_type_die): Likewise.
16572 (gen_descr_array_type_die: Likewise.
16573 (gen_enumeration_type_die): Likewise.
16574 (gen_subprogram_die): Likewise.
16575 (gen_variable_die): Likewise.
16576 (gen_field_die): Likewise.
16577 (gen_ptr_to_mbr_type_die): Likewise.
16578 (gen_struct_or_union_type_die): Likewise.
16579 (gen_subroutine_type_die): Likewise.
16580 (gen_typedef_die): Likewise.
16581 (base_type_cmp): Compare alignment attribute.
16583 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16586 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
16587 (setb_unsigned) New pattern for setb with CCUNS.
16588 * config/rs6000/rs6000.c (expand_block_compare): Use a different
16589 subfc./subfe sequence to avoid overflow problems. Generate a
16590 shorter sequence with cmpld/setb for power9.
16591 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
16592 for generating subfc. instruction.
16593 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
16594 now uses this instruction.
16596 2017-01-30 Ian Lance Taylor <iant@google.com>
16599 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
16600 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
16602 2017-01-30 Martin Sebor <msebor@redhat.com>
16604 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
16605 Move constant to the right of a relational operator.
16606 (get_mpfr_format_length, format_character, format_string): Ditto.
16607 (should_warn_p, maybe_warn): Same.
16609 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
16611 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
16614 * asan.c (get_translation_unit_decl): Remove function.
16615 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
16617 2017-01-30 Martin Liska <mliska@suse.cz>
16619 PR gcov-profile/79259
16620 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
16621 -fprofile-generate.
16623 2017-01-30 Martin Liska <mliska@suse.cz>
16626 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
16627 Initialize variables with NULL value.
16629 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
16632 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
16634 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
16636 2017-01-30 Richard Biener <rguenther@suse.de>
16638 PR tree-optimization/79276
16639 * tree-vrp.c (process_assert_insertions): Properly adjust common
16640 when removing a duplicate.
16642 * gcc.dg/torture/pr79276.c: New testcase.
16644 2017-01-30 Richard Biener <rguenther@suse.de>
16646 PR tree-optimization/79256
16647 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
16648 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
16650 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
16652 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
16655 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
16656 ("*r<noxa>sbg_<mode>_sll_bitmask")
16657 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
16658 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
16659 Use contiguous_bitmask_nowrap_operand.
16661 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16664 * config/rs6000/altivec.h (vec_xl): Revise #define.
16665 (vec_xst): Likewise.
16667 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
16669 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
16671 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
16673 PR rtl-optimization/79194
16674 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
16675 traps before call to bypass_conditional_jumps.
16677 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16679 PR tree-optimization/71374
16680 * lra-constraints.c (check_conflict_input_operands): New.
16681 (match_reload): Use it.
16683 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16686 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
16687 account to calculate conflict_set.
16689 2017-01-27 Bin Cheng <bin.cheng@arm.com>
16691 PR rtl-optimization/78559
16692 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
16693 other_insn in combine.
16695 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16697 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
16698 uint16_type_node for BT_UINT16.
16700 2017-01-27 David Malcolm <dmalcolm@redhat.com>
16702 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
16703 "RTL Tests" to menu.
16704 (GIMPLE Tests): New node.
16705 (RTL Tests): New node.
16707 2017-01-27 Richard Biener <rguenther@suse.de>
16709 PR tree-optimization/79245
16710 * tree-loop-distribution.c (distribute_loop): Apply cost
16711 modeling also to detected patterns.
16713 2017-01-27 Richard Biener <rguenther@suse.de>
16715 PR tree-optimization/71433
16716 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
16717 (compare_assert_loc): New function.
16718 (process_assert_insertions): Sort and optimize assert locations
16719 to remove duplicates and push down identical assertions on
16720 edges to their destination block.
16722 2017-01-27 Richard Biener <rguenther@suse.de>
16724 PR tree-optimization/79244
16725 * tree-vrp.c (remove_range_assertions): Forcefully propagate
16726 out SSA names even if abnormal.
16728 2017-01-27 Jakub Jelinek <jakub@redhat.com>
16730 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
16731 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
16732 instead of MPFR_RNDN.
16734 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
16737 * arm.c (arm_option_override): Don't call build_target_option_node
16738 until after doing all option overrides.
16739 (arm_valid_target_attribute_tree): Likewise.
16741 2017-01-27 Martin Liska <mliska@suse.cz>
16743 * doc/invoke.texi (-fprofile-arcs): Document profiling support
16744 for {cd}tors and C++ {cd}tors.
16746 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
16748 * config/s390/s390.md ("*setmem_long_and")
16749 ("*setmem_long_and_31z"): Use zero_extend instead of and.
16751 2017-01-26 Martin Sebor <msebor@redhat.com>
16753 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
16756 2017-01-26 Martin Sebor <msebor@redhat.com>
16758 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
16759 HAVE_DFmode before using XFmode or DFmode.
16760 (parse_directive): Avoid using the z length modifier to avoid
16761 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
16763 PR middle-end/78703
16764 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
16765 to accept adjustment as an array.
16766 (get_int_range): New function.
16767 (struct directive): Make width and prec arrays.
16768 (directive::set_width, directive::set_precision): Call get_int_range.
16769 (format_integer, format_floating): Handle width and precision ranges.
16770 (format_string, parse_directive): Same.
16772 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16775 * dwarf2out.c (generate_skeleton_bottom_up): For children with
16776 comdat_type_p set, just clone them, but keep the children in the
16780 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
16781 which have direct callers with -fvar-tracking-assignments enabled
16783 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
16784 inside of type units.
16786 2017-01-26 Martin Sebor <msebor@redhat.com>
16788 PR middle-end/78703
16789 * gimple-ssa-sprintf.c (struct result_range): Add likely and
16791 (struct format_result): Replace number_chars, number_chars_min,
16792 and number_chars_max with a single member of struct result_range.
16794 (format_result::operator+=): Adjust.
16795 (struct fmtresult): Remove bounded. Handle likely and unlikely
16797 (fmtresult::adjust_for_width_or_precision): New function.
16798 (fmtresult:type_max_digits): New function.
16799 (bytes_remaining): Handle likely and unlikely counters.
16800 (min_bytes_remaining): Remove.
16801 (format_percent): Simplify.
16802 (format_integer, format_floating): Set likely and unlikely counters.
16803 (get_string_length, format_character, format_string): Same.
16804 (format_plain, should_warn_p): New function.
16805 (maybe_warn): Call should_warn_p. Update diagnostic messages
16806 and handle those for all directives, including plain strings.
16807 (format_directive): Handle likely and unlikely counters.
16808 Remove unnecessary quoting from diagnostics. Add an informational
16810 (add_bytes): Remove.
16811 (pass_sprintf_length::compute_format_length): Simplify.
16812 (try_substitute_return_value): Handle likely and unlikely counters.
16814 2017-01-26 Carl Love <cel@us.ibm.com>
16816 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
16817 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
16819 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
16822 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
16823 endianess for subregs into account.
16824 * lra-constraints.c (lra_constraints): Do risky transformations
16825 always on the first iteration.
16826 * lra-lives.c (check_pseudos_live_through_calls): Add arg
16827 last_call_used_reg_set.
16828 (process_bb_lives): Define and use last_call_used_reg_set.
16829 * lra.c (lra): Always continue after lra_constraints on the first
16832 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
16834 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
16836 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
16838 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16840 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
16841 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
16842 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
16843 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
16844 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
16845 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
16846 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
16847 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
16848 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
16850 2017-01-26 Marek Polacek <polacek@redhat.com>
16853 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
16854 for the third operand.
16856 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16858 PR middle-end/79236
16859 * omp-low.c (struct omp_context): Add simt_stmt field.
16860 (scan_omp_for): Return omp_context *.
16861 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
16862 context to the _simt_ SIMD stmt.
16863 (lower_omp_for): For combined SIMD with sibling _simt_
16864 SIMD, make sure to use the same decls in _looptemp_
16865 clauses as in the sibling.
16867 2017-01-26 David Sherwood <david.sherwood@arm.com>
16869 PR middle-end/79212
16870 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
16873 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16876 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
16877 emit fld b; fld a; if possible.
16879 * brig-builtins.def: Update copyright years.
16880 * config/arm/arm_acle_builtins.def: Update copyright years.
16882 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16885 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
16886 constraint instead of o for the stxsd instruction.
16888 2017-01-25 Carl Love <cel@us.ibm.com>
16890 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
16891 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
16893 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
16895 * doc/invoke.texi (C++ Dialect Options): Fix typo.
16897 2017-01-25 Richard Biener <rguenther@suse.de>
16899 PR tree-optimization/69264
16900 * target.def (vector_alignment_reachable): Improve documentation.
16901 * doc/tm.texi: Regenerate.
16902 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
16904 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
16905 earlier changes with respect to TYPE_USER_ALIGN.
16906 (vector_alignment_reachable_p): Likewise. Improve dumping.
16908 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16911 * config/arm/arm.md (xordi3): Force constant operand into a register
16914 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16916 * doc/invoke.texi (-fstore-merging): Correct default optimization
16917 levels at which it is enabled.
16918 (-O): Move -fstore-merging from list to...
16921 2017-01-25 Richard Biener <rguenther@suse.de>
16924 * omp-expand.c: Include debug.h.
16925 (expand_omp_taskreg): Make sure to generate early debug before
16926 outlining anything from a function.
16927 (expand_omp_target): Likewise.
16928 (grid_expand_target_grid_body): Likewise.
16930 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
16933 * asan.c (get_translation_unit_decl): New function.
16934 (asan_add_global): Extract modules file name from globals
16935 TRANSLATION_UNIT_DECL name.
16937 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
16940 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
16941 for long calls with APCS frame and VFP.
16943 2017-01-24 David Malcolm <dmalcolm@redhat.com>
16945 * cfg.c (original_copy_tables_initialized_p): New function.
16946 * cfg.h (original_copy_tables_initialized_p): New decl.
16947 * cfgrtl.c (relink_block_chain): Guard the call to
16948 free_original_copy_tables with a call to
16949 original_copy_tables_initialized_p.
16950 * cgraph.h (symtab_node::native_rtl_p): New decl.
16951 * cgraphunit.c (symtab_node::native_rtl_p): New function.
16952 (symtab_node::needed_p): Don't assert for early assembly output
16953 for __RTL functions.
16954 (cgraph_node::finalize_function): Set "force_output" for __RTL
16956 (cgraph_node::analyze): Bail out early for __RTL functions.
16957 (analyze_functions): Update assertion to support __RTL functions.
16958 (cgraph_node::expand): Bail out early for __RTL functions.
16959 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
16961 * function.h (struct function): Update comment for field
16963 * gimple-expr.c: Include "tree-pass.h".
16964 (gimple_has_body_p): Return false for __RTL functions.
16965 * Makefile.in (OBJS): Add run-rtl-passes.o.
16966 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
16968 (gcc::pass_manager::get_clean_slate): New accessor.
16969 * passes.c: Include "insn-addr.h".
16970 (should_skip_pass_p): Add logging. Update logic for running
16971 "expand" to be compatible with both __GIMPLE and __RTL. Guard
16972 property-provider override so it is only done for gimple passes.
16974 (skip_pass): New function.
16975 (execute_one_pass): Call skip_pass when skipping passes.
16976 * read-md.c (md_reader::read_char): Support filtering
16977 the input to a subset of line numbers.
16978 (md_reader::md_reader): Initialize fields
16979 m_first_line and m_last_line.
16980 (md_reader::read_file_fragment): New function.
16981 * read-md.h (md_reader::read_file_fragment): New decl.
16982 (md_reader::m_first_line): New field.
16983 (md_reader::m_last_line): New field.
16984 * read-rtl-function.c (function_reader::create_function): Only
16985 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
16986 curr_properties. Set DECL_INITIAL to a dummy block.
16987 (read_rtl_function_body_from_file_range): New function.
16988 * read-rtl-function.h (read_rtl_function_body_from_file_range):
16990 * run-rtl-passes.c: New file.
16991 * run-rtl-passes.h: New file.
16993 2017-01-24 Jeff Law <law@redhat.com>
16995 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
16998 2017-01-24 Bin Cheng <bin.cheng@arm.com>
17000 PR tree-optimization/79159
17001 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
17002 (record_nonwrapping_iv): Improve boundary using above function if no
17003 value range information.
17005 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
17006 Martin Jambor <mjambor@suse.cz>
17008 * brig-builtins.def: New file.
17009 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
17010 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
17011 (DEF_HSAIL_SAT_BUILTIN): Likewise.
17012 (DEF_HSAIL_INTR_BUILTIN): Likewise.
17013 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
17014 * builtin-types.def (BT_INT8): New.
17015 (BT_INT16): Likewise.
17016 (BT_UINT8): Likewise.
17017 (BT_UINT16): Likewise.
17018 (BT_FN_ULONG): Likewise.
17019 (BT_FN_UINT_INT): Likewise.
17020 (BT_FN_UINT_ULONG): Likewise.
17021 (BT_FN_UINT_LONG): Likewise.
17022 (BT_FN_UINT_PTR): Likewise.
17023 (BT_FN_ULONG_PTR): Likewise.
17024 (BT_FN_INT8_FLOAT): Likewise.
17025 (BT_FN_INT16_FLOAT): Likewise.
17026 (BT_FN_UINT32_FLOAT): Likewise.
17027 (BT_FN_UINT16_FLOAT): Likewise.
17028 (BT_FN_UINT8_FLOAT): Likewise.
17029 (BT_FN_UINT64_FLOAT): Likewise.
17030 (BT_FN_UINT16_UINT32): Likewise.
17031 (BT_FN_UINT32_UINT16): Likewise.
17032 (BT_FN_UINT16_UINT16_UINT16): Likewise.
17033 (BT_FN_INT_PTR_INT): Likewise.
17034 (BT_FN_UINT_PTR_UINT): Likewise.
17035 (BT_FN_LONG_PTR_LONG): Likewise.
17036 (BT_FN_ULONG_PTR_ULONG): Likewise.
17037 (BT_FN_VOID_UINT64_UINT64): Likewise.
17038 (BT_FN_UINT8_UINT8_UINT8): Likewise.
17039 (BT_FN_INT8_INT8_INT8): Likewise.
17040 (BT_FN_INT16_INT16_INT16): Likewise.
17041 (BT_FN_INT_INT_INT): Likewise.
17042 (BT_FN_UINT_FLOAT_UINT): Likewise.
17043 (BT_FN_FLOAT_UINT_UINT): Likewise.
17044 (BT_FN_ULONG_UINT_UINT): Likewise.
17045 (BT_FN_ULONG_UINT_PTR): Likewise.
17046 (BT_FN_ULONG_ULONG_ULONG): Likewise.
17047 (BT_FN_UINT_UINT_UINT): Likewise.
17048 (BT_FN_VOID_UINT_PTR): Likewise.
17049 (BT_FN_UINT_UINT_PTR: Likewise.
17050 (BT_FN_UINT32_UINT64_PTR): Likewise.
17051 (BT_FN_INT_INT_UINT_UINT): Likewise.
17052 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
17053 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
17054 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
17055 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
17056 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
17057 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
17058 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
17059 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
17060 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
17061 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
17062 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
17063 * doc/frontends.texi: List BRIG FE.
17064 * doc/install.texi (Testing): Add BRIG tesring requirements.
17065 * doc/invoke.texi (Overall Options): Mention BRIG.
17066 * doc/standards.texi (Standards): Doucment BRIG HSA version.
17068 2017-01-24 Richard Biener <rguenther@suse.de>
17070 PR translation/79208
17071 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
17073 2017-01-24 Martin Jambor <mjambor@suse.cz>
17076 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
17077 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
17078 and known_contexts.
17080 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
17082 PR middle-end/79123
17083 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
17084 casts from signed to unsigned really don't have a range.
17086 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
17088 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
17089 GMP_RNDx for compatiblity.
17091 2017-01-24 Martin Liska <mliska@suse.cz>
17094 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
17095 that would prevent us to call alloca with -1 as argument.
17097 2017-01-24 Jakub Jelinek <jakub@redhat.com>
17099 * dwarf2out.c (output_compilation_unit_header, output_file_names):
17100 Avoid -Wformat-security warning.
17102 2017-01-23 Andrew Pinski <apinski@cavium.com>
17104 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
17107 2017-01-23 Martin Sebor <msebor@redhat.com>
17109 PR middle-end/78703
17110 * gimple-ssa-sprintf.c (warn_level): New global.
17111 (format_integer): Use it here and throughout the rest of the file.
17112 Use the same switch to compute sign as base.
17113 (maybe_warn): New function.
17114 (format_directive): Factor out warnings into maybe_warn.
17115 Add debugging output. Use warn_level.
17116 (add_bytes): Use warn_level.
17117 (pass_sprintf_length::compute_format_length): Add debugging output.
17118 (try_substitute_return_value): Same.
17119 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
17121 PR middle-end/78703
17122 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
17123 (struct fmtresult, format_integer, format_floating): Adjust.
17124 (fmtresult::fmtresult): Set max correctly in two argument ctor.
17125 (get_string_length, format_string,format_directive): Same.
17126 (pass_sprintf_length::compute_format_length): Same.
17127 (try_substitute_return_value): Simplify slightly.
17129 PR middle-end/78703
17130 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
17131 (fmtresult::operator+=): Outlined.
17132 (struct fmtresult): Add ctors.
17133 (struct conversion_spec): Rename...
17134 (struct directive): ...to this. Add and remove data members.
17135 (directive::set_width, directive::set_precision): New functions.
17136 (format_percent): Use fmtresult ctor.
17137 (get_width_and_precision): Remove.
17138 (format_integer): Make naming changes. Avoid computing width and
17140 (format_floating): Same. Adjust indentation.
17141 (format_character, format_none): New functions.
17142 (format_string): Moved character handling to format_character.
17143 (format_directive): Remove arguments, change return type.
17144 (parse_directive): New function.
17145 (pass_sprintf_length::compute_format_length): Move directive
17146 parsing to parse_directive.
17148 2017-01-23 Jakub Jelinek <jakub@redhat.com>
17150 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
17151 (assign_assembler_name_if_needed): ... this.
17152 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
17153 (assign_assembler_name_if_needed): ... this.
17154 (free_lang_data_in_cgraph): Adjust callers.
17155 * cgraphunit.c (cgraph_node::analyze): Likewise.
17156 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
17159 2017-01-23 Richard Biener <rguenther@suse.de>
17161 PR tree-optimization/79088
17162 PR tree-optimization/79188
17163 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
17164 resetting loop bounds after last path deletion. Reset loop
17165 bounds of the target loop, make code match the comments.
17166 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
17167 Make sure loops need no fixups.
17169 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
17171 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
17172 exponent support with double type for first argument.
17173 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
17174 type returned by __builtin_vec_extract_sig,
17175 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
17176 functions from "vector int" to "vector unsigned int" or from
17177 "vector long long int" to "vector unsigned long long int".
17178 Changed type returned by __builtin_vec_extract_exp,
17179 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
17180 functions from "vector int" to "vector unsigned int" or from
17181 "vector long long int" to "vector unsigned long long int".
17182 Changed return type of __builtin_vec_test_data_class,
17183 __builtin_vec_test_data_class_sp, and
17184 __builtin_vec_test_data_class_dp from "vector int" to
17185 "vector bool int" or from "vector long long int" to "vector bool
17186 long long int" and changed second argument type from "unsigned
17187 int" to "int". Added new overloaded function forms "vector float
17188 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
17189 "vector float __builtin_vec_insert_exp_sp (vector float, vector
17190 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
17191 double, vector unsigned long long int)" and "vector double
17192 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
17193 long int)". Changed return type of
17194 __builtin_scalar_test_data_class and
17195 __builtin_scalar_test_data_class_sp and
17196 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
17197 int" and changed second argument from "unsigned int" to "int".
17198 Changed type returned by __builtin_scalar_test_neg,
17199 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
17200 from "int" to "bool int". Added new overloaded function form
17201 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
17202 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
17203 exponent double-precision with floating point first argument.
17204 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
17205 documentation of scalar_test_data_class, scalar_test_neg,
17206 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
17207 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
17208 vec_test_data_class built-in functions to reflect refinements in
17209 their type signatures.
17211 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
17213 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
17215 (aarch64_elf_asm_destructor): Likewise.
17217 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
17219 PR rtl-optimization/78634
17220 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
17221 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
17222 * ifcvt.c (noce_try_cmove): Add missing cost check.
17224 PR rtl-optimization/71724
17225 * combine.c (if_then_else_cond): Look for situations where it is
17226 beneficial to undo the work of one of the recursive calls.
17228 2017-01-23 Bin Cheng <bin.cheng@arm.com>
17230 PR tree-optimization/70754
17231 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
17232 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
17233 combined stmt before it if not NULL.
17234 (combine_chains): Process refs reversely and compute dominance point
17237 2017-01-23 Martin Liska <mliska@suse.cz>
17239 PR tree-optimization/79196
17240 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
17241 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
17243 (strlen_optimize_stmt): Call the renamed function.
17245 2017-01-23 Michael Matz <matz@suse.de>
17247 PR tree-optimization/78384
17248 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
17250 2017-01-23 Richard Biener <rguenther@suse.de>
17252 PR tree-optimization/79186
17253 * tree-vrp.c (register_new_assert_for): Make sure we've seen
17254 both incoming edges before moving an assert.
17256 2017-01-23 Martin Jambor <mjambor@suse.cz>
17258 * ipa-prop.c (load_from_param_1): Removed.
17259 (load_from_unmodified_param): Bits from load_from_param_1 put back
17261 (load_from_param): Removed.
17262 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
17263 with stmt. Reverted back to use of load_from_unmodified_param.
17265 2017-01-23 Martin Jambor <mjambor@suse.cz>
17268 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
17269 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
17270 field a pointer to garbage collected vector, mark lattices and
17271 ipcp_orig_node with GTY((skip)).
17272 (ipa_get_param_count): Adjust to descriptors being a pointer.
17273 (ipa_get_param): Likewise.
17274 (ipa_get_type): Likewise.
17275 (ipa_get_param_move_cost): Likewise.
17276 (ipa_set_param_used): Likewise.
17277 (ipa_get_controlled_uses): Likewise.
17278 (ipa_set_controlled_uses): Likewise.
17279 (ipa_is_param_used): Likewise.
17280 (ipa_node_params_t): Move into garbage collector. New methods insert
17282 (ipa_node_params_sum): Annotate wth GTY(()).
17283 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
17285 (ipa_load_from_parm_agg): Adjust declaration.
17286 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
17287 * ipa-profile.c (ipa_profile): Likewise.
17288 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
17289 (ipa_populate_param_decls): Make descriptors parameter garbage
17291 (ipa_dump_param): Adjust to descriptors being a pointer.
17292 (ipa_alloc_node_params): Likewise.
17293 (ipa_initialize_node_params): Likewise.
17294 (load_from_param_1): Make descriptors parameter garbage collected.
17295 (load_from_unmodified_param): Likewise.
17296 (load_from_param): Likewise.
17297 (ipa_load_from_parm_agg): Likewise.
17298 (ipa_node_params::~ipa_node_params): Removed.
17299 (ipa_free_all_node_params): Remove call to delete operator.
17300 (ipa_node_params_t::insert): New.
17301 (ipa_node_params_t::remove): Likewise.
17302 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
17303 copy known_csts and known_contexts vectors.
17304 (ipa_read_node_info): Adjust to descriptors being a pointer.
17305 (ipcp_modif_dom_walker): Make m_descriptors field garbage
17307 (ipcp_transform_function): Make descriptors variable garbage
17310 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
17312 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
17313 * config/i386/avx512dqintrin.h: Ditto.
17314 * config/i386/avx512fintrin.h: Ditto.
17315 * config/i386/i386.c: Handle new builtins.
17316 * config/i386/i386-builtin.def: Add new builtins.
17317 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
17318 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
17320 2017-01-23 Jakub Jelinek <jakub@redhat.com>
17321 Martin Liska <mliska@suse.cz>
17323 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
17324 * asan.c (asan_expand_poison_ifn): Support stores and use
17325 appropriate ASAN report function.
17326 * internal-fn.c (expand_ASAN_POISON_USE): New function.
17327 * internal-fn.def (ASAN_POISON_USE): Declare.
17328 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
17329 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
17330 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
17331 ASAN_POISON calls w/o LHS.
17332 * tree-ssa.c (execute_update_addresses_taken): Create clobber
17333 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
17334 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
17335 * gimplify.c (asan_poison_variables): Add attribute
17336 use_after_scope_memory to variables that really needs to live
17338 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
17339 having the attribute.
17341 2017-01-23 Martin Liska <mliska@suse.cz>
17343 * asan.c (create_asan_shadow_var): New function.
17344 (asan_expand_poison_ifn): Likewise.
17345 * asan.h (asan_expand_poison_ifn): New declaration.
17346 * internal-fn.c (expand_ASAN_POISON): Likewise.
17347 * internal-fn.def (ASAN_POISON): New builtin.
17348 * sanopt.c (pass_sanopt::execute): Expand
17349 asan_expand_poison_ifn.
17350 * tree-inline.c (copy_decl_for_dup_finish): Make function
17352 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
17353 * tree-ssa.c (is_asan_mark_p): New function.
17354 (execute_update_addresses_taken): Rewrite local variables
17355 (identified just by use-after-scope as addressable) into SSA.
17357 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
17359 * doc/install.texi (Specific): opensource.apple.com uses https
17360 now. Remove trailing slash.
17362 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
17364 * README.Portability: Remove note on an Irix compatibility issue.
17366 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
17368 * gcov.c (INCLUDE_ALGORITHM): Define.
17369 (INCLUDE_VECTOR): Define.
17370 No longer include <vector> and <algorithm> directly.
17372 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
17374 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
17376 * doc/invoke.texi (Code Gen Options): Ditto.
17378 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
17381 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
17383 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
17385 rtl-optimization/79125
17386 * cprop.c (local_cprop_pass): Handle cases where we make an
17387 unconditional trap.
17389 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
17393 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
17394 read from, for big endian.
17396 2017-01-20 Jiong Wang <jiong.wang@arm.com>
17398 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
17399 register pauth builtins for LP64 only.
17401 2017-01-20 Marek Polacek <polacek@redhat.com>
17404 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
17407 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
17409 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
17411 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
17412 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
17413 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
17415 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17418 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
17419 in big-endian mode when they are not a single duplicated value.
17421 2017-01-20 Richard Biener <rguenther@suse.de>
17423 * BASE-VER: Bump to 7.0.1.
17425 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
17427 * omp-low.c (omplow_simd_context): New struct. Use it...
17428 (lower_rec_simd_input_clauses): ...here and...
17429 (lower_rec_input_clauses): ...here to hold common data. Adjust all
17430 references to idx, lane, max_vf, is_simt.
17432 2017-01-20 Graham Markall <graham.markall@embecosm.com>
17434 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
17437 2017-01-20 Martin Jambor <mjambor@suse.cz>
17439 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
17440 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
17442 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
17443 (GTFILES): Rename hsa.c to hsa-common.c.
17444 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
17445 * hsa-dump.c: Likewise.
17446 * hsa-gen.c: Likewise.
17447 * hsa-regalloc.c: Likewise.
17448 * ipa-hsa.c: Likewise.
17449 * omp-expand.c: Likewise.
17450 * omp-low.c: Likewise.
17451 * toplev.c: Likewise.
17453 2017-01-20 Marek Polacek <polacek@redhat.com>
17456 * doc/invoke.texi: Document -Wduplicated-branches.
17457 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
17458 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
17459 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
17460 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
17461 return 0 only when not OEP_LEXICOGRAPHIC.
17462 (fold_build_cleanup_point_expr): Use the expression
17463 location when building CLEANUP_POINT_EXPR.
17464 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
17465 * tree.c (add_expr): Handle error_mark_node.
17467 2017-01-20 Martin Liska <mliska@suse.cz>
17470 * tree-profile.c (init_ic_make_global_vars): Do not call
17472 (gimple_init_gcov_profiler): Likewise.
17474 2017-01-20 Martin Liska <mliska@suse.cz>
17477 * cgraph.h (maybe_create_reference): Remove argument and
17479 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
17481 * ipa-cp.c (create_specialized_node): Likewise.
17482 * symtab.c (symtab_node::maybe_create_reference): Handle
17483 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
17485 2017-01-20 Martin Liska <mliska@suse.cz>
17487 * read-rtl-function.c (function_reader::create_function): Use
17488 build_decl instread of build_decl_stat.
17490 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
17492 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
17493 * config/i386/avx512dqintrin.h: Ditto.
17494 * config/i386/avx512fintrin.h: Ditto.
17495 * config/i386/i386-builtin-types.def: Add new types.
17496 * config/i386/i386.c: Handle new types.
17497 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
17498 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
17499 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
17500 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
17501 (__builtin_ia32_kshiftridi): New.
17502 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
17504 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
17508 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
17509 define to rs6000_init_stack_protect_guard.
17510 (rs6000_init_stack_protect_guard): New function.
17512 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
17513 Yunqiang Su <yunqiang.su@imgtec.com>
17515 * config.gcc (supported_defaults): Add madd4.
17516 (with_madd4): Add validation.
17517 (all_defaults): Add madd4.
17518 * config/mips/mips.opt (mmadd4): New option.
17519 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17521 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17523 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
17524 (ISA_HAS_FUSED_MADD4): Likewise.
17525 * doc/invoke.texi (-mmadd4): Document the new option.
17526 * doc/install.texi (--with-madd4): Document the new option.
17528 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17530 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
17531 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
17532 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
17533 (aarch64_init_pauth_hint_builtins): New.
17534 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
17535 (aarch64_expand_builtin): Expand new builtins.
17537 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17539 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
17540 * combine-stack-adj.c (no_unhandled_cfa): Handle
17541 REG_CFA_TOGGLE_RA_MANGLE.
17542 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
17543 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
17544 info for return address signing.
17545 (aarch64_expand_epilogue): Likewise.
17547 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17549 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
17550 * config/aarch64/aarch64-protos.h
17551 (aarch64_return_address_signing_enabled): New declaration.
17552 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
17554 (aarch64_expand_prologue): Sign return address before it's pushed onto
17556 (aarch64_expand_epilogue): Authenticate return address fetched from
17558 (aarch64_override_options): Sanity check for ILP32 and ISA level.
17559 (aarch64_attributes): New function attributes for "sign-return-address".
17560 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
17561 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
17562 ("*do_return"): Generate combined instructions according to key index.
17563 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
17564 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
17566 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
17567 * config/aarch64/aarch64.opt (msign-return-address=): New.
17568 * doc/extend.texi (AArch64 Function Attributes): Documents
17569 "sign-return-address=".
17570 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
17572 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
17574 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
17575 overall option summary.
17577 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17579 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
17580 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
17581 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
17582 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
17584 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
17586 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
17587 -mpower9-minmax by default for -mcpu=power9.
17588 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
17589 128-bit floating point.
17591 2017-01-20 Alan Modra <amodra@gmail.com>
17593 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
17594 optimizing for size.
17596 2017-01-20 Alan Modra <amodra@gmail.com>
17599 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
17600 for strcmp and strncmp from corresponding builtin decl.
17602 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17604 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
17605 instead of i386/rtems-64.h.
17606 * config/i386/rtems-64.h: Remove.
17608 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17612 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
17614 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
17616 2017-01-19 Tamar Christina <tamar.christina@arm.com>
17618 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
17619 Change int to HOST_WIDE_INT.
17620 * config/aarch64/aarch64-protos.h
17621 (aarch64_simd_gen_const_vector_dup): Likewise.
17622 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
17624 2017-01-19 David Malcolm <dmalcolm@redhat.com>
17626 * langhooks-def.h (lhd_type_for_size): New decl.
17627 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
17628 * langhooks.c (lhd_type_for_size): New function, taken from
17631 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
17633 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
17634 define_bypass for CR latency.
17635 (power9-cracked-alu): Update bypass latency and remove power9-branch.
17636 (power9-alu2): Add define_bypass for CR latency.
17638 (power9-mul): Update insn latency.
17639 (power9-mul-compare): Update insn latency, bypass latency and remove
17642 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17644 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
17646 * config/aarch64/aarch64.md
17647 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
17648 aarch64_nopcrelative_literal_loads.
17649 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
17651 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
17653 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
17654 TARGET_LOONGSON_3A.
17655 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
17657 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
17660 * config.gcc (supported_defaults): Add lxc1-sxc1.
17661 (with_lxc1_sxc1): Add validation.
17662 (all_defaults): Add lxc1-sxc1.
17663 * config/mips/mips.opt (mlxc1-sxc1): New option.
17664 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17666 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17667 __mips_no_lxc1_sxc1.
17668 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
17669 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
17670 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
17672 2017-01-19 Richard Biener <rguenther@suse.de>
17674 PR tree-optimization/72488
17675 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
17676 sure to restore SSA info.
17677 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
17679 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
17681 PR rtl-optimization/79121
17682 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
17683 of the inner type when shifting an extended value.
17685 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
17688 * symtab.c (symtab_node::equal_address_to): Fix comparing of
17689 interposable aliases.
17691 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
17694 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
17695 Use the evmergelohi instruction.
17696 (mov_si<mode>_e500_subreg4_2_le): Likewise.
17697 (mov_sitf_e500_subreg8_2_be): Likewise.
17698 (mov_sitf_e500_subreg12_2_le): Likewise.
17699 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
17700 (mov_si<mode>_e500_subreg4_2_be): Likewise.
17701 (mov_sitf_e500_subreg8_2_le): Likewise.
17702 (mov_sitf_e500_subreg12_2_be): Likewise.
17704 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17706 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
17707 attribute from vecsimple to vecperm.
17708 (altivec_vbpermq2): Likewise.
17710 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17713 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
17715 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17716 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
17717 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
17718 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
17719 case where N arg is SIZE_MAX.
17720 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
17721 (cmpstrsi): Add pattern.
17723 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17725 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17726 __builtin_vec_revb builtins.
17727 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
17728 built-in functions to support generation of the ISA 3.0 XXBR<x>
17729 vector byte reverse instructions.
17730 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
17731 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
17732 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
17733 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
17734 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
17735 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
17736 (P9V_BUILTIN_VEC_REVB): Likewise.
17737 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
17738 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
17739 (p9_xxbrq_v16qi): Likewise.
17740 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
17741 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
17742 (p9_xxbrh_v8hi): Likewise.
17743 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
17744 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
17745 vec_revb built-in functions.
17747 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
17749 PR rtl-optimization/78952
17750 * config/i386/i386.md (any_extract): New code iterator.
17751 (*insvqi_2): Use any_extract for source operand.
17752 (*insvqi_3): Use any_shiftrt for source operand.
17754 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
17756 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
17758 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
17760 2017-01-18 Matthias Klose <doko@ubuntu.com>
17762 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
17764 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17766 * config/rs6000/altivec.h (vec_bperm): Change #define.
17767 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
17768 (altivec_vbpermq2): New define_insn.
17769 (altivec_vbpermd): Likewise.
17770 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
17771 function interface.
17772 (VBPERMD): Likewise.
17773 (VBPERM): New polymorphic function interface.
17774 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
17775 Add entries for P9V_BUILTIN_VEC_VBPERM.
17776 * doc/extend.texi: Add interfaces for vec_bperm.
17778 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17780 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
17781 first letter of error messages.
17782 (s390_resolve_overloaded_builtin): Likewise.
17783 * config/s390/s390.c (s390_expand_builtin): Likewise.
17784 (s390_invalid_arg_for_unprototyped_fn): Likewise.
17785 (s390_valid_target_attribute_inner_p): Likewise.
17786 * config/s390/s390.md ("tabort"): Likewise.
17788 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
17790 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
17791 (ISA_AVOID_DIV_HILO): New macro.
17792 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
17793 (ISA_HAS_DDIV): Likewise.
17795 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17797 * doc/invoke.texi (fabi-version): Correct number of occurrences.
17799 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17801 * doc/invoke.texi (fabi-version): Spelling fix.
17803 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17806 * doc/invoke.texi (fabi-version): Mention mangling fix for
17809 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17812 * doc/invoke.texi (fabi-version): Document discriminator mangling.
17814 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
17817 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
17818 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
17820 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
17822 (stack_protect_test): Ditto.
17823 * config/rs6000/rs6000.opt (mstack-protector-guard=,
17824 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
17826 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
17827 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
17828 -mstack-protector-guard-offset=.
17829 (RS/6000 and PowerPC Options): Ditto.
17831 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
17833 * config/i386/i386.h (MASK_CLASS_P): New define.
17834 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
17835 there are no registers from different register sets also when
17836 mask registers are used. Update function comment.
17837 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
17838 to (*k/*r) and (*k/*km) alternatives.
17840 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
17842 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
17843 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
17844 (EH_RETURN_HANDLER_RTX): New define.
17845 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
17846 Force frame pointer in EH return functions.
17847 (aarch64_expand_epilogue): Add barrier for eh_return.
17848 (aarch64_final_eh_return_addr): Remove.
17849 (aarch64_eh_return_handler_rtx): New function.
17850 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
17852 (aarch64_eh_return_handler_rtx): New prototype.
17854 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17856 * config/rs6000/altivec.h (vec_rlmi): New #define.
17857 (vec_vrlnm): Likewise.
17858 (vec_rlnm): Likewise.
17859 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
17860 (UNSPEC_VRLNM): Likewise.
17861 (VIlong): New mode iterator.
17862 (altivec_vrl<VI_char>mi): New define_insn.
17863 (altivec_vrl<VI_char>nm): Likewise.
17864 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
17866 (VRLDNM): Likewise.
17867 (RLNM): New polymorphic function entry.
17868 (VRLWMI): New monomorphic function entry.
17869 (VRLDMI): Likewise.
17870 (RLMI): New polymorphic function entry.
17871 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
17872 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
17873 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
17876 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17879 * dwarf2out.c (field_byte_offset): Restore the
17880 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
17881 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
17882 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
17885 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17888 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
17890 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17893 * dwarf2out.c (add_data_member_location_attribute): For constant
17894 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
17895 instead of DW_AT_data_member_location, DW_AT_bit_offset and
17896 DW_AT_byte_size attributes.
17898 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17900 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
17901 after forcing to constant memory when the code model is medium.
17903 2017-01-17 Julia Koval <julia.koval@intel.com>
17906 * config/i386/avx512fintrin.h
17907 (_mm512_i32gather_ps): Change __addr type to void const*.
17908 (_mm512_mask_i32gather_ps): Ditto.
17909 (_mm512_i32gather_pd): Ditto.
17910 (_mm512_mask_i32gather_pd): Ditto.
17911 (_mm512_i64gather_ps): Ditto.
17912 (_mm512_mask_i64gather_ps): Ditto.
17913 (_mm512_i64gather_pd): Ditto.
17914 (_mm512_mask_i64gather_pd): Ditto.
17915 (_mm512_i32gather_epi32): Ditto.
17916 (_mm512_mask_i32gather_epi32): Ditto.
17917 (_mm512_i32gather_epi64): Ditto.
17918 (_mm512_mask_i32gather_epi64): Ditto.
17919 (_mm512_i64gather_epi32): Ditto.
17920 (_mm512_mask_i64gather_epi32): Ditto.
17921 (_mm512_i64gather_epi64): Ditto.
17922 (_mm512_mask_i64gather_epi64): Ditto.
17923 (_mm512_i32scatter_ps): Change __addr type to void*.
17924 (_mm512_mask_i32scatter_ps): Ditto.
17925 (_mm512_i32scatter_pd): Ditto.
17926 (_mm512_mask_i32scatter_pd): Ditto.
17927 (_mm512_i64scatter_ps): Ditto.
17928 (_mm512_mask_i64scatter_ps): Ditto.
17929 (_mm512_i64scatter_pd): Ditto.
17930 (_mm512_mask_i64scatter_pd): Ditto.
17931 (_mm512_i32scatter_epi32): Ditto.
17932 (_mm512_mask_i32scatter_epi32): Ditto.
17933 (_mm512_i32scatter_epi64): Ditto.
17934 (_mm512_mask_i32scatter_epi64): Ditto.
17935 (_mm512_i64scatter_epi32): Ditto.
17936 (_mm512_mask_i64scatter_epi32): Ditto.
17937 (_mm512_i64scatter_epi64): Ditto.
17938 (_mm512_mask_i64scatter_epi64): Ditto.
17939 * config/i386/avx512pfintrin.h
17940 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
17941 (_mm512_mask_prefetch_i32gather_ps): Ditto.
17942 (_mm512_mask_prefetch_i64gather_pd): Ditto.
17943 (_mm512_mask_prefetch_i64gather_ps): Ditto.
17944 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
17945 (_mm512_prefetch_i32scatter_ps): Ditto.
17946 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
17947 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
17948 (_mm512_prefetch_i64scatter_pd): Ditto.
17949 (_mm512_prefetch_i64scatter_ps): Ditto.
17950 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
17951 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
17952 * config/i386/avx512vlintrin.h
17953 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
17954 (_mm_mmask_i32gather_ps): Ditto.
17955 (_mm256_mmask_i32gather_pd): Ditto.
17956 (_mm_mmask_i32gather_pd): Ditto.
17957 (_mm256_mmask_i64gather_ps): Ditto.
17958 (_mm_mmask_i64gather_ps): Ditto.
17959 (_mm256_mmask_i64gather_pd): Ditto.
17960 (_mm_mmask_i64gather_pd): Ditto.
17961 (_mm256_mmask_i32gather_epi32): Ditto.
17962 (_mm_mmask_i32gather_epi32): Ditto.
17963 (_mm256_mmask_i32gather_epi64): Ditto.
17964 (_mm_mmask_i32gather_epi64): Ditto.
17965 (_mm256_mmask_i64gather_epi32): Ditto.
17966 (_mm_mmask_i64gather_epi32): Ditto.
17967 (_mm256_mmask_i64gather_epi64): Ditto.
17968 (_mm_mmask_i64gather_epi64): Ditto.
17969 (_mm256_i32scatter_ps): Change __addr type to void*.
17970 (_mm256_mask_i32scatter_ps): Ditto.
17971 (_mm_i32scatter_ps): Ditto.
17972 (_mm_mask_i32scatter_ps): Ditto.
17973 (_mm256_i32scatter_pd): Ditto.
17974 (_mm256_mask_i32scatter_pd): Ditto.
17975 (_mm_i32scatter_pd): Ditto.
17976 (_mm_mask_i32scatter_pd): Ditto.
17977 (_mm256_i64scatter_ps): Ditto.
17978 (_mm256_mask_i64scatter_ps): Ditto.
17979 (_mm_i64scatter_ps): Ditto.
17980 (_mm_mask_i64scatter_ps): Ditto.
17981 (_mm256_i64scatter_pd): Ditto.
17982 (_mm256_mask_i64scatter_pd): Ditto.
17983 (_mm_i64scatter_pd): Ditto.
17984 (_mm_mask_i64scatter_pd): Ditto.
17985 (_mm256_i32scatter_epi32): Ditto.
17986 (_mm256_mask_i32scatter_epi32): Ditto.
17987 (_mm_i32scatter_epi32): Ditto.
17988 (_mm_mask_i32scatter_epi32): Ditto.
17989 (_mm256_i32scatter_epi64): Ditto.
17990 (_mm256_mask_i32scatter_epi64): Ditto.
17991 (_mm_i32scatter_epi64): Ditto.
17992 (_mm_mask_i32scatter_epi64): Ditto.
17993 (_mm256_i64scatter_epi32): Ditto.
17994 (_mm256_mask_i64scatter_epi32): Ditto.
17995 (_mm_i64scatter_epi32): Ditto.
17996 (_mm_mask_i64scatter_epi32): Ditto.
17997 (_mm256_i64scatter_epi64): Ditto.
17998 (_mm256_mask_i64scatter_epi64): Ditto.
17999 (_mm_i64scatter_epi64): Ditto.
18000 (_mm_mask_i64scatter_epi64): Ditto.
18001 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
18002 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
18003 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
18004 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
18005 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
18006 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
18007 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
18008 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
18009 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
18010 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
18011 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
18012 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
18013 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
18014 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
18015 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
18016 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
18017 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
18018 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
18019 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
18020 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
18021 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
18022 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
18023 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
18024 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
18025 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
18026 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
18027 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
18028 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
18029 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
18030 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
18031 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
18032 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
18033 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
18034 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
18035 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
18036 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
18037 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
18038 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
18039 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
18040 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
18041 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
18042 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
18043 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
18044 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
18045 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
18046 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
18047 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
18048 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
18049 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
18050 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
18051 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
18052 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
18053 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
18054 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
18055 definitions accordingly.
18057 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
18058 Kuan-Lin Chen <kuanlinchentw@gmail.com>
18061 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
18064 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
18067 * ira-conflicts.c (ira_build_conflicts): Update total conflict
18068 hard regs for inner regno.
18070 2017-01-17 Martin Liska <mliska@suse.cz>
18073 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
18074 assumption and add comment.
18076 2017-01-17 Nathan Sidwell <nathan@acm.org>
18078 * ipa-visibility.c (localize_node): New function, broken out of ...
18079 (function_and_variable_visibility): ... here. Call it.
18081 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
18083 PR middle-end/77445
18084 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
18085 correctly set frequency of oudgoing edge.
18086 (duplicate_thread_path): Fix profile updating.
18088 2017-01-17 Jakub Jelinek <jakub@redhat.com>
18091 * configure.ac: Add GCC_BASE_VER.
18092 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
18093 version from BASE-VER file.
18094 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
18095 (gcc.o): Depend on $(BASEVER).
18096 * common.opt (dumpfullversion): New option.
18097 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
18098 * doc/invoke.texi: Document -dumpfullversion.
18099 * doc/install.texi: Document --with-gcc-major-version-only.
18100 * configure: Regenerated.
18102 2017-01-17 Richard Biener <rguenther@suse.de>
18104 PR tree-optimization/71433
18105 * tree-vrp.c (register_new_assert_for): Merge same asserts
18106 on all incoming edges.
18107 (process_assert_insertions_for): Handle insertions at the
18110 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
18112 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
18113 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
18115 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
18118 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
18121 2017-01-17 Alan Modra <amodra@gmail.com>
18124 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
18125 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
18126 symbolic stack limit when pic.
18128 2017-01-16 Martin Sebor <msebor@redhat.com>
18130 PR tree-optimization/78608
18131 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
18133 2017-01-16 Jeff Law <law@redhat.com>
18136 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
18137 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
18138 for several include directories that may be relative to sysroot.
18139 * config/i386/x-mingw32 (gplus_includedir): Define.
18140 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
18141 (native_system_includedir): Likewise.
18142 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
18143 override if TARGET_SYSTEM_ROOT is defined.
18144 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
18146 PR tree-optimization/79090
18147 PR tree-optimization/33562
18148 PR tree-optimization/61912
18149 PR tree-optimization/77485
18150 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
18151 and computed trims into the dump file.
18153 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
18155 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
18157 2017-01-16 Jakub Jelinek <jakub@redhat.com>
18160 * gimplify.c (gimplify_init_constructor): If want_value and
18161 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
18165 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
18166 sequence. Formatting fixes.
18167 (doloop_optimize): Formatting fixes.
18170 * gcc.c (debug_level_greater_than_spec_func): New function.
18171 (static_spec_functions): Add debug-level-gt spec function.
18172 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
18174 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
18175 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
18176 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
18177 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
18178 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
18179 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
18181 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
18183 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
18184 QImode fixups to general and mask registers only.
18186 2017-01-16 Carl Love <cel@us.ibm.com>
18188 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
18189 for built-in functions
18190 vector signed char vec_nabs (vector signed char)
18191 vector signed short vec_nabs (vector signed short)
18192 vector signed int vec_nabs (vector signed int)
18193 vector signed long long vec_nabs (vector signed long long)
18194 vector float vec_nabs (vector float)
18195 vector double vec_nabs (vector double)
18196 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
18198 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
18199 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
18200 * doc/extend.texi: Update the documentation file for the new built-in
18203 2017-01-16 Martin Sebor <msebor@redhat.com>
18205 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
18208 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18210 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
18211 UNSPEC_VSX__XXSPLTD to require special splat handling.
18213 2017-01-16 David Malcolm <dmalcolm@redhat.com>
18216 * system.h: Poison strndup.
18218 2017-01-16 Alan Modra <amodra@gmail.com>
18221 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
18224 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
18226 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
18228 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
18230 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
18231 call recog here. Assert that INSN_CODE (insn) is non-negative.
18233 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
18236 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
18238 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
18239 in the currently scheduled RTL fragment.
18241 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
18243 PR rtl-optimization/78751
18244 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
18247 2017-01-14 Jeff Law <law@redhat.com>
18249 PR tree-optimization/79090
18250 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
18251 variable length stores.
18252 (compute_trims): Delete dead assignment to *trim_tail.
18253 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
18256 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
18258 PR rtl-optimization/78626
18259 PR rtl-optimization/78727
18260 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
18261 of a block, and split such blocks after everything else is finished.
18263 2017-01-14 Alan Modra <amodra@gmail.com>
18266 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
18267 target legitimate_combined_insn.
18268 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
18269 (rs6000_legitimate_combined_insn): New function.
18270 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
18272 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
18273 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
18274 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
18276 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
18278 * doc/frontends.texi (G++ and GCC): Remove references to Java.
18280 2017-01-13 Jeff Law <law@redhat.com>
18282 PR tree-optimization/33562
18283 PR tree-optimization/61912
18284 PR tree-optimization/77485
18285 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
18287 (delete_dead_assignment): Likewise.
18288 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
18289 statement to delete_dead_call and delete_dead_assignment.
18291 2017-01-13 David Malcolm <dmalcolm@redhat.com>
18294 * substring-locations.c (format_warning_va): Strengthen case 1 so
18295 that both endpoints of the substring must be within the format
18296 range for just the substring to be printed.
18298 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
18300 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
18301 * config/i386/i386.c (ix86_target_string): Add missing options
18302 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
18303 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
18304 flags_other and ix86_target_other to flags2_other. Display unknown
18306 (ix86_valid_target_attribute_inner_p): Add missing options and
18307 reorder options by implied ISAs, as in ix86_target_string.
18309 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
18311 * hash-table.h (hash_table::too_empty_p): New function.
18312 (hash_table::expand): Use it.
18313 (hash_table::traverse): Likewise.
18314 (hash_table::empty_slot): Use sizeof (value_type) instead of
18315 sizeof (PTR) to convert bytes to elements. Shrink the table
18316 if the current size is excessive for the current number of
18319 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
18321 * ira-costs.c (record_reg_classes): Break from the inner loop
18322 early once alt_fail is known to be true. Update outer loop
18323 handling accordingly.
18325 2017-01-13 Jeff Law <law@redhat.com>
18327 * tree-ssa-dse.c (decrement_count): New function.
18328 (increment_start_addr, maybe_trim_memstar_call): Likewise.
18329 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
18330 when we know the partially dead statement is a mem* function.
18332 PR tree-optimization/61912
18333 PR tree-optimization/77485
18334 * tree-ssa-dse.c: Include expr.h.
18335 (maybe_trim_constructor_store): New function.
18336 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
18338 PR tree-optimization/33562
18339 PR tree-optimization/61912
18340 PR tree-optimization/77485
18341 * doc/invoke.texi: Document new dse-max-object-size param.
18342 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
18343 * tree-ssa-dse.c: Include params.h.
18344 (dse_store_status): New enum.
18345 (initialize_ao_ref_for_dse): New, partially extracted from
18347 (valid_ao_ref_for_dse, normalize_ref): New.
18348 (setup_live_bytes_from_ref, compute_trims): Likewise.
18349 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
18350 (maybe_trim_partially_dead_store): Likewise.
18351 (maybe_trim_complex_store): Likewise.
18352 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
18353 Track what bytes live from the original store. Return tri-state
18354 for dead, partially dead or live.
18355 (dse_dom_walker): Add constructor, destructor and new private members.
18356 (delete_dead_call, delete_dead_assignment): New extracted from
18358 (dse_optimize_stmt): Make a member of dse_dom_walker.
18359 Use initialize_ao_ref_for_dse.
18361 PR tree-optimization/33562
18362 PR tree-optimization/61912
18363 PR tree-optimization/77485
18364 * sbitmap.h (bitmap_count_bits): Prototype.
18365 (bitmap_clear_range, bitmap_set_range): Likewise.
18366 * sbitmap.c (bitmap_clear_range): New function.
18367 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
18369 2017-01-13 Martin Liska <mliska@suse.cz>
18372 * function.c (set_cfun): Add new argument force.
18373 * function.h (set_cfun): Likewise.
18374 * ipa-inline-transform.c (inline_call): Use the function when
18375 strict alising from is dropped for function we inline to.
18377 2017-01-13 Richard Biener <rguenther@suse.de>
18379 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
18380 for dumping GIMPLE INTEGER_CSTs.
18382 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18384 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
18385 to 201112L since C++17.
18387 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
18390 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
18391 if -fsanitize=kernel-address is present.
18393 2017-01-13 Richard Biener <rguenther@suse.de>
18395 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
18396 as _Literal ( type ) number in case usual suffixes do not
18397 preserve all information.
18399 2017-01-13 Richard Biener <rguenther@suse.de>
18401 PR tree-optimization/77283
18402 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
18403 and ssa-iterators.h.
18404 (is_feasible_trace): Implement a cost model based on joiner
18407 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
18410 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
18411 char or short to __float128/_Float128 directly.
18413 2017-01-12 Martin Sebor <msebor@redhat.com>
18415 to -Wformat-overflow.
18416 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
18417 (min_bytes_remaining): Same.
18418 (get_string_length): Same.
18419 (format_string): Same.
18420 (format_directive): Same.
18422 (pass_sprintf_length::handle_gimple_call): Same.
18424 2017-01-12 Jakub Jelinek <jakub@redhat.com>
18426 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
18427 info.nowrite calls with no lhs that can't throw. Return bool
18428 whether gsi_remove has been called or not.
18429 (pass_sprintf_length::handle_gimple_call): Return bool whether
18430 try_substitute_return_value called gsi_remove. Formatting fix.
18431 (pass_sprintf_length::execute): Don't use gsi_remove if
18432 handle_gimple_call returned true.
18435 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
18436 be removed due to side-effects, don't remove following barrier nor
18437 turn the successor edge into fallthru edge.
18439 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18442 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
18443 element-reversing loads and stores as not swappable.
18445 2017-01-12 Nathan Sidwell <nathan@acm.org>
18446 Nicolai Stange <nicstange@gmail.com>
18448 * combine.c (try_combine): Don't ignore result of overlap checking
18449 loop. Combine overlap & asm check into single loop.
18451 2017-01-12 Richard Biener <rguenther@suse.de>
18453 * tree-pretty-print.c (dump_generic_node): Provide -gimple
18454 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
18456 2017-01-12 Richard Biener <rguenther@suse.de>
18458 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
18459 and TS_TARGET_OPTION directly derive from TS_BASE.
18460 * tree-core.h (tree_optimization_option): Derive from tree_base.
18461 (tree_target_option): Likewise.
18463 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
18465 * config/i386/i386.c (memory_address_length): Increase len
18466 only when rip_relative_addr_p returns false.
18468 2017-01-11 Julia Koval <julia.koval@intel.com>
18470 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
18471 (OPTION_MASK_ISA_SGX_SET): New.
18472 (ix86_handle_option): Handle OPT_msgx.
18473 * config.gcc: Added sgxintrin.h.
18474 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
18475 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
18476 * config/i386/i386.c (ix86_target_string): Add -msgx.
18478 (ix86_option_override_internal): Handle new options.
18479 (ix86_valid_target_attribute_inner_p): Add sgx.
18480 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
18481 * config/i386/i386.opt: Add msgx.
18482 * config/i386/sgxintrin.h: New file.
18483 * config/i386/x86intrin.h: Add sgxintrin.h.
18485 2017-01-11 Jakub Jelinek <jakub@redhat.com>
18488 * fold-const.c (maybe_nonzero_address): Return 1 for function
18490 (tree_single_nonzero_warnv_p): Don't handle function local objects
18494 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
18497 2017-01-11 David Malcolm <dmalcolm@redhat.com>
18500 * opts.c: Include "spellcheck.h"
18501 (struct string_fragment): New struct.
18502 (struct edit_distance_traits<const string_fragment &>): New
18504 (get_closest_sanitizer_option): New function.
18505 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
18507 2017-01-11 Jakub Jelinek <jakub@redhat.com>
18509 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
18511 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
18512 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
18513 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
18514 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
18515 for initial die_offset if dwarf_split_debug_info.
18516 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
18517 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
18519 (output_skeleton_debug_sections): Formatting fix. Use
18520 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
18521 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
18523 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
18525 * config/arm/cortex-a53.md: Add bypasses for
18526 cortex_a53_r2f_cvt.
18527 (cortex_a53_r2f): Only use for transfers.
18528 (cortex_a53_f2r): Likewise.
18529 (cortex_a53_r2f_cvt): Add reservation for conversions.
18530 (cortex_a53_f2r_cvt): Likewise.
18532 2017-01-11 Tamar Christina <tamar.christina@arm.com>
18534 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
18535 to all inlined functions, change static to extern.
18537 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
18540 * config/arm/arm.c (legitimize_pic_address): Handle reference to
18542 (arm_assemble_integer): Likewise.
18544 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
18546 * config.gcc: Use new awk script to check CPU, FPU and architecture
18547 parameters for --with-... options.
18548 * config/arm/parsecpu.awk: New file
18549 * config/arm/arm-cpus.in: New file.
18550 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
18552 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
18554 * config/arm/t-arm: Update dependency rules.
18555 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
18556 of processing .def files.
18557 * config/arm/genopt.sh: Deleted.
18558 * config/arm/gentune.sh: Deleted.
18559 * config/arm/arm-cores.def: Deleted.
18560 * config/arm/arm-arches.def: Deleted.
18561 * config/arm/arm-fpus.def: Deleted.
18562 * config/arm/arm-tune.md: Regenerated.
18563 * config/arm/arm-tables.opt: Regenerated.
18564 * config/arm/arm-cpu.h: New generated file.
18565 * config/arm/arm-cpu-data.h: New generated file.
18566 * config/arm/arm-cpu-cdata.h: New generated file.
18568 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
18571 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
18573 (input_varpool_node): Unpack dynamically_initialized bit.
18575 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
18577 PR rtl-optimization/79032
18578 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
18579 the alignment of the adjusted memory reference against that of MODE,
18580 instead of the alignment of the original memory reference.
18582 2017-01-11 Martin Jambor <mjambor@suse.cz>
18584 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
18586 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
18587 decorated functions.
18589 2017-01-11 Richard Biener <rguenther@suse.de>
18591 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
18592 set range/nonnull info for PHI results. Do not set it on
18593 stmts marked for removal.
18595 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
18597 * expr.c (store_field): In the bitfield case, fetch the return value
18598 from the registers before applying a single big-endian adjustment.
18599 Always do a final load for a BLKmode value not larger than a word.
18601 2017-01-10 David Malcolm <dmalcolm@redhat.com>
18604 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18605 that we correctly handle column numbers greater than
18606 LINE_MAP_MAX_COLUMN_NUMBER.
18608 2017-01-10 Martin Sebor <msebor@redhat.com>
18610 PR middle-end/78245
18611 * gimple-ssa-sprintf.c (get_destination_size): Call
18612 {init,fini}object_sizes.
18613 * tree-object-size.c (addr_object_size): Adjust.
18614 (pass_through_call): Adjust.
18615 (pass_object_sizes::execute): Adjust.
18616 * tree-object-size.h (fini_object_sizes): Declare.
18618 2017-01-10 Martin Sebor <msebor@redhat.com>
18620 PR tree-optimization/78775
18621 * builtins.c (get_size_range): Move...
18622 * calls.c: ...to here.
18623 (alloc_max_size): Accept zero argument.
18624 (operand_signed_p): Remove.
18625 (maybe_warn_alloc_args_overflow): Call get_size_range.
18626 * calls.h (get_size_range): Declare.
18628 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
18630 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
18631 from TI's devices.csv file as of September 2016.
18632 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
18634 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
18636 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
18637 * doc/invoke.texi: Likewise.
18638 * doc/md.texi: Likewise.
18639 * doc/objc.texi: Likewise.
18641 2017-01-10 Joshua Conner <joshconner@google.com>
18643 * config/arm/fuchsia-elf.h: New file.
18644 * config/fuchsia.h: New file.
18645 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
18646 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
18648 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
18650 2016-01-10 Richard Biener <rguenther@suse.de>
18652 PR tree-optimization/79034
18653 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18654 Propagate out degenerate PHIs in the joiner.
18656 2017-01-10 Martin Liska <mliska@suse.cz>
18658 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
18659 (sort_congruence_classes_by_decl_uid): Likewise.
18660 (sort_congruence_class_groups_by_decl_uid): Likewise.
18661 (sem_item_optimizer::merge_classes): Sort class, groups in these
18662 classes and members in the groups by DECL_UID of declarations.
18663 This would make merge operations stable.
18665 2017-01-10 Martin Liska <mliska@suse.cz>
18667 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
18668 usage of m_classes_vec.
18669 (sem_item_optimizer::~sem_item_optimizer): Likewise.
18670 (sem_item_optimizer::get_group_by_hash): Likewise.
18671 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
18672 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
18673 (sem_item_optimizer::verify_classes): Likewise.
18674 (sem_item_optimizer::process_cong_reduction): Likewise.
18675 (sem_item_optimizer::dump_cong_classes): Likewise.
18676 (sem_item_optimizer::merge_classes): Likewise.
18677 * ipa-icf.h (congruence_class_hash): Rename from
18678 congruence_class_group_hash. Remove declaration of m_classes_vec.
18680 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
18682 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
18683 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
18684 * config.gcc: Add avx512vpopcntdqintrin.h.
18685 * config/i386/avx512vpopcntdqintrin.h: New.
18686 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
18687 * config/i386/i386-builtin-types.def: Add new types.
18688 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
18689 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
18690 __builtin_ia32_vpopcountq_v8di_mask): New.
18691 * config/i386/i386-c.c (ix86_target_macros_internal): Define
18692 __AVX512VPOPCNTDQ__.
18693 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
18694 (PTA_AVX512VPOPCNTDQ): Define.
18695 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
18696 TARGET_AVX512VPOPCNTDQ_P): Define.
18697 * config/i386/i386.opt: Add mavx512vpopcntdq.
18698 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
18699 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
18701 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18703 PR middle-end/77484
18704 * predict.def (PRED_CALL): Set to 67.
18706 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
18708 * expr.c (store_field): In the bitfield case, if the value comes from
18709 a function call and is of an aggregate type returned in registers, do
18710 not modify the field mode; extract the value in all cases if the mode
18711 is BLKmode and the size is not larger than a word.
18713 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
18716 * config/i386/cpuid.h: Fix undefined behavior.
18718 2017-01-04 Jeff Law <law@redhat.com>
18720 PR tree-optimization/79007
18721 PR tree-optimization/67955
18722 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
18723 conservative for pt.null when flag_non_call_exceptions is on.
18725 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18727 PR translation/79019
18728 PR translation/79020
18729 * params.def (PARAM_INLINE_MIN_SPEEDUP,
18730 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
18731 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
18733 * config/avr/avr.opt (maccumulate-args): Likewise.
18734 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
18735 * common.opt (freport-bug): Likewise.
18736 * cif-code.def (CIF_FINAL_ERROR): Likewise.
18737 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
18738 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
18739 translatable string.
18740 * config/i386/i386.c (function_value_32): Likewise.
18741 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
18742 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
18744 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
18745 * common/config/msp430/msp430-common.c (msp430_handle_option):
18747 * symtab.c (symtab_node::verify_base): Likewise.
18748 * opts.c (set_debug_level): Likewise.
18749 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
18750 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
18751 missing whitespace to translatable strings.
18752 * config/avr/avr.md (bswapsi2): Fix typo in comment.
18753 * config/sh/superh.h: Likewise.
18754 * config/i386/xopintrin.h: Likewise.
18755 * config/i386/znver1.md: Likewise.
18756 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
18757 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
18758 * double-int.h (struct double_int): Likewise.
18759 * double-int.c (div_and_round_double): Likewise.
18760 * wide-int.cc: Likewise.
18761 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
18762 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
18763 * cfgcleanup.c (crossjumps_occured): Renamed to ...
18764 (crossjumps_occurred): ... this.
18765 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
18768 PR tree-optimization/78899
18769 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
18770 returning bool return struct loop *, NULL for failure and the new
18772 (versionable_outer_loop_p): Don't version outer loop if it has
18773 dont_vectorized bit set.
18774 (tree_if_conversion): When versioning outer loop, ensure
18775 tree_if_conversion is performed also on the inner loop of the
18776 non-vectorizable outer loop copy.
18777 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
18778 LOOP_VECTORIZED in inner loop of the scalar outer loop and
18779 prevent vectorization of it.
18780 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
18781 the outer loop vectorization of the non-scalar version is attempted
18782 before vectorization of the inner loop in scalar version. If
18783 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
18784 vectorization of its inner loop.
18785 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
18786 has 2 inner loops, rename also on edges from bb whose single pred
18787 is outer_loop->header. Fix typo in function comment.
18789 2017-01-09 Martin Sebor <msebor@redhat.com>
18792 * asan.c (asan_emit_stack_protection): Increase local buffer size
18793 to avoid snprintf truncation warning.
18795 2017-01-09 Andrew Pinski <apinski@cavium.com>
18797 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
18798 to reference thunderx2t99 for the tuning structure
18799 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
18801 (thunderx2t99_extra_costs): This.
18802 * config/aarch64/aarch64-tune.md: Regenerate.
18803 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
18804 (vulcan_addrcost_table): This.
18805 (vulcan_regmove_cost): Rename to ...
18806 (thunderx2t99_regmove_cost): This.
18807 (vulcan_vector_cost): Rename to ...
18808 (thunderx2t99_vector_cost): this.
18809 (vulcan_branch_cost): Rename to ...
18810 (thunderx2t99_branch_cost): This.
18811 (vulcan_tunings): Rename to ...
18812 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
18813 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
18815 2017-01-09 Martin Jambor <mjambor@suse.cz>
18819 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
18820 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
18821 (propagate_vr_accross_jump_function): Use the above function for all
18822 value range computations for pass-through jump functions and type
18823 converasion from explicit value range values.
18824 (ipcp_propagate_stage): Do not attempt to deduce types of formal
18825 parameters from TYPE_ARG_TYPES.
18826 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
18827 (ipa_write_node_info): Stream type of the actual argument.
18828 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
18830 2017-01-09 Martin Liska <mliska@suse.cz>
18833 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
18834 (lookup_compiler): Do not show error message with have_E.
18836 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18838 PR tree-optimization/78938
18839 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
18840 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
18841 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
18842 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
18845 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18847 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
18850 2017-01-09 Richard Biener <rguenther@suse.de>
18852 PR tree-optimization/78997
18853 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
18854 name condition properly.
18856 2017-01-09 Richard Biener <rguenther@suse.de>
18859 * dwarf2out.c (is_cxx): New overload with context.
18860 (is_naming_typedef_decl): Use it.
18862 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18864 * invoke.texi (Option Summary): Correct spacing in option lists
18865 and add line breaks to fix over-long lines.
18867 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18869 PR middle-end/17660
18871 * extend.texi (Common Variable Attributes): Add xref to GCC
18872 Internals manual to explain mode attribute keywords.
18874 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18877 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
18878 and Preprocessor Options.
18879 (Options for Linking): Document -pthread here....
18880 (RS/6000 and PowerPC Options): ...not here.
18881 (Solaris 2 Options): ...or here.
18882 * doc/cppopts.texi: Document -pthread.
18884 2017-01-08 Martin Sebor <msebor@redhat.com>
18886 PR middle-end/77708
18887 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
18888 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
18889 New member functions.
18890 (format_directive): Used them.
18892 (pass_sprintf_length::handle_gimple_call): Same.
18893 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
18894 to avoid truncation for any argument.
18895 (extract_affine_mul): Same.
18896 * tree.c (get_file_function_name): Same.
18898 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18900 PR middle-end/77484
18901 * predict.def (PRED_INDIR_CALL): Set to 86.
18903 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18905 PR preprocessor/54124
18906 * doc/cppopts.texi: Reformat -d subtable to list the full name
18907 of the options. Add cross-reference to the docs for the general
18908 compiler -d options.
18909 * doc/invoke.texi (Developer Options): Add cross-reference to the
18910 preprocessor-specific -d option documentation.
18912 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18914 PR preprocessor/13498
18915 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
18916 redudant material, and reflect new command-line options.
18917 (System Headers): Likewise.
18919 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18921 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
18922 -isystem, and -idirafter. Copy-edit.
18923 * doc/cppopts.texi: Copy-edit. Remove contradiction about
18924 default for -ftrack-macro-expansion. Delete obsolete and
18925 badly-formatted implementation details about -fdebug-cpp output.
18926 * doc/cppwarnopts.texi: Copy-edit.
18928 2017-01-07 David Malcolm <dmalcolm@redhat.com>
18931 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18932 that the transition from a max line width >= 1<<10 to narrower
18933 lines works correctly.
18935 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
18937 * doc/options.texi (PerFunction): New.
18938 * opt-functions.awk (switch_flags): Map both Optimization and
18939 PerFunction to CL_OPTIMIZATION.
18940 * opth-gen.awk: Test for PerFunction flag along with
18942 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
18943 it only when the latter is present. Skip those that don't in
18944 the hash function generator.
18945 * common.opt (fvar-tracking): Mark as PerFunction instead of
18947 (fvar-tracking-assignments): Likewise.
18948 (fvar-tracking-assignments-toggle): Likewise.
18949 (fvar-tracking-uninit): Likewise.
18951 2017-01-07 Jakub Jelinek <jakub@redhat.com>
18953 PR translation/79018
18954 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
18957 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
18960 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
18961 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
18962 TARGET_LONG_JUMP_TABLE_OFFSETS.
18963 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
18964 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
18965 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
18966 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
18967 * config/m68k/m68k.md (tablejump expander): Likewise.
18968 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
18969 TARGET_LONG_JUMP_TABLE_OFFSETS.
18970 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
18971 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
18973 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
18974 David Holsgrove <david.holsgrove@xilinx.com>
18976 * common/config/microblaze/microblaze-common.c
18977 (TARGET_EXCEPT_UNWIND_INFO): Remove.
18978 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
18980 * config/microblaze/microblaze.c (microblaze_must_save_register)
18981 (microblaze_expand_epilogue, microblaze_return_addr): Handle
18983 (microblaze_eh_return): New function.
18984 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
18985 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
18986 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
18987 * config/microblaze/microblaze.md (eh_return): New pattern.
18989 2017-01-06 Jakub Jelinek <jakub@redhat.com>
18991 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
18992 GCC_DIAGNOSTIC_STRINGIFY): Define.
18994 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
18996 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18998 * config/arm/arm.md (<mcrr>): New.
19000 * config/arm/arm.c (arm_arch5te): New.
19001 (arm_option_override): Set arm_arch5te.
19002 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
19004 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
19005 (arm_mcrr_qualifiers): ... this. New.
19006 (MRRC_QUALIFIERS): Define to...
19007 (arm_mrrc_qualifiers): ... this. New.
19008 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
19010 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
19011 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
19012 (MRRCI, mrrc, MRRC): New.
19013 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
19014 VUNSPEC_MRRC2): New.
19016 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19018 * config/arm/arm.md (<mcr>): New.
19020 * config/arm/arm.c (arm_coproc_builtin_available): Add
19021 support for mcr, mrc, mcr2 and mrc2.
19022 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
19023 (arm_mcr_qualifiers): ... this. New.
19024 (MRC_QUALIFIERS): Define to ...
19025 (arm_mrc_qualifiers): ... this. New.
19026 (MCR_QUALIFIERS): Define to ...
19027 (arm_mcr_qualifiers): ... this. New.
19028 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
19030 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
19031 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
19032 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
19033 VUNSPEC_MRC2): New.
19035 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19037 * config/arm/arm.md (*ldc): New.
19041 * config/arm/arm.c (arm_coproc_builtin_available): Add
19042 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
19043 (arm_coproc_ldc_stc_legitimate_address): New.
19044 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
19045 'qualifier_const_pointer'.
19046 (LDC_QUALIFIERS): Define to...
19047 (arm_ldc_qualifiers): ... this. New.
19048 (STC_QUALIFIERS): Define to...
19049 (arm_stc_qualifiers): ... this. New.
19050 * config/arm/arm-protos.h
19051 (arm_coproc_ldc_stc_legitimate_address): New.
19052 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
19053 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
19054 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
19055 stc2, stcl, stc2l): New.
19056 * config/arm/constraints.md (Uz): New.
19057 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
19058 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
19059 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
19060 VUNSPEC_STC2L): New.
19062 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19064 * config/arm/arm.md (<cdp>): New.
19065 * config/arm/arm.c (neon_const_bounds): Rename this ...
19066 (arm_const_bounds): ... this.
19067 (arm_coproc_builtin_available): New.
19068 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
19069 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
19070 (CDP_QUALIFIERS): Define to...
19071 (arm_cdp_qualifiers): ... this. New.
19073 (arm_expand_builtin_args): Add case for 6 arguments.
19074 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
19075 (arm_const_bounds): ... this.
19076 (arm_coproc_builtin_available): New.
19077 * config/arm/arm_acle.h (__arm_cdp): New.
19079 * config/arm/arm_acle_builtins.def (cdp): New.
19081 * config/arm/iterators.md (CDPI,CDP,cdp): New.
19082 * config/arm/neon.md: Rename all 'neon_const_bounds' to
19083 'arm_const_bounds'.
19084 * config/arm/types.md (coproc): New.
19085 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
19086 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
19087 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
19088 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
19090 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19092 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
19093 (UBINOP_QUALIFIERS): New.
19095 (acle_builtin_data): New. Change comment.
19096 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
19097 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
19098 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
19099 arm_acle_builtins.def.
19100 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
19101 (arm_init_acle_builtins): New.
19102 (CRC32_BUILTIN): Remove.
19103 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
19104 crc32cb, crc32ch and crc32cw.
19105 (arm_init_crc32_builtins): Remove.
19106 (arm_init_builtins): Use arm_init_acle_builtins rather
19107 than arm_init_crc32_builtins.
19108 (arm_expand_acle_builtin): New.
19109 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
19110 * config/arm/arm_acle_builtins.def: New.
19112 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19114 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
19115 (arm_builtin_datum): ... this.
19116 (arm_init_neon_builtin): Rename to ...
19117 (arm_init_builtin): ... this. Add a new parameters PREFIX
19118 and USE_SIG_IN_NAME.
19119 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
19120 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
19121 'arm_builtin_datum'.
19122 (arm_init_vfp_builtins): Likewise.
19123 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
19124 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
19125 (arm_expand_neon_args): Rename to ...
19126 (arm_expand_builtin_args): ... this. Rename builtin_arg
19127 enum values and differentiate between ARG_BUILTIN_MEMORY
19128 and ARG_BUILTIN_NEON_MEMORY.
19129 (arm_expand_neon_builtin_1): Rename to ...
19130 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
19131 values, arm_expand_builtin_args and add bool parameter NEON.
19132 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
19133 (arm_expand_vfp_builtin): Likewise.
19134 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
19136 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19138 PR middle-end/77484
19139 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
19140 * predict.c (tree_estimate_probability_bb): Reverse direction of
19141 polymorphic call predictor.
19143 2017-01-06 David Malcolm <dmalcolm@redhat.com>
19145 * passes.c (execute_one_pass): Split out pass-skipping logic into...
19146 (determine_pass_name_match): ...this new function and...
19147 (should_skip_pass_p): ...this new function.
19149 2017-01-06 Nathan Sidwell <nathan@acm.org>
19151 * ipa-visibility.c (function_and_variable_visibility): Reformat
19152 comments and long lines. Remove extrneous if.
19153 * symtab.c (symtab_node::make_decl_local): Fix code format.
19154 (symtab_node::set_section_for_node): Fix comment typo.
19156 2017-01-06 Martin Liska <mliska@suse.cz>
19159 * lra-constraints.c: Rename invariant to lra_invariant.
19160 * predict.c (set_even_probabilities): Initialize e to NULL.
19162 2017-01-05 Martin Sebor <msebor@redhat.com>
19164 PR tree-optimization/78910
19165 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
19166 (format_integer): Correct off-by-one error in the handling
19167 of precision with negative numbers in signed conversions..
19169 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
19171 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
19173 2017-01-05 Jakub Jelinek <jakub@redhat.com>
19175 PR tree-optimization/71016
19176 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
19177 factor_out_conditional_conversion. Formatting fix.
19178 (factor_out_conditional_conversion): Add cond_stmt argument.
19179 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
19180 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
19183 2017-01-05 David Malcolm <dmalcolm@redhat.com>
19185 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
19186 read-rtl-function.o, and selftest-rtl.o.
19187 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
19188 (selftest::aarch64_test_loading_full_dump): New function.
19189 (selftest::aarch64_run_selftests): New function.
19190 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
19191 selftest::aarch64_run_selftests.
19192 * config/i386/i386.c
19193 (selftest::ix86_test_loading_dump_fragment_1): New function.
19194 (selftest::ix86_test_loading_call_insn): New function.
19195 (selftest::ix86_test_loading_full_dump): New function.
19196 (selftest::ix86_test_loading_unspec): New function.
19197 (selftest::ix86_run_selftests): Call the new functions.
19198 * emit-rtl.c (maybe_set_max_label_num): New function.
19199 * emit-rtl.h (maybe_set_max_label_num): New decl.
19200 * function.c (instantiate_decls): Guard call to
19201 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
19202 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
19204 * gensupport.c (gen_reader::gen_reader): Pass "false"
19205 for new "compact" param of rtx_reader.
19206 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
19207 rather than an empty string for NULL strings.
19208 * read-md.c: Potentially include config.h rather than bconfig.h.
19209 Wrap include of errors.h with #ifdef GENERATOR_FILE.
19210 (have_error): New global, copied from errors.c.
19211 (md_reader::read_name): Rename to...
19212 (md_reader::read_name_1): ...this, adding "out_loc" param,
19213 and converting "missing name or number" to returning false, rather
19215 (md_reader::read_name): Reimplement in terms of read_name_1.
19216 (md_reader::read_name_or_nil): New function.
19217 (md_reader::read_string): Handle "(nil)" by returning NULL.
19218 (md_reader::md_reader): Add new param "compact".
19219 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
19220 (md_reader::read_file): New method.
19221 * read-md.h (md_reader::md_reader): Add new param "compact".
19222 (md_reader::read_file): New method.
19223 (md_reader::is_compact): New accessor.
19224 (md_reader::read_name): Convert return type from void to file_location.
19225 (md_reader::read_name_or_nil): New decl.
19226 (md_reader::read_name_1): New decl.
19227 (md_reader::m_compact): New field.
19228 (noop_reader::noop_reader): Pass "false" for new "compact" param
19230 (rtx_reader::rtx_reader): Add new "compact" param.
19231 (rtx_reader::read_rtx_operand): Make virtual and convert return
19232 type from void to rtx.
19233 (rtx_reader::read_until): New decl.
19234 (rtx_reader::handle_any_trailing_information): New virtual function.
19235 (rtx_reader::postprocess): New virtual function.
19236 (rtx_reader::finalize_string): New virtual function.
19237 (rtx_reader::m_in_call_function_usage): New field.
19238 (rtx_reader::m_reuse_rtx_by_id): New field.
19239 * read-rtl-function.c: New file.
19240 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
19241 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
19242 (selftest::verify_three_block_rtl_cfg): New decl.
19243 * read-rtl-function.h: New file.
19244 * read-rtl.c: Potentially include config.h rather than bconfig.h.
19245 For host, include function.h, memmodel.h, and emit-rtl.h.
19246 (one_time_initialization): New function.
19247 (struct compact_insn_name): New struct.
19248 (compact_insn_names): New array.
19249 (find_code): Handle insn codes in compact dumps.
19250 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
19251 (bind_subst_iter_and_attr): Likewise.
19252 (add_condition_to_string): Likewise.
19253 (add_condition_to_rtx): Likewise.
19254 (apply_attribute_uses): Likewise.
19255 (add_current_iterators): Likewise.
19256 (apply_iterators): Likewise.
19257 (initialize_iterators): Guard usage of apply_subst_iterator with
19258 #ifdef GENERATOR_FILE.
19259 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
19260 (md_reader::read_mapping): Likewise.
19261 (add_define_attr_for_define_subst): Likewise.
19262 (add_define_subst_attr): Likewise.
19263 (read_subst_mapping): Likewise.
19264 (check_code_iterator): Likewise.
19265 (rtx_reader::read_rtx): Likewise. Move one-time initialization
19267 (one_time_initialization): New function.
19268 (rtx_reader::read_until): New method.
19269 (read_flags): New function.
19270 (parse_reg_note_name): New function.
19271 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
19272 Handle reuse_rtx ids.
19273 Wrap iterator lookup within #ifdef GENERATOR_FILE.
19274 Add parsing support for RTL dumps, mirroring the special-cases in
19275 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
19276 values, and calling handle_any_trailing_information.
19277 (rtx_reader::read_rtx_operand): Convert return type from void
19278 to rtx, returning return_rtx. Handle case 'e'. Call
19279 finalize_string on XSTR and XTMPL fields.
19280 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
19281 "(nil)" values were omitted. Call the postprocess vfunc on the
19283 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
19284 class ctor. Initialize m_in_call_function_usage. Call
19285 one_time_initialization.
19286 * rtl-tests.c (selftest::test_uncond_jump): Call
19287 set_new_first_and_last_insn.
19288 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
19289 * selftest-rtl.c: New file.
19290 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
19291 (selftest::get_insn_by_uid): New decl.
19292 * selftest-run-tests.c (selftest::run_tests): Call
19293 read_rtl_function_c_tests.
19294 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
19295 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
19298 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
19300 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
19301 operands in a special way. Assert that pos+len <= mode precision.
19303 2017-01-05 Jakub Jelinek <jakub@redhat.com>
19305 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
19306 3 argument Alias with unlimited for the negative form.
19307 (fno-vect-cost-model): Removed.
19309 2017-01-05 Martin Liska <mliska@suse.cz>
19311 * hsa-gen.c (gen_hsa_divmod): New function.
19312 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
19314 2017-01-05 Martin Liska <mliska@suse.cz>
19317 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
19320 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19322 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
19323 small constant length operands.
19325 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19327 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
19328 between loop iterations.
19330 2017-01-05 Martin Liska <mliska@suse.cz>
19333 * gimplify.c (gimplify_decl_expr): Compare to
19334 asan_poisoned_variables instread of checking flags.
19335 (gimplify_target_expr): Likewise.
19336 (gimplify_expr): Likewise.
19337 (gimplify_function_tree): Conditionally initialize
19338 asan_poisoned_variables.
19340 2017-01-04 Jeff Law <law@redhat.com>
19342 PR tree-optimizatin/78812
19343 * rtl.h (contains_mem_rtx_p): Prototype.
19344 * ifcvt.c (containts_mem_rtx_p): Move from here to...
19345 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
19346 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
19347 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
19348 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
19350 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19352 * input.c (assert_char_at_range): Default-initialize actual_range.
19354 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19356 * df-scan.c (df_ref_create_structure): Make regno unsigned,
19357 to match the caller.
19359 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19361 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
19362 insns after final jump in test to emit dummy move.
19364 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19366 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
19367 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
19369 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19371 * multiple_target.c (create_dispatcher_calls): Init e_next.
19372 * tree-ssa-loop-split.c (split_loop): Init border.
19373 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
19376 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
19381 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
19382 (altivec_register_operand): Do not return true if the operand
19383 contains a SUBREG mixing SImode and SFmode.
19384 (vsx_register_operand): Likewise.
19385 (vsx_reg_sfsubreg_ok): New predicate.
19386 (vfloat_operand): Do not return true if the operand contains a
19387 SUBREG mixing SImode and SFmode.
19388 (vint_operand): Likewise.
19389 (vlogical_operand): Likewise.
19390 (gpc_reg_operand): Likewise.
19391 (int_reg_operand): Likewise.
19392 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
19393 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
19394 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
19396 (rs6000_emit_move_si_sf_subreg): New helper function.
19397 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
19398 fixup SUBREGs involving SImode and SFmode.
19399 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
19400 numbers for the new peephole2 optimization.
19401 (peephole2 for SFmode unions): New peephole2 to optimize cases in
19402 the GLIBC math library that do AND/IOR/XOR operations on single
19403 precision floating point.
19404 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
19405 target macros to say whether we need to avoid SUBREGs mixing
19407 (TARGET_ALLOW_SF_SUBREG): Likewise.
19408 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
19409 (UNSPEC_SI_FROM_SF): Likewise.
19410 (iorxor): Change spacing.
19411 (and_ior_xor): New iterator for AND, IOR, and XOR.
19412 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
19413 (movdi_from_sf_zero_ext): Likewise.
19414 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
19415 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
19416 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
19417 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
19418 (fms<mode>4): Likewise.
19419 (fnma<mode>4): Likewise.
19420 (fnms<mode>4): Likewise.
19421 (nfma<mode>4): Likewise.
19422 (nfms<mode>4): Likewise.
19424 2017-01-04 Marek Polacek <polacek@redhat.com>
19427 * doc/invoke.texi: Document -Wpointer-compare.
19429 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19431 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
19434 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
19435 descriptions for -gdwarf-5 and emit them as uleb128 instead of
19438 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
19441 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
19442 documentation of the powerpc_popcntb_ok attribute.
19443 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19444 code to issue warning messages if a requested CPU configuration is
19445 not supported by the binary (assembler and loader) toolchain.
19446 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
19447 made to define a built-in function that has been disabled.
19448 (paired_init_builtins): Add assertion to prevent ICE if attempt is
19449 made to define a built-in function that has been disabled.
19450 (altivec_init_builtins): Add comment explaining why definition
19451 of the DST built-in functions is not preceded by an assertion
19452 check. Add assertions to prevent ICE if attempts are made to
19453 define an altivec predicate or an abs* built-in function that has
19455 (htm_init_builtins): Add comment explaining why definition of the
19456 htm built-in functions is not preceded by an assertion check.
19458 2017-01-04 Jeff Law <law@redhat.com>
19460 PR tree-optimizatin/67955
19461 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
19462 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
19463 the points-to solution does not include pt_null. Use DECL_PT_UID
19466 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
19468 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
19469 Use gen_int_mode instead of gen_lopwart for const_int operands.
19471 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19473 PR tree-optimization/71563
19474 * match.pd: Simplify X << Y into X if Y is known to be 0 or
19475 out of range value - has low bits known to be zero.
19477 2017-01-04 Alan Modra <amodra@gmail.com>
19479 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
19480 * configure: Regenerate.
19481 * config.in: Regenerate.
19483 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19486 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
19487 a substring of the message, but strcmp with the whole message. Ifdef
19488 ENABLE_NLS, translate the message first using dgettext.
19490 2017-01-03 Jeff Law <law@redhat.com>
19492 PR tree-optimizatin/78856
19493 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
19494 (mark_threaded_blocks): Remove code to truncate thread paths that
19495 cross multiple loop headers. Instead invalidate the cached loop
19496 iteration information and handle case of a thread path walking
19497 into an irreducible region.
19499 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
19502 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
19503 assertions. Add support for doing the signbit if the IEEE 128-bit
19504 floating point value is in a GPR.
19505 * config/rs6000/rs6000.md (Fsignbit): Delete.
19506 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
19507 Update the length attribute if the value is in a GPR.
19508 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
19509 the sign or zero extension instruction, since the value is always 0/1.
19510 (signbit<mode>2_dm2): Delete using <Fsignbit>.
19513 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
19514 extracting SImode to a GPR register so that we can generate a
19515 store, limit the vector to be in a traditional Altivec register
19516 for the vextuwrx instruction.
19518 2017-01-03 Ian Lance Taylor <iant@google.com>
19520 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
19522 2017-01-03 Martin Sebor <msebor@redhat.com>
19524 PR tree-optimization/78696
19525 * gimple-ssa-sprintf.c (format_floating): Correct handling of
19526 precision. Use MPFR for %f for greater fidelity. Correct handling
19528 (pass_sprintf_length::compute_format_length): Set width and precision
19529 specified by asrerisk to void_node for vararg functions.
19530 (try_substitute_return_value): Adjust dump output.
19532 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
19534 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
19536 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
19538 * doc/invoke.texi (SPARC options): Document -mlra as the default.
19539 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
19540 -mlra/-mno-lra was passed to the compiler.
19542 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
19544 PR rtl-optimization/65618
19545 * emit-rtl.c (try_split): Move initialization of "before" and
19546 "after" to just before the call to emit_insn_after_setloc.
19548 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
19550 * doc/md.texi (Standard Names): Remove reference to Java frontend.
19552 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
19554 * dwarf2out.c (gen_enumeration_type_die): When
19555 -gno-strict-dwarf, add a DW_AT_encoding attribute.
19557 2017-01-03 Jakub Jelinek <jakub@redhat.com>
19559 PR tree-optimization/78965
19560 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
19561 Change first argument from const call_info & to call_info &. For %n
19562 set info.nowrite to false.
19564 PR middle-end/78901
19565 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
19566 possibly throwing calls.
19568 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
19569 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
19570 and fns handling, rather than in a separate case SSA_NAME.
19572 2017-01-02 Jeff Law <law@redhat.com>
19574 * config/darwin-driver.c (darwin_driver_init): Const-correctness
19575 fixes for first_period and second_period variables.
19577 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
19580 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
19581 (*insvqi_1): New insn pattern.
19582 (*insvqi_1_mem_rex64): Ditto.
19583 (*insvqi_2): Ditto.
19584 (*insvqi_3): Rename from *insvqi.
19586 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
19588 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
19590 * doc/cfg.texi (Edges): Remove reference to Java.
19591 (Maintaining the CFG): Ditto.
19593 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19595 PR middle-end/77674
19596 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
19597 transparent aliases.
19599 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19601 PR middle-end/77484
19602 * predict.def (PRED_CALL): Update hitrate.
19603 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
19604 * predict.c (tree_estimate_probability_bb): Split CALL predictor
19605 into direct/indirect/polymorphic variants.
19607 2017-01-01 Jakub Jelinek <jakub@redhat.com>
19609 Update copyright years.
19611 * gcc.c (process_command): Update copyright notice dates.
19612 * gcov-dump.c (print_version): Ditto.
19613 * gcov.c (print_version): Ditto.
19614 * gcov-tool.c (print_version): Ditto.
19615 * gengtype.c (create_file): Ditto.
19616 * doc/cpp.texi: Bump @copying's copyright year.
19617 * doc/cppinternals.texi: Ditto.
19618 * doc/gcc.texi: Ditto.
19619 * doc/gccint.texi: Ditto.
19620 * doc/gcov.texi: Ditto.
19621 * doc/install.texi: Ditto.
19622 * doc/invoke.texi: Ditto.
19624 Copyright (C) 2017 Free Software Foundation, Inc.
19626 Copying and distribution of this file, with or without modification,
19627 are permitted in any medium without royalty provided the copyright
19628 notice and this notice are preserved.