1 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
4 (build_tree_list_stat): Likewise.
5 * tree.h (build_tree_list): Remove macro.
6 (build_tree_list_vec): Likewise.
8 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10 * tree.c (make_vector_stat): Remove _stat from name.
11 (build_vector_stat): Likewise.
12 * tree.h (make_vector_stat): Remove macro.
13 (build_vector_stat): Likewise.
15 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
17 * tree.h (build_var_debug_value): Remove prototype.
19 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21 * tree.c (tree_cons_stat): Remove _stat from name.
22 * tree.h (tree_cons): Remove macro.
24 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
26 * tree.c (build_vl_exp_stat): Remove _stat from name.
27 * tree.h (build_vl_exp): Remove macro.
29 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31 * tree.c (build_decl_stat): Remove _stat from name.
32 * tree.h (build_decl): Remove macro.
34 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
36 * gimple.c (gimple_build_with_ops_stat): Adjust.
37 (gimple_alloc_stat): Remove _stat from name.
38 * gimple.h (gimple_alloc): Remove macro.
40 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
42 * tree.c (make_tree_vec_stat): Remove _stat from name.
43 (grow_tree_vec_stat): Likewise.
44 * tree.h (make_tree_vec_stat): Adjust prototype.
45 (grow_tree_vec_stat): Likewise.
46 (make_tree_vec): Remove macro.
47 (grow_tree_vec): Likewise.
49 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
51 * fold-const.c (fold_build1_stat_loc): Adjust.
52 (fold_build2_stat_loc): Likewise.
53 (fold_build3_stat_loc): Likewise.
54 * tree.c (build0_stat): Remove _stat from name.
55 (build1_stat): Likewise.
56 (build2_stat): Likewise.
57 (build3_stat): Likewise.
58 (build4_stat): Likewise.
59 (build5_stat): Likewise.
60 * tree.h (build1_loc): Remove macro, and rename _stat function
62 (build2_loc): Likewise.
63 (build3_loc): Likewise.
64 (build4_loc): Likewise.
65 (build5_loc): Likewise.
67 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
69 * tree.c (make_int_cst_stat): Remove _stat from name.
70 * tree.h (make_int_cst_stat): Adjust prototype.
71 (make_int_cst): Remove macro.
73 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
75 * tree.c (make_tre_binfo_stat): Remove _stat from name.
76 * tree.h (make_tree_binfo_stat): Adjust prototype.
77 (make_tree_binfo): Remove.
79 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
81 * tree.c (copy_node_stat): Rename to copy_node.
82 (build_distinct_type_copy): Adjust.
83 * tree.h (copy_node_stat): Adjust prototype.
84 (copy_node): Remove macro.
86 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
88 * tree.c (make_node_stat): rename to make_node.
89 (build_tree_list_stat): Adjust.
90 (build0_stat): Likewise.
91 (build2_stat): Likewise.
92 (build3_stat): Likewise.
93 (build4_stat): Likewise.
94 (build5_stat): Likewise.
95 (build_decl_stat): Likewise.
96 * tree.h (make_node_stat): Adjust prototype.
97 (make_node): remove macro.
99 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
101 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
102 (PPC_FEATURE2_SCV): Likewise.
103 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
105 2017-07-28 Tamar Christina <tamar.christina@arm.com>
107 * config/aarch64/aarch64.c
108 (aarch64_internal_mov_immediate): Add new special pattern.
109 * config/aarch64/aarch64.md (*movdi_aarch64):
110 Add reg/32bit const mov case.
112 2017-07-28 Tamar Christina <tamar.christina@arm.com>
113 Richard Sandiford <richard.sandiford@linaro.org>
115 * config/aarch64/aarch64.md (mov<mode>): Generalize.
116 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
117 Add integer and movi cases.
118 (movi-split-hf-df-sf split, fp16): New.
119 (enabled): Added TARGET_FP_F16INST.
120 * config/aarch64/iterators.md (GPF_HF): New.
121 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
123 2017-07-28 Tamar Christina <tamar.christina@arm.com>
125 * config/aarch64/aarch64.c
126 (aarch64_simd_container_mode): Add prototype.
127 (aarch64_expand_mov_immediate): Add HI support.
128 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
129 (aarch64_can_const_movi_rtx_p): New.
130 (aarch64_preferred_reload_class):
131 Remove restrictions of using FP registers for certain SIMD operations.
132 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
133 (aarch64_valid_floating_const): Add integer move validation.
134 (aarch64_simd_imm_scalar_p): Remove.
135 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
136 (aarch64_legitimate_constant_p): Expand list of supported cases.
137 * config/aarch64/aarch64-protos.h
138 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
139 (aarch64_reinterpret_float_as_int): New.
140 (aarch64_simd_imm_scalar_p): Remove.
141 * config/aarch64/constraints.md (Uvi): New.
142 (Dd): Split into Ds and new Dd.
143 * config/aarch64/aarch64.md (*movsi_aarch64):
145 (*movdi_aarch64): Add SIMD mov case.
147 2017-07-28 Bin Cheng <bin.cheng@arm.com>
149 * tree-predcom.c: (struct chain): Handle store-store chain in which
150 stores for elimination only store loop invariant values.
151 (execute_pred_commoning_chain): Ditto.
152 (prepare_initializers_chain_store_elim): Ditto.
153 (prepare_finalizers): Ditto.
154 (is_inv_store_elimination_chain): New function.
155 (initialize_root_vars_store_elim_1): New function.
157 2017-07-28 Bin Cheng <bin.cheng@arm.com>
159 * tree-predcom.c: Revise general description of the pass.
160 (enum chain_type): New enum type for store elimination.
161 (struct chain): New field supporting store elimination.
162 (struct component): Ditto.
163 (dump_chain): Dump store-stores chain.
164 (release_chain): Release resources.
165 (split_data_refs_to_components): Compute and create component
166 contains only stores for elimination.
167 (get_chain_last_ref_at): New function.
168 (make_invariant_chain): Initialization.
169 (make_rooted_chain): Specify chain type in parameter and record it.
170 (add_looparound_copies): Skip for store-stores chain.
171 (determine_roots_comp): Compute type of chain and pass it to
173 (initialize_root_vars_store_elim_2): New function.
174 (finalize_eliminated_stores): New function.
175 (remove_stmt): Handle store for elimination.
176 (execute_pred_commoning_chain): Execute predictive commoning on
178 (determine_unroll_factor): Skip unroll for store-stores chain.
179 (prepare_initializers_chain_store_elim): New function.
180 (prepare_initializers_chain): Hanlde store-store chain.
181 (prepare_finalizers_chain, prepare_finalizers): New function.
182 (tree_predictive_commoning_loop): Return integer value indicating
183 if loop is unrolled or lcssa form is corrupted.
184 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
186 2017-07-28 Bin Cheng <bin.cheng@arm.com>
188 * tree-predcom.c (initialize_root): Delete.
189 (execute_pred_commoning_chain): Initialize root vars and replace
190 reference of non-combined chain directly, rather than call above
193 2017-07-28 Bin Cheng <bin.cheng@arm.com>
195 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
196 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
198 2017-07-28 Bin Cheng <bin.cheng@arm.com>
200 * tree-predcom.c (struct chain): New field init_seq.
201 (release_chain): Release init_seq.
202 (prepare_initializers_chain): Record intialization stmts in above
204 (insert_init_seqs): New function.
205 (tree_predictive_commoning_loop): Call insert_init_seqs.
207 2017-07-28 Bin Cheng <bin.cheng@arm.com>
209 * tree-predcom.c (determine_roots_comp): Skip trivial components.
211 2017-07-28 Richard Biener <rguenther@suse.de>
213 * match.pd: Remove superfluous :c.
214 * genmatch.c (simplify::id): Add member.
215 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
217 (current_id): New global.
218 (dt_node::parent): Move from ...
219 (dt_operand::parent): ... here. Add for_id member.
220 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
221 (decision_tree::find_node): Relax order requirement when
222 merging DT_TRUE nodes to ones inbetween the current simplify
223 and the one we try to merge with. Add diagnostic whenever
224 we need to enforce pattern order by not merging.
225 (decision_tree::insert): Set current_id.
226 (decision_tree::print_node): Dump parent node and for_id.
227 (parser::last_id): Add member.
228 (parser::push_simplify): Assign unique id.
229 (parser::parser): Initialize last_id.
231 2017-07-28 Martin Liska <mliska@suse.cz>
234 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
235 gimple_build_debug_bind.
237 2017-07-28 Richard Biener <rguenther@suse.de>
239 PR tree-optimization/81502
240 * match.pd: Add pattern combining BIT_INSERT_EXPR with
242 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
244 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
245 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
246 for BIT_FIELD_REF args.
247 * fold-const.c (make_bit_field_ref): Likewise.
248 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
250 2017-07-28 Jakub Jelinek <jakub@redhat.com>
253 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
254 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
255 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
256 Or it into SANITIZER_UNDEFINED.
257 * ubsan.c: Include gimple-fold.h and varasm.h.
258 (ubsan_expand_ptr_ifn): New function.
259 (instrument_pointer_overflow): New function.
260 (maybe_instrument_pointer_overflow): New function.
261 (instrument_object_size): Formatting fix.
262 (pass_ubsan::execute): Call instrument_pointer_overflow
263 and maybe_instrument_pointer_overflow.
264 * internal-fn.c (expand_UBSAN_PTR): New function.
265 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
266 * sanitizer.def (__ubsan_handle_pointer_overflow,
267 __ubsan_handle_pointer_overflow_abort): New builtins.
268 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
269 * internal-fn.def (UBSAN_PTR): New internal function.
270 * opts.c (sanitizer_opts): Add pointer-overflow.
271 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
272 * fold-const.c (build_range_check): Compute pointer range check in
273 integral type if pointer arithmetics would be needed. Formatting
276 2017-07-28 Martin Liska <mliska@suse.cz>
279 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
280 parameters that are of a variable-length.
282 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
284 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
286 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
287 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
288 (CRT_CALL_STATIC_FUNCTION): Likewise.
289 (ASM_DEFAULT_SPEC): New define.
290 (ASM_SPEC32): Likewise.
291 (ASM_SPEC64): Likewise.
292 (ASM_SPEC_COMMON): Likewise.
293 (ASM_SPEC): Likewise.
294 (INVALID_64BIT): Likewise.
295 (LINK_OS_DEFAULT_SPEC): Likewise.
296 (LINK_OS_SPEC32): Likewise.
297 (LINK_OS_SPEC64): Likewise.
298 (POWERPC_LINUX): Likewise.
299 (PTRDIFF_TYPE): Likewise.
300 (RESTORE_FP_PREFIX): Likewise.
301 (RESTORE_FP_SUFFIX): Likewise.
302 (SAVE_FP_PREFIX): Likewise.
303 (SAVE_FP_SUFFIX): Likewise.
304 (SIZE_TYPE): Likewise.
305 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
306 (TARGET_64BIT): Likewise.
307 (TARGET_64BIT): Likewise.
308 (TARGET_AIX): Likewise.
309 (WCHAR_TYPE_SIZE): Likewise.
310 (WCHAR_TYPE): Undefine.
311 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
312 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
313 (CPP_OS_RTEMS_SPEC): Delete.
314 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
315 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
317 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
319 2017-07-28 Jakub Jelinek <jakub@redhat.com>
321 PR tree-optimization/81578
322 * tree-parloops.c (build_new_reduction): Bail out if
323 reduction_code isn't one of the standard OpenMP reductions.
324 Move the details printing after that decision.
326 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
328 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
329 related to reload_in_progress.
330 (splat_input_operand): Likewise.
331 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
333 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
335 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
336 (TARGET_INSTANTIATE_DECLS): Likewise.
337 (legitimate_indexed_address_p): Delete reload_in_progress code.
338 (rs6000_debug_legitimate_address_p): Likewise.
339 (rs6000_eliminate_indexed_memrefs): Likewise.
340 (rs6000_emit_le_vsx_store): Likewise.
341 (rs6000_emit_move_si_sf_subreg): Likewise.
342 (rs6000_emit_move): Likewise.
343 (register_to_reg_type): Likewise.
344 (rs6000_pre_atomic_barrier): Likewise.
345 (rs6000_machopic_legitimize_pic_address): Likewise.
346 (rs6000_allocate_stack_temp): Likewise.
347 (rs6000_address_for_fpconvert): Likewise.
348 (rs6000_address_for_altivec): Likewise.
349 (rs6000_secondary_memory_needed_rtx): Delete function.
350 (rs6000_check_sdmode): Likewise.
351 (rs6000_alloc_sdmode_stack_slot): Likewise.
352 (rs6000_instantiate_decls): Likewise.
353 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
354 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
355 Delete reload_in_progress.
356 (*vec_reload_and_plus_<mptrsize>): Likewise.
357 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
358 (vsx_div_v2di): Likewise.
359 (vsx_udiv_v2di): Likewise.
361 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
363 * config/rs6000/rs6000.opt (mlra): Replace with stub.
364 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
365 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
366 (rs6000_debug_reg_global): Delete print of LRA status.
367 (rs6000_option_override_internal): Delete dead LRA related code.
368 (rs6000_lra_p): Delete function.
369 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
371 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
373 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
374 * config/riscv/rtems.h: New file.
376 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
377 Sudakshina Das <sudi.das@arm.com>
379 * config/aarch64/aarch64.md
380 (define_split for and<mode>3nr_compare): Move
381 non aarch64_logical_operand to a register.
382 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
383 register immediate operand to a register.
384 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
386 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
389 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
391 2017-07-27 Richard Biener <rguenther@suse.de>
393 PR tree-optimization/81573
394 PR tree-optimization/81494
395 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
396 multi defuse cycle case.
398 2017-07-27 Richard Biener <rguenther@suse.de>
400 PR tree-optimization/81571
401 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
404 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
406 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
407 earlier and only if MASK_FPU is set. Adjust formatting.
409 2017-07-27 Martin Liska <mliska@suse.cz>
411 * opt-functions.awk: Add validation of value of Init.
412 * optc-gen.awk: Pass new argument.
414 2017-07-27 Martin Liska <mliska@suse.cz>
416 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
419 2017-07-27 Martin Liska <mliska@suse.cz>
421 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
422 BBs and edges seen by autoFDO.
424 2017-07-27 Richard Biener <rguenther@suse.de>
426 PR tree-optimization/81502
427 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
428 with incompatible but same sized type.
429 (execute_update_addresses_taken): Likewise.
431 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
433 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
434 flag_tree_loop_vectorize rather than flag_tree_vectorize.
436 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
439 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
440 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
441 Change s_operand to memory_operand.
443 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
445 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
446 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
447 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
448 Emit instructions rather than returning an expression. Handle TFmode
449 and KFmode by casting to TImode.
450 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
451 (rs6000_emit_le_vsx_store): Likewise.
452 * config/rs6000/vsx.md (VSX_TI): New iterator.
453 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
454 (*vsx_le_undo_permute_<mode>): Likewise.
455 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
456 emit the split sequence.
457 (*vsx_le_perm_store_<mode>): Likewise.
459 2017-07-27 Jakub Jelinek <jakub@redhat.com>
461 PR tree-optimization/81555
462 PR tree-optimization/81556
463 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
464 if true, force CHANGED for the recursive invocation.
465 (reassociate_bb): Remember original length of ops array, pass
466 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
468 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
469 attributes for noipa attribute. For naked attribute use
470 lookup_attribute first before lookup_attribute_spec.
471 * final.c (rest_of_handle_final): Disable IPA RA for functions with
473 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
474 for functions with noipa attribute.
475 (cgraph_externally_visible_p): Return true for functions with noipa
477 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
478 for functions with noipa attribute.
479 * doc/extend.texi: Document noipa function attribute.
480 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
481 also for functions with noipa attribute.
482 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
484 2017-07-26 Andrew Pinski <apinski@cavium.com>
486 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
487 vec_unalign_load_cost and vec_unalign_store_cost.
489 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
491 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
492 -mvsx-small-integer option.
493 (ISA_3_0_MASKS_IEEE): Likewise.
494 (OTHER_VSX_VECTOR_MASKS): Likewise.
495 (POWERPC_MASKS): Likewise.
496 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
497 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
498 code, only testing for DImode being allowed in non-VSX floating
500 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
501 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
503 (rs6000_option_override_internal): Delete -mvsx-small-integer.
504 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
505 TARGET_P8_VECTOR test.
506 (rs6000_secondary_reload_simple_move): Likewise.
507 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
508 since TARGET_P9_VECTOR was already tested.
509 (rs6000_opt_masks): Remove -mvsx-small-integer.
510 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
511 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
513 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
514 test for TARGET_VEXTRACTUB was used, and that uses
516 (p9 extract splitter): Likewise.
517 (vsx_extract_<mode>_di_p9): Likewise.
518 (vsx_extract_<mode>_store_p9): Likewise.
519 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
520 for TARGET_P9_VECTOR was used. Delete code that is now dead with
521 the elimination of TARGET_VSX_SMALL_INTEGER.
522 (vsx_extract_<mode>_p8): Likewise.
523 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
524 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
525 (vsx_set_<mode>_p9): Likewise.
526 (vsx_set_v4sf_p9): Likewise.
527 (vsx_set_v4sf_p9_zero): Likewise.
528 (vsx_insert_extract_v4sf_p9): Likewise.
529 (vsx_insert_extract_v4sf_p9_2): Likewise.
530 * config/rs6000/rs6000.md (sign extend splitter): Change
531 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
532 (floatsi<mode>2_lfiwax_mem): Likewise.
533 (floatunssi<mode>2_lfiwzx_mem): Likewise.
534 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
535 since a test for TARGET_P9_VECTOR was used.
536 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
537 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
538 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
539 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
540 TARGET_P8_VECTOR test.
541 (fix_trunc<mode>si2_stfiwx): Likewise.
542 (fix_trunc<mode>si2_internal): Likewise.
543 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
544 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
546 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
547 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
548 TARGET_P8_VECTOR test.
549 (fixuns_trunc<mode>si2_stfiwx): Likewise.
550 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
551 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
553 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
554 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
555 since a test for TARGET_P9_VECTOR was used.
556 (splitter for loading small constants): Likewise.
558 2017-07-26 Andrew Pinski <apinski@cavium.com>
560 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
563 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
566 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
567 (fp_valid_at): Likewise.
569 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
571 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
572 (qdf24xx_addrcost_table): Likewise.
573 (cortexa57_tunings): Update to use generic_branch_cost.
574 (cortexa72_tunings): Likewise.
575 (cortexa73_tunings): Likewise.
576 (qdf24xx_tunings): Likewise.
578 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
580 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
581 (thunderx2t99_branch_cost): Likewise.
582 (cortexa35_tunings): Update to use generic_branch_cost.
583 (cortexa53_tunings): Likewise.
584 (cortexa57_tunings): Likewise.
585 (cortexa72_tunings): Likewise.
586 (cortexa73_tunings): Likewise.
587 (thunderx2t99_tunings): Likewise.
589 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
591 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
592 (sparc_option_override): Honour MASK_FSMULD.
593 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
594 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
595 * config/sparc/sparc.opt (mfsmuld): New option.
596 * doc/invoke.texi (mfsmuld): Document option.
598 2017-07-26 Marek Polacek <polacek@redhat.com>
601 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
603 2017-07-26 Richard Biener <rguenther@suse.de>
605 * gimple-match-head.c (do_valueize): Return OP if valueize
607 (get_def): New helper to get at the def stmt of a SSA name
609 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
610 do_valueize to get at the def stmt.
611 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
613 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
616 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
618 2017-07-26 Martin Liska <mliska@suse.cz>
621 * function.c (expand_function_start): Make expansion of
622 nonlocal_goto_save_area after parm_birth_insn.
624 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
626 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
627 from all CPU target flags enable members.
629 2017-07-26 Richard Biener <rguenther@suse.de>
631 * genmatch.c (dt_simplify::gen): Make iterator vars const.
632 (decision_tree::gen): Make 'type' const.
633 (write_predicate): Likewise.
635 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
637 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
638 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
639 (rs6000_option_override_internal): Likewise.
640 (rs6000_expand_vector_set): Likewise.
641 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
642 (TARGET_UPPER_REGS_SF): Likewise.
643 (TARGET_UPPER_REGS_DI): Likewise.
644 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
645 (TARGET_DIRECT_MOVE_64BIT): Likewise.
646 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
647 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
648 (Splitters for DI constants in Altivec registers): Likewise.
649 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
650 (vsx_set_v4sf_p9): Likewise.
651 (vsx_set_v4sf_p9_zero): Likewise.
652 (vsx_insert_extract_v4sf_p9): Likewise.
653 (vsx_insert_extract_v4sf_p9_2): Likewise.
655 2017-07-25 Carl Love <cel@us.ibm.com>
657 * doc/extend.texi: Update the built-in documentation file for the
658 existing built-in functions
659 vector signed char vec_cnttz (vector signed char);
660 vector unsigned char vec_cnttz (vector unsigned char);
661 vector signed short vec_cnttz (vector signed short);
662 vector unsigned short vec_cnttz (vector unsigned short);
663 vector signed int vec_cnttz (vector signed int);
664 vector unsigned int vec_cnttz (vector unsigned int);
665 vector signed long long vec_cnttz (vector signed long long);
666 vector unsigned long long vec_cnttz (vector unsigned long long);
668 2017-07-25 Andrew Pinski <apinski@cavium.com>
670 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
671 accesses where the use is for the first operand of a BIT_INSERT.
673 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
676 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
677 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
679 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
681 * config/i386/gstabs.h: Delete.
682 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
684 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
686 * config/i386/i386.c (ix86_decompose_address): Do not check for
687 register RTX when looking at index_reg or base_reg.
688 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
690 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
692 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
693 to update EH info here.
695 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
697 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
699 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
701 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
703 2017-07-25 Torsten Duwe <duwe@suse.de>
705 * common.opt: Introduce -fpatchable-function-entry
706 command line option, and its variables function_entry_patch_area_size
707 and function_entry_patch_area_start.
708 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
709 including a two-value parser.
710 * target.def (print_patchable_function_entry): New target hook.
711 * targhooks.h (default_print_patchable_function_entry): New function.
712 * targhooks.c (default_print_patchable_function_entry): Likewise.
713 * toplev.c (process_options): Switch off IPA-RA if
714 patchable function entries are being generated.
715 * varasm.c (assemble_start_function): Look at the
716 patchable-function-entry command line switch and current
717 function attributes and maybe generate NOP instructions by
718 calling the print_patchable_function_entry hook.
719 * doc/extend.texi: Document patchable_function_entry attribute.
720 * doc/invoke.texi: Document -fpatchable_function_entry
722 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
724 * doc/tm.texi: Re-generate.
726 2017-07-25 Jakub Jelinek <jakub@redhat.com>
729 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
730 TARGET_AVX512DQ rather than TARGET_AVX512BW.
732 2017-07-25 Tamar Christina <tamar.christina@arm.com>
734 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
736 2017-07-25 Richard Biener <rguenther@suse.de>
738 PR tree-optimization/81455
739 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
740 not walk in cycles when looking for guards.
742 2017-07-25 Richard Biener <rguenther@suse.de>
744 PR tree-optimization/81529
745 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
746 when optimizing backedge uses.
748 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
750 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
752 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
753 to dl_section_ref. On AIX, append an expression to subtract
754 the size of the section length to dl_section_ref.
756 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
758 * configure.ac: If any of the config.* scripts fail, exit 1.
759 * configure: Regenerate.
761 2017-07-25 Richard Biener <rguenther@suse.de>
764 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
765 of immediate uses, be more verbose on errors.
767 2017-07-25 Richard Biener <rguenther@suse.de>
769 PR tree-optimization/81510
770 * tree-vect-loop.c (vect_is_simple_reduction): When the
771 reduction stmt is not inside the loop bail out.
773 2017-07-25 Richard Biener <rguenther@suse.de>
775 PR tree-optimization/81303
776 * tree-vect-loop-manip.c (vect_loop_versioning): Build
777 profitability check against LOOP_VINFO_NITERSM1.
779 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
781 * domwalk.c (cmp_bb_postorder): Simplify.
782 (sort_bbs_postorder): New function. Use it...
783 (dom_walker::walk): ...here to optimize common cases.
785 2017-07-25 Martin Liska <mliska@suse.cz>
788 * ipa-visibility.c (function_and_variable_visibility): Make the
789 redirection just on target that supports aliasing.
790 Fix GNU coding style.
792 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
795 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
797 * config/arm/rtems.h: Likewise.
798 * config/bfin/rtems.h: Likewise.
799 * config/i386/rtemself.h: Likewise.
800 * config/lm32/rtems.h: Likewise.
801 * config/m32c/rtems.h: Likewise.
802 * config/m68k/rtemself.h: Likewise.
803 * config/microblaze/rtems.h: Likewise.
804 * config/mips/rtems.h: Likewise.
805 * config/moxie/rtems.h: Likewise.
806 * config/nios2/rtems.h: Likewise.
807 * config/powerpcspe/rtems.h: Likewise.
808 * config/rs6000/rtems.h: Likewise.
809 * config/rtems.h: Likewise.
810 * config/sh/rtems.h: Likewise.
811 * config/sh/rtemself.h: Likewise.
812 * config/sparc/rtemself.h: Likewise.
814 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
817 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
818 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
819 * tree-ssa-structalias.c (alias_get_name): Same.
821 2017-07-25 Bin Cheng <bin.cheng@arm.com>
824 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
825 instructions if no du chain is found.
827 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
829 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
831 2017-07-25 Richard Biener <rguenther@suse.de>
834 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
837 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
839 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
841 (ISA_2_7_MASKS_SERVER): Likewise.
842 (ISA_3_0_MASKS_IEEE): Likewise.
843 (OTHER_P8_VECTOR_MASKS): Likewise.
844 (OTHER_VSX_VECTOR_MASKS): Likewise.
845 (POWERPC_MASKS): Likewise.
846 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
847 duplicate list of options.
848 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
849 explicit -mupper-regs options.
850 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
851 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
852 alias for -mupper-regs-df.
853 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
854 (rs6000_init_hard_regno_mode_ok): Likewise.
855 (rs6000_option_override_internal): Likewise.
856 (rs6000_opt_masks): Likewise.
857 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
858 options in terms of whether -mvsx or -mpower8-vector was used.
859 (TARGET_UPPER_REGS_DI): Likewise.
860 (TARGET_UPPER_REGS_SF): Likewise.
861 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
862 -mupper-regs-* options.
864 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
866 * passes.c (emergency_dump_function): Print some empty lines and a
867 header before the RTL dump.
869 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
871 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
873 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
876 * config/aarch64/aarch64.c (aarch64_classify_symbol):
877 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
879 2017-07-24 Carl Love <cel@us.ibm.com>
881 * config/rs6000/rs6000-c.c: Add support for built-in functions
882 vector float vec_extract_fp32_from_shorth (vector unsigned short);
883 vector float vec_extract_fp32_from_shortl (vector unsigned short);
884 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
885 vec_extract_fp_from_shortl): Add defines for the two builtins.
886 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
887 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
889 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
890 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
891 * doc/extend.texi: Update the built-in documentation file for the
892 new built-in function.
894 2017-07-24 Jakub Jelinek <jakub@redhat.com>
897 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
899 * doc/generic.texi: Likewise.
900 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
901 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
903 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
905 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
906 (aarch64_mls_elt_merge<mode>): Likewise.
908 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
910 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
913 2017-07-23 Michael Collison <michael.collison@arm.com>
915 * config/arm/arm.c (arm_option_override): Deprecate
916 use of -mstructure-size-boundary.
917 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
918 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
920 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
923 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
924 Reduce cost estimate for direct moves.
926 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
929 * config/i386/i386.c (ix86_option_override_internal): Disable
930 BMI, BMI2 and TBM instructions for -m16.
932 2017-07-21 Carl Love <cel@us.ibm.com>
934 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
935 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
936 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
937 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
938 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
939 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
940 VMULOSW): New enum "unspec" values.
941 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
942 altivec_vmulosw): New patterns.
943 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
944 VMULOSW): Add definitions.
946 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
948 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
950 * config/aarch64/aarch64-options-extensions.def (rdma); New.
951 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
952 (AARCH64_FL_V8_1): Renumber.
953 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
954 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
955 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
956 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
957 rdma to feature modifiers list.
959 2017-07-21 Yury Gribov <tetra2005@gmail.com>
962 * ipa-visibility (function_and_variable_visibility): Convert
963 recursive PLT call to direct call if appropriate.
965 2017-07-21 Andrew Pinski <apinski@cavium.com>
967 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
968 operand 1 to see if the types precision matches.
969 * fold-const.c (operand_equal_p): Likewise.
971 2017-07-21 Richard Biener <rguenther@suse.de>
973 PR tree-optimization/81303
974 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
975 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
976 (vect_peeling_hash_get_lowest_cost): Adjust.
977 (vect_enhance_data_refs_alignment): Likewise. Use
978 vect_get_peeling_costs_all_drs to compute the penalty for no
979 peeling to match up costs.
981 2017-07-21 Richard Biener <rguenther@suse.de>
983 PR tree-optimization/81500
984 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
985 we didn't identify a reduction path.
987 2017-07-21 Tom de Vries <tom@codesourcery.com>
988 Cesar Philippidis <cesar@codesourcery.com>
990 PR gcov-profile/81442
991 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
994 2017-07-21 Tom de Vries <tom@codesourcery.com>
997 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
999 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
1000 nvptx_override_options_after_change.
1002 2017-07-21 Ulrich Drepper <drepper@redhat.com>
1004 * dwarf2out.c (output_file_names): Avoid double testing for
1007 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
1009 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
1011 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
1013 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
1015 (try_crossjump_to_edge): Do not punt on partitioned functions.
1017 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
1019 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
1020 Put all BBs reachable only via paths crossing cold region to cold
1022 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
1024 2016-07-21 Richard Biener <rguenther@suse.de>
1026 PR tree-optimization/81303
1027 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
1028 into account prologue and epilogue iterations when raising
1029 min_profitable_iters to sth at least covering one vector iteration.
1031 2017-07-21 Tamar Christina <tamar.christina@arm.com>
1033 * config/arm/arm.c (arm_test_cpu_arch_dat):
1036 2017-07-20 Nathan Sidwell <nathan@acm.org>
1038 Remove TYPE_METHODS.
1039 * tree.h (TYPE_METHODS): Delete.
1040 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
1041 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
1042 (dbxout_type_methods): Scan TYPE_FIELDS.
1043 (dbxout_type): Don't check TYPE_METHODS here.
1044 * function.c (use_register_for_decl): Always ignore register for
1045 class types when not optimizing.
1046 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
1047 * tree.c (free_lang_data_in_type): Stitch out member functions and
1048 templates from TYPE_FIELDS.
1049 (build_distinct_type_copy, verify_type_variant,
1050 verify_type): Member fns are on TYPE_FIELDS.
1051 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
1052 * tree-pretty-print.c (dump_generic_node): Likewise.
1054 2017-07-20 Jakub Jelinek <jakub@redhat.com>
1057 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
1058 V2TImode and V4TImode.
1059 (ix86_expand_vector_extract): Likewise.
1060 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
1061 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
1062 (ssescalarmode): Handle V4TImode and V2TImode.
1063 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
1064 (*vec_extractv2ti, *vec_extractv4ti): New insns.
1065 (VEXTRACTI128_MODE): New mode iterator.
1066 (splitter for *vec_extractv?ti first element): New.
1067 (VEC_INIT_MODE): New mode iterator.
1068 (vec_init<mode>): Consolidate 3 expanders into one using
1069 VEC_INIT_MODE mode iterator.
1071 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
1073 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
1074 non_spilled_static_chain_regno_p.
1076 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
1078 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
1080 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
1082 * bb-reorder.c (connect_traces): Allow copying of blocks within
1085 2017-07-20 Richard Biener <rguenther@suse.de>
1087 * gimple.h (gimple_phi_result): Add gphi * overload.
1088 (gimple_phi_result_ptr): Likewise.
1089 (gimple_phi_arg): Likewise. Adjust index assert to only
1090 allow actual argument accesses rather than all slots available
1092 (gimple_phi_arg_def): Add gphi * overload.
1093 * tree-phinodes.c (make_phi_node): Initialize only actual
1095 (resize_phi_node): Clear memory not covered by old node,
1096 do not initialize excess argument slots.
1097 (reserve_phi_args_for_new_edge): Initialize new argument slot
1100 2017-07-20 Bin Cheng <bin.cheng@arm.com>
1102 PR tree-optimization/81388
1104 2016-07-21 Bin Cheng <bin.cheng@arm.com>
1106 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
1107 by removing computation of may_be_zero.
1109 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1110 Tom de Vries <tom@codesourcery.com>
1113 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
1114 when gimple level profile disagrees with what RTL expander did.
1116 2017-07-20 Richard Biener <rguenther@suse.de>
1118 PR tree-optimization/61171
1119 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
1120 (vect_analyze_stmt): Add slp instance parameter.
1121 (vectorizable_reduction): Likewise.
1122 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
1123 (vect_is_simple_reduction): Deal with chains not detected
1124 as SLP reduction chain, specifically not properly associated
1125 chains containing a mix of plus/minus.
1126 (get_reduction_op): Remove.
1127 (get_initial_defs_for_reduction): Simplify, pass in whether
1128 this is a reduction chain, pass in the SLP node for the PHIs.
1129 (vect_create_epilog_for_reduction): Get the SLP instance as
1131 (vectorizable_reduction): Get the SLP instance as arg.
1132 During analysis remember the SLP node with the PHIs in the
1133 instance. Simplify getting at the vectorized reduction PHIs.
1134 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
1135 through SLP instance.
1136 (vect_slp_analyze_operations): Likewise.
1137 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
1138 (vect_transform_stmt): Likewise.
1140 2017-07-20 Tom de Vries <tom@codesourcery.com>
1142 PR tree-optimization/81489
1143 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
1144 read of phi arg location to before loop that modifies phi.
1146 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
1148 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
1151 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1154 * except.c (execute): Fix ordering issue.
1156 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
1158 PR rtl-optimization/81423
1159 * combine.c (make_compound_operation_int): Don't try to optimize
1160 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
1162 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
1164 PR rtl-optimization/81423
1165 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
1166 with a constant that is -1 in the truncated to mode.
1168 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1170 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
1171 (determine_unlikely_bbs): ... here.
1172 * predict.h (propagate_unlikely_bbs_forward): Declare.
1173 * cfgexpand.c (pass_expand::execute): Use it.
1174 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
1176 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
1177 propagate_unlikely_bbs_forward.
1179 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1182 * except.c (maybe_add_nop_after_section_switch): New function.
1185 2017-07-19 Tom de Vries <tom@codesourcery.com>
1187 * gimple.h (gimple_phi_set_arg): Make assert more strict.
1189 2017-07-19 Tom de Vries <tom@codesourcery.com>
1191 * gimple.h (gimple_phi_arg): Make assert more strict.
1193 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
1195 * config.gcc (powerpc*-*-*): Add mmintrin.h.
1196 * config/rs6000/mmintrin.h: New file.
1197 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
1199 2017-07-19 Jakub Jelinek <jakub@redhat.com>
1201 PR tree-optimization/81346
1202 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
1204 2017-07-19 Tom de Vries <tom@codesourcery.com>
1206 * config/nvptx/nvptx.md (VECIM): Add V2DI.
1208 2017-07-19 Tom de Vries <tom@codesourcery.com>
1210 * config/nvptx/nvptx-modes.def: Add V2DImode.
1211 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
1212 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
1213 (nvptx_output_mov_insn): Handle lack of mov.b128.
1214 (nvptx_print_operand): Handle 'H' and 'L' codes.
1215 (nvptx_vector_mode_supported): Allow V2DImode.
1216 (nvptx_preferred_simd_mode): New function.
1217 (nvptx_data_alignment): New function.
1218 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
1219 nvptx_preferred_simd_mode.
1220 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
1222 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
1224 2017-07-19 Tom de Vries <tom@codesourcery.com>
1226 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
1227 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
1228 (nvptx_vector_mode_supported): New function. Allow V2SImode.
1229 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
1230 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
1231 (mov<VECIM>_insn): New define_insn.
1232 (define_expand "mov<VECIM>): New define_expand.
1234 2017-07-19 Tom de Vries <tom@codesourcery.com>
1236 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
1238 2017-07-19 Jakub Jelinek <jakub@redhat.com>
1240 PR tree-optimization/81346
1241 * fold-const.h (fold_div_compare, range_check_type): Declare.
1242 * fold-const.c (range_check_type): New function.
1243 (build_range_check): Use range_check_type.
1244 (fold_div_compare): No longer static, rewritten into
1245 a match.pd helper function.
1246 (fold_comparison): Don't call fold_div_compare here.
1247 * match.pd (X / C1 op C2): New optimization using fold_div_compare
1250 2017-07-19 Nathan Sidwell <nathan@acm.org>
1252 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
1253 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
1254 * tree.c (find_decls_types_r, verify_type): Use
1255 TYPE_{MIN,MAX}_VALUE_RAW.
1256 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
1257 (hash_tree): Likewise.
1258 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
1260 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
1263 2017-07-18 Tom de Vries <tom@codesourcery.com>
1266 * omp-expand.c (expand_omp_for_static_chunk): Handle
1267 equal-argument loop exit phi.
1269 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
1272 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
1273 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
1274 operand 2 predicate.
1275 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
1276 operand 2 predicate.
1277 (ror,rol -> rorx splitters): Use const_int_operand as
1278 operand 2 predicate.
1280 2017-06-18 Richard Biener <rguenther@suse.de>
1282 PR tree-optimization/81410
1283 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
1284 the gap in the ! slp_perm SLP case after each group.
1286 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1289 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
1292 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1295 * predict.c (set_even_probabilities): Cleanup; do not affect
1296 probabilities that are already known.
1297 (combine_predictions_for_bb): Call even when count is set.
1299 2017-07-18 Nathan Sidwell <nathan@acm.org>
1301 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
1304 2017-07-18 Bin Cheng <bin.cheng@arm.com>
1307 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
1308 optimization for loop niter analysis.
1310 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
1313 * config/avr/avr.c (avr_optimize_casesi): Don't use
1314 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
1316 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1318 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
1319 body_cost_vec from _vect_peel_extended_info.
1320 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
1321 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
1324 2017-07-18 Bin Cheng <bin.cheng@arm.com>
1326 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
1328 2017-07-18 Richard Biener <rguenther@suse.de>
1330 PR tree-optimization/80620
1331 PR tree-optimization/81403
1332 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
1333 info when re-using a VN table entry.
1335 2017-07-18 Richard Biener <rguenther@suse.de>
1337 PR tree-optimization/81418
1338 * tree-vect-loop.c (vectorizable_reduction): Properly compute
1339 vectype_in. Verify that with lane-reducing reduction operations
1340 we have a single def-use cycle.
1342 2017-07-17 Carl Love <cel@us.ibm.com>
1344 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
1346 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1347 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
1348 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
1349 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
1350 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
1351 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
1352 VMULOSW): New enum "unspec" values.
1353 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
1354 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
1355 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
1356 altivec_vmulosw): New patterns.
1357 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
1358 VMULOSW): Add definitions.
1359 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
1361 * config/alpha/alpha.c: Include predict.h.
1363 2017-07-17 Yury Gribov <tetra2005@gmail.com>
1365 * tree-vrp.c (compare_assert_loc): Fix comparison function
1366 to return predictable results.
1368 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1370 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
1373 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
1374 * doc/invoke.texi (mexpand-adddi): Update text.
1376 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1378 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
1379 that also clobbers the CC register. The old expand code is moved
1381 (*arc_clzsi2): ... here.
1382 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
1383 the CC register. The old expand code is moved to ...
1384 (arc_ctzsi2): ... here.
1386 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1388 * config/arc/arc.opt (mindexed-loads): Use initial value
1389 TARGET_INDEXED_LOADS_DEFAULT.
1390 (mauto-modify-reg): Use initial value
1391 TARGET_AUTO_MODIFY_REG_DEFAULT.
1392 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
1393 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
1394 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
1395 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
1397 2017-07-17 Martin Liska <mliska@suse.cz>
1400 * opts.c (finish_options): Do not allow -fgnu-tm
1401 w/ -fsanitize={kernel-,}address. Say sorry.
1403 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1406 * tree-loop-distribution.c (classify_partition): Only assert on
1407 numer of iterations.
1408 (merge_dep_scc_partitions): Delete prameter. Update function call.
1409 (distribute_loop): Remove code handling loop with unknown niters.
1410 (pass_loop_distribution::execute): Skip loop with unknown niters.
1412 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1415 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
1416 function sort_partitions_by_post_order.
1418 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1420 PR tree-optimization/81374
1421 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
1422 the max index of basic blocks, rather than number of basic blocks.
1424 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1426 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
1428 (arc_legitimate_pic_operand_p): Likewise.
1429 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
1431 (arc_needs_pcl_p): Likewise.
1432 (arc_legitimate_pc_offset_p): Likewise.
1433 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
1434 function is also used in constrains.md.
1435 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
1436 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
1437 PLUS. Only return true/false in known cases, otherwise assert.
1438 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
1439 is already called in arc_legitimate_constant_p.
1440 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
1442 (LEGITIMATE_PIC_OPERAND_P): Use
1443 arc_raw_symbolic_reference_mentioned_p function.
1444 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
1449 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1450 Andrew Burgess <andrew.burgess@embecosm.com>
1452 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
1453 (arc_return_address_register): New function.
1454 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
1455 (arc_handle_fndecl_attribute): Add naked attribute.
1456 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
1457 (TARGET_WARN_FUNC_RETURN): Likewise.
1458 (arc_allocate_stack_slots_for_args): New function.
1459 (arc_warn_func_return): Likewise.
1460 (machine_function): Change type fn_type.
1461 (arc_compute_function_type): Consider new naked function type,
1462 change function return type.
1463 (arc_must_save_register): Adapt to handle new
1464 arc_compute_function_type's return type.
1465 (arc_expand_prologue): Likewise.
1466 (arc_expand_epilogue): Likewise.
1467 (arc_return_address_regs): Delete.
1468 (arc_return_address_register): New function.
1469 (arc_epilogue_uses): Use above function.
1470 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
1471 (arc_function_type): Change encoding, add naked type.
1472 (ARC_INTERRUPT_P): Change to handle the new encoding.
1473 (ARC_FAST_INTERRUPT_P): Likewise.
1474 (ARC_NORMAL_P): Define.
1475 (ARC_NAKED_P): Likewise.
1476 (arc_compute_function_type): Delete prototype.
1477 * config/arc/arc.md (in_ret_delay_slot): Use
1478 arc_return_address_register function.
1479 (simple_return): Likewise.
1480 (p_return_i): Likewise.
1482 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1484 PR tree-optimization/81428
1485 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
1486 can't be built for those types.
1488 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
1490 Remove stuff dead since r239246.
1492 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
1493 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
1494 (avr_inform_devices): Remove dead stuff.
1496 2017-07-17 Tamar Christina <tamar.christina@arm.com>
1498 * config/arm/arm_neon.h: Fix softp typo.
1500 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1502 PR tree-optimization/81365
1503 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
1504 aggregate moves onto bb predecessor edges, make sure there are no
1505 loads that could alias the lhs in between the start of bb and the
1508 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
1511 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
1512 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
1513 [LSHIFTRT, outer_code = TRUNCATE]: Same.
1515 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1517 PR tree-optimization/81396
1518 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
1519 (init_symbolic_number): Initialize it to 1.
1520 (perform_symbolic_merge): Add n_ops from both operands into the new
1522 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
1523 without base_addr as useless if they need more than one operation.
1524 (bswap_replace): Handle !bswap case for NULL base_addr.
1526 2017-07-17 Tom de Vries <tom@codesourcery.com>
1529 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
1532 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
1534 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
1535 conditional builtin define __FIX_LEON3FT_B2BST.
1537 2017-07-17 Daniel Cederman <cederman@gaisler.com>
1539 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
1540 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
1543 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1545 PR rtl-optimization/81424
1546 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
1547 to remove potential trapping from operands if -fnon-call-exceptions.
1549 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1551 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
1552 profile_proability for scalling.
1553 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
1555 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1557 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
1559 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1561 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
1562 fixpoint arithmetics.
1564 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1566 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
1567 fixpoint arithmetics.
1569 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1571 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
1572 fixpoint arithmetics.
1574 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1576 * profile-count.h (profile_probability::from_reg_br_prob_note,
1577 profile_probability::to_reg_br_prob_note): New functions.
1578 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
1579 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
1580 * predict.c (probability_reliable_p): Update.
1581 (edge_probability_reliable_p): Update.
1582 (br_prob_note_reliable_p): Update.
1583 (invert_br_probabilities): Update.
1584 (add_reg_br_prob_note): New function.
1585 (combine_predictions_for_insn): Update.
1586 * asan.c (asan_clear_shadow): Update.
1587 * cfgbuild.c (compute_outgoing_frequencies): Update.
1588 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
1589 (update_br_prob_note): Update.
1590 (rtl_verify_edges): Update.
1591 (purge_dead_edges): Update.
1592 (fixup_reorder_chain): Update.
1593 * emit-rtl.c (try_split): Update.
1594 * ifcvt.c (cond_exec_process_insns): Update.
1595 (cond_exec_process_if_block): Update.
1596 (dead_or_predicable): Update.
1597 * internal-fn.c (expand_addsub_overflow): Update.
1598 (expand_neg_overflow): Update.
1599 (expand_mul_overflow): Update.
1600 * loop-doloop.c (doloop_modify): Update.
1601 * loop-unroll.c (compare_and_jump_seq): Update.
1602 * optabs.c (emit_cmp_and_jump_insn_1): Update.
1603 * predict.h: Update.
1604 * reorg.c (mostly_true_jump): Update.
1606 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
1607 * config/alpha/alpha.c (emit_unlikely_jump): Update.
1608 * config/arc/arc.c: (emit_unlikely_jump): Update.
1609 * config/arm/arm.c: (emit_unlikely_jump): Update.
1610 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
1611 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
1612 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
1613 (ix86_print_operand): Update.
1614 (ix86_split_fp_branch): Update.
1615 (predict_jump): Update.
1616 * config/ia64/ia64.c (ia64_print_operand): Update.
1617 * config/mmix/mmix.c (mmix_print_operand): Update.
1618 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
1619 (rs6000_expand_split_stack_prologue): Update.
1620 * config/rs6000/rs6000.c: Update.
1621 * config/s390/s390.c (s390_expand_vec_strlen): Update.
1622 (s390_expand_vec_movstr): Update.
1623 (s390_expand_cs_tdsi): Update.
1624 (s390_expand_split_stack_prologue): Update.
1625 * config/sh/sh.c (sh_print_operand): Update.
1626 (expand_cbranchsi4): Update.
1627 (expand_cbranchdi4): Update.
1628 * config/sparc/sparc.c (output_v9branch): Update.
1629 * config/spu/spu.c (get_branch_target): Update.
1630 (ea_load_store_inline): Update.
1631 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
1632 * config/tilepro/tilepro.c: Update.
1634 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1636 * gimplify.c (mostly_copy_tree_r): Revert latest change.
1637 (gimplify_save_expr): Likewise.
1639 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1641 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
1643 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1645 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
1647 * timevar.def (TV_IPA_FNSUMMARY): Define.
1649 2017-07-16 Daniel Cederman <cederman@gaisler.com>
1651 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
1652 to back store errata sensitive sequence from being generated.
1653 (sqrtdf2_fix): Likewise.
1655 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1657 * tree-ssa-threadupdate.c (compute_path_counts,
1658 update_joiner_offpath_counts): Use profile_probability.
1660 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
1663 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1665 * config/arm/arm-c.c (arm_cpu_builtins): Define
1666 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1668 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
1670 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1671 array entries to represent __ieee128 versions of the
1672 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
1673 scalar_extract_sig, and scalar_insert_exp built-in functions.
1674 (altivec_resolve_overloaded_builtin): Add special case handling
1675 for the __builtin_scalar_insert_exp function, as represented by
1676 the P9V_BUILTIN_VEC_VSIEDP constant.
1677 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
1678 exponent support for __ieee128 argument.
1679 (VSESQP): Add scalar extract signature support for __ieee128
1681 (VSTDCNQP): Add scalar test negative support for __ieee128
1683 (VSIEQP): Add scalar insert exponent support for __int128 argument
1684 with __ieee128 result.
1685 (VSIEQPF): Add scalar insert exponent support for __ieee128
1686 argument with __ieee128 result.
1687 (VSTDCQP): Add scalar test data class support for __ieee128
1689 (VSTDCNQP): Add overload support for scalar test negative with
1691 (VSTDCQP): Add overload support for scalar test data class
1693 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
1695 (UNSPEC_VSX_SIEXPQP): New constant.
1696 (xsxexpqp): New insn for VSX scalar extract exponent quad
1698 (xsxsigqp): New insn for VSX scalar extract significand quad
1700 (xsiexpqpf): New insn for VSX scalar insert exponent quad
1701 precision with floating point argument.
1702 (xststdcqp): New expand for VSX scalar test data class quad
1704 (xststdcnegqp): New expand for VSX scalar test negative quad
1706 (xststdcqp): New insn to match expansions for VSX scalar test data
1707 class quad precision and VSX scalar test negative quad precision.
1708 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
1709 special case operand checking to enforce that second operand of
1710 VSX scalar test data class with quad precision argument is a 7-bit
1712 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
1713 prototypes and descriptions of __ieee128 versions of
1714 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
1715 scalar_test_data_class, and scalar_test_neg built-in functions.
1717 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1719 PR tree-optimization/81162
1720 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
1721 replace a negate with an add.
1723 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
1725 * doc/invoke.texi (arm/-mcpu): Document +crypto.
1727 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1729 * config/arm/arm-c.c (arm_cpu_builtins): Define
1730 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1732 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1734 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
1735 (armv8-r): Set ARM Cortex-R52 as default CPU.
1736 * config/arm/arm-tables.opt: Regenerate.
1737 * config/arm/arm-tune.md: Regenerate.
1738 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
1740 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
1741 extension for -mcpu=cortex-r52.
1743 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1745 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
1746 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
1747 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
1748 (fp-armv8): Define it as FP_ARMv8 only.
1749 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
1750 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
1752 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
1753 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
1754 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
1755 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
1756 than TARGET_FPU_ARMV8.
1757 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
1758 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
1759 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
1761 * config/arm/neon.md (neon_vrint): Likewise.
1762 (neon_vcvt): Likewise.
1763 (neon_<fmaxmin_op><mode>): Likewise.
1764 (<fmaxmin><mode>3): Likewise.
1765 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
1766 * config/arm/predicates.md (arm_cond_move_operator): Check against
1767 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
1769 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
1771 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
1774 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1776 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
1777 loop in comment with memset.
1779 2017-07-14 Martin Liska <mliska@suse.cz>
1781 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
1782 * dwarf2out.c (is_java): Remove the function.
1783 (output_pubname): Remove usage of the function.
1784 (lower_bound_default): Remove usage of DW_LANG_Java.
1785 (gen_compile_unit_die): Likewise.
1786 * gcc.c: Remove compiler defaults for .java and .zip files.
1787 * gimple-expr.c (remove_suffix): Change as there's no longer
1788 extension than 4-letter one.
1789 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
1790 (gimplify_save_expr): Likewise.
1791 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
1792 as it's possible even for other languages than Java.
1793 * langhooks.h (struct lang_hooks): Remove Java from a comment.
1794 * lto-opts.c (lto_write_options): Remove reference to Java.
1795 * opts.c (strip_off_ending): Update file extension handling.
1796 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
1797 * tree-eh.c (lower_resx): Likewise.
1798 * tree.c (free_lang_data_in_type): Remove dead code.
1799 (find_decls_types_r): Likewise.
1800 (build_common_builtin_nodes): Remove Java from a comment.
1801 (verify_type): Remove dead code.
1802 * varasm.c (assemble_external): Remove Java from a comment.
1804 2017-07-14 Martin Liska <mliska@suse.cz>
1806 * opts.c (finish_options): Add quotes.
1807 (common_handle_option): Likewise.
1809 2017-07-14 Martin Liska <mliska@suse.cz>
1811 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
1812 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
1814 * dwarf2out.c (lower_bound_default): Do not handle
1816 (gen_compile_unit_die): Likewise.
1817 * gcc.c: Remove default extension binding for GNU Pascal.
1818 * stmt.c: Remove Pascal language from a comment.
1819 * xcoffout.c: Likewise.
1821 2017-07-13 David Malcolm <dmalcolm@redhat.com>
1824 * diagnostic-show-locus.c (fixit_cmp): New function.
1825 (layout::layout): Sort m_fixit_hints.
1826 (column_range::column_range): Assert that the values are valid.
1827 (struct char_span): New struct.
1828 (correction::overwrite): New method.
1829 (struct source_line): New struct.
1830 (line_corrections::add_hint): Add assertions. Reimplement memcpy
1831 calls in terms of classes source_line and char_span, and
1832 correction::overwrite.
1833 (selftest::test_overlapped_fixit_printing_2): New function.
1834 (selftest::diagnostic_show_locus_c_tests): Call it.
1836 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
1838 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
1839 early if there is no lhs.
1841 2017-07-13 Martin Liska <mliska@suse.cz>
1843 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
1844 (gen_reference_type_die): Likewise.
1845 * stor-layout.c: Remove Pascal-related comment.
1847 2017-07-13 Martin Liska <mliska@suse.cz>
1849 * opts.c (finish_options): Add quotes to error messages.
1850 (parse_sanitizer_options): Likewise.
1852 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
1854 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
1856 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
1858 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
1860 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
1862 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
1864 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
1866 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1869 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
1870 provides the hardware capability bits, define the macro
1871 __BUILTIN_CPU_SUPPORTS__.
1872 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
1873 if GLIBC does not provide the hardware capability bits. Add a
1874 gcc_unreachable call if the built-in cpu function is neither
1875 __builtin_cpu_is nor __builtin_cpu_supports.
1876 (rs6000_get_function_versions_dispatcher): Change the warning
1877 that an old GLIBC is used which does not export the capability
1878 bits to be an error.
1879 * doc/extend.texi (target_clones attribute): Document the
1880 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
1881 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
1882 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
1883 the macros defined by GCC if the newer GLIBC is available.
1885 2017-07-12 Jeff Law <law@redhat.com>
1887 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
1888 remaining includes slightly.
1889 * config/riscv/riscv-builtins.c: Include profile-count.h.
1891 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1894 * config/avr/avr.c (avr_set_current_function): In diagnostic
1895 messages: Quote keywords and (parts of) identifiers.
1896 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
1899 2017-07-12 Carl Love <cel@us.ibm.com>
1901 * config/rs6000/rs6000-c.c: Add support for built-in functions
1902 vector bool char vec_revb (vector bool char);
1903 vector bool short vec_revb (vector short char);
1904 vector bool int vec_revb (vector bool int);
1905 vector bool long long vec_revb (vector bool long long);
1906 * doc/extend.texi: Update the built-in documentation file for the
1907 new built-in functions.
1909 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1911 * config/s390/s390.md: Remove movcc splitter.
1913 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1915 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
1916 load/store on condition.
1918 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1921 * config/avr/avr.c (avr_encode_section_info)
1922 [progmem && !TREE_READONLY]: Error if progmem object needs
1925 2017-07-11 Michael Collison <michael.collison@arm.com>
1927 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
1930 2017-07-11 Carl Love <cel@us.ibm.com>
1932 * config/rs6000/rs6000-c.c: Add support for builtins
1933 vector unsigned int vec_parity_lsbb (vector signed int);
1934 vector unsigned int vec_parity_lsbb (vector unsigned int);
1935 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
1936 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
1937 vector unsigned long long vec_parity_lsbb (vector signed long long);
1938 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
1939 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
1940 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
1941 * doc/extend.texi: Update the built-in documentation file for the
1942 new built-in functions.
1944 2017-07-11 David Malcolm <dmalcolm@redhat.com>
1946 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
1947 (layout::m_primary_loc): New field.
1948 (layout::layout): Initialize new field. Move location filtering
1949 logic from here to...
1950 (layout::maybe_add_location_range): ...this new method. Add
1951 support for filtering to just the lines already specified by other
1953 (layout::will_show_line_p): New method.
1954 (gcc_rich_location::add_location_if_nearby): New method.
1955 (selftest::test_add_location_if_nearby): New test function.
1956 (selftest::diagnostic_show_locus_c_tests): Call it.
1957 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
1960 2017-07-11 Tom de Vries <tom@codesourcery.com>
1962 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
1963 (bb_first_real_insn): New function.
1964 (nvptx_single): Add extra initialization of broadcasted condition
1967 2017-07-11 Nathan Sidwell <nathan@acm.org>
1969 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
1971 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
1973 * doc/extend.texi (AVR Function Attributes): Remove weblink to
1974 Binutils doc as TEXI will mess them up.
1975 * doc/invoke.texi (AVR Options): Same here.
1977 2017-07-11 Daniel Cederman <cederman@gaisler.com>
1979 * config/sparc/sparc.opt (mfix-ut700): New option.
1980 (mfix-gr712rc): Likewise.
1981 (sparc_fix_b2bst): New variable.
1982 * doc/invoke.texi (SPARC options): Document them.
1983 (ARM options): Fix warnings.
1984 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
1985 instructions to prevent sequences that can trigger the store-store
1986 errata for certain LEON3FT processors.
1987 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
1988 (sparc_option_override): Set sparc_fix_b2bst appropriately.
1989 * config/sparc/sparc.md (fix_b2bst): New attribute.
1990 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
1992 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
1995 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
1997 (*rsqrtsf2_sse): Ditto.
1999 (div<mode>3): Macroize insn from divdf3 and divsf3
2000 using MODEF mode iterator.
2002 2017-07-10 Martin Sebor <msebor@redhat.com>
2004 PR tree-optimization/80397
2005 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
2006 instead of testing for equality to INTEGER_TYPE.
2008 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
2010 * config.gcc: Remove uclibc from arc target spec.
2012 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
2014 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
2016 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
2019 * lto-wrapper.c (remove_option): New function.
2020 (merge_and_complain): Merge PIC/PIE options more realistically.
2022 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
2024 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
2028 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
2029 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
2030 * config.in: Regenerate.
2031 * configure: Regenerate.
2032 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
2033 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
2034 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
2035 (TARGET_GASISR_PROLOGUES): ...target mask.
2036 * common/config/avr/avr-common.c
2037 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
2038 Set -mgas-isr-prologues.
2039 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
2040 INSERT_PASS_BEFORE for it.
2041 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
2042 * config/avr/avr.c (avr_option_override)
2043 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
2044 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
2045 (avr_attribute_table) <no_gccisr>: Add new function attribute.
2046 (avr_set_current_function) <is_no_gccisr>: Init machine field.
2047 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
2049 (make_avr_pass_pre_proep): New function.
2050 (emit_push_sfr) <treg>: Add argument to function and use it
2052 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
2053 and set machine->gasisr.yes.
2054 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
2055 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
2056 __gcc_isr.n_pushed to .L__stack_usage.
2057 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
2058 (avr_asm_final_postscan_insn): ...this new static function.
2059 * config/avr/avr.h (machine_function)
2060 <is_no_gccisr, use_L__stack_usage>: New fields.
2061 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
2062 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
2063 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
2064 (gasisr, *gasisr): New expander and insn.
2065 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
2066 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
2067 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
2069 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
2071 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
2074 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
2076 Move jump-tables out of .text again.
2079 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
2080 (ASM_OUTPUT_ADDR_VEC): New function.
2081 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
2082 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
2083 INSN_ADDRESSes as asm comment.
2084 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
2085 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
2086 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
2087 * config/avr/avr.md (*tablejump): Adjust comment.
2088 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
2089 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
2091 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
2092 (avr_output_addr_vec): New proto.
2093 (avr_log_t) <insn_addresses>: New field.
2095 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
2098 * config/i386/i386.c (ix86_function_arg_advance): Set
2099 outgoing_args_on_stack to true if there are outgoing arguments
2101 (ix86_function_arg): Likewise.
2102 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
2103 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
2104 * config/i386/i386.h (machine_function): Add
2105 outgoing_args_on_stack.
2107 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
2109 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
2111 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
2113 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
2115 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
2116 (REAL_H): Remove $(MACHMODE_H).
2117 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
2119 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
2120 $(MACHMODE_H) and double-int.h.
2121 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
2123 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
2126 2017-07-07 Andrew Pinski <apinski@cavium.com>
2128 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
2129 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
2131 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
2133 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
2134 Add warning if GCC was not configured to link against a GLIBC that
2135 exports the hardware capability bits.
2136 (make_resolver_func): Make resolver function private and not a
2137 COMDAT function. Create the name with clone_function_name instead
2138 of make_unique_name.
2141 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
2142 correct operand in doing the split.
2144 2017-07-07 Carl Love <cel@us.ibm.com>
2146 * config/rs6000/rs6000-c: Add support for built-in function
2147 vector unsigned short vec_pack_to_short_fp32 (vector float,
2149 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
2150 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
2151 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
2152 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
2153 (convert_4f32_8i16): Add define_expand.
2154 * doc/extend.texi: Update the built-in documentation file for the
2155 new built-in function.
2157 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2159 * config/sparc/m8.md: New file.
2160 * config/sparc/sparc.md: Include m8.md.
2162 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2164 * config/sparc/sparc.opt: New option -mvis4b.
2165 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
2166 (sparc_option_override): Handle VIS4B.
2167 (enum sparc_builtins): Define
2168 SPARC_BUILTIN_DICTUNPACK{8,16,32},
2169 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
2170 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
2171 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
2172 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
2173 (check_constant_argument): New function.
2174 (sparc_vis_init_builtins): Define builtins
2175 __builtin_vis_dictunpack{8,16,32},
2176 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
2177 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
2178 __builtin_vis_fpcmpde{8,16,32}shl and
2179 __builtin_vis_fpcmpur{8,16,32}shl.
2180 (sparc_expand_builtin): Check that the constant operands to
2181 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
2182 constant and in range.
2183 * config/sparc/sparc-c.c (sparc_target_macros): Handle
2185 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
2186 (SPARC_IMM5_P): Likewise.
2187 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
2188 (enabled): Handle vis4b.
2189 (UNSPEC_DICTUNPACK): New unspec.
2190 (UNSPEC_FPCMPSHL): Likewise.
2191 (UNSPEC_FPUCMPSHL): Likewise.
2192 (UNSPEC_FPCMPDESHL): Likewise.
2193 (UNSPEC_FPCMPURSHL): Likewise.
2194 (cpu_feature): New CPU feature `vis4b'.
2195 (dictunpack{8,16,32}): New insns.
2196 (FPCSMODE): New mode iterator.
2197 (fpcscond): New code iterator.
2198 (fpcsucond): Likewise.
2199 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
2200 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
2201 (fpcmpde{8,16,32}{si,di}shl): Likewise.
2202 (fpcmpur{8,16,32}{si,di}shl): Likewise.
2203 * config/sparc/constraints.md: Define constraints `q' for unsigned
2204 2-bit integer constants and `t' for unsigned 5-bit integer
2206 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
2208 (imm5_operand_dictunpack16): Likewise.
2209 (imm5_operand_dictunpack32): Likewise.
2210 (imm2_operand): Likewise.
2211 * doc/invoke.texi (SPARC Options): Document -mvis4b.
2212 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
2213 ditunpack* and fpcmp*shl builtins.
2215 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2217 * config.gcc: Handle m8 in --with-{cpu,tune} options.
2218 * config.in: Add HAVE_AS_SPARC6 define.
2219 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
2221 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
2223 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
2224 (CPP_CPU_SPEC): Handle m8.
2225 (ASM_CPU_SPEC): Likewise.
2226 * config/sparc/sparc-opts.h (enum processor_type): Add
2228 * config/sparc/sparc.c (m8_costs): New struct.
2229 (sparc_option_override): Handle TARGET_CPU_m8.
2230 (sparc32_initialize_trampoline): Likewise.
2231 (sparc64_initialize_trampoline): Likewise.
2232 (sparc_issue_rate): Likewise.
2233 (sparc_register_move_cost): Likewise.
2234 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
2235 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
2236 (ASM_CPU64_DEFAULT_SPEC): Likewise.
2237 (CPP_CPU_SPEC): Handle M8.
2238 (ASM_CPU_SPEC): Likewise.
2239 (AS_M8_FLAG): Define.
2240 * config/sparc/sparc.md: Add m8 to the cpu attribute.
2241 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
2242 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
2244 * configure: Regenerate.
2245 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
2248 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2250 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
2252 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
2253 ("*movdi_insn_sp32"): Do not set v3pipe.
2254 ("*movsi_insn"): Likewise.
2255 ("*movdi_insn_sp64"): Likewise.
2256 ("*movsf_insn"): Likewise.
2257 ("*movdf_insn_sp32"): Likewise.
2258 ("*movdf_insn_sp64"): Likewise.
2259 ("*zero_extendsidi2_insn_sp64"): Likewise.
2260 ("*sign_extendsidi2_insn"): Likewise.
2261 ("*mov<VM32:mode>_insn"): Likewise.
2262 ("*mov<VM64:mode>_insn_sp64"): Likewise.
2263 ("*mov<VM64:mode>_insn_sp32"): Likewise.
2264 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
2265 ("<vlop:code><VL:mode>3"): Likewise.
2266 ("*not_<vlop:code><VL:mode>3"): Likewise.
2267 ("*nand<VL:mode>_vis"): Likewise.
2268 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
2269 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
2270 ("one_cmpl<VL:mode>2"): Likewise.
2271 ("faligndata<VM64:mode>_vis"): Likewise.
2272 ("alignaddrsi_vis"): Likewise.
2273 ("alignaddrdi_vis"): Likweise.
2274 ("alignaddrlsi_vis"): Likewise.
2275 ("alignaddrldi_vis"): Likewise.
2276 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
2277 ("bmaskdi_vis"): Likewise.
2278 ("bmasksi_vis"): Likewise.
2279 ("bshuffle<VM64:mode>_vis"): Likewise.
2280 ("cmask8<P:mode>_vis"): Likewise.
2281 ("cmask16<P:mode>_vis"): Likewise.
2282 ("cmask32<P:mode>_vis"): Likewise.
2283 ("pdistn<P:mode>_vis"): Likewise.
2284 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
2286 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2288 * config/sparc/sparc.md ("subtype"): New insn attribute.
2289 ("*wrgsr_sp64"): Set insn subtype.
2290 ("*rdgsr_sp64"): Likewise.
2291 ("alignaddrsi_vis"): Likewise.
2292 ("alignaddrdi_vis"): Likewise.
2293 ("alignaddrlsi_vis"): Likewise.
2294 ("alignaddrldi_vis"): Likewise.
2295 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
2296 ("fexpand_vis"): Likewise.
2297 ("fpmerge_vis"): Likewise.
2298 ("faligndata<VM64:mode>_vis"): Likewise.
2299 ("bshuffle<VM64:mode>_vis"): Likewise.
2300 ("cmask8<P:mode>_vis"): Likewise.
2301 ("cmask16<P:mode>_vis"): Likewise.
2302 ("cmask32<P:mode>_vis"): Likewise.
2303 ("fchksm16_vis"): Likewise.
2304 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
2305 ("fmean16_vis"): Likewise.
2306 ("fp<plusminus_insn>64_vis"): Likewise.
2307 ("<plusminus_insn>v8qi3"): Likewise.
2308 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
2309 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
2310 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
2311 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
2312 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
2313 ("*movqi_insn"): Likewise.
2314 ("*movhi_insn"): Likewise.
2315 ("*movsi_insn"): Likewise.
2316 ("movsi_pic_gotdata_op"): Likewise.
2317 ("*movdi_insn_sp32"): Likewise.
2318 ("*movdi_insn_sp64"): Likewise.
2319 ("movdi_pic_gotdata_op"): Likewise.
2320 ("*movsf_insn"): Likewise.
2321 ("*movdf_insn_sp32"): Likewise.
2322 ("*movdf_insn_sp64"): Likewise.
2323 ("*zero_extendhisi2_insn"): Likewise.
2324 ("*zero_extendqihi2_insn"): Likewise.
2325 ("*zero_extendqisi2_insn"): Likewise.
2326 ("*zero_extendqidi2_insn"): Likewise.
2327 ("*zero_extendhidi2_insn"): Likewise.
2328 ("*zero_extendsidi2_insn_sp64"): Likewise.
2329 ("ldfsr"): Likewise.
2330 ("prefetch_64"): Likewise.
2331 ("prefetch_32"): Likewise.
2332 ("tie_ld32"): Likewise.
2333 ("tie_ld64"): Likewise.
2334 ("*tldo_ldub_sp32"): Likewise.
2335 ("*tldo_ldub1_sp32"): Likewise.
2336 ("*tldo_ldub2_sp32"): Likewise.
2337 ("*tldo_ldub_sp64"): Likewise.
2338 ("*tldo_ldub1_sp64"): Likewise.
2339 ("*tldo_ldub2_sp64"): Likewise.
2340 ("*tldo_ldub3_sp64"): Likewise.
2341 ("*tldo_lduh_sp32"): Likewise.
2342 ("*tldo_lduh1_sp32"): Likewise.
2343 ("*tldo_lduh_sp64"): Likewise.
2344 ("*tldo_lduh1_sp64"): Likewise.
2345 ("*tldo_lduh2_sp64"): Likewise.
2346 ("*tldo_lduw_sp32"): Likewise.
2347 ("*tldo_lduw_sp64"): Likewise.
2348 ("*tldo_lduw1_sp64"): Likewise.
2349 ("*tldo_ldx_sp64"): Likewise.
2350 ("*mov<VM32:mode>_insn"): Likewise.
2351 ("*mov<VM64:mode>_insn_sp64"): Likewise.
2352 ("*mov<VM64:mode>_insn_sp32"): Likewise.
2354 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2356 * config/sparc/sparc.md ("type"): New insn type viscmp.
2357 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
2359 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
2360 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
2361 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
2362 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
2364 ("n7_vis_logical_11cycle"): Likewise.
2365 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
2366 * config/sparc/niagara2.md ("niag3_vis": Likewise.
2367 * config/sparc/niagara.md ("niag_vis"): Likewise.
2368 * config/sparc/ultra3.md ("us3_fga"): Likewise.
2369 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
2371 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2373 * config/sparc/sparc.md: New instruction type `bmask'.
2374 (bmaskdi_vis): Use the `bmask' type.
2375 (bmasksi_vis): Likewise.
2376 * config/sparc/ultra3.md (us3_array): Likewise.
2377 * config/sparc/niagara7.md (n7_array): Likewise.
2378 * config/sparc/niagara4.md (n4_array): Likewise.
2379 * config/sparc/niagara2.md (niag2_vis): Likewise.
2380 (niag3_vis): Likewise.
2381 * config/sparc/niagara.md (niag_vis): Likewise.
2383 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2385 * ipa-comdats.c: Remove optimize check from gate.
2386 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
2387 for functions not optimized.
2388 (ipa_fn_summary_read): Skip optimize check.
2389 (ipa_fn_summary_write): Likewise.
2390 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
2392 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
2394 (can_inline_edge_p): Check flag_pcc_struct_return for match.
2395 (check_callers): Give up on caller which is not optimized.
2396 (inline_small_functions): Likewise.
2397 (ipa_inline): Do not give up when not optimizing.
2398 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
2399 away unoptimizes cdtors.
2400 (whole_program_function_and_variable_visibility): Do
2401 ipa_discover_readonly_nonaddressable_vars in LTO mode.
2402 * ipa.c (process_references): Do not check optimize.
2403 (symbol_table::remove_unreachable_nodes): Update optimize check.
2404 (set_writeonly_bit): Update optimize check.
2405 (pass_ipa_cdtor_merge::gate): Do not check optimize.
2406 (pass_ipa_single_use::gate): Remove.
2408 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2410 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
2411 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
2412 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
2413 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
2414 permute_load, permute_store, adjust_extract, adjust_splat,
2415 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
2416 replace_swap_with_copy, dump_swap_insn_table,
2417 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
2418 recombine_lvx_pattern, recombine_stvx_pattern,
2419 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
2420 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
2421 to file rs6000-p8swap.c.
2422 * config/rs6000/rs6000-p8swap.c: New file.
2423 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
2424 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
2425 and rs6000*-*-* targets.
2427 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2429 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
2431 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2433 * lto-wrapper.c (merge_and_complain): Do not merge
2434 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
2435 fsigned_zeros, ftrapping_math, fwrapv.
2436 (append_compiler_options): Do not track these options.
2437 (append_linker_options): Likewie
2439 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2441 * cgraphunit.c (cgraph_node::finalize_function): When
2442 !flag_toplevel_reorde set no_reorder flag.
2443 (varpool_node::finalize_decl): Likewise.
2444 (symbol_table::compile): Drop no toplevel reorder path.
2446 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2448 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
2449 edges; zero probability is not better than uninitialized.
2451 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
2453 * asan.h (asan_sanitize_allocas_p): Declare.
2454 * asan.c (asan_sanitize_allocas_p): New function.
2455 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
2456 (handle_builtin_alloca): Likewise.
2457 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
2458 if !asan_sanitize_allocas_p.
2459 * params.def (asan-instrument-allocas): Add new option.
2460 * params.h (ASAN_PROTECT_ALLOCAS): Define.
2461 * opts.c (common_handle_option): Disable allocas sanitization for
2464 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
2466 * asan.c: Include gimple-fold.h.
2467 (get_last_alloca_addr): New function.
2468 (handle_builtin_stackrestore): Likewise.
2469 (handle_builtin_alloca): Likewise.
2470 (asan_emit_allocas_unpoison): Likewise.
2471 (get_mem_refs_of_builtin_call): Add new parameter, remove const
2472 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
2473 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
2474 (instrument_builtin_call): Pass gimple iterator to
2475 get_mem_refs_of_builtin_call.
2476 (last_alloca_addr): New global.
2477 * asan.h (asan_emit_allocas_unpoison): Declare.
2478 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
2479 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
2480 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
2481 if function calls alloca.
2482 * gimple-fold.c (replace_call_with_value): Remove static keyword.
2483 * gimple-fold.h (replace_call_with_value): Declare.
2484 * internal-fn.c: Include asan.h.
2485 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
2486 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
2488 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2490 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
2491 (C_SELFTEST_FLAGS): New.
2492 (CPP_SELFTEST_FLAGS): New.
2493 (SELFTEST_DEPS): New, from deps of s-selftest.
2494 (C_SELFTEST_DEPS): New, from deps of s-selftest.
2495 (CPP_SELFTEST_DEPS): New.
2496 (selftest): Add dependency on s-selftest-c++.
2497 (s-selftest): Rename to...
2498 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
2499 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
2500 than SELFTEST_FLAGS.
2501 (selftest-gdb): Rename to...
2502 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
2504 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
2505 (selftest-valgrind): Rename to...
2506 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
2508 (selftest-valgrind): Reintroduce as an alias for
2509 selftest-c-valgrind.
2510 (s-selftest-c++): New.
2511 (selftest-c++-gdb): New.
2512 (selftest-c++-valgrind): New.
2514 2017-07-06 Olivier Hainque <hainque@adacore.com>
2516 * gcc.c (process_command): When deciding if undefined variables
2517 should be ignored when processing specs, accept "gcc -v" as well.
2519 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2521 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
2522 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
2524 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
2526 * config/arm/arm-cpus.in (armv8-r): Add new entry.
2527 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
2528 * config/arm/arm-tables.opt: Regenerate.
2529 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
2531 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
2533 2017-07-06 Carl Love <cel@us.ibm.com>
2535 * ChangeLog: Clean up from mid air collision
2537 2017-07-06 Carl Love <cel@us.ibm.com>
2539 * config/rs6000/rs6000-c.c: Add support for built-in functions
2540 vector signed int vec_subc (vector signed int, vector signed int);
2541 vector signed __int128 vec_subc (vector signed __int128,
2542 vector signed __int128);
2543 vector unsigned __int128 vec_subc (vector unsigned __int128,
2544 vector unsigned __int128);
2545 vector signed int vec_sube (vector signed int, vector signed int,
2547 vector unsigned int vec_sube (vector unsigned int,
2548 vector unsigned int,
2549 vector unsigned int);
2550 vector signed __int128 vec_sube (vector signed __int128,
2551 vector signed __int128,
2552 vector signed__int128);
2553 vector unsigned __int128 vec_sube (vector unsigned __int128,
2554 vector unsigned __int128,
2555 vector unsigned __int128);
2556 vector signed int vec_subec (vector signed int, vector signed int,
2558 vector unsigned int vec_subec (vector unsigned int,
2559 vector unsigned int,
2560 vector unsigned int);
2561 vector signed __int128 vec_subec (vector signed __int128,
2562 vector signed __int128,
2563 vector signed__int128);
2564 vector unsigned __int128 vec_subec (vector unsigned __int128,
2565 vector unsigned __int128,
2566 vector unsigned __int128);
2567 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
2568 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
2569 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
2570 BU_ALTIVEC_OVERLOAD_X definitions.
2571 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
2572 * doc/extend.texi: Update the built-in documentation file for the new
2575 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2578 * diagnostic-show-locus.c (layout::layout): Use start and finish
2579 spelling location for the start and finish of each range.
2580 * genmatch.c (linemap_client_expand_location_to_spelling_point):
2581 Add unused aspect param.
2582 * input.c (expand_location_1): Add "aspect" param, and use it
2583 to access the correct part of the location.
2584 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
2586 (expand_location_to_spelling_point): Likewise.
2587 (linemap_client_expand_location_to_spelling_point): Add "aspect"
2588 param, and pass it to expand_location_1.
2590 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
2592 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
2593 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
2594 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
2595 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
2596 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
2597 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
2598 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
2599 _mm_maskz_getmant_ss): New intrinsics.
2600 (__builtin_ia32_getexpss128_mask): Changed to ...
2601 __builtin_ia32_getexpss128_round ... this.
2602 (__builtin_ia32_getexpsd128_mask): Changed to ...
2603 __builtin_ia32_getexpsd128_round ... this.
2604 * config/i386/i386-builtin-types.def
2605 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
2606 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
2607 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
2608 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
2609 __builtin_ia32_getmantss_mask_round): New builtins.
2610 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
2611 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
2612 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
2613 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
2614 * config/i386/sse.md
2615 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
2616 avx512f_sgetexp<mode><mask_scalar_name>
2617 <round_saeonly_scalar_name> ... this.
2618 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
2619 %0, %1, %2<round_saeonly_op3>}): Changed to ...
2620 vgetexp<ssescalarmodesuffix>
2621 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2622 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
2623 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
2624 avx512f_vgetmant<mode><mask_scalar_name>
2625 <round_saeonly_scalar_name> ... this.
2626 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
2627 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
2628 vgetmant<ssescalarmodesuffix>
2629 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
2630 %0<mask_scalar_operand4>, %1, %2
2631 <round_saeonly_scalar_mask_op4>, %3} ... this.
2632 * config/i386/subst.md (mask_scalar_operand4,
2633 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
2634 round_saeonly_scalar_nimm_predicate): New subst attributes.
2636 2017-07-06 Julia Koval <julia.koval@intel.com>
2638 * config/i386/i386.c (ix86_erase_embedded_rounding):
2639 Remove code for old rounding pattern.
2641 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
2643 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
2645 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
2647 * doc/sourcebuild.texi (Test Directives, Variants of
2648 dg-require-support): Add documentation for dg-require-stack-check.
2650 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
2652 * config/i386/subst.md (mask_scalar, round_scalar,
2653 round_saeonly_scalar): New meta-templates.
2654 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
2655 round_scalar_mask_operand3, round_scalar_mask_op3,
2656 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
2657 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
2658 round_saeonly_scalar_constraint,
2659 round_saeonly_scalar_prefix): New subst attribute.
2660 * config/i386/sse.md
2661 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
2662 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
2663 <round_scalar_name> ... this.
2664 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
2665 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
2666 <round_scalar_name> ... this.
2667 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
2668 <sse>_vm<code><mode>3<mask_scalar_name>
2669 <round_saeonly_scalar_name> ... this.
2670 (v<plusminus_mnemonic><ssescalarmodesuffix>
2671 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2672 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2673 v<plusminus_mnemonic><ssescalarmodesuffix>
2674 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2675 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2676 (v<multdiv_mnemonic><ssescalarmodesuffix>
2677 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2678 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2679 v<multdiv_mnemonic><ssescalarmodesuffix>
2680 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2681 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2682 (v<maxmin_float><ssescalarmodesuffix>
2683 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
2684 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
2685 v<maxmin_float><ssescalarmodesuffix>
2686 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2687 %0<mask_scalar_operand3>, %1, %<iptr>2
2688 <round_saeonly_scalar_mask_op3>} ... this.
2690 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
2692 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
2693 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
2695 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
2696 Alan Hayward <alan.hayward@arm.com>
2697 David Sherwood <david.sherwood@arm.com>
2699 * combine.c (simplify_if_then_else): Remove "enum" before
2701 * compare-elim.c (can_eliminate_compare): Likewise.
2702 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
2704 (aarch64_lookup_simd_builtin_type): Likewise.
2705 (aarch64_simd_builtin_type): Likewise.
2706 (aarch64_init_simd_builtin_types): Likewise.
2707 (aarch64_simd_expand_args): Likewise.
2708 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
2710 (aarch64_reverse_mask): Likewise.
2711 (aarch64_simd_emit_reg_reg_move): Likewise.
2712 (aarch64_gen_adjusted_ldpstp): Likewise.
2713 (aarch64_ccmp_mode_to_code): Likewise.
2714 (aarch64_operands_ok_for_ldpstp): Likewise.
2715 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2716 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
2718 (aarch64_min_divisions_for_recip_mul): Likewise.
2719 (aarch64_reassociation_width): Likewise.
2720 (aarch64_get_condition_code_1): Likewise.
2721 (aarch64_simd_emit_reg_reg_move): Likewise.
2722 (aarch64_simd_attr_length_rglist): Likewise.
2723 (aarch64_reverse_mask): Likewise.
2724 (aarch64_operands_ok_for_ldpstp): Likewise.
2725 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2726 (aarch64_gen_adjusted_ldpstp): Likewise.
2727 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
2729 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
2730 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
2731 (arm_lookup_simd_builtin_type): Likewise.
2732 (arm_simd_builtin_type): Likewise.
2733 (arm_init_simd_builtin_types): Likewise.
2734 (arm_expand_builtin_args): Likewise.
2735 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
2736 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
2737 (ft32_setup_incoming_varargs): Likewise.
2738 (ft32_function_arg): Likewise.
2739 (ft32_function_arg_advance): Likewise.
2740 (ft32_pass_by_reference): Likewise.
2741 (ft32_arg_partial_bytes): Likewise.
2742 (ft32_valid_pointer_mode): Likewise.
2743 (ft32_addr_space_pointer_mode): Likewise.
2744 (ft32_addr_space_legitimate_address_p): Likewise.
2745 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
2747 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
2748 (ix86_emit_outlined_ms2sysv_restore): Likewise.
2749 (iamcu_alignment): Likewise.
2750 (canonicalize_vector_int_perm): Likewise.
2751 (ix86_noce_conversion_profitable_p): Likewise.
2752 (ix86_mpx_bound_mode): Likewise.
2753 (ix86_operands_ok_for_move_multiple): Likewise.
2754 * config/microblaze/microblaze-protos.h
2755 (microblaze_expand_conditional_branch_reg): Likewise.
2756 * config/microblaze/microblaze.c
2757 (microblaze_expand_conditional_branch_reg): Likewise.
2758 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
2760 (rs6000_reassociation_width): Likewise.
2761 (rs6000_invalid_binary_op): Likewise.
2762 (fusion_p9_p): Likewise.
2763 (emit_fusion_p9_load): Likewise.
2764 (emit_fusion_p9_store): Likewise.
2765 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
2767 (riscv_hard_regno_mode_ok_p): Likewise.
2768 (riscv_address_insns): Likewise.
2769 (riscv_split_symbol): Likewise.
2770 (riscv_legitimize_move): Likewise.
2771 (riscv_function_value): Likewise.
2772 (riscv_hard_regno_nregs): Likewise.
2773 (riscv_expand_builtin): Likewise.
2774 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
2775 (riscv_build_integer): Likewise.
2776 (riscv_split_integer): Likewise.
2777 (riscv_legitimate_constant_p): Likewise.
2778 (riscv_cannot_force_const_mem): Likewise.
2779 (riscv_regno_mode_ok_for_base_p): Likewise.
2780 (riscv_valid_base_register_p): Likewise.
2781 (riscv_valid_offset_p): Likewise.
2782 (riscv_valid_lo_sum_p): Likewise.
2783 (riscv_classify_address): Likewise.
2784 (riscv_legitimate_address_p): Likewise.
2785 (riscv_address_insns): Likewise.
2786 (riscv_load_store_insns): Likewise.
2787 (riscv_force_binary): Likewise.
2788 (riscv_split_symbol): Likewise.
2789 (riscv_force_address): Likewise.
2790 (riscv_legitimize_address): Likewise.
2791 (riscv_move_integer): Likewise.
2792 (riscv_legitimize_const_move): Likewise.
2793 (riscv_legitimize_move): Likewise.
2794 (riscv_address_cost): Likewise.
2795 (riscv_subword): Likewise.
2796 (riscv_output_move): Likewise.
2797 (riscv_canonicalize_int_order_test): Likewise.
2798 (riscv_emit_int_order_test): Likewise.
2799 (riscv_function_arg_boundary): Likewise.
2800 (riscv_pass_mode_in_fpr_p): Likewise.
2801 (riscv_pass_fpr_single): Likewise.
2802 (riscv_pass_fpr_pair): Likewise.
2803 (riscv_get_arg_info): Likewise.
2804 (riscv_function_arg): Likewise.
2805 (riscv_function_arg_advance): Likewise.
2806 (riscv_arg_partial_bytes): Likewise.
2807 (riscv_function_value): Likewise.
2808 (riscv_pass_by_reference): Likewise.
2809 (riscv_setup_incoming_varargs): Likewise.
2810 (riscv_print_operand): Likewise.
2811 (riscv_elf_select_rtx_section): Likewise.
2812 (riscv_save_restore_reg): Likewise.
2813 (riscv_for_each_saved_reg): Likewise.
2814 (riscv_register_move_cost): Likewise.
2815 (riscv_hard_regno_mode_ok_p): Likewise.
2816 (riscv_hard_regno_nregs): Likewise.
2817 (riscv_class_max_nregs): Likewise.
2818 (riscv_memory_move_cost): Likewise.
2819 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
2820 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
2821 (rl78_addr_space_address_mode): Likewise.
2822 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2824 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2825 (rs6000_reassociation_width): Likewise.
2826 (rs6000_invalid_binary_op): Likewise.
2827 (fusion_p9_p): Likewise.
2828 (emit_fusion_p9_load): Likewise.
2829 (emit_fusion_p9_store): Likewise.
2830 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
2831 (ok_for_simple_move_operands): Likewise.
2832 (ok_for_simple_move_strict_operands): Likewise.
2833 (ok_for_simple_arith_logic_operands): Likewise.
2834 (visium_legitimize_reload_address): Likewise.
2835 (visium_select_cc_mode): Likewise.
2836 (output_cbranch): Likewise.
2837 (visium_split_double_move): Likewise.
2838 (visium_expand_copysign): Likewise.
2839 (visium_expand_int_cstore): Likewise.
2840 (visium_expand_fp_cstore): Likewise.
2841 * config/visium/visium.c (visium_pass_by_reference): Likewise.
2842 (visium_function_arg): Likewise.
2843 (visium_function_arg_advance): Likewise.
2844 (visium_libcall_value): Likewise.
2845 (visium_setup_incoming_varargs): Likewise.
2846 (visium_legitimate_constant_p): Likewise.
2847 (visium_legitimate_address_p): Likewise.
2848 (visium_legitimize_address): Likewise.
2849 (visium_secondary_reload): Likewise.
2850 (visium_register_move_cost): Likewise.
2851 (visium_memory_move_cost): Likewise.
2852 (prepare_move_operands): Likewise.
2853 (ok_for_simple_move_operands): Likewise.
2854 (ok_for_simple_move_strict_operands): Likewise.
2855 (ok_for_simple_arith_logic_operands): Likewise.
2856 (visium_function_value_1): Likewise.
2857 (rtx_ok_for_offset_p): Likewise.
2858 (visium_legitimize_reload_address): Likewise.
2859 (visium_split_double_move): Likewise.
2860 (visium_expand_copysign): Likewise.
2861 (visium_expand_int_cstore): Likewise.
2862 (visium_expand_fp_cstore): Likewise.
2863 (visium_split_cstore): Likewise.
2864 (visium_select_cc_mode): Likewise.
2865 (visium_split_cbranch): Likewise.
2866 (output_cbranch): Likewise.
2867 (visium_print_operand_address): Likewise.
2868 * expmed.c (flip_storage_order): Likewise.
2869 * expmed.h (emit_cstore): Likewise.
2870 (flip_storage_order): Likewise.
2871 * genrecog.c (validate_pattern): Likewise.
2872 * hsa-gen.c (gen_hsa_addr): Likewise.
2873 * internal-fn.c (expand_arith_overflow): Likewise.
2874 * ira-color.c (allocno_copy_cost_saving): Likewise.
2875 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2876 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
2877 (process_invariant_for_inheritance): Likewise.
2878 * lra-eliminations.c (move_plus_up): Likewise.
2879 * omp-low.c (lower_oacc_reductions): Likewise.
2880 * simplify-rtx.c (simplify_subreg): Likewise.
2881 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
2882 (TARGET_CHKP_BOUND_MODE): Likewise..
2883 * targhooks.c (default_chkp_bound_mode): Likewise.
2884 (default_setup_incoming_vararg_bounds): Likewise.
2885 * targhooks.h (default_chkp_bound_mode): Likewise.
2886 (default_setup_incoming_vararg_bounds): Likewise.
2887 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
2888 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
2889 (have_whole_vector_shift): Likewise.
2890 * tree-vect-stmts.c (vectorizable_load): Likewise.
2891 * doc/tm.texi: Regenerate.
2893 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2895 Graceful degrade if Binutils PR21472 is not available.
2898 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
2899 .rodata in flash test fails.
2900 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
2901 * confgure: Regenerate.
2902 * config.in: Regenerate.
2903 * config/avr/avr.c (avr_asm_named_section)
2904 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
2905 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
2906 (avr_asm_init_sections): Same.
2908 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2910 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
2911 (fma<VH:mode>4_intrinsic): Likewise.
2912 (*fmsub<VCVTF:mode>4): Likewise.
2913 (*fmsub<VH:mode>4_intrinsic): Likewise.
2915 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2918 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
2919 Don't depend on "optimize > 0".
2920 (out_movhi_r_mr, out_movqi_mr_r): Same.
2921 (out_movhi_mr_r, out_movqi_r_mr): Same.
2922 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
2923 io_address_operand on "optimize > 0".
2925 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2927 * tree-loop-distribution.c: Add general explanantion on the pass.
2928 (generate_loops_for_partition): Mark distributed loop.
2929 (pg_add_dependence_edges): New parameter. Handle alias data
2930 dependence specially and record it in the parameter if asked.
2931 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
2932 (init_partition_graph_vertices, add_partition_graph_edge): New.
2933 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
2934 (free_partition_graph_vdata, build_partition_graph): New.
2935 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
2936 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
2937 (data_ref_segment_size, latch_dominated_by_data_ref): New.
2938 (compute_alias_check_pairs, version_loop_by_alias_check): New.
2939 (version_for_distribution_p, finalize_partitions): New.
2940 (distribute_loop): Handle alias data dependence specially. Factor
2941 out loop fusion code as functions and call these functions.
2943 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2945 * tree-loop-distribution.c (classify_partition): New parameter and
2946 better handle reduction statement.
2947 (rdg_build_partitions): Revise comment.
2948 (distribute_loop): Compute statements in all partitions and pass it
2949 to classify_partition.
2951 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2953 * tree-loop-distribution.c (enum partition_type): New.
2954 (struct partition): New field type.
2955 (partition_merge_into): Add parameter. Update partition type.
2956 (data_dep_in_cycle_p, update_type_for_merge): New functions.
2957 (build_rdg_partition_for_vertex): Compute partition type.
2958 (rdg_build_partitions): Dump partition type.
2959 (distribute_loop): Update calls to partition_merge_into.
2961 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2963 * tree-loop-distribution.c (struct ddr_hasher): New.
2964 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
2966 (classify_partition): Call get_data_dependence.
2967 (pg_add_dependence_edges): Ditto.
2968 (distribute_loop): Release data dependence hash table.
2970 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2972 * tree-loop-distribution.c (ref_base_address): Delete.
2973 (similar_memory_accesses): Rename ...
2974 (share_memory_accesses): ... to this. Check if partitions access
2975 the same memory reference.
2976 (distribute_loop): Call share_memory_accesses.
2978 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2980 * tree-loop-distribution.c (struct partition): New field recording
2982 (partition_alloc, partition_free): Init and release data refs.
2983 (partition_merge_into): Merge data refs.
2984 (build_rdg_partition_for_vertex): Collect data refs for partition.
2985 (pg_add_dependence_edges): Change parameters from vector to bitmap.
2987 (distribute_loop): Remve data refs from vertice data of partition
2990 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2992 * tree-loop-distribution.c (params.h): Include header file.
2993 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
2994 (datarefs_vec): New global var.
2995 (create_rdg_vertices): Use datarefs_vec directly.
2996 (free_rdg): Don't free data references.
2997 (build_rdg): Update use. Don't free data references.
2998 (distribute_loop): Compute global variable for data references.
2999 Bail out if there are too many data references.
3001 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3003 * tree-loop-distribution.c (loop_nest): New global var.
3004 (build_rdg): Use loop directly, rather than loop nest.
3005 (pg_add_dependence_edges): Remove loop nest parameter. Use global
3007 (distribute_loop): Compute global variable loop nest. Update use.
3009 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3011 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
3012 (partition_merge_into): New parameter. Dump reason for fusion.
3013 (distribute_loop): Update use of partition_merge_into.
3015 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3017 * tree-loop-distribution.c (bb_top_order_index): New.
3018 (bb_top_order_index_size, bb_top_order_cmp): New.
3019 (stmts_from_loop): Use topological order.
3020 (pass_loop_distribution::execute): Compute and release topological
3021 order for basic blocks.
3023 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3025 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
3028 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3030 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
3031 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
3032 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
3033 * internal-fn.def (LOOP_DIST_ALIAS): New.
3034 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
3035 (fold_loop_internal_call): ... this.
3036 (vect_loop_dist_alias_call): New function.
3037 (set_uid_loop_bbs): Call fold_loop_internal_call.
3038 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
3041 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
3044 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
3045 Require dead FLAGS_REG at the beginning of a peephole.
3047 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
3050 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
3051 arguments in the call to __builtin_ia32_sbb_u32.
3052 (_subborrow_u64): Swap _X and _Y arguments in the call to
3053 __builtin_ia32_sbb_u64.
3055 2017-07-04 Jakub Jelinek <jakub@redhat.com>
3058 * tree-vrp.c (compare_assert_loc): Turn into a function template
3059 with stable template parameter. Only test if a->e is NULL,
3060 !a->e == !b->e has been verified already. Use e == NULL or
3061 e != NULL instead of e or ! e tests. If stable is true, don't use
3062 iterative_hash_expr, on the other side allow a or b or both NULL
3063 and sort the NULLs last.
3064 (process_assert_insertions): Sort using compare_assert_loc<false>
3065 instead of compare_assert_loc, later sort using
3066 compare_assert_loc<true> before calling process_assert_insertions_for
3067 in a loop. Use break instead of continue once seen NULL pointer.
3069 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
3071 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
3072 Cortex-R7 and Cortex-R8 processors.
3074 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
3076 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
3077 uninitialized while src is not.
3079 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
3081 * common/config/arm/arm-common.c: Adjust include path for
3083 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
3084 (arm-cpu.h): Create in build directory. Adjust dependency rules.
3085 (arm-cpu-data.h): Likewise.
3086 (arm-cpu-cdata.h): Likewise.
3087 * config/arm/arm-cpu.h: Delete.
3088 * config/arm/arm-cpu-cdata.h: Delete.
3089 * config/arm/arm-cpu-data.h: Delete.
3091 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
3093 * config/arm/arm-cpus.in (cortex-a55): New.
3094 (cortex-a75): Likewise.
3095 (cortex-a75.cortex-a55): Likewise.
3096 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
3098 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
3099 * config/arm/arm-cpu-cdata.h: Regenerate.
3100 * config/arm/arm-cpu-data.h: Regenerate.
3101 * config/arm/arm-cpu.h: Regenerate.
3102 * config/arm/arm-tables.opt: Regenerate.
3103 * config/arm/arm-tune.md: Regenerate.
3105 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
3107 * haifa-sched.c (sched_create_recovery_edges): Update profile.
3109 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
3111 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
3114 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
3116 PR tree-optimization/81292
3117 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
3118 full_string_p, also call adjust_related_strinfos if the adjustment
3119 is simple, otherwise invalidate related strinfos.
3121 2017-07-04 Martin Liska <mliska@suse.cz>
3124 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
3125 newly created variable as DECL_IGNORED_P.
3127 2017-07-04 Martin Liska <mliska@suse.cz>
3130 * ipa-inline.c (inline_small_functions):
3131 Use xstrdup_for_dump.
3133 2017-07-04 Tom de Vries <tom@codesourcery.com>
3135 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
3137 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
3140 * config/darwin.c (darwin_function_switched_text_sections):
3143 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
3145 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
3147 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
3149 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
3151 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3153 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
3154 min_profitable_iters, and th as inclusive lower bounds.
3155 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
3156 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
3157 for min_profitable_iters and min_profitable_estimate.
3158 (vect_transform_loop): Treat th as an inclusive lower bound.
3159 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
3161 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
3164 * config/darwin.c (darwin_function_switched_text_sections):
3165 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
3166 in two pieces, and suppress the use of buf.
3168 2017-07-03 Nathan Sidwell <nathan@acm.org>
3170 * hash-table.h (hash_table_mod1): Fix indentation.
3172 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3175 * predict.c (force_edge_cold): Be more careful about propagation
3177 * profile-count.h (profile_probability::guessed,
3178 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
3180 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
3182 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
3184 * doc/invoke.texi (rcpc architecture extension): Document it.
3186 2017-07-03 Richard Biener <rguenther@suse.de>
3188 PR tree-optimization/60510
3189 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
3190 the scalar reduction PHI and use it.
3191 (vectorizable_reduction): Properly guard the single_defuse_cycle
3192 path for non-SLP reduction chains where we cannot use it.
3193 Rework reduc_def/index and vector type deduction. Rework
3194 vector operand gathering during reduction op code-gen.
3195 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
3196 chains dissolve the chain and leave it to non-SLP reduction
3199 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3201 * tree-data-ref.h (dr_alignment): Declare.
3202 * tree-data-ref.c (dr_alignment): New function.
3203 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
3204 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
3206 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
3208 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3210 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
3211 and base_misalignment fields.
3212 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
3213 * tree-data-ref.c: Include builtins.h.
3214 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
3215 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
3216 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
3217 * tree-vect-data-refs.c: Include tree-cfg.h.
3218 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
3219 fields instead of calculating an alignment here.
3220 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
3221 innermost_loop_behavior fields.
3223 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3225 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
3227 (DR_STEP_ALIGNMENT): New macro.
3228 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
3229 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
3230 (create_data_ref): Print it.
3231 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
3232 to tell whether the step preserves vector (mis)alignment.
3233 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
3234 Move the check for an integer step and generalise to all INTEGER_CST.
3235 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
3236 Print the outer step alignment.
3238 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3240 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
3241 with offset_alignment.
3242 (DR_ALIGNED_TO): Delete.
3243 (DR_OFFSET_ALIGNMENT): New macro.
3244 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
3245 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
3246 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
3247 (create_data_ref): Likewise.
3248 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
3249 (vect_analyze_data_refs): Likewise.
3250 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
3251 creating dummy innermost behavior.
3253 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3255 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
3256 with a "innermost_loop_behavior *" and refeence tree.
3257 * tree-data-ref.c (dr_analyze_innermost): Likewise.
3258 (create_data_ref): Update call accordingly.
3259 * tree-predcom.c (find_looparound_phi): Likewise.
3261 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3263 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
3264 fields with dr_wrt_vec_loop.
3265 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
3266 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
3267 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
3268 (vect_dr_behavior): New function.
3269 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
3270 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
3271 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
3272 track whether the step preserves the misalignment.
3273 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
3274 Use vect_dr_behavior.
3275 (vect_setup_realignment): Update call accordingly.
3276 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
3277 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
3278 call to vect_create_addr_base_for_vector_ref.
3279 (vect_create_cond_for_align_checks): Likewise.
3280 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
3281 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
3282 (vect_recog_mask_conversion_pattern): Likewise.
3283 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
3284 (new_stmt_vec_info): Remove redundant zeroing.
3286 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
3288 * common/config/arm/arm-common.c (arm_be8_option): New function.
3289 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
3290 (ISA_ARMv6): Add isa_bit_be8.
3291 * config/arm/arm.h (arm_be8_option): Add prototype.
3292 (BE8_SPEC_FUNCTION): New define.
3293 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
3294 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
3295 (mlittle-endian): Similarly.
3296 (mbe8, mbe32): New options.
3297 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
3298 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
3300 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3302 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
3304 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3306 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
3307 (cleanup_tree_cfg_bb): Use it.
3308 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
3310 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
3312 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3315 * loop-doloop.c (add_test): Update profile.
3317 2017-07-03 Martin Liska <mliska@suse.cz>
3320 * sanopt.c (rewrite_usage_of_param): New function.
3321 (sanitize_rewrite_addressable_params): Likewise.
3322 (pass_sanopt::execute): Call rewrite_usage_of_param.
3324 2017-07-03 Richard Biener <rguenther@suse.de>
3326 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
3327 back to using VIEW_CONVERT_EXPR.
3329 2017-07-03 Martin Liska <mliska@suse.cz>
3332 * doc/extend.texi: Document when a resolver function is
3333 generated for target_clones.
3335 2017-07-03 Martin Liska <mliska@suse.cz>
3337 * asan.c (asan_emit_stack_protection): Unpoison just red zones
3338 and shadow memory of auto variables which are subject of
3339 use-after-scope sanitization.
3340 (asan_expand_mark_ifn): Add do set only when is_poison.
3342 2016-07-03 Richard Biener <rguenther@suse.de>
3344 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
3346 (vect_force_simple_reduction): Record reduction def -> phi mapping.
3347 (vectorizable_reduction): Perform reduction PHI creation when
3348 visiting a reduction PHI and adjust and simplify code generation
3349 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
3350 (vect_transform_loop): Visit reduction PHIs.
3351 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
3352 defs into the SLP tree.
3353 (vect_build_slp_tree): Reduction defs terminate the recursion.
3354 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
3356 (vect_get_vec_defs_for_stmt_copy): Export.
3357 (vect_get_vec_defs): Likewise.
3358 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
3360 (vect_get_vec_defs_for_stmt_copy): Declare.
3361 (vect_get_vec_defs): Likewise.
3363 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3365 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
3366 parameter with a "loop" parameter and use it instead of the
3367 loop containing DR_STMT. Don't check simple_iv when doing
3368 BB analysis. Describe the two analysis modes in the comment.
3370 2017-07-03 Tom de Vries <tom@codesourcery.com>
3372 PR tree-optimization/69468
3373 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
3374 (find_same_succ_bb): Handle ignore_edge_flags.
3376 2017-07-03 Tom de Vries <tom@codesourcery.com>
3378 PR tree-optimization/81192
3379 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
3381 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
3383 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
3385 2017-07-03 Tom de Vries <tom@codesourcery.com>
3387 PR tree-optimization/81192
3388 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
3389 BB_SAME_SUCC (bb) == NULL.
3391 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3393 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
3396 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3398 * dumpfile.c: Include profile-count.h
3399 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
3401 (insert_cond_bb): Update profile.
3402 * tree-cfg.h (insert_cond_bb): Update prototype.
3403 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
3404 * tree-dump.c: Do not include tree-cfg.
3406 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3408 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
3410 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3412 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
3415 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3417 * tree-complex.c (expand_complex_div_wide): update profile.
3419 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3420 Alan Hayward <alan.hayward@arm.com>
3421 David Sherwood <david.sherwood@arm.com>
3423 * Makefile.in (MACHMODE_H): Remove insn-modes.h
3424 (CORETYPES_H): New define.
3425 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
3426 (insn-modes-inline.h, s-modes-inline-h): New rules.
3427 (generated_files): Add insn-modes-inline.h.
3428 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
3429 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
3430 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
3431 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
3432 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
3433 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
3434 (build/gencodes.o, build/genconditions.o): Likewise.
3435 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
3436 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
3437 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
3438 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
3439 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
3440 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
3441 * coretypes.h: Include everything up to real.h for generators.
3442 Include insn-modes.h first. Include wide-int-print.h after
3443 wide-int.h. Include insn-modes-inline.h and then machmode.h.
3444 * machmode.h: Don't include insn-modes.h here.
3445 * function-tests.c: Remove includes of signop.h, machmode.h,
3446 double-int.h and wide-int.h.
3448 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
3450 * optc-save-gen.awk: Likewise.
3451 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
3452 * godump.c: Remove include of wide-int-print.h.
3453 * pretty-print.h: Likewise.
3454 * wide-int-print.cc: Likewise.
3455 * wide-int.cc: Likewise.
3456 * hash-map-tests.c: Remove include of signop.h.
3457 * hash-set-tests.c: Likewise.
3458 * rtl-tests.c: Likewise.
3459 * mkconfig.sh: Remove include of machmode.h.
3460 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
3462 (emit_insn_modes_inline_h): ...this new function. Emit the code
3463 into an insn-modes-inline.h header file, adding appropriate
3464 include guards and end comments.
3465 (emit_insn_modes_c_header): Remove include of machmode.h.
3466 (emit_min_insn_modes_c_header): Include coretypes.h rather than
3468 (main): Handle -i flag and call emit_insn_modes_inline_h when
3471 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3473 * tree-ssa-strlen.c (strinfo): Rename the length field to
3474 nonzero_chars. Add a full_string_p field.
3475 (compare_nonzero_chars, zero_length_string_p): New functions.
3476 (get_addr_stridx): Add an offset_out parameter.
3477 Use compare_nonzero_chars.
3478 (get_stridx): Update accordingly. Use compare_nonzero_chars.
3479 (new_strinfo): Update after above changes to strinfo.
3480 (set_endptr_and_length): Set full_string_p.
3481 (get_string_length): Update after above changes to strinfo.
3482 (unshare_strinfo): Update call to new_strinfo.
3483 (maybe_invalidate): Likewise.
3484 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
3485 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
3486 as a uhwi instead of an shwi. Update after above changes to
3487 strinfo and new_strinfo.
3488 (zero_length_string): Assert that chainsi contains full strings.
3489 Use zero_length_string_p. Update call to new_strinfo.
3490 (adjust_related_strinfos): Update after above changes to strinfo.
3491 Copy full_string_p from origsi.
3492 (adjust_last_stmt): Use zero_length_string_p.
3493 (handle_builtin_strlen): Update after above changes to strinfo and
3494 new_strinfo. Install the lhs as the string length if the previous
3495 entry didn't describe a full string.
3496 (handle_builtin_strchr): Update after above changes to strinfo
3498 (handle_builtin_strcpy): Likewise.
3499 (handle_builtin_strcat): Likewise.
3500 (handle_builtin_malloc): Likewise.
3501 (handle_pointer_plus): Likewise.
3502 (handle_builtin_memcpy): Likewise. Track nonzero characters
3503 that aren't necessarily followed by a nul terminator.
3504 (handle_char_store): Likewise.
3506 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3508 PR tree-optimization/80769
3509 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
3510 for malloc and calloc. Document the new invariant that all related
3511 strinfos have delayed lengths or none do.
3512 (verify_related_strinfos): Move earlier in file.
3513 (set_endptr_and_length): New function, split out from...
3514 (get_string_length): ...here. Also set the lengths of related
3516 (zero_length_string): Assert that chainsi has known (rather than
3518 (adjust_related_strinfos): Likewise.
3520 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3522 PR tree-optimization/81136
3523 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
3524 assert that two references with the same misalignment have the same
3525 compile-time misalignment if those compile-time misalignments
3528 2017-07-01 Andi Kleen <ak@linux.intel.com>
3530 * print-tree.c (print_node): Print all attributes.
3532 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3534 * cfg.c (scale_bbs_frequencies): New function.
3535 * cfg.h (scale_bbs_frequencies): Declare it.
3536 * cfgloopanal.c (single_likely_exit): Cleanup.
3537 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
3539 (scale_loop_profile): Likewise.
3540 (loop_version): Likewise.
3541 (create_empty_loop_on_edge): Update.
3542 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
3543 scale_loop_frequencies, scale_loop_profile, loopify,
3544 loop_version): Update prototypes.
3545 * modulo-sched.c (sms_schedule): Update.
3546 * predict.c (unlikely_executed_edge_p): Also check probability.
3547 (probably_never_executed_edge_p): Fix typo.
3548 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3549 * tree-parloops.c (gen_parallel_loop): Update.
3550 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
3551 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3552 * tree-ssa-loop-split.c (split_loop): Update.
3553 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3554 * tree-vect-loop-manip.c (vect_do_peeling): Update.
3555 (vect_loop_versioning): Update.
3556 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3558 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3560 * trans-mem.c (split_bb_make_tm_edge): Update profile.
3562 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3564 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
3565 to keep profile consistent.
3567 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3569 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
3570 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
3571 * profile-count.h (max_safe_multiplier): Make unsigned.
3572 (profile_count::guessed_zero): New.
3574 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3576 * bb-reorder.c (fix_up_crossing_landing_pad,
3577 fix_crossing_conditional_branches): Use make_single_succ_edge
3578 to keep profile consistent.
3580 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3582 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
3585 2017-07-01 Jakub Jelinek <jakub@redhat.com>
3588 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
3589 the right scopes, make sure cond_jump isn't preserved between multiple
3590 iterations. Search for fallthru edge whenever there are 3+ edges and
3591 use find_fallthru_edge for it.
3593 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3595 Patch by Alexander Monakov <amonakov@ispras.ru>
3596 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
3597 probabilities consistently.
3599 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3601 * pa.c (pa_expand_compare_and_swap_loop): Update call of
3602 emit_cmp_and_jump_insns.
3604 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3607 * tree-inline.c (expand_call_inline): Combine profile statuses.
3609 2017-06-30 Andrew Pinski <apinski@cavium.com>
3611 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
3612 fold_stmt returned true.
3614 2017-06-30 Nathan Sidwell <nathan@acm.org>
3616 * ggc.h (empty_string): Delete.
3617 * cfgexpand.c (expand_asm_stmt): Use plain "".
3618 * optabs.c (expand_asm_memory_barrier): Likewise.
3619 * stringpool.c (empty_string): Delete.
3620 (digit_vector, digit_string): Delete.
3621 (ggc_alloc_string): Use plain "", don't optimize single digit
3622 strings. Use ggc_alloc_atomic.
3624 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
3626 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
3627 comparison set and one other set, use the cost of the non-comparison
3630 2017-06-30 Nathan Sidwell <nathan@acm.org>
3632 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
3635 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
3637 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
3638 loops. Remove now unneeded calls to gimple_switch_set_label() that
3639 just set removed labels to NULL_TREE.
3641 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
3643 * tree-ssanames.c (set_range_info_raw): Abstract from ...
3644 (set_range_info): ...here. Only call set_range_info_raw if domain
3646 (set_nonzero_bits): Call set_range_info_raw.
3647 * tree-ssanames.h (set_range_info_raw): New.
3649 2017-06-30 Jakub Jelinek <jakub@redhat.com>
3652 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
3653 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
3654 of nonimmediate_operand and <store_mask_constraint> instead of m
3655 for the input operand. For V8FI iterator, always split if input
3656 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
3657 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
3658 <store_mask_predicate> instead of register_operand and
3659 <store_mask_constraint> instead of v for the input operand. Make
3660 sure both operands aren't MEMs for if not <mask_applied>.
3662 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
3664 * lto-wrapper.c (copy_file) Close both file descriptors before
3667 2017-06-30 Martin Liska <mliska@suse.cz>
3670 * multiple_target.c (create_dispatcher_calls): Make ifunc
3671 also for function that don't have calls or are not referenced.
3673 2017-06-30 Richard Biener <rguenther@suse.de>
3675 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
3676 analyze the first scalar stmt. Move vector type computation
3677 for the BB case here from ...
3678 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
3679 live operation processing in the SLP case properly.
3681 2017-06-30 Richard Biener <rguenther@suse.de>
3683 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
3685 2017-06-30 Martin Liska <mliska@suse.cz>
3688 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
3689 before BUILT_IN_UNWIND_RESUME when ASAN is used.
3691 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
3693 * doc/invoke.texi (AArch64): Add missing options and remove redundant
3696 2017-06-30 Richard Biener <rguenther@suse.de>
3698 PR tree-optimization/81249
3699 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
3700 condition reduction result to original scalar type.
3702 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3704 * profile-count.h (enum profile_quality): Fix typos and whitespace
3707 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3709 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
3710 type for branch probabilities.
3712 2017-06-29 Julian Brown <julian@codesourcery.com>
3713 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3715 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
3716 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
3717 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
3718 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
3720 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3722 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
3723 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
3724 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
3725 CC usage from generic code to here.
3726 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
3727 CC usage into the target macros.
3729 2017-06-29 Maya Rashish <coypu@sdf.org>
3731 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
3734 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3736 * arm/arm-builtins.c: Include profile-count.h
3737 * except.c (sjlj_emit_function_enter): Use
3738 profile_probability::unlikely.
3740 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3742 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
3743 and tocrel_offset be pointer args rather than implicitly using
3745 (legitimate_constant_pool_address_p, rs6000_emit_move,
3746 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
3747 tocrel_offset and use in toc_relative_expr_p call.
3748 (print_operand, print_operand_address): Use static tocrel_base_oac
3749 and tocrel_offset_oac.
3750 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
3753 2017-06-29 Maya Rashish <coypu@sdf.org>
3755 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
3757 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
3759 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
3760 objects, take into account only the alignment of 'op0' and 'mode1' if
3763 2017-06-29 Steve Ellcey <sellcey@cavium.com>
3765 * ccmp.c (ccmp_tree_comparison_p): New function.
3766 (ccmp_candidate_p): Update to use above function.
3767 (get_compare_parts): New function.
3768 (expand_ccmp_next): Update to use new functions.
3769 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
3771 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
3772 take mode as argument.
3773 * ccmp.h (expand_ccmp_expr): Add mode as argument.
3774 * expr.c (expand_expr_real_1): Pass mode as argument.
3776 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
3778 * combine.c (combine_instructions): Print insns to dump_file, together
3781 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3783 * asan.c (asan_emit_stack_protection): Update.
3784 (create_cond_insert_point): Update.
3785 * auto-profile.c (afdo_propagate_circuit): Update.
3786 * basic-block.h (struct edge_def): Turn probability to
3787 profile_probability.
3788 (EDGE_FREQUENCY): Update.
3789 * bb-reorder.c (find_traces_1_round): Update.
3790 (better_edge_p): Update.
3791 (sanitize_hot_paths): Update.
3792 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
3793 (make_single_succ_edge): Update.
3794 (check_bb_profile): Update.
3795 (dump_edge_info): Update.
3796 (update_bb_profile_for_threading): Update.
3797 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
3798 probabilitycount to 0.
3799 * cfgbuild.c (compute_outgoing_frequencies): Update.
3800 * cfgcleanup.c (try_forward_edges): Update.
3801 (outgoing_edges_match): Update.
3802 (try_crossjump_to_edge): Update.
3803 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
3804 (expand_gimple_tailcall): Update.
3805 (construct_init_block): Use make_single_succ_edge.
3806 (construct_exit_block): Use make_single_succ_edge.
3807 * cfghooks.c (verify_flow_info): Update.
3808 (redirect_edge_succ_nodup): Update.
3809 (split_edge): Update.
3810 (account_profile_record): Update.
3811 * cfgloopanal.c (single_likely_exit): Update.
3812 * cfgloopmanip.c (scale_loop_profile): Update.
3813 (set_zero_probability): Remove.
3814 (duplicate_loop_to_header_edge): Update.
3815 * cfgloopmanip.h (loop_version): Update prototype.
3816 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3817 (force_nonfallthru_and_redirect): Update.
3818 (update_br_prob_note): Update.
3819 (rtl_verify_edges): Update.
3820 (purge_dead_edges): Update.
3821 (rtl_lv_add_condition_to_bb): Update.
3822 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
3823 * cgraphunit.c (init_lowered_empty_function): Update.
3824 (cgraph_node::expand_thunk): Update.
3825 * cilk-common.c: Include profile-count.h
3826 * dojump.c (inv): Remove.
3827 (jumpifnot): Update.
3828 (jumpifnot_1): Update.
3829 (do_jump_1): Update.
3831 (do_jump_by_parts_greater_rtx): Update.
3832 (do_compare_rtx_and_jump): Update.
3833 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
3834 do_jump_1. do_compare_rtx_and_jump): Update prototype.
3835 * dwarf2cfi.c: Include profile-count.h
3836 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
3837 (sjlj_emit_dispatch_table): Likewise.
3838 * explow.c: Include profile-count.h
3839 * expmed.c (emit_store_flag_force): Update.
3840 (do_cmp_and_jump): Update.
3841 * expr.c (compare_by_pieces_d::generate): Update.
3842 (compare_by_pieces_d::finish_mode): Update.
3843 (emit_block_move_via_loop): Update.
3844 (store_expr_with_bounds): Update.
3845 (store_constructor): Update.
3846 (expand_expr_real_2): Update.
3847 (expand_expr_real_1): Update.
3848 * expr.h (try_casesi, try_tablejump): Update prototypes.
3849 * gimple-pretty-print.c (dump_probability): Update.
3850 (dump_profile): New.
3851 (dump_gimple_label): Update.
3852 (dump_gimple_bb_header): Update.
3853 * graph.c (draw_cfg_node_succ_edges): Update.
3854 * hsa-gen.c (convert_switch_statements): Update.
3855 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
3856 (find_if_case_1): Update.
3857 (find_if_case_2): Update.
3858 * internal-fn.c (expand_arith_overflow_result_store): Update.
3859 (expand_addsub_overflow): Update.
3860 (expand_neg_overflow): Update.
3861 (expand_mul_overflow): Update.
3862 (expand_vector_ubsan_overflow): Update.
3863 * ipa-cp.c (good_cloning_opportunity_p): Update.
3864 * ipa-split.c (split_function): Use make_single_succ_edge.
3865 * ipa-utils.c (ipa_merge_profiles): Update.
3866 * loop-doloop.c (add_test): Update.
3867 (doloop_modify): Update.
3868 * loop-unroll.c (compare_and_jump_seq): Update.
3869 (unroll_loop_runtime_iterations): Update.
3870 * lra-constraints.c (lra_inheritance): Update.
3871 * lto-streamer-in.c (input_cfg): Update.
3872 * lto-streamer-out.c (output_cfg): Update.
3873 * mcf.c (adjust_cfg_counts): Update.
3874 * modulo-sched.c (sms_schedule): Update.
3875 * omp-expand.c (expand_omp_for_init_counts): Update.
3876 (extract_omp_for_update_vars): Update.
3877 (expand_omp_ordered_sink): Update.
3878 (expand_omp_for_ordered_loops): Update.
3879 (expand_omp_for_generic): Update.
3880 (expand_omp_for_static_nochunk): Update.
3881 (expand_omp_for_static_chunk): Update.
3882 (expand_cilk_for): Update.
3883 (expand_omp_simd): Update.
3884 (expand_omp_taskloop_for_outer): Update.
3885 (expand_omp_taskloop_for_inner): Update.
3886 * omp-simd-clone.c (simd_clone_adjust): Update.
3887 * optabs.c (expand_doubleword_shift): Update.
3888 (expand_abs): Update.
3889 (emit_cmp_and_jump_insn_1): Update.
3890 (expand_compare_and_swap_loop): Update.
3891 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
3892 * predict.c (predictable_edge_p): Update.
3893 (edge_probability_reliable_p): Update.
3894 (set_even_probabilities): Update.
3895 (combine_predictions_for_insn): Update.
3896 (combine_predictions_for_bb): Update.
3897 (propagate_freq): Update.
3898 (estimate_bb_frequencies): Update.
3899 (force_edge_cold): Update.
3900 * profile-count.c (profile_count::dump): Add missing space into dump.
3901 (profile_count::debug): Add newline.
3902 (profile_count::differs_from_p): Explicitly convert to unsigned.
3903 (profile_count::stream_in): Update.
3904 (profile_probability::dump): New member function.
3905 (profile_probability::debug): New member function.
3906 (profile_probability::differs_from_p): New member function.
3907 (profile_probability::differs_lot_from_p): New member function.
3908 (profile_probability::stream_in): New member function.
3909 (profile_probability::stream_out): New member function.
3910 * profile-count.h (profile_count_quality): Rename to ...
3911 (profile_quality): ... this one.
3912 (profile_probability): New.
3913 (profile_count): Update.
3914 * profile.c (compute_branch_probabilities): Update.
3915 * recog.c (peep2_attempt): Update.
3916 * sched-ebb.c (schedule_ebbs): Update.
3917 * sched-rgn.c (find_single_block_region): Update.
3918 (compute_dom_prob_ps): Update.
3919 (schedule_region): Update.
3920 * sel-sched-ir.c (compute_succs_info): Update.
3921 * stmt.c (struct case_node): Update.
3922 (do_jump_if_equal): Update.
3923 (get_outgoing_edge_probs): Update.
3924 (conditional_probability): Update.
3925 (emit_case_dispatch_table): Update.
3926 (expand_case): Update.
3927 (expand_sjlj_dispatch_table): Update.
3928 (emit_case_nodes): Update.
3929 * targhooks.c: Update.
3930 * tracer.c (better_p): Update.
3931 (find_best_successor): Update.
3932 * trans-mem.c (expand_transaction): Update.
3933 * tree-call-cdce.c: Update.
3934 * tree-cfg.c (gimple_split_edge): Upate.
3935 (move_sese_region_to_fn): Upate.
3936 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
3937 * tree-eh.c (lower_resx): Upate.
3938 (cleanup_empty_eh_move_lp): Upate.
3939 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3940 * tree-inline.c (copy_edges_for_bb): Update.
3941 (copy_cfg_body): Update.
3942 * tree-parloops.c (gen_parallel_loop): Update.
3943 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
3944 (gimple_gen_time_profiler): Update.
3945 * tree-ssa-dce.c (remove_dead_stmt): Update.
3946 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3947 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3948 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3949 (unloop_loops): Update.
3950 (try_peel_loop): Update.
3951 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3952 * tree-ssa-loop-split.c (connect_loops): Update.
3953 (split_loop): Update.
3954 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3955 (hoist_guard): Update.
3956 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3957 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3958 (value_replacement): Update.
3959 * tree-ssa-reassoc.c (branch_fixup): Update.
3960 * tree-ssa-tail-merge.c (replace_block_by): Update.
3961 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3962 (create_edge_and_update_destination_phis): Update.
3963 (compute_path_counts): Update.
3964 (recompute_probabilities): Update.
3965 (update_joiner_offpath_counts): Update.
3966 (freqs_to_counts_path): Update.
3967 (duplicate_thread_path): Update.
3968 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3969 (struct switch_conv_info): Update.
3970 (gen_inbound_check): Update.
3971 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3972 (vect_do_peeling): Update.
3973 (vect_loop_versioning): Update.
3974 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3975 (optimize_mask_stores): Update.
3976 * ubsan.c (ubsan_expand_null_ifn): Update.
3977 * value-prof.c (gimple_divmod_fixed_value): Update.
3978 (gimple_divmod_fixed_value_transform): Update.
3979 (gimple_mod_pow2): Update.
3980 (gimple_mod_pow2_value_transform): Update.
3981 (gimple_mod_subtract): Update.
3982 (gimple_mod_subtract_transform): Update.
3983 (gimple_ic): Update.
3984 (gimple_stringop_fixed_value): Update.
3985 (gimple_stringops_transform): Update.
3986 * value-prof.h: Update.
3988 2017-06-29 Carl Love <cel@us.ibm.com>
3990 * config/rs6000/rs6000-c.c: Add support for built-in functions
3991 vector signed int vec_signed (vector float);
3992 vector signed long long vec_signed (vector double);
3993 vector signed int vec_signed2 (vector double, vector double);
3994 vector signed int vec_signede (vector double);
3995 vector signed int vec_signedo (vector double);
3996 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
3997 instruction generator.
3998 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3999 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
4000 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
4002 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
4003 vunsignede_v2df): Add define_expands.
4004 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
4005 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
4006 VEC_UNSIGNEDO): Add definitions.
4007 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
4008 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
4009 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
4010 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
4011 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
4012 * config/rs6000/altivec.h (vec_signed, vec_signed2,
4013 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
4014 vec_unsignede, vec_unsignedo): Add builtin defines.
4015 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
4017 * doc/extend.texi: Update the built-in documentation file for the
4018 new built-in functions.
4020 2017-06-29 Richard Biener <rguenther@suse.de>
4022 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
4023 reduction chains to LOOP_VINFO_REDUCTIONS.
4024 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
4025 SLP reductions after processing reduction chains.
4027 2017-06-29 Nathan Sidwell <nathan@acm.org>
4029 * builtins.c (fold_builtin_FUNCTION): Use
4030 lang_hooks.decl_printable_name.
4032 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
4035 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
4036 with only one label.
4037 * stmt.c (expand_case): Assert NCASES is greater than one.
4039 2017-06-29 Richard Biener <rguenther@suse.de>
4041 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
4043 (group_case_labels): Likewise.
4044 (find_taken_edge): Push sanity checking on val to workers...
4045 (find_taken_edge_cond_expr): ... here
4046 (find_taken_edge_switch_expr): ... and here, handle cases
4047 with just a default label.
4048 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
4049 (group_case_labels): Likewise.
4050 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
4051 group_case_labels does anything cleanup the CFG again.
4053 2017-06-29 Bin Cheng <bin.cheng@arm.com>
4055 PR tree-optimization/81196
4056 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
4057 exit condition comparing two IVs.
4059 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
4061 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
4062 profile to the dummy entry at the end of the list of architectures.
4063 * config/arm/arm-cpu-cdata.h: Regenerated.
4065 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4066 Michael Collison <michael.collison@arm.com>
4069 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
4071 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
4072 (*aarch64_reg_<mode>3_minus_mask): New pattern.
4073 (*aarch64_<optab>_reg_di3_mask2): New pattern.
4074 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
4075 of shift when the shift amount is masked with constant equal to
4076 the size of the mode.
4077 * config/aarch64/predicates.md (subreg_lowpart_operator): New
4080 2017-06-29 Martin Liska <mliska@suse.cz>
4082 * config/i386/i386.opt: Change range from [1,5] to [0,5].
4084 2017-06-29 Yury Gribov <tetra2005@gmail.com>
4087 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
4090 (edge_growth_cache_entry::edge_growth_cache_entry): New
4092 (reset_edge_growth_cache): Update to use constructor.
4094 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4096 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
4097 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
4098 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
4100 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
4102 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
4103 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
4105 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
4107 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
4108 (*-linux-uclibc*): Add t-uclibc tmake_file.
4109 * config/t-musl: New.
4110 * config/t-uclibc: New.
4112 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
4114 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
4116 (gen_comm_data): Emit architectural setting of arch_prof.
4117 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
4119 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
4120 (armv8-m.base, armv8-m.main): Likewise.
4121 * arm-protos.h (arm_build_target): Add profile field.
4122 (arch_option): Likewise.
4123 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
4125 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
4126 arm_active_target.profile.
4128 2017-06-28 Richard Biener <rguenther@suse.de>
4131 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
4132 TYPE_OVERFLOW_WRAPS.
4133 * match.pd (negate_expr_p): Likewise.
4134 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
4135 fold_build2, not fold_binary.
4137 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
4139 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
4140 Convert memory address to Pmode.
4141 (aarch64_print_operand): Assert MEM operands are always Pmode.
4143 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
4146 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
4147 Remove redundant if.
4148 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
4149 * config/arm/aarch-common-protos.h
4150 (aarch_forward_to_shift_is_not_shifted_re): Remove.
4151 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
4153 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
4156 * multiple_target.c (create_dispatcher_calls): Set the default
4157 clone to be static, not public.
4159 2017-06-28 Richard Biener <rguenther@suse.de>
4161 * tree-vect-loop.c (vectorizable_reduction): Move special
4162 cond reduction IV var creation ...
4163 (vect_create_epilog_for_reduction): ... here. Remove induction_index
4164 parameter. Use STMT_VINFO_VECTYPE.
4165 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
4168 2017-06-28 Martin Liska <mliska@suse.cz>
4171 * ipa-visibility.c (non_local_p): Handle visibility.
4173 2017-06-28 Martin Liska <mliska@suse.cz>
4176 * common.opt: Add IntegerRange to various options.
4177 * opt-functions.awk (integer_range_info): New function.
4178 * optc-gen.awk: Add integer_range_info to cl_options struct.
4179 * opts-common.c (decode_cmdline_option): Handle
4180 CL_ERR_INT_RANGE_ARG.
4181 (cmdline_handle_error): Likewise.
4182 * opts.c (print_filtered_help): Show valid interval in
4183 when --help is provided.
4184 * opts.h (struct cl_option): Add range_min and range_max fields.
4185 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
4187 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
4189 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
4190 (x * C EQ/NE y * C): New transformation.
4192 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
4194 * genmultilib (combination_space): Accept '+' in option names.
4196 2017-06-28 Martin Liska <mliska@suse.cz>
4199 * asan.c (instrument_derefs): Bail out inner references
4200 that are hard register variables.
4202 2017-06-28 Jakub Jelinek <jakub@redhat.com>
4205 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
4206 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
4208 2017-06-28 Richard Biener <rguenther@suse.de>
4210 * tree-vectorizer.h (vect_get_vec_defs): Remove.
4211 (vect_get_slp_defs): Adjust.
4212 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
4214 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
4216 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4217 get_initial_defs_for_reduction instead of vect_get_vec_defs.
4218 (vectorizable_reduction): Adjust.
4219 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
4221 (vect_get_slp_defs): Likewise.
4222 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
4223 (vectorizable_bswap): Adjust.
4224 (vectorizable_call): Likewise.
4225 (vectorizable_conversion): Likewise.
4226 (vectorizable_assignment): Likewise.
4227 (vectorizable_shift): Likewise.
4228 (vectorizable_operation): Likewise.
4229 (vectorizable_store): Likewise.
4230 (vectorizable_condition): Likewise.
4231 (vectorizable_comparison): Likewise.
4233 2017-06-28 Michael Collison <michael.collison@arm.com>
4236 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
4238 (arm_gen_movmemqi): Removed unused variable 'i'.
4239 Convert 'for' loop into 'while' loop.
4240 (arm_expand_prologue): Remove last unnecessary set of insn.
4241 (thumb_pop): Remove unused variable 'pushed_words'.
4242 (thumb_exit): Remove last unnecessary set of regs_to_pop.
4244 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4246 * config/s390/predicates.md: Use s390_rel_address_ok_p.
4247 * config/s390/s390-protos.h: Add prototype of
4248 s390_rel_address_ok_p.
4249 * config/s390/s390.c (s390_got_symbol): New function.
4250 (s390_rel_address_ok_p): New function.
4251 (legitimize_pic_address): Use s390_rel_address_ok_p.
4252 (s390_load_got): Use s390_got_symbol.
4253 (s390_option_override): Issue error if
4254 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
4255 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
4257 * config/s390/s390.opt: New option mpic-data-is-text-relative.
4259 2017-06-27 Andrew Pinski <apinski@cavium.com>
4261 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
4262 (X * copysign (1.0, X)): New pattern.
4263 (X * copysign (1.0, -X)): New pattern.
4264 (copysign (-1.0, CST)): New pattern.
4266 2017-06-27 Joseph Myers <joseph@codesourcery.com>
4268 * genmultilib (combination_space): Remove variable.
4269 Validate reuse rules against regular expression for any sequence
4270 of multilib options in any order.
4272 2017-06-27 Michael Collison <michael.collison@arm.com>
4274 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
4275 call aarch64_split_simd_combine.
4276 * (aarch64_combine_internal<mode>): Delete pattern.
4277 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
4278 Allow register and subreg operands.
4280 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4282 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
4283 specific need, just fallback on defaults.
4284 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
4286 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4287 Olivier Hainque <hainque@adacore.com>
4289 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
4291 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
4292 targets. Pick a default if no particular attempt applied.
4293 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
4296 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4298 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
4299 (x86_64-wrs-vxworks7): Likewise.
4301 2017-06-27 Marek Polacek <polacek@redhat.com>
4304 * ubsan.c (instrument_null): Check get_base_address's result for null.
4306 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
4308 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
4310 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
4312 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
4313 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
4314 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
4315 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
4317 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
4318 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
4319 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
4320 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
4321 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
4322 BUILT_IN_FEUPDATEENV): New builtins.
4323 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
4324 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
4325 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
4326 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
4328 (builtin_structptr_types): Adjust size.
4329 * tree.c (builtin_structptr_types): Add four entries.
4331 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4332 Olivier Hainque <hainque@adacore.com>
4334 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
4335 (TLS_SYM): New local macro, forcing reference to __tls__ on
4336 link command lines for VxWorks 7 RTPs, triggering initialization
4338 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
4339 OS features TLS support, true for RTPs on VxWorks 7.
4340 * config/vxworks.c (vxworks_override_options): Setup emutls
4343 2017-06-27 Jakub Jelinek <jakub@redhat.com>
4345 * predict.c (test_prediction_value_range): Use -1U instead of -1
4346 to avoid narrowing conversion warning.
4347 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
4348 to avoid narrowing conversion warning.
4349 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
4351 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
4353 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4355 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
4356 64bit configurations.
4357 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
4358 (SIZE_TYPE): Likewise.
4359 * config/vxworks.c (vxworks_emutls_var_fields): Use
4360 long_unsigned_type_node instead of unsigned_type_node as the offset
4361 field type, which is "pointer" mode in emutls.c.
4363 2017-06-27 Jakub Jelinek <jakub@redhat.com>
4366 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
4369 * gimple-fold.c (replace_call_with_call_and_fold): Handle
4370 gimple_vuse copying separately from gimple_vdef copying.
4372 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4374 * value-prof.c (free_hist): Remove call to memset and the enclosing if
4377 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
4378 Olivier Hainque <hainque@adacore.com>
4380 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
4381 for all vxworks7 targets.
4382 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
4383 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
4384 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
4385 variations for VX6/VX7 and 32/64bits later on in ...
4386 (VXWORKS_LIB_SPEC): Leverage new macros.
4387 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
4388 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
4390 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
4392 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
4393 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
4395 2017-06-26 Carl Love <cel@us.ibm.com>
4397 * config/rs6000/rs6000-c.c: Add support for built-in functions
4398 vector bool char vec_reve (vector bool char);
4399 vector signed char vec_reve (vector signed char);
4400 vector unsigned char vec_reve (vector unsigned char);
4401 vector bool int vec_reve (vector bool int);
4402 vector signed int vec_reve (vector signed int);
4403 vector unsigned int vec_reve (vector unsigned int);
4404 vector bool long long vec_reve (vector bool long long);
4405 vector signed long long vec_reve (vector signed long long);
4406 vector unsigned long long vec_reve (vector unsigned long long);
4407 vector bool short vec_reve (vector bool short);
4408 vector signed short vec_reve (vector signed short);
4409 vector double vec_reve (vector double);
4410 vector float vec_reve (vector float);
4411 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
4412 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
4413 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
4414 (altivec_vreve): New pattern.
4415 * config/rs6000/altivec.h (vec_reve): New define.
4416 * doc/extend.texi (vec_rev): Update the built-in documentation file
4417 for the new built-in functions.
4419 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4421 PR tree-optimization/71815
4422 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
4424 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
4426 (slsr_process_phi): Likewise.
4427 (replace_uncond_cands_and_profitable_phis): Don't replace a
4428 multiply candidate with a stride of 1 (copy or cast).
4429 (phi_incr_cost): Call uses_consumed_by_stmt rather than
4431 (lowest_cost_path): Likewise.
4432 (total_savings): Likewise.
4434 2017-06-26 Richard Biener <rguenther@suse.de>
4437 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
4438 Use def_builtin_pure for all gather builtins.
4440 2017-06-26 Richard Biener <rguenther@suse.de>
4442 PR tree-optimization/81203
4443 * tree-tailcall.c (find_tail_calls): Do not move stmts into
4446 2017-06-26 Marek Polacek <polacek@redhat.com>
4449 * doc/invoke.texi: Document -Wmultistatement-macros.
4451 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
4453 * doc/sourcebuild.texi (ARM-specific attributes): Document new
4454 arm_neon_ok_no_float_abi effective target.
4456 2017-06-26 Richard Biener <rguenther@suse.de>
4458 PR tree-optimization/80928
4459 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
4460 (copy_bbs): Set BB_DUPLICATED flag early.
4461 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
4463 (execute_on_shrinking_pred): Likewise.
4464 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
4465 BB_DUPLICATED blocks.
4466 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
4467 iterate over all PHIs considering removal of *gsi.
4469 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
4471 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
4474 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4476 * config/rs6000/rs6000-string.c: (expand_block_clear,
4477 do_load_for_compare, select_block_compare_mode,
4478 compute_current_alignment, expand_block_compare,
4479 expand_strncmp_align_check, expand_strn_compare,
4480 expand_block_move, rs6000_output_load_multiple)
4481 Move functions related to string/block move/compare
4483 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
4484 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
4485 for this function which is now used in two files.
4486 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
4487 * config.gcc: Add rs6000-string.o to extra_objs for
4488 targets powerpc*-*-* and rs6000*-*-*.
4490 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4493 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
4494 32-bit, since indexed is not valid for DImode.
4495 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
4496 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
4497 (define_peephole2 for Altivec d-form load): Add 32-bit support.
4498 (define_peephole2 for Altivec d-form store): Likewise.
4501 * multiple_target.c (create_dispatcher_calls): Only create the
4502 dispatcher call if the function is the default clone of a
4505 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
4508 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
4509 a call, force the call to not be a tail call.
4511 2017-06-23 Jeff Law <law@redhat.com>
4513 * doc/contrib.texi: Add entry for Steven Pemberton's work on
4516 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
4518 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
4519 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
4520 handling for early expansion of vector shifts (sl,sr,sra,rl).
4521 (builtin_function_type): Add vector shift right instructions
4522 to the unsigned argument list.
4524 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
4526 rtl-optimizatoin/79286
4527 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
4528 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
4529 trap. PIC register plus a const unspec without offset can never trap.
4531 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
4533 * tree.h (builtin_structptr_type): New type.
4534 (builtin_structptr_types): Declare new array.
4535 * tree.c (builtin_structptr_types): New array.
4536 (free_lang_data, build_common_tree_nodes): Use it.
4538 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
4541 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
4544 2017-06-22 Matt Turner <mattst88@gmail.com>
4546 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
4547 Lake models to skylake case. Assume skylake for unknown
4548 models with clflushopt.
4550 2017-06-22 Jeff Law <law@redhat.com>
4552 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
4553 frame sizes that do not satisfy aarch64_uimm12_shift.
4555 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
4557 * profile-count.h (apply_probability,
4558 apply_scale, probability_in): Fix checks for zero.
4560 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4562 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
4563 * doc/cppdiropts.texi (-I @var{dir}): Document it.
4565 2016-06-22 Richard Biener <rguenther@suse.de>
4567 * tree-vect-loop.c (vect_model_reduction_cost): Handle
4568 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
4569 REDUC_MAX_EXPR support.
4570 (vectorizable_reduction): Likewise.
4571 (vect_create_epilog_for_reduction): Likewise.
4573 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
4575 * match.pd (A / (1 << B) -> A >> B): New.
4576 * generic-match-head.c: Include optabs-tree.h.
4577 * gimple-match-head.c: Likewise.
4578 * optabs-tree.h (target_supports_op_p): New.
4579 * optabs-tree.c (target_supports_op_p): New.
4581 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4583 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
4584 $gcc_cv_ld --help output.
4585 (gcc_cv_ld_demangle): Likewise.
4586 (gcc_cv_ld_eh_frame_hdr): Likewise.
4587 (gcc_cv_ld_pie): Likewise.
4588 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
4589 (gcc_cv_ld_buildid): Likewise.
4590 (gcc_cv_ld_sysroot): Likewise.
4591 (ld_bndplt_support): Likewise.
4592 (ld_pushpopstate_support): Likewise.
4593 * configure: Regenerate.
4594 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
4596 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4599 * config/i386/sse.md (round<mode>2): Renumber match_dup and
4600 operands indexes to avoid gap between operands and match_dups.
4602 2017-06-21 Andrew Pinski <apinski@cavium.com>
4604 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
4605 Increment Arith_shift and Arith_shift_reg by 1.
4606 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
4608 * config/aarch64/aarch64.c (thunderx_tunings): Enable
4609 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
4610 (aarch64_strip_extend): Add new argument and test for it.
4611 (aarch64_cheap_mult_shift_p): New function.
4612 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
4613 add a cost if it is true.
4614 Update calls to aarch64_strip_extend.
4615 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
4617 2017-06-21 Andrew Pinski <apinski@cavium.com>
4619 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
4621 (thunderxt88): Likewise.
4622 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
4623 (thunderx_prefetch_tune): New variable.
4624 (thunderx2t99_prefetch_tune): Update for the correct values.
4625 (thunderxt88_tunings): New variable.
4626 (thunderx_tunings): Use thunderx_prefetch_tune instead of
4627 generic_prefetch_tune.
4628 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
4630 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4632 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
4633 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
4634 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
4635 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
4636 (aarch64_atomic_cas<mode>, GPI): Likewise.
4638 2017-06-21 Martin Liska <mliska@suse.cz>
4640 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
4641 statements on cold and hot labels.
4642 * predict.c (tree_estimate_probability_bb): Remove the
4643 prediction from this place.
4645 2017-06-21 Martin Liska <mliska@suse.cz>
4647 PR tree-optimization/79489
4648 * gimplify.c (maybe_add_early_return_predict_stmt): New
4650 (gimplify_return_expr): Call the function.
4651 * predict.c (tree_estimate_probability_bb): Remove handling
4653 * predict.def: Update comment about early return predictor.
4654 * gimple-predict.h (is_gimple_predict): New function.
4655 * predict.def: Change default value of early return to 66.
4656 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
4658 * passes.def: Put pass_strip_predict_hints to the beginning of
4661 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
4663 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
4664 FUNCTION_DECL declarations.
4665 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
4667 (dwaf2out_decl): Likewise.
4668 * godump.c (go_early_global_decl): Skip call to the real debug hook
4669 for FUNCTION_DECL declarations.
4670 * passes.c (rest_of_decl_compilation): Skip call to the
4671 early_global_decl debug hook for FUNCTION_DECL declarations, unless
4672 -fdump-go-spec is passed.
4674 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4676 * config/i386/i386.c (struct builtin_isa): New field pure_p.
4677 Reorder for compactness.
4678 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
4679 (def_builtin_pure, def_builtin_pure2): New functions.
4680 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
4682 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4684 * match.pd (nop_convert): New predicate.
4685 ((A +- CST1) +- CST2): Allow some NOP conversions.
4687 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4690 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
4691 with ctors/dtors if GOVD_SHARED is set.
4693 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4695 * config/aarch64/aarch64.md (movti_aarch64):
4696 Emit mov rather than orr.
4697 (movtf_aarch64): Likewise.
4698 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
4699 Emit mov rather than orr.
4701 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4703 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
4704 Swap alternatives, make integer dup more expensive.
4706 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4708 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
4709 Return true for non-tls symbols.
4711 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
4713 * config/aarch64/aarch64-cores.def (cortex-a55): New.
4714 (cortex-a75): Likewise.
4715 (cortex-a75.cortex-a55): Likewise.
4716 * config/aarch64/aarch64-tune.md: Regenerate.
4717 * doc/invoke.texi (-mtune): Document new values for -mtune.
4719 2017-06-21 Tom de Vries <tom@codesourcery.com>
4721 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
4723 (Effective-Target Keywords, Other attributes): Suggest using
4724 dg-add-options stack_size feature to get stack limit in stack_size
4725 effective target documentation.
4727 2017-06-21 Julian Brown <julian@codesourcery.com>
4728 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4730 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
4731 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
4732 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
4734 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
4735 attribute type list for neon_multiply.
4736 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
4737 attribute type list for neon_multiply.
4738 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
4739 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
4740 attribute type list for neon_multiply.
4741 * config/arm/types.md (crypto_pmull): Add.
4742 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
4743 attribute type list.
4745 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
4747 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
4750 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4752 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
4753 to make sure not to dereference a NULL cost_classes_ptr pointer.
4755 2017-06-20 Carl Love <cel@us.ibm.com>
4757 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4758 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4759 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4760 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4761 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4762 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4763 VMULOSW): New enum "unspec" values.
4764 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
4765 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
4766 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4767 altivec_vmulosw): New patterns.
4768 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4769 VMULOSW): Add definitions.
4771 2017-06-20 Julia Koval <julia.koval@intel.com>
4773 * config/i386/i386.c: Fix rounding expand for new pattern.
4774 * config/i386/subst.md: Fix pattern (parallel -> unspec).
4776 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4778 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
4779 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
4781 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4783 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
4786 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4788 * config/aarch64/aarch64-cores.def: Rearrange to sort by
4789 architecture, then by implementer ID.
4790 * config/aarch64/aarch64-tune.md: Regenerate.
4792 2017-06-20 Richard Biener <rguenther@suse.de>
4795 * fold-const.c (split_tree): Fold to type before negating.
4797 2017-06-20 David Malcolm <dmalcolm@redhat.com>
4799 * diagnostic-show-locus.c
4800 (selftest::test_fixit_deletion_affecting_newline): New function.
4801 (selftest::diagnostic_show_locus_c_tests): Call it.
4803 2017-06-20 Andreas Schwab <schwab@suse.de>
4806 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
4809 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
4811 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
4812 __ARM_FEATURE_COPROC according to support.
4814 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4816 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
4817 Rewritten to avoid overflow for > 32-bit pointers.
4820 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
4821 by removing enum keyword.
4822 (ubsan_type_descriptor): Likewise. Formatting fix.
4825 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
4826 splitter): Require TARGET_SSE2 in the condition.
4828 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
4831 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
4832 for doing vector set of SFmode on ISA 3.0.
4833 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
4834 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
4836 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
4837 SFmode value into a V4SF variable that was extracted from another
4838 V4SF variable without converting the element to double precision
4839 and back to single precision vector format.
4840 (vsx_insert_extract_v4sf_p9_2): Likewise.
4842 2017-06-19 Jakub Jelinek <jakub@redhat.com>
4844 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
4845 in UWHI to avoid undefined overflow.
4848 * ubsan.h (enum ubsan_encode_value_phase): New.
4849 (ubsan_encode_value): Change second argument to
4850 enum ubsan_encode_value_phase with default value of
4851 UBSAN_ENCODE_VALUE_GENERIC.
4852 * ubsan.c (ubsan_encode_value): Change second argument to
4853 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
4854 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
4855 create_tmp_var_raw instead of create_tmp_var and use a
4857 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
4858 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
4859 ubsan_encode_value callers.
4862 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
4863 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
4864 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
4866 2017-06-19 Richard Biener <rguenther@suse.de>
4869 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
4870 estimates if we changed anything.
4872 2017-06-19 Richard Biener <rguenther@suse.de>
4874 PR tree-optimization/80887
4875 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
4876 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
4877 simplified lookups, then reset mprts_hook.
4878 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
4880 (try_to_simplify): Likewise.
4882 2017-06-19 Martin Liska <mliska@suse.cz>
4885 * gimplify.c (gimplify_switch_expr):
4886 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
4888 2017-06-19 Martin Liska <mliska@suse.cz>
4890 * doc/install.texi: Document that PGO runs in 4 stages.
4892 2017-06-19 Martin Liska <mliska@suse.cz>
4895 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
4896 to dispatcher function name.
4897 * multiple_target.c (replace_function_decl): New function.
4898 (create_dispatcher_calls): Redirect both edges and references.
4900 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
4902 * profile-count.c (profile_count::dump): Dump quality.
4903 (profile_count::differs_from_p): Update for unsigned val.
4904 * profile-count.h (profile_count_quality): New enum.
4905 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
4907 2017-06-19 Richard Biener <rguenther@suse.de>
4909 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
4910 struct function as arg.
4911 (estimate_numbers_of_iterations): Export overload with loop arg.
4912 (free_numbers_of_iterations_estimates_loop): Use an overload of
4913 free_numbers_of_iterations_estimates instead.
4914 * tree-cfg.c (remove_bb): Adjust.
4915 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
4916 * tree-parloops.c (gen_parallel_loop): Likewise.
4917 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
4919 (tree_unroll_loops_completely): Likewise.
4920 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
4921 Use an overload instead and export.
4922 (estimated_loop_iterations): Adjust.
4923 (max_loop_iterations): Likewise.
4924 (likely_max_loop_iterations): Likewise.
4925 (estimate_numbers_of_iterations): Take struct function as arg
4927 (loop_exits_before_overflow): Adjust.
4928 (free_numbers_of_iterations_estimates_loop): Use an overload.
4929 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
4930 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
4932 2017-06-19 Richard Biener <rguenther@suse.de>
4935 * ipa-prop.c (find_constructor_constant_at_offset): Handle
4936 RANGE_EXPR conservatively.
4938 2017-06-16 Carl Love <cel@us.ibm.com>
4940 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4941 definitions for vec_float, vec_float2, vec_floato,
4942 vec_floate built-ins.
4943 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
4944 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
4946 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
4947 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
4948 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
4949 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
4950 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
4951 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
4952 vec_floato): Add builtin defines.
4953 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
4954 Update the built-in documentation file for the new built-in
4957 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4959 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
4960 (mthumb): Mark as the negative of -marm.
4962 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4964 * doc/invoke.texi (ARM Options, -mcpu): Document supported
4966 (ARM Options, -mtune): Document that this accepts the same
4967 extension options as -mcpu.
4968 (ARM Options, -mfpu): Document addition of -mfpu=auto.
4970 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4972 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
4973 permitted extensions.
4975 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4977 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
4978 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
4979 (armv8-m.main): Add option +nodsp.
4980 * config/arm/arm-cpu-cdata.h: Regenerated.
4982 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4984 * config/arm/t-fuchsia: New file.
4985 * config.gcc (arm*-*-fuchsia*): Use it.
4987 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4989 * config/arm/t-symbian: Rewrite for new option infrastructure.
4991 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4993 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
4994 (MULTILIB_REQUIRED): Likewise.
4996 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4998 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
4999 (MULTILIB_RESUE): Likewise.
5000 (MULTILIB_MATCHES): Likewise.
5001 (MULTLIB_REQUIRED): Likewise.
5003 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5005 * config/arm/t-rtems: Rewrite for new option framework.
5007 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5009 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
5010 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
5011 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
5012 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
5013 * config/arm/t-multilib: ... here.
5014 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
5015 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
5016 armv7-a and armv8*-a when A-profile libraries have not been built.
5017 * config/arm/t-rmprofile: Rewrite.
5019 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5021 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
5022 with a backslash. Remove the backslash after substituting unescaped
5024 * doc/fragments.texi (MULTILIB_REUSE): Document it.
5026 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5028 * config.gcc: (arm*-*-*): When building a-profile libraries, force
5029 the driver to pass through the default setting of -mfloat-abi.
5030 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
5032 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
5033 (all_feat_combs): New rule.
5034 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
5036 * config/arm/t-aprofile: Rewrite.
5038 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5040 * config/arm/arm.h (FPUTYPE_AUTO): Define.
5041 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
5042 fpu is not specified by the user/command-line.
5043 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
5044 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
5045 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
5046 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
5047 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
5048 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
5050 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5052 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
5053 * config/arm/t-arm-elf: Rewritten.
5055 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5057 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
5058 have some floating-point instructions.
5059 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
5060 (TARGET_MAYBE_HARD_FLOAT): New macro.
5061 * config/arm/arm-builtins.c (arm_init_builtins): Use
5062 TARGET_MAYBE_HARD_FLOAT.
5063 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
5065 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5067 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
5068 (configargs.h): Include it.
5069 (arm_print_hint_for_fpu_option): New function.
5070 (arm_parse_fpu_option): New function.
5071 (candidate_extension): New class.
5072 (arm_canon_for_multilib): New function.
5073 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
5074 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
5075 (ARCH_CANONICAL_SPECS): New macro.
5076 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
5078 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5080 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
5081 are set after handling multilib fragments. Set target_cpu_default2
5084 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5086 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
5088 (arm*-*-*): Set target_cpu_default2 to a quoted string.
5089 * config/arm/parsecpu.awk (check_cpu): Validate any extension
5091 (check_arch): Likewise.
5092 * config/arm/arm.c (arm_configure_build_target): Handle
5093 TARGET_CPU_DEFAULT being a string constant. Scan any feature
5094 options in the default.
5096 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5098 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
5099 when an option is an alias of another.
5100 * config/arm/parsecpu.awk (optalias): New parser token.
5101 (gen_comm_data): Mark non-alias options as such. Emit entries
5102 for extension aliases.
5103 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
5104 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
5105 (armv6kz, armv6zk, armv6t2): Likewise.
5106 (armv7): Make vfpv3-d16 an alias.
5107 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
5109 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
5110 Sort in canonical order.
5111 (armv8-a): Sort in canonical order.
5112 (armv8.1-a, armv8.2-a): Likewise.
5113 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
5115 (cortex-a9): Sort in canonical order.
5116 * config/arm/arm.c (selftests.h): Include it.
5117 (arm_test_cpu_arch_data): New function.
5118 (arm_run_self_tests): New function.
5119 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
5120 (targetm): Move declaration to the end of the file.
5121 * arm-cpu-cdata.h: Regenerated.
5123 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5125 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
5126 call to target_mode_check describing the type of option passed.
5127 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
5128 (arm_target_thumb_only): Use arm_parse_arch_option_name or
5129 arm_parse_cpu_option_name to match parameters against list of
5131 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
5132 arm_arch_core_flags data structure.
5133 * config/arm/arm-cpu_cdata.h: Regenerated.
5135 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5137 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
5139 (arm_print_hint_for_cpu_option): Likewise.
5140 (arm_print_hint_for_arch_option): Likewise.
5141 (arm_parse_cpu_option_name): Likewise.
5142 (arm_parse_arch_option_name): Likewise.
5143 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
5144 of entries in the all_fpus list.
5145 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
5146 (arm_parse_cpu_option_name): Declare.
5147 (arm_parse_arch_option_name): Declare.
5148 (arm_parse_option_features): Declare.
5149 (arm_intialize_isa): Declare.
5150 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
5152 (gen_comm_data): ... here. Make definitions non-static.
5153 * config/arm/arm-cpu-data.h: Regenerated.
5154 * config/arm/arm-cpu-cdata.h: Regenerated.
5156 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5158 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
5159 (cpu_arch_extension): New structure.
5160 (cpu_arch_option, arch_option, cpu_option): New structures.
5161 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
5163 (gen_data): Generate new format data tables.
5164 * config/arm/arm.c (cpu_tune): New structure.
5165 (cpu_option, processors): Delete.
5166 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
5167 (arm_print_hint_for_cpu_option): ... this and ...
5168 (arm_print_hint_for_arch_option): ... this.
5169 (arm_parse_arch_cpu_name): Delete. Replace with ...
5170 (arm_parse_cpu_option_name): ... this and ...
5171 (arm_parse_arch_option_name): ... this.
5172 (arm_unrecognized_feature): Change type of target parameter to
5174 (arm_parse_arch_cpu_features): Delete. Replace with ...
5175 (arm_parse_option_features): ... this.
5176 (arm_configure_build_target): Rework to use new configuration data
5178 (arm_print_tune_info): Rework for new configuration data tables.
5179 * config/arm/arm-cpu-data.h: Regenerated.
5180 * config/arm/arm-cpu.h: Regenerated.
5182 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5184 * Makefile.in (OBJS): Move sbitmap.o from here ...
5185 (OBJS-libcommon): ... to here.
5187 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5189 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
5190 (ISA_ALL_CRYPTO): New macro.
5191 (ISA_ALL_SIMD): New macro
5192 (ISA_ALL_FP): New macro.
5193 * config/arm/arm.c (fpu_bitlist): Update initializer.
5194 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
5196 (arm9e): Add fpu. Add option for nofp
5197 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
5198 (arm926ej-s, arm1026ej-s): Likewise.
5199 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
5200 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
5201 neon-fp16, neon-vfpv4, nofp and nosimd.
5202 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
5203 (cortex-a8): Add fpu. Add option for nofp.
5204 (cortex-a9): Add fpu. Add options for nosimd and nofp.
5205 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
5206 (cortex-r4f): Add fpu.
5207 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
5208 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
5210 (cortex-r8): Likewise.
5211 (cortex-m4): Add fpu. Add option for nofp.
5212 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
5213 (cortex-a17.cortex-a7): Likewise.
5214 (cortex-a32): Add fpu. Add options for crypto and nofp.
5215 (cortex-a35, cortex-a53): Likewise.
5216 (cortex-a57): Add fpu. Add option for crypto.
5217 (cortex-a72, cortex-a73): Likewise.
5218 (exynos-m1): Likewise.
5219 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
5220 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
5221 (cortex-m33): Add fpu. Add option for nofp.
5222 * config/arm/arm-cpu-cdata.h: Regenerated
5223 * config/arm/arm-cpu-data.h: Regenerated.
5225 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5227 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
5228 (armv5te, armv5tej): Likewise.
5229 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
5230 (armv7): Add options fp and vfpv3-d16.
5231 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
5232 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
5234 (armv7ve): Likewise.
5235 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
5236 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
5237 (armv8-a): Add nocrypto option.
5238 (armv8.1-a, armv8.2-a): Likewise.
5239 (armv8-m.main): add options fp, fp.dp and nofp.
5241 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5243 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
5245 (armv8-a+crc): Delete.
5246 (armv8.1-a): Add options simd, crypto and nofp.
5247 (armv8.2-a): Add options fp16, simd, crypto and nofp.
5248 (armv8.2-a+fp16): Delete.
5249 (armv8-m.main): Add option dsp.
5250 (armv8-m.main+dsp): Delete.
5251 (cortex-a8): Add fpu. Add nofp option.
5252 (cortex-a9): Add fpu. Add nofp and nosimd options.
5253 * config/arm/parsecpu.awk (gen_data): Generate option tables and
5254 link to main cpu and architecture data structures.
5255 (gen_comm_data): Only put isa attributes from the main architecture
5257 (option): New statement for architecture and CPU entries.
5258 * arm.c (struct cpu_option): New structure.
5259 (struct processors): Add entry for options.
5260 (arm_unrecognized_feature): New function.
5261 (arm_parse_arch_cpu_name): Ignore any characters after the first
5263 (arm_parse_arch_cpu_feature): New function.
5264 (arm_configure_build_target): Separate out any CPU and architecture
5265 features and parse separately. Don't error out if -mfpu=auto is
5266 used with only an architecture string.
5267 (arm_print_asm_arch_directives): New function.
5268 (arm_file_start): Call it.
5269 * config/arm/arm-cpu-cdata.h: Regenerated.
5270 * config/arm/arm-cpu-data.h: Likewise.
5271 * config/arm/arm-tables.opt: Likewise.
5273 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5275 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
5276 assembler when it is not -mfpu=auto.
5278 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5280 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
5281 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
5282 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
5283 (ASM_CPU_SPEC): Rewrite.
5284 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
5285 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
5286 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
5287 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
5288 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
5289 copied string is NUL-terminated. Also strip any characters prefixed
5291 (arm_rewrite_selected_arch): New function.
5292 (arm_rewrite_march): New function.
5294 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5296 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
5297 (x_arm_cpu_string, x_arm_tune_string): Likewise.
5298 (march, mcpu, mtune): Convert to string-based options.
5299 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
5300 (arm_parse_arch_cpu_name): New function.
5301 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
5302 identify selected architecture or CPU.
5303 (arm_option_save): New function.
5304 (TARGET_OPTION_SAVE): Redefine.
5305 (arm_option_restore): Restore string options.
5306 (arm_option_print): Print string options.
5308 2017-06-16 Martin Sebor <msebor@redhat.com>
5310 PR tree-optimization/80933
5311 PR tree-optimization/80934
5312 * builtins.c (fold_builtin_3): Do not handle bcmp here.
5313 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
5314 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
5315 (gimple_fold_builtin): Call them.
5317 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5319 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
5320 as unlikely; update profile.
5322 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5324 * predict.c (force_edge_cold): Handle declaring edges impossible
5327 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5329 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
5331 (try_unroll_loop_completely): Fix reporting.
5333 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5335 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
5337 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
5340 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
5341 if given a non-constant argument for an intrinsic which requires a
5344 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5346 * profile.c (compare_freqs): New function.
5347 (branch_prob): Sort edge list.
5348 (find_spanning_tree): Assume that the list is priority sorted.
5350 2017-06-16 Richard Biener <rguenther@suse.de>
5352 PR tree-optimization/81090
5353 * passes.def (pass_record_bounds): Remove.
5354 * tree-pass.h (make_pass_record_bounds): Likewise.
5355 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
5356 make_pass_record_bounds): Likewise.
5357 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
5358 not free niter estimates at the beginning but at the end.
5359 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
5361 2017-06-16 Richard Biener <rguenther@suse.de>
5363 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
5364 initializer to workaround ICE in host GCC 4.8.
5366 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5368 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
5370 (clone_inlined_nodes): Update.
5372 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5374 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
5375 prefetch settings, and enable prefetching by default at -O3.
5377 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5379 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5380 Set flag_prefetch_loop_arrays according to tuning data.
5382 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5384 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
5386 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
5387 [Unrelated to main purpose of the patch] Place the pointer field last
5388 to enable type checking errors when tune structure are wrongly merged.
5389 * config/aarch64/aarch64.c (generic_prefetch_tune,)
5390 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
5391 (thunderx2t99_prefetch_tune): New tune constants.
5392 (tune_params *_tunings): Update all tunings (no functional change).
5393 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
5394 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
5395 from tunings structures.
5397 2017-06-16 Jakub Jelinek <jakub@redhat.com>
5400 * ubsan.c (instrument_null): Add T argument, use it instead
5401 of computing it based on IS_LHS.
5402 (instrument_object_size): Likewise.
5403 (pass_ubsan::execute): Adjust instrument_null and
5404 instrument_object_size callers to pass gimple_get_lhs or
5405 gimple_assign_rhs1 result to it. Use instrument_null instead of
5406 calling get_base_address and instrument_mem_ref. Handle
5407 aggregate call arguments for object-size sanitization.
5409 2017-06-16 Yury Gribov <tetra2005@gmail.com>
5411 PR tree-optimization/81089
5412 * tree-vrp.c (is_masked_range_test): Validate operands of
5415 2017-06-15 Martin Sebor <msebor@redhat.com>
5418 * dumpfile.c (dump_register): Avoid calling memset to initialize
5419 a class with a default ctor.
5420 * gcc.c (struct compiler): Remove const qualification.
5421 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
5422 * hash-table.h: Ditto.
5423 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
5425 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
5426 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
5428 * params.h (struct param_info): Make struct members non-const.
5429 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
5430 with default initialization.
5431 * vec.h (vec_copy_construct, vec_default_construct): New helper
5433 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
5434 with vec_copy_construct.
5435 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
5436 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
5437 * doc/invoke.texi (-Wclass-memaccess): Document.
5439 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5441 * emit-rtl.h (is_leaf): Update comment about local
5444 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5447 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
5448 for a variable to have a section before checking if the section has a
5450 Set section to.persistent if persistent attribute is set.
5451 Warn if .persistent attribute is used on an automatic variable.
5453 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
5455 PR rtl-optimization/80474
5456 * reorg.c (update_block): Do not ignore instructions in a delay slot.
5458 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
5460 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
5463 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
5465 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
5466 (casesi): Emit bounds checking as RTL.
5467 (casesi_internal_mips16_<mode>): Remove bounds checking.
5469 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
5471 * config/xtensa/xtensa.c (xtensa_option_override): Append
5472 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
5473 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
5474 xtensa_doloop_hooks): Define unconditionally.
5475 (xtensa_reorg_loops): Only call reorg_loops in the presence of
5477 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
5478 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
5479 for it in xtensa_option_override.
5480 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
5481 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
5483 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
5485 * doc/cppopts.texi: Document '-' special value to -MF.
5487 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
5489 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
5490 (cortex_a53_fconst): Likewise.
5491 (cortex_a53_fpmul): Likewise.
5492 (cortex_a53_f_load_64): Likewise.
5493 (cortex_a53_f_load_many): Likewise.
5494 (cortex_a53_advsimd_alu): Likewise.
5495 (cortex_a53_advsimd_alu_q): Likewise.
5496 (cortex_a53_advsimd_mul): Likewise.
5497 (cortex_a53_advsimd_mul_q): Likewise.
5498 (fpmac bypass): Add new bypass for fpmac-fpmac case.
5499 Add missing fmul, r2f_cvt and fconst cases.
5501 2017-06-14 Richard Biener <rguenther@suse.de>
5504 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
5506 (fold_binary_loc): When associating do not treat pre-existing
5507 TREE_OVERFLOW on literal constants as a reason to allow
5508 TREE_OVERFLOW on associated literal constants.
5510 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5512 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
5513 (MASK_FEATURES): New macro.
5514 * config/sparc/sparc.c (sparc_option_override): Remove the special
5515 handling of -mfpu and generalize it to all MASK_FEATURES switches.
5517 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5519 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
5520 a division of 0 if non-call exceptions are enabled.
5522 2017-06-14 Andrew Pinski <apinski@cavium.com>
5523 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5526 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
5527 Improve vector initialization code gen for only variable case.
5529 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5531 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
5533 2017-06-14 Richard Biener <rguenther@suse.de>
5535 PR tree-optimization/81083
5536 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
5539 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5541 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
5542 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
5543 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
5544 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
5545 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
5546 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
5548 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5550 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
5551 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
5553 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5555 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
5557 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5559 * config/rs6000/t-rtems: Don't handle SPE.
5561 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5563 * config/rs6000/t-linux: Don't handle SPE.
5565 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5567 * config/rs6000/eabispe.h: Delete file.
5569 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5571 * config/rs6000/t-spe: Delete file.
5573 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5575 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
5576 (rs6000_legitimate_offset_address_p): Return false for anything in
5577 V2SImode or V2SFmode.
5579 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5581 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
5582 except V2SF and V2SI. Rearrange the vector modes, and add comments.
5583 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
5585 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
5586 (rs6000_legitimate_offset_address_p): Ditto.
5587 (rs6000_emit_move): Ditto.
5588 (rs6000_init_builtins): Remove V4HI_type_node.
5590 2017-06-13 Martin Liska <mliska@suse.cz>
5593 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
5594 (gate_asan): Likewise.
5595 * asan.h (asan_no_sanitize_address_p): Remove the function.
5596 (sanitize_flags_p): New function.
5597 * builtins.def: Fix coding style.
5598 * common.opt: Use renamed enum value.
5599 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
5600 * doc/extend.texi: Document no_sanitize attribute.
5601 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
5602 to SANITIZE_UNDEFINED_NONDEFAULT.
5603 * gcc.c (sanitize_spec_function): Use the renamed enum value.
5604 * gimple-fold.c (optimize_atomic_compare_exchange_p):
5605 Use sanitize_flags_p.
5606 * gimplify.c (gimplify_function_tree): Likewise.
5607 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
5608 * opts.c (parse_no_sanitize_attribute): New function.
5609 (common_handle_option): Use renamed enum value.
5610 * opts.h (parse_no_sanitize_attribute): Declare.
5611 * tree.c (sanitize_flags_p): New function.
5612 * tree.h: Declared here.
5613 * tsan.c: Use sanitize_flags_p.
5614 * ubsan.c (ubsan_expand_null_ifn): Likewise.
5615 (instrument_mem_ref): Likewise.
5616 (instrument_bool_enum_load): Likewise.
5617 (do_ubsan_in_current_function): Remove the function.
5618 (pass_ubsan::execute): Use sanitize_flags_p.
5619 * ubsan.h: Remove do_ubsan_in_current_function
5620 * tree-cfg.c (print_no_sanitize_attr_value): New function.
5621 (dump_function_to_file): Use it here.
5623 2017-06-13 Martin Jambor <mjambor@suse.cz>
5625 PR tree-optimization/80803
5626 PR tree-optimization/81063
5627 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
5628 (propagate_subaccesses_across_link): Enqueue subtree whenever
5629 necessary instead of relying on the caller.
5631 2017-06-13 Martin Jambor <mjambor@suse.cz>
5633 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
5634 that have a first_link.
5635 (sort_and_splice_var_accesses): Do not check first_link before
5637 (subtree_mark_written_and_enqueue): Likewise.
5638 (propagate_all_subaccesses): Likewise and do not stop at first
5639 parent with a first_link.
5641 2017-06-13 Martin Jambor <mjambor@suse.cz>
5643 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
5646 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5648 * match.pd: New pattern.
5650 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5652 * tree-vrp.c (is_masked_range_test): New function.
5653 (register_edge_assert_for): Determine ranges for
5656 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5658 PR tree-optimization/67328
5659 * fold-const.c (maskable_range_p): New function.
5660 (build_range_check): Generate bittests if possible.
5662 2017-06-13 Martin Liska <mliska@suse.cz>
5664 * gimple-pretty-print.c (dump_probability): Add new argument.
5665 (dump_edge_probability): Dump both probability and count.
5666 (dump_gimple_label): Likewise.
5667 (dump_gimple_bb_header): Likewise.
5669 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
5672 * config/avr/avr-devices.c: Fix indentation.
5673 * config/avr/gen-avr-mmcu-specs.c: Dito.
5675 2017-06-13 Richard Biener <rguenther@suse.de>
5677 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
5678 instead get vector type from stmt_info.
5679 (vectorizable_reduction): Adjust. Remove dead code.
5681 2017-06-13 Richard Biener <rguenther@suse.de>
5684 * fold-const.c (extract_muldiv_1): Remove bogus distribution
5685 case of C * (x * C2 + C3).
5686 (fold_addr_of_array_ref_difference): Properly fold index difference.
5688 2017-06-12 David S. Miller <davem@davemloft.net>
5691 * config/sparc/sparc.md (return expander): Emit frame blockage if
5692 function uses alloca.
5694 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
5696 * combine.c (make_field_assignment): Check len rather than the mode
5697 precision when calling force_to_mode.
5699 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
5701 Support multilibs and devices that see flash in RAM address range.
5704 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
5705 (avr_mcu_t) <flash_pm_offset>: New field.
5706 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
5707 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
5708 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
5709 (AVR_TINY_PM_OFFSET): Remove macro.
5710 * config/avr/avr.opt (-mshort-calls): New option.
5711 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
5712 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
5713 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
5714 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
5715 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
5716 instead of avr_arch->have_jmp_call.
5717 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
5718 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
5719 avr_arch->flash_pm_offset to define.
5720 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
5721 new field flash_pm_offset. Add entry for avrxmega3.
5722 (avr_texinfo): Add entry for avrxmega3.
5723 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
5724 attiny212, attiny214,
5725 attiny412, attiny414, attiny416, attiny417,
5726 attiny814, attiny816, attiny817,
5727 attiny1614, attiny1616, attiny1617,
5728 attiny3214, attiny3216, attiny3217.
5729 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
5730 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
5731 (avr_print_operand_address) [AVR_TINY]: Same.
5732 (avr_asm_init_sections) <readonly_data_section>: Only patch
5733 callback if avr_arch->flash_pm_offset = 0.
5734 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
5735 for rodata if avr_arch->flash_pm_offset != 0.
5736 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
5737 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
5738 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
5739 (m_options): Append opt_rcall.
5740 (m_dirnames): Append dir_rcall.
5741 * config/avr/t-multilib: Regenerate.
5743 * configure.ac [target=avr]: Check whether avrxmega3 default
5744 linker description file works as needed.
5745 * configure: Regenerate.
5746 * doc/avr-mmcu.texi: Regenerate.
5747 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
5748 <__AVR_ARCH__>: Document avrxmega3 and 103.
5749 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
5750 <__AVR_SHORT_CALLS__>: Document it.
5751 <__AVR_PM_BASE_ADDRESS__>: Document it.
5752 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
5753 (AVR Variable Attributes) <progmem>: Document this is
5754 not needed for avrxmega3.
5755 (AVR Named Address Spaces) <__flash>: Dito.
5757 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
5759 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
5761 2017-06-12 Doug Rupp <rupp@adacore.com>
5763 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
5764 Append vxworks-stdint.h to the tm_file list.
5765 * config/vxworks-stdint.h: New file.
5767 2017-06-12 Martin Liska <mliska@suse.cz>
5769 PR tree-optimization/81041
5770 * tree-profile.c (gimple_gen_ic_func_profiler):
5771 Create an extra BB in profile-generate
5772 (gimple_gen_time_profiler): Likewise.
5774 2017-06-12 Jakub Jelinek <jakub@redhat.com>
5776 PR tree-optimization/81003
5777 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
5778 (update_range_test): Use it instead of force_gimple_operand_gsi.
5780 2017-06-12 Richard Biener <rguenther@suse.de>
5782 PR tree-optimization/81053
5783 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
5784 with backedge value not defined in loop. Simplify def stmt
5787 2017-06-11 Tom de Vries <tom@codesourcery.com>
5790 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
5792 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
5793 nvptx_cannot_force_const_mem.
5795 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5797 * opts.c (finish_options): Move test for flag_split_stack after
5798 it has been initialized.
5800 2017-06-11 Jason Merrill <jason@redhat.com>
5802 * tree.h (id_equal): New.
5803 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
5804 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
5805 instead of strcmp of IDENTIFIER_POINTER.
5807 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5809 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
5810 (mark_all_inlined_calls_cdtor): Fix formating.
5811 (inline_transform): Rescale profile before inlining.
5813 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5815 * cgraph.h (cgraph_edge::clone): Update prototype.
5816 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
5817 (cgraph_node::create_clone): Update.
5818 (cgraph_node::create_version_clone): Update.
5819 * tree-inline.c (copy_bb): Update.
5820 (expand_call_inline): Update.
5822 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5824 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
5825 factored out from ...
5826 (rs6000_emit_prologue): ... here.
5828 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5830 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
5831 factored out from ...
5832 (rs6000_emit_prologue): ... here.
5834 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5836 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
5838 (handle_missing_profiles): Fix computation of tp_first_run.
5839 (counts_to_freqs): Do not touch freqs when count is 0.
5841 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5843 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
5846 2017-06-10 Tom de Vries <tom@codesourcery.com>
5848 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
5849 attributes): Document signal effective target.
5851 2017-06-10 Tom de Vries <tom@codesourcery.com>
5853 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
5854 Document effective target stack_size.
5856 2017-06-09 David Malcolm <dmalcolm@redhat.com>
5858 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
5859 to the edit_context if they can be auto-applied.
5861 2017-06-9 Ian Lance Taylor <iant@golang.org>
5863 * opts.c (finish_options): If -fsplit-stack, disable implicit
5864 -forder-blocks-and-partition.
5865 * doc/invoke.texi (Optimize Options): Document that when using
5866 -fsplit-stack -forder-blocks-and-partition is not implicitly
5869 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5871 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
5872 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
5873 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
5874 * builtins.def (abort, trap, unreachable): Declare cold.
5875 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
5876 * tree-core.h (ECF_COLD): New.
5877 * tree.c (set_call_expr_flags): Handle ECF_COLD.
5878 (build_common_builtin_nodes): Mark unreachable and abort as cold.
5880 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5882 * predict.c (unlikely_executed_stmt_p): Cleanup.
5884 2017-06-09 Richard Biener <rguenther@suse.de>
5886 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
5887 model if the ref is always written to.
5889 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5891 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
5893 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5895 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
5898 2017-06-09 Tom de Vries <tom@codesourcery.com>
5901 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
5902 "target cannot support label values" when encountering LABEL_REF.
5904 2017-06-09 Martin Liska <mliska@suse.cz>
5906 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
5907 (gimple_gen_ic_func_profiler): Emit direct comparison
5908 of __gcov_indirect_call_callee with NULL.
5909 (gimple_gen_time_profiler): Change probability from
5910 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
5912 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5914 * profile.c (edge_gcov_counts): Turn to pointer.
5915 (compute_branch_probabilities, compute_branch_probabilities): Update.
5916 (branch_prob): Do not clear edge_gcov_count.
5917 * profile.h (edge_gcov_counts): Turn to pointer.
5918 (edge_gcov_count): Update.
5920 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5922 * gimple.h (gimple_check_failed): Mark cold.
5924 2017-06-09 Richard Biener <rguenther@suse.de>
5926 PR tree-optimization/66623
5927 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
5928 refactor check_reduction into two parts, properly computing
5929 whether we have to check reduction validity for outer loop
5932 2017-06-09 Richard Biener <rguenther@suse.de>
5934 PR tree-optimization/79483
5935 * graphite-scop-detection.c (order): New global.
5936 (get_order): Compute bb to order mapping that satisfies code
5937 generation constraints.
5938 (cmp_pbbs): New helper.
5939 (build_scops): Start domwalk at entry block, sort generated
5942 2017-06-09 Richard Biener <rguenther@suse.de>
5945 * ipa-polymorphic-call.c
5946 (ipa_polymorphic_call_context::restrict_to_inner_class):
5947 Skip FIELD_DECLs with error_mark_node type.
5948 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
5951 2017-06-09 Martin Liska <mliska@suse.cz>
5953 * predict.c (struct branch_predictor): New struct.
5954 (test_prediction_value_range): New test.
5955 (predict_c_tests): New function.
5956 * selftest-run-tests.c (selftest::run_tests): Run the function.
5957 * selftest.h: Declare new tests.
5959 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
5962 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
5963 gen_add3_insn did not fail.
5964 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
5965 r0, construct that number in a temporary reg and add that reg to r0.
5966 If asked to put the result in r0 as well, fail.
5968 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
5970 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5971 for early expansion of vec_eqv.
5973 2017-06-08 Jakub Jelinek <jakub@redhat.com>
5976 * ubsan.c (instrument_null): Avoid pointless code temporary.
5977 (pass_ubsan::execute): Instrument aggregate arguments of calls.
5979 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
5983 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
5986 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
5989 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5991 * predict.c (unlikely_executed_edge_p): Move ahead.
5992 (probably_never_executed_edge_p): Use it.
5994 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5997 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
5998 gimple_call_builtin_p call.
6000 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6002 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
6003 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
6004 rtl_check_failed_type2, rtl_check_failed_code1,
6005 rtl_check_failed_code2, rtl_check_failed_code_mode,
6006 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
6007 rtvec_check_failed_bounds, rtl_check_failed_flag,
6008 _fatal_insn_not_found, _fatal_insn): Likewise.
6009 * tree.h (tree_contains_struct_check_failed,
6010 tree_check_failed, tree_not_check_failed,
6011 tree_class_check_failed, tree_range_check_failed,
6012 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
6013 tree_vec_elt_check_failed, phi_node_elt_check_failed,
6014 tree_operand_check_failed, omp_clause_check_failed,
6015 omp_clause_operand_check_failed, omp_clause_range_check_failed):
6018 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6020 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
6021 flag_branch_probabilities.
6022 * ipa-inline.c (edge_badness): Likewise.
6023 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
6024 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
6025 * predict.c (maybe_hot_frequency_p): Likewise.
6026 (probably_never_executed): Likewise.
6027 * sched-ebb.c (schedule_ebbs): Likewise.
6028 * sched-rgn.c (find_single_block_region): Likewise.
6029 * tracer.c (tail_duplicate): Likewise.
6031 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6033 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
6034 longer requires x_flag_profile_use.
6036 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6038 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
6039 instead of flag_reorder_blocks_and_partition.
6040 * dbxout.c (dbxout_function_end): Likewise.
6041 * dwarf2out.c (gen_subprogram_die): Likewise.
6042 * haifa-sched.c (sched_create_recovery_edges): Likewise.
6043 * hw-doloop.c (reorg_loops): Likewise.
6044 * varasm.c (assemble_start_function,
6045 assemble_end_function): Likewise.
6046 (decide_function_section): Do not check for
6047 flag_reorder_blocks_and_partition.
6049 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
6051 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
6053 (chkp_get_hard_register_fake_addr_expr): Ditto.
6054 (chkp_build_addr_expr): Add check for hard reg case.
6055 (chkp_parse_array_and_component_ref): Ditto.
6056 (chkp_find_bounds_1): Ditto.
6057 (chkp_process_stmt): Don't generate bounds store for
6060 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6062 * predict.c (maybe_hot_bb_p): Do not check profile status.
6063 (maybe_hot_edge_p): Likewise.
6064 (probably_never_executed): Check for zero counts even if profile
6066 (unlikely_executed_edge_p): New function.
6067 (unlikely_executed_stmt_p): New function.
6068 (unlikely_executed_bb_p): New function.
6069 (set_even_probabilities): Use unlikely predicates.
6070 (combine_predictions_for_bb): Likewise.
6071 (predict_paths_for_bb): Likewise.
6072 (predict_paths_leading_to_edge): Likewise.
6073 (determine_unlikely_bbs): New function.
6074 (estimate_bb_frequencies): Use it.
6075 (compute_function_frequency): Use zero counts even if profile is
6077 * profile-count.h: Fix typo.
6079 2017-08-08 Julia Koval <julia.koval@intel.com>
6081 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
6082 _mm512_mask_cvtsepi16_storeu_epi8,
6083 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
6084 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
6085 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
6086 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
6087 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
6088 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
6089 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
6090 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
6091 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
6092 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
6093 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
6094 __builtin_ia32_pmovuswb256mem_mask,
6095 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
6096 __builtin_ia32_pmovwb512mem_mask): New builtins.
6098 2017-08-08 Julia Koval <julia.koval@intel.com>
6100 PR target/73350,80862
6101 * config/i386/subst.md (round): Fix round pattern.
6102 * config/i386/i386.c (ix86_erase_embedded_rounding):
6103 Fix erasing rounding for the fixed pattern.
6105 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6107 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
6109 2017-06-08 Martin Liska <mliska@suse.cz>
6111 PR gcov-profile/80911
6112 * gcov.c (block_info::block_info): New constructor.
6114 2017-06-07 Carl Love <cel@us.ibm.com>
6116 * config/rs6000/rs6000-c: The return type of the following
6117 built-in functions was implemented as int not long long. Fix sign
6118 of return value for the unsigned version of vec_mulo and vec_mule.
6119 vector unsigned long long vec_bperm (vector unsigned long long,
6120 vector unsigned char)
6121 vector signed long long vec_mule (vector signed int,
6123 vector unsigned long long vec_mule (vector unsigned int,
6124 vector unsigned int)
6125 vector signed long long vec_mulo (vector signed int,
6127 vector unsigned long long vec_mulo (vector unsigned int,
6128 vector unsigned int)
6129 * doc/extend.texi: Fix the documentation for the built-in
6132 2017-06-07 Carl Love <cel@us.ibm.com>
6135 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
6138 2017-06-07 Carl Love <cel@us.ibm.com>
6140 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
6141 support, Generate doublehv for signed int/float for BE case only.
6143 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
6145 * doc/invoke.texi (mcx16): Rewrite.
6147 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6149 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
6150 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
6151 *mov<mode>_softfloat, and an anonymous splitter): Use
6152 nonimmediate_operand instead of rs6000_nonimmediate_operand.
6154 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6156 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
6158 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
6159 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
6160 (rs6000_debug_reg_global): Adjust.
6161 (rs6000_init_hard_regno_mode_ok): Adjust.
6162 (rs6000_dbx_register_number): Adjust.
6163 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
6164 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
6165 Remove SPE_ACC and SPEFSCR.
6166 (REG_ALLOC_ORDER): Ditto.
6167 (FRAME_POINTER_REGNUM): Change to 111.
6168 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
6169 (REG_CLASS_NAMES): Ditto.
6170 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
6171 (REGISTER_NAMES): Ditto.
6172 (ADDITIONAL_REG_NAMES): Ditto.
6173 (rs6000_reg_names): Ditto.
6174 * config/rs6000/rs6000.md: Renumber some register number
6177 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6179 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
6181 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
6182 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
6184 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
6185 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
6186 Delete the SPE high registers.
6187 (REG_ALLOC_ORDER): Ditto.
6188 (enum reg_class): Remove SPE_HIGH_REGS.
6189 (REG_CLASS_NAMES): Ditto.
6190 (REG_CLASS_CONTENTS): Delete the SPE high registers.
6191 (REGISTER_NAMES): Ditto.
6192 (rs6000_reg_names): Ditto.
6193 * doc/tm.texi.in: Remove SPE as example.
6194 * doc/tm.texi: Regenerate.
6196 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6198 * config/rs6000/8540.md (ppc8540_brinc): Delete.
6199 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
6200 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
6201 * config/rs6000/rs6000.md (type): Remove "brinc".
6203 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6205 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
6206 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
6207 * config/rs6000/linuxspe.h: Delete file.
6208 * config/rs6000/rs6000.md: Don't include spe.md.
6209 * config/rs6000/spe.h: Delete file.
6210 * config/rs6000/spe.md: Delete file.
6211 * config/rs6000/t-rs6000: Remove spe.md.
6213 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6215 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
6216 (reg_or_none500mem_operand): Delete.
6217 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
6218 instead of reg_or_none500mem_operand.
6220 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6222 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
6223 handling of SPE flags.
6224 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
6226 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6228 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
6230 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
6231 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
6232 paired_divv2sf3): Similar.
6233 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
6234 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
6235 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
6237 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
6238 Rename the paired_* instruction patterns.
6239 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
6241 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
6242 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
6243 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
6244 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
6245 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
6247 (struct machine_function): Delete field spe_insn_chain_scanned_p.
6248 (spe_func_has_64bit_regs_p): Delete.
6249 (spe_expand_predicate_builtin): Delete.
6250 (spe_expand_evsel_builtin): Delete.
6251 (TARGET_DWARF_REGISTER_SPAN): Do not define.
6252 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
6253 (invalid_e500_subreg): Delete.
6254 (rs6000_legitimize_address): Always force_reg op2 as well, for
6255 paired single memory accesses.
6256 (rs6000_member_type_forces_blk): Delete.
6257 (rs6000_spe_function_arg): Delete.
6258 (rs6000_expand_unop_builtin): Delete SPE handling.
6259 (rs6000_expand_binop_builtin): Ditto.
6260 (spe_expand_stv_builtin): Delete.
6261 (bdesc_2arg_spe): Delete.
6262 (spe_expand_builtin): Delete.
6263 (spe_expand_predicate_builtin): Delete.
6264 (spe_expand_evsel_builtin): Delete.
6265 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
6266 (spe_init_builtins): Delete.
6267 (spe_func_has_64bit_regs_p): Delete.
6268 (savres_routine_name): Delete "info" parameter. Adjust callers.
6269 (rs6000_emit_stack_reset): Ditto.
6270 (rs6000_dwarf_register_span): Delete.
6271 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
6272 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
6273 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
6275 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
6277 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
6278 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
6279 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
6280 mulv2sf3, divv2sf3): Delete expanders.
6282 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6284 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
6286 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6288 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
6289 * config/rs6000/rs6000.c: Ditto.
6291 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6293 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
6294 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
6295 comparison_operator.
6297 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6299 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
6300 * config/rs6000/rs6000.opt: Ditto.
6301 * config/rs6000/t-rtems: Ditto.
6303 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6305 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
6306 TARGET_E500_SINGLE by 0, simplify.
6307 * config/rs6000/rs6000.c: Ditto.
6308 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
6309 (spe_build_register_parallel): Delete.
6310 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
6311 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
6312 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
6313 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
6314 (E500_CONVERT): Delete.
6315 * config/rs6000/spe.md: Remove many patterns and all define_constants.
6317 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6319 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
6320 * config/rs6000/dfp.md: Ditto.
6321 (negdd2, *negdd2_fpr): Merge.
6322 (absdd2, *absdd2_fpr): Merge.
6323 (negtd2, *negtd2_fpr): Merge.
6324 (abstd2, *abstd2_fpr): Merge.
6325 * config/rs6000/e500.h: Delete file.
6326 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
6327 TARGET_FPRS by 1 and simplify.
6328 * config/rs6000/rs6000-c.c: Ditto.
6329 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
6331 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
6333 * config/rs6000/rs6000.md: Ditto.
6334 (floatdidf2, *floatdidf2_fpr): Merge.
6335 (move_from_CR_gt_bit): Delete.
6336 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
6337 (E500_CR_IOR_COMPARE): Delete.
6338 (All patterns that require !TARGET_FPRS): Delete.
6339 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
6341 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6343 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
6345 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6347 * graphds.c (add_edge): Intitialize edge's attached data.
6348 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
6349 pointer parameter. Call pointed function on each edge during
6350 graph traversing. Skip traversing the edge when the function
6352 (graphds_dfs, graphds_scc): Ditto.
6353 (for_each_edge): New parameter. Pass the new parameter to callback
6355 * graphds.h (skip_edge_callback): New function pointer type.
6356 (graphds_dfs, graphds_scc): New function pointer parameter.
6357 (graphds_edge_callback, for_each_edge): New parameter.
6359 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6361 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
6362 out code checking if runtime alias check is possible to below ...
6363 Call the new function.
6364 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
6365 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
6367 2017-06-07 Marek Polacek <polacek@redhat.com>
6370 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
6371 TYPE_OVERFLOW_WRAPS check.
6373 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6375 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
6376 if versioning is required.
6377 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
6378 peeling with the check for versioning.
6380 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6382 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
6383 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
6384 Set true to new parameter if new ssa variable is defined.
6385 (vect_gen_vector_loop_niters): Refactor. Set range information
6386 for the new vector loop bound variable.
6387 (vect_do_peeling): Ditto.
6389 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6391 * tree-affine.c (ssa.h): Include header file.
6392 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
6393 has wrapping overflow behavior.
6395 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6397 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
6399 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6401 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
6402 (tree_to_aff_combination): ... here.
6404 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6406 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
6407 reg_pressure model function.
6408 (ivopts_global_cost_for_size): Delete.
6409 (determine_set_costs, iv_ca_recount_cost): Call new model function
6410 ivopts_estimate_reg_pressure.
6412 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6414 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
6415 expensive than udiv. Remove floating point cases from mod.
6417 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6419 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
6422 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6424 * config/aarch64/aarch64.md
6425 (copysignsf3): Fix mask generation.
6427 2017-06-07 Jakub Jelinek <jakub@redhat.com>
6429 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
6431 (class dump_manager): Add register_dumps method.
6432 * dumpfile.c: Include langhooks.h.
6433 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
6434 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
6435 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
6436 (dump_manager::dump_register): Start with 512 entries instead of 32.
6437 (dump_manager::register_dumps): New method.
6438 * toplev.c (general_init): Instead of invoking register_dumps
6439 langhook, invoke register_dumps method on the dump manager.
6440 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
6443 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
6445 * doc/md.texi: Clarify the restrictions on a define_insn condition.
6446 Say that # requires an associated define_split to exist, and that
6447 the define_split must be suitable for use after register allocation.
6449 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6451 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
6452 (compute_outgoing_frequencies): Also initialize zero counts.
6453 (find_many_sub_basic_blocks): Do not produce uninitialized profile
6454 around loops; preserve more of profile when nothing changes.
6456 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
6458 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
6460 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
6461 * config/arm/arm-cpu-cdata.h: Regenerate.
6462 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
6463 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
6464 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
6465 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
6466 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
6468 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
6470 * config/arm/t-rmprofile: Likewise.
6471 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
6473 2017-06-06 David S. Miller <davem@davemloft.net>
6476 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
6477 blockage if function uses alloca.
6479 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
6481 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
6482 New "uid" fields to hold pretty-print IDs of group and ref.
6483 Memory references are now identified as <group_id>:<ref_id>
6484 instead of using [random] addresses.
6485 (dump_mem_details): Simplify, no functional change.
6486 (dump_mem_ref): Simplify and make output more concise.
6487 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
6488 (find_or_create_group): Initialize group uid.
6489 (record_ref): Initialize ref uid. Improve debug output.
6490 (prune_group_by_reuse, should_issue_prefetch_p,)
6491 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
6492 (mark_nontemporal_store, determine_loop_nest_reuse):
6493 Improve debug output.
6495 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
6497 * dbgcnt.def (prefetch): New debug counter.
6498 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
6499 (schedule_prefetches): Stop issueing prefetches if debug counter
6502 2017-06-06 Tom de Vries <tom@codesourcery.com>
6504 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
6505 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
6507 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6509 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
6510 Use aarch64_reg_or_zero predicate for operand 4.
6511 (aarch64_compare_and_swap<mode> define_insn_and_split):
6512 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
6513 (aarch64_store_exclusive<mode>): Likewise for operand 2.
6515 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
6517 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
6518 (arm_compute_save_core_reg_mask): This.
6519 (thumb1_compute_save_reg_mask): Rename into ...
6520 (thumb1_compute_save_core_reg_mask): This.
6521 (arm_compute_save_reg0_reg12_mask): Adapt comment.
6522 (arm_compute_frame_layout): Likewise.
6524 2017-06-06 Richard Biener <rguenther@suse.de>
6526 PR tree-optimization/80974
6527 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
6528 keep or clear leaders SSA info.
6530 2017-06-06 Tom de Vries <tom@codesourcery.com>
6532 * config/nvptx/nvptx.c (split_mode_p): New function.
6533 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
6535 2017-06-06 Tom de Vries <tom@codesourcery.com>
6537 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
6539 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6542 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
6545 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6547 * shrink-wrap.c (handle_simple_exit): Update profile.
6548 (try_shrink_wrapping): Upate profile.
6550 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6552 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
6553 (tree_guess_outgoing_edge_probabilities): New.
6554 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
6555 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
6557 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6559 * ipa-split.c (split_function): Initialize return bb profile.
6561 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6563 * profile.c (compute_branch_probabilities): Also initialize
6566 2017-06-06 Richard Biener <rguenther@suse.de>
6568 PR tree-optimization/80928
6569 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
6570 (vect_analyze_loop_operations): Properly guard analysis for
6572 (vect_transform_loop): Likewise.
6573 (vect_analyze_loop_2): Also reset SLP type on PHIs.
6574 (vect_model_induction_cost): Do not cost for pure SLP.
6575 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
6576 of induction in inner loop vectorization.
6577 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
6578 (vect_get_and_check_slp_defs): Handle vect_induction_def.
6579 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
6581 (vect_analyze_slp_cost_1): Cost induction.
6582 (vect_detect_hybrid_slp_stmts): Handle PHIs.
6583 (vect_get_slp_vect_defs): Likewise.
6584 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
6585 (vect_transform_stmt): Handle SLP reductions.
6586 * tree-vectorizer.h (vectorizable_induction): Adjust.
6588 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6590 * config/rs6000/rs6000.c (make_resolver_func): Update
6591 init_lowered_empty_function call.
6593 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
6595 * doc/invoke.texi: Document the -fprofile-abs-path option.
6596 * common.opt (fprofile-abs-path): New option.
6597 * gcov-io.h (gcov_write_filename): Declare.
6598 * gcov-io.c (gcov_write_filename): New function.
6599 * coverage.c (coverage_begin_function): Use gcov_write_filename.
6600 * profile.c (output_location): Likewise.
6602 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6604 * shring-wrap.c: Revert accidental commit.
6606 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
6608 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
6610 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6612 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
6614 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
6615 profile in callgraph edge.
6616 * profile-count.h (apply_probability): If THIS is 0, then result is 0
6617 (apply_scale): Likewise.
6618 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
6619 Also scale profile when inlining function with zero profile.
6620 (initialize_cfun): Update exit block profile even when it is zero.
6621 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
6622 when profile is read.
6624 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6626 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
6627 (CLONE_*): New constants to define the processors we can generate
6628 code for with the target_clone attribute.
6629 (rs6000_clone_map): New array to identify which clone processors
6630 the current program is running on.
6631 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
6632 target_clone attribute.
6633 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
6634 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
6635 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
6636 (cpu_expand_builtin): Add support for target_clone attribute.
6637 (rs6000_valid_attribute_p): Allow "default" attribute.
6638 (get_decl_name): New debug function to simplify printing the
6639 current function name in debugging statements.
6640 (rs6000_clone_priority): New functions to support the target_clone
6641 attribute, and be able to generate code to switch between ISA 2.05
6642 through ISA 3.0 (power6 through power9).
6643 (rs6000_compare_version_priority): Likewise.
6644 (rs6000_get_function_versions_dispatcher): Likewise.
6645 (make_resolver_func): Likewise.
6646 (add_condition_to_bb): Likewise.
6647 (dispatch_function_versions): Likewise.
6648 (rs6000_generate_version_dispatcher_body): Likewise.
6649 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
6650 (fusion_gpr_load_p): Fix a spacing issue.
6651 * doc/extend.texi (Common Function Attributes): Document that the
6652 PowerPC supports the target_clone attribute.
6654 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
6656 * config/arm/arm.h: explain F symbol found in description of ARM
6657 register allocation in its legend.
6659 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6661 * config/mips/frame-header-opt.c: Include profile-count.h.
6662 * config/riscv/riscv.c: Include profile-count.h
6664 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6666 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
6668 (sm_set_flag_if_changed): Add bbs field.
6669 (execute_sm_if_changed_flag_set): Pass BBS.
6670 (execute_sm): Update.
6672 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6674 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6677 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6679 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
6680 (peephole2): New peephole2 to emit the above.
6681 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
6683 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6685 * config/aarch64/aarch64.c (define_peephole2 above
6686 *sub_<shift>_<mode>): New peephole.
6688 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
6690 * config/i386/i386.c (make_resolver_func): Update.
6691 * Makefile.in: Add profile-count.h and profile-count.o
6692 * auto-profile.c (afdo_indirect_call): Update to new API.
6693 (afdo_set_bb_count): Update.
6694 (afdo_propagate_edge): Update.
6695 (afdo_propagate_circuit): Update.
6696 (afdo_calculate_branch_prob): Update.
6697 (afdo_annotate_cfg): Update.
6698 * basic-block.h: Include profile-count.h
6699 (struct edge_def): Turn count to profile_count.
6700 (struct basic_block_def): Likewie.
6701 (REG_BR_PROB_BASE): Move to profile-count.h
6702 (RDIV): Move to profile-count.h
6703 * bb-reorder.c (max_entry_count): Turn to profile_count.
6704 (find_traces): Update.
6705 (rotate_loop):Update.
6706 (connect_traces):Update.
6707 (sanitize_hot_paths):Update.
6708 * bt-load.c (migrate_btr_defs): Update.
6709 * cfg.c (RDIV): Remove.
6710 (init_flow): Use alloc_block.
6711 (alloc_block): Uninitialize count.
6712 (unchecked_make_edge): Uninitialize count.
6713 (check_bb_profile): Update.
6714 (dump_edge_info): Update.
6715 (dump_bb_info): Update.
6716 (update_bb_profile_for_threading): Update.
6717 (scale_bbs_frequencies_int): Update.
6718 (scale_bbs_frequencies_gcov_type): Update.
6719 (scale_bbs_frequencies_profile_count): New.
6720 * cfg.h (update_bb_profile_for_threading): Update.
6721 (scale_bbs_frequencies_profile_count): Declare.
6722 * cfgbuild.c (compute_outgoing_frequencies): Update.
6723 (find_many_sub_basic_blocks): Update.
6724 * cfgcleanup.c (try_forward_edges): Update.
6725 (try_crossjump_to_edge): Update.
6726 * cfgexpand.c (expand_gimple_tailcall): Update.
6727 (construct_exit_block): Update.
6728 * cfghooks.c (verify_flow_info): Update.
6729 (dump_bb_for_graph): Update.
6730 (split_edge): Update.
6731 (make_forwarder_block): Update.
6732 (duplicate_block): Update.
6733 (account_profile_record): Update.
6734 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
6735 (get_estimated_loop_iterations): Update.
6736 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
6737 (single_likely_exit): Update.
6738 * cfgloopmanip.c (scale_loop_profile): Update.
6740 (set_zero_probability): Update.
6741 (lv_adjust_loop_entry_edge): Update.
6742 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
6743 (purge_dead_edges): Update.
6744 (rtl_account_profile_record): Update.
6745 * cgraph.c (cgraph_node::create): Uninitialize count.
6746 (symbol_table::create_edge): Uninitialize count.
6747 (cgraph_update_edges_for_call_stmt_node): Update.
6748 (cgraph_edge::dump_edge_flags): Update.
6749 (cgraph_node::dump): Update.
6750 (cgraph_edge::maybe_hot_p): Update.
6751 * cgraph.h: Include profile-count.h
6752 (create_clone), create_edge, create_indirect_edge): Update.
6753 (cgraph_node): Turn count to profile_count.
6754 (cgraph_edge0: Likewise.
6755 (make_speculative, clone): Update.
6756 (create_edge): Update.
6757 (init_lowered_empty_function): Update.
6758 * cgraphclones.c (cgraph_edge::clone): Update.
6759 (duplicate_thunk_for_node): Update.
6760 (cgraph_node::create_clone): Update.
6761 * cgraphunit.c (cgraph_node::analyze): Update.
6762 (cgraph_node::expand_thunk): Update.
6763 * final.c (dump_basic_block_info): Update.
6764 * gimple-streamer-in.c (input_bb): Update.
6765 * gimple-streamer-out.c (output_bb): Update.
6766 * graphite.c (print_global_statistics): Update.
6767 (print_graphite_scop_statistics): Update.
6768 * hsa-brig.c: Include basic-block.h.
6769 * hsa-dump.c: Include basic-block.h.
6770 * hsa-gen.c (T sum_slice): Update.
6771 (convert_switch_statements):Update.
6772 * hsa-regalloc.c: Include basic-block.h.
6773 * ipa-chkp.c (chkp_produce_thunks): Update.
6774 * ipa-cp.c (struct caller_statistics): Update.
6775 (init_caller_stats): Update.
6776 (gather_caller_stats): Update.
6777 (ipcp_cloning_candidate_p): Update.
6778 (good_cloning_opportunity_p): Update.
6779 (get_info_about_necessary_edges): Update.
6780 (dump_profile_updates): Update.
6781 (update_profiling_info): Update.
6782 (update_specialized_profile): Update.
6783 (perhaps_add_new_callers): Update.
6784 (decide_about_value): Update.
6785 (ipa_cp_c_finalize): Update.
6786 * ipa-devirt.c (struct odr_type_warn_count): Update.
6787 (struct decl_warn_count): Update.
6788 (struct final_warning_record): Update.
6789 (possible_polymorphic_call_targets): Update.
6790 (ipa_devirt): Update.
6791 * ipa-fnsummary.c (redirect_to_unreachable): Update.
6792 * ipa-icf.c (sem_function::merge): Update.
6793 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
6794 * ipa-inline.c (compute_uninlined_call_time): Update.
6795 (compute_inlined_call_time): Update.
6796 (want_inline_small_function_p): Update.
6797 (want_inline_self_recursive_call_p): Update.
6798 (edge_badness): Update.
6799 (lookup_recursive_calls): Update.
6800 (recursive_inlining): Update.
6801 (inline_small_functions): Update.
6802 (dump_overall_stats): Update.
6803 (dump_inline_stats): Update.
6804 * ipa-profile.c (ipa_profile_generate_summary): Update.
6805 (ipa_propagate_frequency): Update.
6806 (ipa_profile): Update.
6807 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
6808 * ipa-utils.c (ipa_merge_profiles): Update.
6809 * loop-doloop.c (doloop_modify): Update.
6810 * loop-unroll.c (report_unroll): Update.
6811 (unroll_loop_runtime_iterations): Update.
6812 * lto-cgraph.c (lto_output_edge): Update.
6813 (lto_output_node): Update.
6814 (input_node): Update.
6815 (input_edge): Update.
6816 (merge_profile_summaries): Update.
6817 * lto-streamer-in.c (input_cfg): Update.
6818 * lto-streamer-out.c (output_cfg): Update.
6819 * mcf.c (create_fixup_graph): Update.
6820 (adjust_cfg_counts): Update.
6821 (sum_edge_counts): Update.
6822 * modulo-sched.c (sms_schedule): Update.
6823 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
6824 * predict.c (maybe_hot_count_p): Update.
6825 (probably_never_executed): Update.
6826 (dump_prediction): Update.
6827 (combine_predictions_for_bb): Update.
6828 (propagate_freq): Update.
6829 (handle_missing_profiles): Update.
6830 (counts_to_freqs): Update.
6831 (rebuild_frequencies): Update.
6832 (force_edge_cold): Update.
6833 * predict.h: Include profile-count.h
6834 (maybe_hot_count_p, counts_to_freqs): UPdate.
6835 * print-rtl-function.c: Do not include cfg.h
6836 * print-rtl.c: Include basic-block.h
6837 * profile-count.c: New file.
6838 * profile-count.h: New file.
6839 * profile.c (is_edge_inconsistent): Update.
6840 (correct_negative_edge_counts): Update.
6841 (is_inconsistent): Update.
6842 (set_bb_counts): Update.
6843 (read_profile_edge_counts): Update.
6844 (compute_frequency_overlap): Update.
6845 (compute_branch_probabilities): Update; Initialize and deinitialize
6847 (branch_prob): Update.
6848 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
6849 (edge_gcov_count): New.
6850 (bb_gcov_count): New.
6851 * shrink-wrap.c (try_shrink_wrapping): Update.
6852 * tracer.c (better_p): Update.
6853 * trans-mem.c (expand_transaction): Update.
6854 (ipa_tm_insert_irr_call): Update.
6855 (ipa_tm_insert_gettmclone_call): Update.
6856 * tree-call-cdce.c: Update.
6857 * tree-cfg.c (gimple_duplicate_sese_region): Update.
6858 (gimple_duplicate_sese_tail): Update.
6859 (gimple_account_profile_record): Update.
6860 (execute_fixup_cfg): Update.
6861 * tree-inline.c (copy_bb): Update.
6862 (copy_edges_for_bb): Update.
6863 (initialize_cfun): Update.
6864 (freqs_to_counts): Update.
6865 (copy_cfg_body): Update.
6866 (expand_call_inline): Update.
6867 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
6868 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
6869 (try_unroll_loop_completely): Update.
6870 (try_peel_loop): Update.
6871 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6872 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
6873 * tree-ssa-loop-split.c (connect_loops): Update.
6874 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
6875 * tree-ssa-reassoc.c (branch_fixup): Update.
6876 * tree-ssa-tail-merge.c (replace_block_by): Update.
6877 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
6878 (compute_path_counts): Update.
6879 (update_profile): Update.
6880 (recompute_probabilities): Update.
6881 (update_joiner_offpath_counts): Update.
6882 (estimated_freqs_path): Update.
6883 (freqs_to_counts_path): Update.
6884 (clear_counts_path): Update.
6885 (ssa_fix_duplicate_block_edges): Update.
6886 (duplicate_thread_path): Update.
6887 * tree-switch-conversion.c (case_bit_test_cmp): Update.
6888 (struct switch_conv_info): Update.
6889 * tree-tailcall.c (decrease_profile): Update.
6890 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
6891 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6892 * value-prof.c (check_counter): Update.
6893 (gimple_divmod_fixed_value): Update.
6894 (gimple_mod_pow2): Update.
6895 (gimple_mod_subtract): Update.
6896 (gimple_ic_transform): Update.
6897 (gimple_stringop_fixed_value): Update.
6898 * value-prof.h (gimple_ic): Update.
6900 2017-06-02 Carl Love <cel@us.ibm.com>
6902 * config/rs6000/rs6000-c: Add support for built-in functions
6903 vector double vec_doublee (vector signed int);
6904 vector double vec_doublee (vector unsigned int);
6905 vector double vec_doublee (vector float);
6906 vector double vec_doubleh (vector signed int);
6907 vector double vec_doubleh (vector unsigned int);
6908 vector double vec_doubleh (vector float);
6909 vector double vec_doublel (vector signed int);
6910 vector double vec_doublel (vector unsigned int);
6911 vector double vec_doublel (vector float);
6912 vector double vec_doubleo (vector signed int);
6913 vector double vec_doubleo (vector unsigned int);
6914 vector double vec_doubleo (vector float);.
6915 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
6916 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
6918 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
6919 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
6920 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
6922 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
6923 vec_doublel, vec_doubleh.
6924 * doc/extend.texi: Update the built-in documentation file for the
6925 new built-in functions.
6927 2017-06-02 David Malcolm <dmalcolm@redhat.com>
6930 * ipa-inline-analysis.c (free_growth_caches): Set
6931 edge_removal_hook_holder to NULL after removing it.
6933 2017-06-02 Sudakshina Das <sudi.das@arm.com>
6935 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
6936 comparision with zero.
6938 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6939 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6940 for early expansion of vec_min and vec_max builtins.
6941 (builtin_function_type): Add min/max unsigned variants to those
6942 identified as having unsigned arguments.
6944 2017-06-02 Olivier Hainque <hainque@adacore.com>
6946 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
6948 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6950 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
6951 Use VALL_F16 iterator rather than VALL.
6953 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6955 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
6956 Emit CBNZ inside loop when doing a strong exchange and comparing
6957 against zero. Generate the CC flags after the loop.
6959 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
6961 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
6962 (dl_section_ref): New.
6963 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
6964 On AIX, append an expression to subtract the size of the
6965 section length to dl_section_ref.
6967 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6969 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6970 for early expansion of vector absolute builtins.
6972 2017-06-02 Richard Biener <rguenther@suse.de>
6974 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
6975 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
6977 2017-06-02 Richard Biener <rguenther@suse.de>
6979 PR tree-optimization/80948
6980 * tree-tailcall.c (find_tail_calls): Track stmts to move in
6983 2017-06-02 Richard Biener <rguenther@suse.de>
6985 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
6987 * tree-vect-stmts.c (process_use): Do not mark backedge defs
6988 for inductions as relevant.
6990 2017-06-02 Richard Biener <rguenther@suse.de>
6992 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
6993 (vectorizable_induction): ... this. Remove dead code.
6995 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
6997 * builtins. (expand_builtin_alloca): Remove second parameter and
6998 infer its value from the first parameter instead.
6999 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
7001 2017-06-02 Jakub Jelinek <jakub@redhat.com>
7003 PR rtl-optimization/80903
7004 * loop-doloop.c (add_test): Unshare sequence.
7006 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
7008 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
7010 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
7012 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
7014 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
7015 xlogue_layout::get_instance, logue_layout::xlogue_layout,
7016 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
7017 (xlogue_layout::get_stub_rtx): Make static.
7018 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
7019 (xlogue_layout::compute_stub_managed_regs): Rename to...
7020 (xlogue_layout::count_stub_managed_regs): ...this.
7021 (xlogue_layout::is_stub_managed_reg): New function.
7022 (xlogue_layout::m_stub_names): Rename to...
7023 (xlogue_layout::s_stub_names): ...this, make static.
7024 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
7025 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
7026 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
7027 xlogue_layout::s_stub_names): Instantiate statics.
7028 (stub_managed_regs): Remove.
7029 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
7030 (disable_call_ms2sysv_xlogues): Rename to...
7031 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
7032 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
7034 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
7035 change after reload_completed.
7036 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
7038 (ix86_expand_prologue): Likewise.
7039 (ix86_expand_epilogue): Likewise.
7040 (ix86_expand_split_stack_prologue): Likewise.
7041 (ix86_compute_frame_layout): Remove frame parameter ...
7042 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
7043 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
7045 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
7046 (ix86_frame): Move from here ...
7047 * config/i386/i386.h (ix86_frame): ... to here.
7048 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
7049 complete ix86_frame data structure instead. Remove some_ld_name.
7051 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
7053 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
7054 symbols that hold a DECL_VALUE_EXPR.
7056 2017-06-01 Martin Jambor <mjambor@suse.cz>
7058 PR tree-optimization/80898
7059 * tree-sra.c (process_subtree_disqualification): Removed.
7060 (disqualify_candidate): Do not acll
7061 process_subtree_disqualification.
7062 (subtree_mark_written_and_enqueue): New function.
7063 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
7064 RHS has been disqualified and re-queue LHS if necessary. Apart
7065 from that, ignore disqualified RHS.
7067 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7069 * config/s390/s390.c (s390_emit_epilogue): Disable early return
7070 address fetch for z10 or later.
7072 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7074 * config/arc/arc.md (tst_movb): Add guard when splitting.
7076 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7078 * config/arc/arc.c (arc_can_eliminate): Test against
7079 arc_frame_pointer_needed.
7081 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7083 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
7084 to prevent store reordering.
7085 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
7086 (type): Add block type.
7087 (stack_tie): Define special instruction to be used in
7090 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7092 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
7093 constraint. It is not valid for the pattern.
7094 (noncommutative_binary_comparison): Likewise.
7096 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7098 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
7101 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7103 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
7104 be used by the reg-alloc.
7106 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7108 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
7109 reg-alloc when having mul64 or mul32x16 instructions.
7110 (mulsidi3): Likewise.
7111 (umulsidi3): Likewise.
7112 (mulsi32x16): New pattern.
7113 (mulsi64): Likewise.
7114 (mulsidi64): Likewise.
7115 (umulsidi64): Likewise.
7116 (MUL32x16_REG): Define.
7117 (mul64_600): Use MUL32x16_REG.
7118 (mac64_600): Likewise.
7119 (umul64_600): Likewise.
7120 (umac64_600): Likewise.
7122 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7124 * config/arc/arc.md (mulsi3_700): Make it commutative.
7126 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
7128 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
7130 (*sign_extendsidi2_insn): Likewise for movstosw.
7132 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
7134 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
7135 the type of the input discriminant value. Convert the
7136 discriminant value of signedness vary.
7138 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
7140 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
7141 Add to -Wall section.
7143 2017-06-01 Richard Biener <rguenther@suse.de>
7146 * fold-const.c (fold_plusminus_mult_expr): If the factored
7147 factor may be zero use a wrapping type for the inner operation.
7148 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
7149 and handle moved defs.
7150 (process_assignment): Properly guard the unary op case. Return a
7151 tri-state indicating that moving the stmt before the call may allow
7152 to continue. Pass through to_move.
7153 (find_tail_calls): Handle moving unrelated defs before
7156 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
7159 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
7160 splitter result in the canonical way.
7162 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
7164 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
7165 also for 32bit target. Update insn attributes.
7166 (zero-extendsidi2 splitter): Allow all registers for operand 1.
7168 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
7170 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
7171 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
7172 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
7173 (_mm_maskz_min_ss): New intrinsics.
7175 2017-05-31 Martin Liska <mliska@suse.cz>
7177 * tree-vect-loop.c (vect_create_epilog_for_reduction):
7178 Change comment style to one we normally use.
7179 (vectorizable_reduction): Likewise.
7180 (vectorizable_induction): Likewise.
7181 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7182 (vectorizable_call): Likewise.
7183 (vectorizable_simd_clone_call): Likewise.
7184 (vectorizable_conversion): Likewise.
7185 (vectorizable_assignment): Likewise.
7186 (vectorizable_shift): Likewise.
7187 (vectorizable_operation): Likewise.
7188 (vectorizable_store): Likewise.
7189 (vectorizable_load): Likewise.
7190 * tree-vectorizer.h: Likewise.
7192 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
7194 * passes.c (emergency_dump_function): New.
7195 * tree-pass.h (emergency_dump_function): Declare.
7196 * plugin.c (plugins_internal_error_function): Remove.
7197 * plugin.h (plugins_internal_error_function): Remove declaration.
7198 * toplev.c (internal_error_function): New static function. Use it...
7199 (general_init): ...here.
7201 2017-05-31 Graham Markall <graham.markall@embecosm.com>
7203 * config/arc/arc.c (arc_print_operand): Handle constant operands.
7204 (arc_rtx_costs): Add costs for new patterns.
7205 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
7206 * config/arc/predicates.md: Add _1_2_3_operand predicate.
7208 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
7210 * tree-ssa-strlen.c (get_next_strinfo): New function.
7211 (get_stridx_plus_constant): Use it.
7212 (zero_length_string): Likewise.
7213 (adjust_related_strinfos): Likewise.
7214 (adjust_last_stmt): Likewise.
7216 2017-05-31 Richard Biener <rguenther@suse.de>
7219 * config/i386/i386.c (ix86_expand_builtin): Remove assert
7220 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
7222 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
7224 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
7225 loop_vinfo argument and use of dependence distance vectors.
7226 Check instead whether the two references differ only in their
7227 initial value and assume that they have the same alignment if the
7228 difference is a multiple of the vector alignment.
7229 (vect_analyze_data_refs_alignment): Update call accordingly.
7231 2017-05-31 Martin Liska <mliska@suse.cz>
7234 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
7236 2017-05-31 Bin Cheng <bin.cheng@arm.com>
7238 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
7239 (create_intersect_range_checks): Move from ...
7240 * tree-data-ref.c (create_intersect_range_checks_index)
7241 (create_intersect_range_checks): ... to here.
7242 (create_runtime_alias_checks): New function factored from ...
7243 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
7244 here. Call above function.
7245 * tree-data-ref.h (create_runtime_alias_checks): New function.
7247 2017-05-31 Bin Cheng <bin.cheng@arm.com>
7249 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
7250 segment length for dr_b and compute it in wide_int.
7252 2017-05-31 Richard Biener <rguenther@suse.de>
7254 PR tree-optimization/80906
7255 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
7256 and pass through iv_map.
7257 (copy_bb_and_scalar_dependences): Adjust.
7258 (translate_pending_phi_nodes): Likewise.
7259 (copy_loop_close_phi_args): Handle code-generating IVs instead
7262 2017-05-30 David Malcolm <dmalcolm@redhat.com>
7264 * diagnostic-color.c (color_dict): Add "type-diff".
7265 (parse_gcc_colors): Update comment.
7266 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
7267 -fdiagnostics-show-template-tree and -fno-elide-type.
7268 (GCC_COLORS): Add type-diff to example.
7270 (-fdiagnostics-show-template-tree): New.
7271 (-fno-elide-type): New.
7272 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
7273 the pp_format_decoder callback. Call any m_format_postprocessor's
7275 (pretty_printer::pretty_printer): Initialize
7276 m_format_postprocessor.
7277 (pretty_printer::~pretty_printer): Delete any
7278 m_format_postprocessor.
7279 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
7280 (class format_postprocessor): New class.
7281 (struct pretty_printer::format_decoder): Document the new parameters.
7282 (struct pretty_printer::m_format_postprocessor): New field.
7283 * tree-diagnostic.c (default_tree_printer): Update for new
7284 bool and const char ** params.
7285 * tree-diagnostic.h (default_tree_printer): Likewise.
7287 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
7289 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
7290 (lwa_operand): Delete rs6000_gen_cell_microcode test.
7291 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
7292 rs6000_gen_cell_microcode code.
7293 (rs6000_final_prescan_insn): Delete.
7294 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
7295 "warn-cell-microcode" entries.
7296 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
7297 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
7298 throughout. Change cc_reg_not_micro_cr0_operand to
7299 cc_reg_not_cr0_operand throughout.
7300 (*extendhi<mode>2_noload): Delete.
7301 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
7302 (mwarn-cell-microcode): Delete.
7303 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
7304 -mgen-cell-microcode and -mwarn-cell-microcode.
7306 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
7309 * config/i386/constraints.md (Yd): New constraint.
7311 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
7312 and (?Yd, r) alternatives. Update insn attributes.
7313 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
7314 and (?*Yd, r) alternatives. Update insn attributes.
7315 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
7317 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
7319 * gimplify.c (gimplify_modify_expr): Don't create a
7320 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
7323 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
7325 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
7327 2017-05-30 Richard Biener <rguenther@suse.de>
7329 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
7330 and reduc_def fields.
7331 (STMT_VINFO_REDUC_TYPE): New define.
7332 (STMT_VINFO_REDUC_DEF): Likewise.
7333 (vect_force_simple_reduction): Adjust prototype.
7334 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
7335 (vect_is_simple_reduction): Remove check_reduction argument.
7336 (vect_force_simple_reduction): Adjust and set
7337 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
7338 (vectorizable_reduction): Do not re-do reduction analysis
7339 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
7340 * tree-parloops.c (gather_scalar_reductions): Adjust.
7342 2017-05-30 Richard Biener <rguenther@suse.de>
7345 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
7348 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
7350 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
7351 Introduce unknown_misalignment parameter and remove vf.
7352 (vect_peeling_hash_get_lowest_cost):
7353 Pass unknown_misalignment parameter.
7354 (vect_enhance_data_refs_alignment):
7355 Fix unsupportable data ref treatment.
7357 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7359 * tree-vect-data-refs.c (vect_get_data_access_cost):
7360 Workaround for SLP handling.
7361 (vect_enhance_data_refs_alignment):
7362 Compute costs for doing no peeling at all, compare to the best
7363 peeling costs so far and avoid peeling if cheaper.
7365 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7367 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
7368 Return peeling info and set costs to zero for unlimited cost
7370 (vect_enhance_data_refs_alignment): Also inspect all datarefs
7371 with unknown misalignment. Compute and costs for unknown
7372 misalignment, compare them to the costs for known misalignment
7373 and choose the cheapest for peeling.
7375 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7377 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
7378 (vect_get_peeling_costs_all_drs): Create function.
7379 (vect_peeling_hash_get_lowest_cost):
7380 Use vect_get_peeling_costs_all_drs.
7381 (vect_peeling_supportable): Create function.
7382 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
7384 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7386 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
7387 DR_HAS_NEGATIVE_STEP.
7388 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
7389 (vect_enhance_data_refs_alignment): Use.
7390 (vect_duplicate_ssa_name_ptr_info): Use.
7391 * tree-vectorizer.h (dr_misalignment): Use.
7392 (known_alignment_for_access_p): Use.
7394 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7397 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
7399 (has_section_name): New function.
7401 2017-05-30 Martin Liska <mliska@suse.cz>
7404 * auto-profile.c (get_function_decl_from_block): Fix
7407 2017-05-30 Richard Biener <rguenther@suse.de>
7410 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
7412 2017-05-30 Martin Liska <mliska@suse.cz>
7414 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
7415 * dumpfile.h (struct dump_file_info): Remove ctors.
7417 2017-05-30 Martin Liska <mliska@suse.cz>
7419 * predict.def: Fix GNU coding style.
7421 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
7423 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
7424 Mark 'to' argument with ATTRIBUTE_UNUSED.
7426 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
7428 * config/xtensa/xtensa.c (xtensa_emit_call): Use
7429 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
7430 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
7433 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
7435 * doc/install.texi (Options specification): Restore entry of
7436 --enable-sjlj-exceptions.
7438 2017-05-27 Michael Eager <eager@eagercon.com>
7441 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
7443 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
7445 * config/microblaze/microblaze.h
7446 (FIXED_REGISTERS): Update in macro.
7447 (CALL_USED_REGISTERS): Update in macro.
7449 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
7451 * doc/install.texi: Add links to macOS binary distributions.
7453 2017-05-27 Jakub Jelinek <jakub@redhat.com>
7457 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7459 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
7461 2017-05-26 Martin Liska <mliska@suse.cz>
7463 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
7465 2017-05-26 Martin Liska <mliska@suse.cz>
7467 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
7468 always leading ';; '.
7469 (dump_bb_info): Likewise.
7470 (brief_dump_cfg): Likewise.
7471 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
7472 * dumpfile.c: Remove usage of TDF_VERBOSE.
7473 * dumpfile.h (enum dump_kind): Likewise.
7474 (dump_gimple_bb_header): Do not use TDF_COMMENT.
7475 * print-tree.c (debug_verbose): Remove.
7476 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
7477 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
7478 * tree-diagnostic.c (default_tree_printer): Replace
7479 TDF_DIAGNOSTIC with TDF_SLIM.
7481 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7483 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
7484 in parameter loop, rather than loop_vinfo.
7485 (create_intersect_range_checks): Ditto.
7486 (vect_create_cond_for_alias_checks): Update call to above functions.
7488 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7490 PR tree-optimization/80815
7491 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
7492 for merging runtime alias checks. Handle negative DR_STEPs.
7494 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7496 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
7498 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
7499 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
7500 out code pruning runtime alias checks.
7501 * tree-data-ref.c (prune_runtime_alias_test_list): New function
7502 factored out from above.
7503 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
7505 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
7507 (prune_runtime_alias_test_list): New decalaration.
7509 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7511 * tree-vect-data-refs.c (compare_tree): Rename and move ...
7512 * tree-data-ref.c (data_ref_compare_tree): ... to here.
7513 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
7514 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
7515 (operator==, comp_dr_with_seg_len_pair): Ditto.
7516 (vect_prune_runtime_alias_test_list): Ditto.
7518 2017-05-26 Martin Liska <mliska@suse.cz>
7521 * params.def: Bound partial-inlining-entry-probability param.
7523 2017-05-26 Marek Polacek <polacek@redhat.com>
7526 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
7529 2017-05-26 Richard Biener <rguenther@suse.de>
7531 PR tree-optimization/80842
7532 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
7535 2017-05-26 Richard Biener <rguenther@suse.de>
7537 PR tree-optimization/80844
7538 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
7540 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
7542 * doc/md.texi (Machine Constraints): Update x86 family
7543 machine constraints section to match 'config/i386/constraints.md'.
7545 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
7547 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
7549 2017-05-25 Nathan Sidwell <nathan@acm.org>
7551 * doc/invoke.texi (--enable-languages): Update documentation.
7553 2017-05-25 Martin Liska <mliska@suse.cz>
7555 * dumpfile.c: Add TDF_FOLDING.
7556 * dumpfile.h (enum dump_kind): Likewise.
7557 * genmatch.c (dt_simplify::gen_1): Use it.
7559 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7561 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
7563 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7565 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
7566 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
7568 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7570 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
7571 * match.pd (X == C): Rewrite it here.
7572 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
7573 with_certain_nonzero_bits2): New predicates.
7574 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
7576 2017-05-24 Nathan Sidwell <nathan@acm.org>
7578 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
7581 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
7584 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
7586 * config/powerpcspe: New port. Files are copied from the rs6000
7587 port, with "rs6000" in filenames replaced by "powerpcspe".
7589 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
7591 PR rtl-optimization/80754
7592 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
7594 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
7596 * config/sparc/sparc.md (length): Return the correct value for -mflat
7597 sibcalls to match output_sibcall.
7599 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
7603 * config/rs6000/rs6000.c (struct machine_function): Add new field
7605 (rs6000_get_separate_components): Init that field, use it.
7606 (rs6000_components_for_bb): Use the field.
7608 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
7610 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
7612 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
7615 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
7617 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7620 * config/s390/s390.c (s390_check_qrst_address): Check incoming
7621 address against address_operand predicate.
7622 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
7624 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
7626 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
7627 parameters passed indirectly.
7629 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
7631 * config/i386/i386.md (*movdi_internal): Remove SSE4
7632 alternative 18 (?r, *v). Update insn attributes.
7633 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
7634 Update insn attributes.
7635 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
7636 Update insn attributes.
7637 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
7638 alternative 1 (r, v). Remove isa attribute.
7639 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
7640 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
7641 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
7643 2017-05-23 Tom de Vries <tom@codesourcery.com>
7645 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
7648 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7650 * cgraphunit.c (symbol_table::process_new_functions): Update.
7651 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
7652 (inline_generate_summary): Rename to ...
7653 (ipa_fn_summary_generate): ... this one.
7654 (inline_read_summary): Rename to ...
7655 (ipa_fn_summary_read): ... this one.
7656 (inline_write_summary): Rename to ...
7657 (ipa_fn_summary_write): ... this one.
7658 (inline_free_summary): Rename to ...
7659 (ipa_free_fn_summary): ... this one.
7660 (pass_data_local_fn_summary, pass_local_fn_summary,
7661 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
7662 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
7663 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
7664 make_pass_ipa_fn_summary): New.
7665 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
7666 inline_write_summary, inline_free_summary): Remove.
7667 (ipa_free_fn_summary) : New.
7668 * ipa-inline.c (ipa_inline): Update.
7669 (pass_ipa_inline): Do not generate summaries.
7670 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
7672 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
7673 and add pass_ipa_fn_summary.
7674 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
7676 (make_pass_inline_parameters): Remove.
7678 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
7680 * omp-low.c (struct omp_context): Remove "default_kind" member.
7683 * omp-offload.c (execute_oacc_device_lower): Remove the
7684 parallelism dimensions function attributes for unparallelized
7685 OpenACC kernels constructs.
7687 2017-05-23 Martin Liska <mliska@suse.cz>
7689 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
7691 (cgraph_edge::make_speculative): Likewise.
7692 (cgraph_edge::resolve_speculation): Likewise.
7693 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
7694 (cgraph_node::dump): Likewise.
7695 * cgraph.h: Likewise.
7696 * cgraphunit.c (analyze_functions): Likewise.
7697 (symbol_table::compile): Likewise.
7698 * ipa-cp.c (print_all_lattices): Likewise.
7699 (determine_versionability): Likewise.
7700 (initialize_node_lattices): Likewise.
7701 (ipcp_verify_propagated_values): Likewise.
7702 (estimate_local_effects): Likewise.
7703 (update_profiling_info): Likewise.
7704 (create_specialized_node): Likewise.
7705 (perhaps_add_new_callers): Likewise.
7706 (decide_about_value): Likewise.
7707 (decide_whether_version_node): Likewise.
7708 (identify_dead_nodes): Likewise.
7709 (ipcp_store_bits_results): Likewise.
7710 * ipa-devirt.c (dump_targets): Likewise.
7711 (ipa_devirt): Likewise.
7712 * ipa-icf.c (sem_item::dump): Likewise.
7713 (sem_function::equals): Likewise.
7714 (sem_variable::equals): Likewise.
7715 (sem_item_optimizer::read_section): Likewise.
7716 (sem_item_optimizer::execute): Likewise.
7717 (congruence_class::dump): Likewise.
7718 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
7719 (dump_inline_summary): Likewise.
7720 (estimate_node_size_and_time): Likewise.
7721 (inline_analyze_function): Likewise.
7722 * ipa-inline-transform.c (inline_call): Likewise.
7723 * ipa-inline.c (report_inline_failed_reason): Likewise.
7724 (want_early_inline_function_p): Likewise.
7725 (edge_badness): Likewise.
7726 (update_edge_key): Likewise.
7727 (inline_small_functions): Likewise.
7728 * ipa-profile.c (ipa_profile): Likewise.
7729 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
7730 (ipa_make_edge_direct_to_target): Likewise.
7731 (remove_described_reference): Likewise.
7732 (ipa_impossible_devirt_target): Likewise.
7733 (propagate_controlled_uses): Likewise.
7734 (ipa_print_node_params): Likewise.
7735 (ipcp_transform_function): Likewise.
7736 * ipa-pure-const.c (pure_const_read_summary): Likewise.
7737 (propagate_pure_const): Likewise.
7738 * ipa-reference.c (generate_summary): Likewise.
7739 (read_write_all_from_decl): Likewise.
7740 (propagate): Likewise.
7741 (ipa_reference_read_optimization_summary): Likewise.
7742 * ipa-utils.c (ipa_merge_profiles): Likewise.
7743 * ipa.c (walk_polymorphic_call_targets): Likewise.
7744 (symbol_table::remove_unreachable_nodes): Likewise.
7745 (ipa_single_use): Likewise.
7746 * passes.c (execute_todo): Likewise.
7747 * predict.c (drop_profile): Likewise.
7748 * symtab.c (symtab_node::get_dump_name): New function.
7749 (symtab_node::dump_name): Likewise.
7750 (symtab_node::dump_asm_name): Likewise.
7751 (symtab_node::dump_references): Likewise.
7752 (symtab_node::dump_referring): Likewise.
7753 (symtab_node::dump_base): Likewise.
7754 (symtab_node::debug_symtab): Likewise.
7755 * tree-sra.c (convert_callers_for_node): Likewise.
7756 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
7757 * value-prof.c (init_node_map): Likewise.
7759 2017-05-23 Martin Liska <mliska@suse.cz>
7761 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
7762 and symtab_node::debug_symtab to symbol_table::debug.
7763 * cgraphunit.c (analyze_functions): Use the renamed function.
7764 (symbol_table::compile): Likewise.
7765 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
7766 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
7767 * passes.c (execute_todo): Likewise.
7768 * symtab.c (symbol_table::dump): New function.
7769 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
7771 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7773 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
7774 that nonconst implies exec.
7776 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7778 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7779 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7780 (inline_edge_summary_vec): Turn into ...
7781 (ipa_call_summaries): ... this one.
7782 (redirect_to_unreachable, edge_set_predicate,
7783 evaluate_properties_for_edge, inline_summary_alloc,
7784 reset_ipa_call_summary, reset_inline_summary,
7785 inline_summary_t::duplicate): Update.
7786 (inline_edge_duplication_hook): Turn to ...
7787 (ipa_call_summary_t::duplicate): ... this one.
7788 (inline_edge_removal_hook): Turn to ...
7789 (ipa_call_summary_t::remove): ... this one.
7790 (dump_inline_edge_summary): Turn to ...
7791 (dump_ipa_call_summary): ... this one.
7792 (estimate_function_body_sizes): Update.
7793 (inline_update_callee_summaries): Update.
7794 (remap_edge_change_prob): Update.
7795 (remap_edge_summaries): Update.
7796 (inline_merge_summary): Update.
7797 (do_estimate_edge_time): Update.
7798 (inline_generate_summary): Update.
7799 (inline_read_section): Update.
7800 (inline_read_summary): Update.
7801 (inline_free_summary): Update.
7802 * ipa-inline.c (can_inline_edge_p): Update.
7803 (compute_inlined_call_time): Update.
7804 (want_inline_small_function_p): Update.
7805 (edge_badness): Update.
7806 (early_inliner): Update.
7807 * ipa-inline.h (inline_edge_summary): Turn to ...
7808 (ipa_call_summary): ... this one.
7809 (ipa_call_summary_t): New class.
7810 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
7811 (ipa_call_summaries): New.
7812 (inline_edge_summary): Remove.
7813 (estimate_edge_growth): Update.
7814 * ipa-profile.c (ipa_propagate_frequency_1): Update.
7815 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7816 * ipa-split.c (execute_split_functions): Update.
7817 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
7819 2017-05-23 Tom de Vries <tom@codesourcery.com>
7821 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7822 attributes): Document rdrand effective target.
7824 2017-05-23 Tom de Vries <tom@codesourcery.com>
7826 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7827 attributes): Sort alphabetically.
7829 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
7831 * config/avr/genmultilib.awk: Use gsub instead of gensub.
7833 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
7836 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
7837 V2DF/V2DI splat into two separate patterns, one that handles
7838 registers, and the other that only handles memory. Drop support
7839 for splatting from a GPR on ISA 2.07 and then splitting the
7840 splat into direct move and splat.
7841 (vsx_splat_<mode>_reg): Likewise.
7842 (vsx_splat_<mode>_mem): Likewise.
7844 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
7846 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
7848 2017-05-22 Jakub Jelinek <jakub@redhat.com>
7851 * omp-low.c (finish_taskreg_remap): New function.
7852 (finish_taskreg_scan): If unit size of ctx->record_type
7853 is non-constant, unshare the size expression and replace
7854 decls in it with possible outer var refs.
7857 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
7858 GOVD_SHARED rather than GOVD_PRIVATE with it.
7859 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
7860 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
7863 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
7864 as last argument to build_outer_var_ref for pointer bases of array
7867 2017-05-19 Martin Sebor <msebor@redhat.com>
7869 * print-tree.c (print_node): Print DECL_READ_P flag.
7871 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7873 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
7874 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
7875 * cgraph.c: Likewise.
7876 * cgraphunit.c: Likewise.
7877 * gengtype.c: Likewise.
7878 * ipa-cp.c: Likewise.
7879 * ipa-devirt.c: Likewise.
7880 * ipa-icf.c: Likewise.
7881 * ipa-predicate.c: Likewise.
7882 * ipa-profile.c: Likewise.
7883 * ipa-prop.c: Likewise.
7884 * ipa-split.c: Likewise.
7886 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
7887 edge_predicate_pool, dump_inline_hints,
7888 inline_summary::account_size_time, redirect_to_unreachable,
7889 edge_set_predicate, set_hint_predicate,
7890 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
7891 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
7892 inline_summary_t::remove, remap_hint_predicate_after_duplication,
7893 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
7894 ipa_call_summary_t::remove, initialize_growth_caches,
7895 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
7896 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
7897 mark_modified, unmodified_parm_1, unmodified_parm,
7898 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
7899 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
7900 compute_bb_predicates, will_be_nonconstant_expr_predicate,
7901 will_be_nonconstant_predicate, record_modified_bb_info,
7902 get_minimal_bb, record_modified, param_change_prob,
7903 phi_result_unknown_predicate, predicate_for_phi_result,
7904 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
7905 estimate_function_body_sizes, compute_inline_parameters,
7906 compute_inline_parameters_for_curren, pass_data_inline_parameters,
7907 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
7908 inline_update_callee_summaries, remap_edge_change_prob,
7909 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
7910 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
7911 inline_analyze_function, inline_summary_t::insert,
7912 inline_generate_summary, read_ipa_call_summary, inline_read_section,
7913 inline_read_summary, write_ipa_call_summary, inline_write_summary,
7914 inline_free_summary): Move to ipa-fnsummary.h
7915 (predicate_t): Remove.
7916 * ipa-fnsummary.c: New file.
7917 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
7918 (enum inline_hints_vals, inline_hints, agg_position_info,
7919 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
7920 inline_summaries, ipa_call_summary, ipa_call_summary_t,
7921 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
7922 dump_inline_summary, dump_inline_hints, inline_generate_summary,
7923 inline_read_summary, inline_write_summary, inline_free_summary,
7924 inline_analyze_function, initialize_inline_failed,
7925 inline_merge_summary, inline_update_overall_summary,
7926 compute_inline_parameters): Move to ipa-fnsummary.h
7927 * ipa-fnsummary.h: New file.
7928 * ipa-inline-transform.h: Include ipa-inline.h.
7929 * ipa-inline.c: LIkewise.
7931 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7933 * ipa-inline.c (edge_badness): Use inlined_time instead of
7934 inline_summaries->get.
7936 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7938 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
7940 2017-05-22 Nathan Sidwell <nathan@acm.org>
7942 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
7943 (fdump-lang): Document 'raw' option.
7944 * dumpfile.h (TDI_tu): Delete.
7945 * dumpfile.c (dump_files): Remove translation-unit.
7946 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
7948 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
7950 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
7951 command option from $(AWK) call.
7952 * config/avr/genmultilib.awk: Simplify and rewrite so that it
7953 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
7954 [FORMAT]: Remove handling of variable.
7955 * config/avr/t-multilib: Regenerate.
7957 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7959 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
7961 (dump_inline_summary): Do not print self_time.
7962 (estimate_function_body_sizes): Do not set self_time.
7963 (compute_inline_parameters): Likewise.
7964 (inline_read_section, inline_write_summary): Do not stream self_time.
7965 * ipa-inline.h (inline_summary): Drop self_time.
7967 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7969 * ipa-inline-analysis.c (account_size_time): Rename to ...
7970 (inline_summary::account_size_time): ... this one.
7971 (reset_ipa_call_summary): Turn to ...
7972 (ipa_call_summary::reset): ... this one.
7973 (reset_inline_summary): Turn to ...
7974 (inline_summary::reset): ... this one.
7975 (inline_summary_t::remove): Update.
7976 (inline_summary_t::duplicate): Update.
7977 (ipa_call_summary_t::remove): Update.
7978 (dump_inline_summary): Update.
7979 (estimate_function_body_sizes): Update.
7980 (compute_inline_parameters): Update.
7981 (estimate_node_size_and_time): Update.
7982 (inline_merge_summary): Update.
7983 (inline_update_overall_summary): Update.
7984 (inline_read_section): Update.
7985 (inline_write_summary): Update.
7986 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
7987 add account_size_time and reset member functions.
7988 (ipa_call_summary): Add reset function.
7989 * ipa-predicate.h (predicate::operator &): Constify.
7991 2017-05-22 Richard Biener <rguenther@suse.de>
7993 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
7995 2017-05-19 Jason Merrill <jason@redhat.com>
7997 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
7999 2017-05-19 Marek Polacek <polacek@redhat.com>
8002 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
8003 TYPE_OVERFLOW_WRAPS checks.
8005 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
8007 * tree-core.h (enum omp_clause_default_kind): Add
8008 "OMP_CLAUSE_DEFAULT_PRESENT".
8009 * tree-pretty-print.c (dump_omp_clause): Handle it.
8010 * gimplify.c (enum gimplify_omp_var_data): Add
8011 "GOVD_MAP_FORCE_PRESENT".
8012 (gimplify_adjust_omp_clauses_1): Map it to
8013 "GOMP_MAP_FORCE_PRESENT".
8014 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
8016 * gimplify.c (oacc_default_clause): Clarify.
8018 2017-05-19 Nathan Sidwell <nathan@acm.org>
8020 LANG_HOOK_REGISTER_DUMPS
8021 * toplev.c (general_init): Call register dump lang hook.
8022 * doc/invoke.texi: Document -fdump-lang option family.
8023 * dumpfile.c (dump_files): Remove class dump here.
8024 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
8025 * dumpfile.h (tree_dump_index): Remove TDI_class.
8026 * langhooks-def.h (lhd_register_dumps): Declare.
8027 (LANG_HOOKS_REGISTER_DUMPS): Define.
8028 (LANG_HOOKS_INITIALIZER): Add it.
8029 * langhooks.c (lhd_register_dumps): Define.
8030 * langhooks.h (struct lang_hooks): Add register_dumps.
8032 2017-05-19 Nathan Sidwell <nathan@acm.org>
8034 * context.h (context::set_passes): New.
8035 * context.c (context::context): Do not create pass manager.
8036 * toplev.c (general_init): Create pass manager here.
8038 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
8040 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
8041 use this splitter if two add or or instructions would also work for
8042 the constant we want to generate.
8044 2017-05-19 Richard Biener <rguenther@suse.de>
8047 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
8048 predicate evaluation.
8050 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
8052 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
8054 * ipa-inline.c (want_inline_small_function_p): Do not cast to
8057 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
8059 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
8060 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
8061 (inline_edge_summary_vec): Turn into ...
8062 (ipa_call_summaries): ... this one.
8063 (redirect_to_unreachable, edge_set_predicate,
8064 evaluate_properties_for_edge, inline_summary_alloc,
8065 reset_ipa_call_summary, reset_inline_summary,
8066 inline_summary_t::duplicate): Update.
8067 (inline_edge_duplication_hook): Turn to ...
8068 (ipa_call_summary_t::duplicate): ... this one.
8069 (inline_edge_removal_hook): Turn to ...
8070 (ipa_call_summary_t::remove): ... this one.
8071 (dump_inline_edge_summary): Turn to ...
8072 (dump_ipa_call_summary): ... this one.
8073 (estimate_function_body_sizes): Update.
8074 (inline_update_callee_summaries): Update.
8075 (remap_edge_change_prob): Update.
8076 (remap_edge_summaries): Update.
8077 (inline_merge_summary): Update.
8078 (do_estimate_edge_time): Update.
8079 (inline_generate_summary): Update.
8080 (inline_read_section): Update.
8081 (inline_read_summary): Update.
8082 (inline_free_summary): Update.
8083 * ipa-inline.c (can_inline_edge_p): Update.
8084 (compute_inlined_call_time): Update.
8085 (want_inline_small_function_p): Update.
8086 (edge_badness): Update.
8087 (early_inliner): Update.
8088 * ipa-inline.h (inline_edge_summary): Turn to ...
8089 (ipa_call_summary): ... this one.
8090 (ipa_call_summary_t): New class.
8091 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
8092 (ipa_call_summaries): New.
8093 (inline_edge_summary): Remove.
8094 (estimate_edge_growth): Update.
8095 * ipa-profile.c (ipa_propagate_frequency_1): Update.
8096 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
8097 * ipa-split.c (execute_split_functions): Update.
8098 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
8100 2017-05-19 Richard Biener <rguenther@suse.de>
8103 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
8105 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
8107 * config/rs6000/rs6000.c (struct machine_function): Add field
8108 fpr_is_wrapped_separately.
8109 (rs6000_get_separate_components): Use 64 components. Handle the
8111 (rs6000_components_for_bb): Handle the FPR components.
8112 (rs6000_emit_prologue_components): Handle the FPR components.
8113 (rs6000_emit_epilogue_components): Handle the FPR components.
8114 (rs6000_set_handled_components): Handle the FPR components.
8115 (rs6000_emit_prologue): Don't output prologue code for those FPRs
8116 that are already separately shrink-wrapped.
8117 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
8118 that are already separately shrink-wrapped.
8120 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
8123 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
8126 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
8127 (define_peephole2 for Altivec d-form load): Add peepholes to catch
8128 cases where the register allocator uses a move and an offsettable
8129 memory operation to/from a FPR register on ISA 2.06/2.07.
8130 (define_peephole2 for Altivec d-form store): Likewise.
8132 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
8135 * config/i386/mmx.md (*mov<mode>_internal): Enable
8136 alternatives 11, 12, 13 and 14 also for 32bit targets.
8137 Remove alternatives 15, 16, 17 and 18.
8138 * config/i386/sse.md (vec_concatv2di): Change
8139 alternative (!x, *y) to (x, ?!*Yn).
8141 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
8143 * dumpfile.h (enum dump_kind): Remove stray comma.
8145 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
8147 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
8148 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
8149 predicate::num_conditions
8150 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
8151 (CHANGED): turn into predicate::changed.
8152 (agg_position_info): Move to ipa-predicate.h
8153 (add_condition, predicate::add_clause, predicate::operator &=,
8154 predicate::or_with, predicate::evaluate, predicate::probability,
8155 dump_condition, dump_clause, predicate::dump,
8156 predicate::remap_after_duplication, predicate::remap_after_inlining,
8157 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
8158 (evaluate_conditions_for_known_args): Update.
8159 (set_cond_stmt_execution_predicate): Update.
8160 * ipa-inline.h: Include ipa-predicate.h
8161 (condition, inline_param_summary, conditions, agg_position_info,
8162 predicate): Move to ipa-predicate.h
8163 * ipa-predicate.c: New file.
8164 * ipa-predicate.h: New file.
8166 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
8168 * final.c (leaf_function_p): Check we are not in a sequence.
8170 2017-05-18 Martin Liska <mliska@suse.cz>
8172 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
8173 * dumpfile.c (dump_register): Use new enum dump_kind.
8174 (get_dump_file_name): Likewise.
8175 (dump_enable_all): Likewise.
8176 (dump_switch_p_1): Likewise.
8177 (enable_rtl_dump_file): Remove usage of TDF_RTL.
8178 * dumpfile.h (enum dump_kind): New enum type.
8179 (struct dump_file_info): Create constructor and
8180 format fields and comments.
8181 * passes.c (pass_manager::register_one_dump_file):
8183 * statistics.c (statistics_early_init): Likewise.
8184 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
8185 TDF_TREE with TDF_SLIM.
8186 (gather_memory_references_ref): Likewise.
8188 2017-05-18 Martin Liska <mliska@suse.cz>
8190 * vec.h (struct vnull): Use it.
8192 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
8194 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
8195 (true_predicate, false_predicate, true_predicate_p,
8196 false_predicate_p): Remove.
8197 (single_cond_predicate, not_inlined_predicate): Turn to member function
8199 (add_condition): Update.
8200 (add_clause): Turn to...
8201 (predicate::add_clause): ... this one; update; allow passing NULL
8203 (and_predicates): Turn to ...
8204 (predicate::operator &=): ... this one.
8205 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
8206 (or_predicates): Turn to ...
8207 (predicate::or_with): ... this one.
8208 (evaluate_predicate): Turn to ...
8209 (predicate::evaluate): ... this one.
8210 (predicate_probability): Turn to ...
8211 (predicate::probability): ... this one.
8212 (dump_condition): Update.
8213 (dump_predicate): Turn to ...
8214 (predicate::dump): ... this one.
8215 (account_size_time): Update.
8216 (edge_set_predicate): Update.
8217 (set_hint_predicate): UPdate.
8218 (evaluate_conditions_for_known_args): Update.
8219 (evaluate_properties_for_edge): Update.
8220 (remap_predicate_after_duplication): Turn to...
8221 (predicate::remap_after_duplication): ... this one.
8222 (remap_hint_predicate_after_duplication): Update.
8223 (inline_summary_t::duplicate): UPdate.
8224 (dump_inline_edge_summary): Update.
8225 (dump_inline_summary): Update.
8226 (set_cond_stmt_execution_predicate): Update.
8227 (set_switch_stmt_execution_predicate): Update.
8228 (compute_bb_predicates): Update.
8229 (will_be_nonconstant_expr_predicate): Update.
8230 (will_be_nonconstant_predicate): Update.
8231 (phi_result_unknown_predicate): Update.
8232 (predicate_for_phi_result): Update.
8233 (array_index_predicate): Update.
8234 (estimate_function_body_sizes): Update.
8235 (estimate_node_size_and_time): Update.
8236 (estimate_ipcp_clone_size_and_time): Update.
8237 (remap_predicate): Rename to ...
8238 (predicate::remap_after_inlining): ... this one.
8239 (remap_hint_predicate): Update.
8240 (inline_merge_summary): Update.
8241 (inline_update_overall_summary): Update.
8242 (estimate_size_after_inlining): Update.
8243 (read_predicate): Rename to ...
8244 (predicate::stream_in): ... this one.
8245 (read_inline_edge_summary): Update.
8246 (write_predicate): Rename to ...
8247 (predicate::stream_out): ... this one.
8248 (write_inline_edge_summary): Update.
8249 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
8250 (clause_t): Turn to uint32_t
8251 (predicate): Turn to class; implement constructor and operators
8253 (size_time_entry): Update.
8254 (inline_summary): Update.
8255 (inline_edge_summary): Update.
8257 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
8259 * fold-const.c (fold_binary_loc): Move transformation...
8260 * match.pd (C - X CMP X): ... here.
8262 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
8264 * config/sparc/sparc.c (sparc_option_override): Set function
8265 alignment for -mcpu=niagara7 to 64 to match the I$ line.
8266 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
8268 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
8270 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
8272 2017-05-18 Marek Polacek <polacek@redhat.com>
8275 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
8276 (pass_ubsan::execute): Call gimple_assign_single_p instead of
8277 gimple_assign_load_p.
8279 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
8282 * real.c (do_compare): Give decimal_do_compare preference over
8283 comparing just the signs.
8285 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
8287 * doc/md.texi (Canonicalization of Instructions): Describe the
8288 canonical form of instructions that inherently set a condition
8291 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
8294 * tree-cfg.c: Move deletion of unreachable case statements to after
8295 the merging of consecutive case labels.
8297 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
8299 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
8300 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
8301 restoring of callee-saved registers.
8303 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
8305 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
8306 * config/visium/visium.c (single_set_and_flags): Likewise.
8307 * config/visium/visium.md (Substitutions): Likewise.
8309 2017-05-17 Martin Liska <mliska@suse.cz>
8311 * cfg.c: Introduce dump_flags_t type and
8312 use it instead of int type.
8314 * cfghooks.c: Likewise.
8315 * cfghooks.h (struct cfg_hooks): Likewise.
8316 * cfgrtl.c: Likewise.
8317 * cfgrtl.h: Likewise.
8318 * cgraph.c (cgraph_node::get_body): Likewise.
8319 * coretypes.h: Likewise.
8320 * domwalk.c: Likewise.
8321 * domwalk.h: Likewise.
8322 * dumpfile.c (struct dump_option_value_info): Likewise.
8323 (dump_enable_all): Likewise.
8324 (dump_switch_p_1): Likewise.
8325 (opt_info_switch_p): Likewise.
8326 * dumpfile.h (enum tree_dump_index): Likewise.
8327 (struct dump_file_info): Likewise.
8328 * genemit.c: Likewise.
8329 * generic-match-head.c: Likewise.
8330 * gengtype.c (open_base_files): Likewise.
8331 * gimple-pretty-print.c: Likewise.
8332 * gimple-pretty-print.h: Likewise.
8333 * graph.c (print_graph_cfg): Likewise.
8334 * graphite-scop-detection.c (dot_all_sese): Likewise.
8335 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
8336 * loop-unroll.c (report_unroll): Likewise.
8337 * passes.c (pass_manager::register_one_dump_file): Likewise.
8338 * print-tree.c: Likewise.
8339 * statistics.c: Likewise.
8340 * tree-cfg.c: Likewise.
8341 * tree-cfg.h: Likewise.
8342 * tree-dfa.c: Likewise.
8343 * tree-dfa.h: Likewise.
8344 * tree-dump.c (dump_function): Likewise.
8345 * tree-dump.h (struct dump_info): Likewise.
8346 * tree-pretty-print.c: Likewise.
8347 * tree-pretty-print.h: Likewise.
8348 * tree-ssa-live.c: Likewise.
8349 * tree-ssa-live.h: Likewise.
8350 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
8351 * tree-vect-loop.c: Likewise.
8352 * tree-vect-slp.c: Likewise.
8354 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
8355 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8357 PR tree-optimization/80457
8358 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
8359 of all arguments to a statement as scalar_to_vec operations.
8360 (vectorizable_call): Adjust call to vect_model_simple_cost for
8362 (vectorizable_conversion): Likewise.
8363 (vectorizable_assignment): Likewise.
8364 (vectorizable_shift): Likewise.
8365 (vectorizable_operation): Likewise.
8366 (vectorizable_comparison): Likewise.
8367 (vect_is_simple_cond): Record the def types for operands.
8368 (vectorizable_condition): Likewise, call vect_model_simple_cost.
8369 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
8370 for statement argument count.
8372 2017-05-16 Carl Love <cel@us.ibm.com>
8374 * config/rs6000/rs6000-c: Add support for built-in functions
8375 vector unsigned long long vec_bperm (vector unsigned long long,
8376 vector unsigned char)
8377 vector signed long long vec_mule (vector signed int,
8379 vector unsigned long long vec_mule (vector unsigned int,
8380 vector unsigned int)
8381 vector signed long long vec_mulo (vector signed int,
8383 vector unsigned long long vec_mulo (vector unsigned int,
8384 vector unsigned int)
8385 vector signed char vec_sldw (vector signed char,
8388 vector unsigned char vec_sldw (vector unsigned char,
8389 vector unsigned char,
8391 vector signed short vec_sldw (vector signed short,
8392 vector signed short,
8394 vector unsigned short vec_sldw (vector unsigned short,
8395 vector unsigned short,
8397 vector signed int vec_sldw (vector signed int,
8400 vector unsigned int vec_sldw (vector unsigned int,
8401 vector unsigned int,
8403 vector signed long long vec_sldw (vector signed long long,
8404 vector signed long long,
8406 vector unsigned long long vec_sldw (vector unsigned long long,
8407 vector unsigned long long,
8409 * config/rs6000/rs6000-c: Add support for built-in functions
8410 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
8411 * config/rs6000/altivec.h: Add defintion for vec_sldw.
8412 * doc/extend.texi: Update the built-in documentation for the
8413 new built-in functions.
8415 2017-05-16 Marek Polacek <polacek@redhat.com>
8419 * tree.c (save_expr): Don't fold the expression.
8421 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
8423 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
8424 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
8425 and (?*y,m). Update insn attributes.
8427 2017-05-16 Martin Liska <mliska@suse.cz>
8429 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
8430 flags argument of print_gimple_stmt, print_gimple_expr,
8431 print_generic_stmt and print_generic_expr.
8432 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
8433 * coretypes.h: Likewise.
8434 * except.c (dump_eh_tree): Likewise.
8435 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8436 * gimple-pretty-print.h: Likewise.
8437 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
8438 (backprop::push_to_worklist): Likewise.
8439 (backprop::pop_from_worklist): Likewise.
8440 (backprop::process_use): Likewise.
8441 (backprop::intersect_uses): Likewise.
8442 (note_replacement): Likewise.
8443 * gimple-ssa-store-merging.c
8444 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
8445 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
8446 (pass_store_merging::execute): Likewise.
8447 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
8448 (ssa_base_cand_dump_callback): Likewise.
8449 (dump_incr_vec): Likewise.
8450 (replace_refs): Likewise.
8451 (replace_mult_candidate): Likewise.
8452 (create_add_on_incoming_edge): Likewise.
8453 (create_phi_basis): Likewise.
8454 (insert_initializers): Likewise.
8455 (all_phi_incrs_profitable): Likewise.
8456 (introduce_cast_before_cand): Likewise.
8457 (replace_one_candidate): Likewise.
8458 * gimplify.c (gimplify_expr): Likewise.
8459 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
8460 (set_rename): Likewise.
8461 (rename_uses): Likewise.
8462 (copy_loop_phi_nodes): Likewise.
8463 (add_close_phis_to_merge_points): Likewise.
8464 (copy_loop_close_phi_args): Likewise.
8465 (copy_cond_phi_args): Likewise.
8466 (graphite_copy_stmts_from_block): Likewise.
8467 (translate_pending_phi_nodes): Likewise.
8468 * graphite-poly.c (print_pdr): Likewise.
8469 (dump_gbb_cases): Likewise.
8470 (dump_gbb_conditions): Likewise.
8471 (print_scop_params): Likewise.
8472 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
8473 (build_cross_bb_scalars_use): Likewise.
8474 (gather_bbs::before_dom_children): Likewise.
8475 * hsa-dump.c (dump_hsa_immed): Likewise.
8476 * ipa-cp.c (print_ipcp_constant_value): Likewise.
8477 (get_replacement_map): Likewise.
8478 * ipa-inline-analysis.c (dump_condition): Likewise.
8479 (estimate_function_body_sizes): Likewise.
8480 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
8481 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
8482 * ipa-prop.c (ipa_dump_param): Likewise.
8483 (ipa_print_node_jump_functions_for_edge): Likewise.
8484 (ipa_modify_call_arguments): Likewise.
8485 (ipa_modify_expr): Likewise.
8486 (ipa_dump_param_adjustments): Likewise.
8487 (ipa_dump_agg_replacement_values): Likewise.
8488 (ipcp_modif_dom_walker::before_dom_children): Likewise.
8489 * ipa-pure-const.c (check_stmt): Likewise.
8490 (pass_nothrow::execute): Likewise.
8491 * ipa-split.c (execute_split_functions): Likewise.
8492 * omp-offload.c (dump_oacc_loop_part): Likewise.
8493 (dump_oacc_loop): Likewise.
8494 * trans-mem.c (tm_log_emit): Likewise.
8495 (tm_memopt_accumulate_memops): Likewise.
8496 (dump_tm_memopt_set): Likewise.
8497 (dump_tm_memopt_transform): Likewise.
8498 * tree-cfg.c (gimple_verify_flow_info): Likewise.
8499 (print_loop): Likewise.
8500 * tree-chkp-opt.c (chkp_print_addr): Likewise.
8501 (chkp_gather_checks_info): Likewise.
8502 (chkp_get_check_result): Likewise.
8503 (chkp_remove_check_if_pass): Likewise.
8504 (chkp_use_outer_bounds_if_possible): Likewise.
8505 (chkp_reduce_bounds_lifetime): Likewise.
8506 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
8507 (chkp_mark_completed_bounds): Likewise.
8508 (chkp_register_incomplete_bounds): Likewise.
8509 (chkp_mark_invalid_bounds): Likewise.
8510 (chkp_maybe_copy_and_register_bounds): Likewise.
8511 (chkp_build_returned_bound): Likewise.
8512 (chkp_get_bound_for_parm): Likewise.
8513 (chkp_build_bndldx): Likewise.
8514 (chkp_get_bounds_by_definition): Likewise.
8515 (chkp_generate_extern_var_bounds): Likewise.
8516 (chkp_get_bounds_for_decl_addr): Likewise.
8517 * tree-chrec.c (chrec_apply): Likewise.
8518 * tree-data-ref.c (dump_data_reference): Likewise.
8519 (dump_subscript): Likewise.
8520 (dump_data_dependence_relation): Likewise.
8521 (analyze_overlapping_iterations): Likewise.
8522 * tree-inline.c (expand_call_inline): Likewise.
8523 (tree_function_versioning): Likewise.
8524 * tree-into-ssa.c (dump_defs_stack): Likewise.
8525 (dump_currdefs): Likewise.
8526 (dump_names_replaced_by): Likewise.
8527 (dump_update_ssa): Likewise.
8528 (update_ssa): Likewise.
8529 * tree-object-size.c (pass_object_sizes::execute): Likewise.
8530 * tree-parloops.c (build_new_reduction): Likewise.
8531 (try_create_reduction_list): Likewise.
8532 (ref_conflicts_with_region): Likewise.
8533 (oacc_entry_exit_ok_1): Likewise.
8534 (oacc_entry_exit_single_gang): Likewise.
8535 * tree-pretty-print.h: Likewise.
8536 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
8537 (get_scalar_evolution): Likewise.
8538 (add_to_evolution): Likewise.
8539 (get_loop_exit_condition): Likewise.
8540 (analyze_evolution_in_loop): Likewise.
8541 (analyze_initial_condition): Likewise.
8542 (analyze_scalar_evolution): Likewise.
8543 (instantiate_scev): Likewise.
8544 (number_of_latch_executions): Likewise.
8545 (gather_chrec_stats): Likewise.
8546 (final_value_replacement_loop): Likewise.
8547 (scev_const_prop): Likewise.
8548 * tree-sra.c (dump_access): Likewise.
8549 (disqualify_candidate): Likewise.
8550 (create_access): Likewise.
8552 (maybe_add_sra_candidate): Likewise.
8553 (create_access_replacement): Likewise.
8554 (analyze_access_subtree): Likewise.
8555 (analyze_all_variable_accesses): Likewise.
8556 (sra_modify_assign): Likewise.
8557 (initialize_constant_pool_replacements): Likewise.
8558 (find_param_candidates): Likewise.
8559 (decide_one_param_reduction): Likewise.
8560 (replace_removed_params_ssa_names): Likewise.
8561 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
8562 * tree-ssa-copy.c (dump_copy_of): Likewise.
8563 (copy_prop_visit_cond_stmt): Likewise.
8564 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
8565 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
8566 (record_equivalences_from_stmt): Likewise.
8567 * tree-ssa-dse.c (compute_trims): Likewise.
8568 (delete_dead_call): Likewise.
8569 (delete_dead_assignment): Likewise.
8570 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
8571 (forward_propagate_into_cond): Likewise.
8572 (pass_forwprop::execute): Likewise.
8573 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8574 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
8576 (move_computations_worker): Likewise.
8577 (execute_sm): Likewise.
8578 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
8579 (remove_exits_and_undefined_stmts): Likewise.
8580 (remove_redundant_iv_tests): Likewise.
8581 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
8582 (adjust_iv_update_pos): Likewise.
8583 * tree-ssa-math-opts.c (bswap_replace): Likewise.
8584 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
8585 (value_replacement): Likewise.
8586 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
8587 * tree-ssa-pre.c (print_pre_expr): Likewise.
8588 (get_representative_for): Likewise.
8589 (create_expression_by_pieces): Likewise.
8590 (insert_into_preds_of_block): Likewise.
8591 (eliminate_insert): Likewise.
8592 (eliminate_dom_walker::before_dom_children): Likewise.
8593 (eliminate): Likewise.
8594 (remove_dead_inserted_code): Likewise.
8595 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
8596 * tree-ssa-reassoc.c (get_rank): Likewise.
8597 (eliminate_duplicate_pair): Likewise.
8598 (eliminate_plus_minus_pair): Likewise.
8599 (eliminate_not_pairs): Likewise.
8600 (undistribute_ops_list): Likewise.
8601 (eliminate_redundant_comparison): Likewise.
8602 (update_range_test): Likewise.
8603 (optimize_range_tests_var_bound): Likewise.
8604 (optimize_vec_cond_expr): Likewise.
8605 (rewrite_expr_tree): Likewise.
8606 (rewrite_expr_tree_parallel): Likewise.
8607 (linearize_expr): Likewise.
8608 (break_up_subtract): Likewise.
8609 (linearize_expr_tree): Likewise.
8610 (attempt_builtin_powi): Likewise.
8611 (attempt_builtin_copysign): Likewise.
8612 (transform_stmt_to_copy): Likewise.
8613 (transform_stmt_to_multiply): Likewise.
8614 (dump_ops_vector): Likewise.
8615 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
8616 (print_scc): Likewise.
8617 (set_ssa_val_to): Likewise.
8618 (visit_reference_op_store): Likewise.
8619 (visit_use): Likewise.
8620 (sccvn_dom_walker::before_dom_children): Likewise.
8621 (run_scc_vn): Likewise.
8622 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
8624 (expr_hash_elt::print): Likewise.
8625 (const_and_copies::pop_to_marker): Likewise.
8626 (const_and_copies::record_const_or_copy_raw): Likewise.
8627 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
8628 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
8629 (dump_predicates): Likewise.
8630 (find_uninit_use): Likewise.
8631 (warn_uninitialized_phi): Likewise.
8632 (pass_late_warn_uninitialized::execute): Likewise.
8633 * tree-ssa.c (verify_vssa): Likewise.
8634 (verify_ssa): Likewise.
8635 (maybe_optimize_var): Likewise.
8636 * tree-vrp.c (dump_value_range): Likewise.
8637 (dump_all_value_ranges): Likewise.
8638 (dump_asserts_for): Likewise.
8639 (register_edge_assert_for_2): Likewise.
8640 (vrp_visit_cond_stmt): Likewise.
8641 (vrp_visit_switch_stmt): Likewise.
8642 (vrp_visit_stmt): Likewise.
8643 (vrp_visit_phi_node): Likewise.
8644 (simplify_cond_using_ranges_1): Likewise.
8645 (fold_predicate_in): Likewise.
8646 (evrp_dom_walker::before_dom_children): Likewise.
8647 (evrp_dom_walker::push_value_range): Likewise.
8648 (evrp_dom_walker::pop_value_range): Likewise.
8649 (execute_early_vrp): Likewise.
8651 2017-05-16 Richard Biener <rguenther@suse.de>
8653 * dwarf2out.c (loc_list_from_tree_1): Do not create
8654 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
8656 2017-05-16 Richard Biener <rguenther@suse.de>
8658 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
8660 (note_variable_value_in_expr): If we resolved the decl ref
8661 do not push to the stack.
8663 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
8665 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
8666 operations in fast-math mode.
8667 (vaddq_f16): Likewise.
8668 (vmul_f16): Likewise.
8669 (vmulq_f16): Likewise.
8670 (vsub_f16): Likewise.
8671 (vsubq_f16): Likewise.
8672 * config/arm/neon.md (add<mode>3): New.
8674 (fma:<VH:mode>3): New. Also remove outdated comment.
8677 2017-05-16 Martin Liska <mliska@suse.cz>
8681 * ipa-devirt.c (warn_types_mismatch): Fix typo.
8682 (odr_types_equivalent_p): Likewise.
8684 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
8686 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
8688 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
8691 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
8692 non-interunit SSE move alternatives with '?'.
8693 (zero-extendsidi peephole2): New peephole to skip intermediate
8694 general register in SSE zero-extend sequence.
8696 2017-05-15 Jeff Law <law@redhat.com>
8698 * reorg.c (relax_delay_slots): Create a new variable to hold
8699 the temporary target rather than clobbering TARGET_LABEL.
8701 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
8702 missing argument to extract_bit_field call.
8703 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
8705 2017-05-15 Martin Liska <mliska@suse.cz>
8708 * gcc.c (process_command): Do not allow empty argument of -o option.
8710 2017-05-15 Renlin Li <renlin.li@arm.com>
8712 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
8713 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
8714 * config/aarch64/constraints.md (Usf): Add long call check.
8715 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
8716 (call_value): Likewise.
8717 (sibcall): Likewise.
8718 (sibcall_value): Likewise.
8720 (call_value_insn): New.
8721 (sibcall_insn): Update rtx pattern.
8722 (sibcall_value_insn): Likewise.
8723 (call_internal): Remove.
8724 (call_value_internal): Likewise.
8725 (sibcall_internal): Likewise.
8726 (sibcall_value_internal): Likewise.
8727 (call_reg): Likewise.
8728 (call_symbol): Likewise.
8729 (call_value_reg): Likewise.
8730 (call_value_symbol): Likewise.
8732 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
8735 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
8737 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
8739 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
8740 compatible with CCGOCmode and with CCZmode.
8742 2017-05-14 Martin Sebor <msebor@redhat.com>
8745 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
8746 (gimple_fold_builtin_snprintf): Same.
8747 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
8748 (gimple_fold_builtin_snprintf): Same.
8749 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
8751 (is_call_safe): New function.
8752 (try_substitute_return_value): Call it.
8753 (try_simplify_call): New function.
8754 (pass_sprintf_length::handle_gimple_call): Call it.
8756 2017-05-14 Martin Sebor <msebor@redhat.com>
8759 * builtins.c (expand_builtin_stpncpy): Simplify.
8761 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
8763 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
8764 * config/i386/i386.h
8765 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
8766 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
8767 (struct machine_function): Add new members call_ms2sysv,
8768 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
8769 (struct machine_frame_state): New fields sp_realigned and
8770 sp_realigned_offset.
8771 * config/i386/i386.c
8772 (enum xlogue_stub): New enum.
8773 (enum xlogue_stub_sets): New enum.
8774 (class xlogue_layout): New class.
8775 (struct ix86_frame): New fields stack_realign_allocate_offset,
8776 stack_realign_offset and outlined_save_offset. Modify comments to
8777 detail stack layout when using out-of-line stubs.
8778 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
8779 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
8780 -mcall-ms2sysv-xlogues.
8781 (stub_managed_regs): New static variable.
8782 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
8783 registers managed by out-of-line stub.
8784 (disable_call_ms2sysv_xlogues): New function.
8785 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
8786 m->call_ms2sysv when appropriate and compute frame layout for
8788 (sp_valid_at, fp_valid_at): New inline functions.
8789 (choose_basereg): New function.
8790 (choose_baseaddr): Add align parameter, use choose_basereg and modify
8792 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
8793 Use align parameter of choose_baseaddr to generated aligned SSE movs
8795 (pro_epilogue_adjust_stack): Modify to track
8796 machine_frame_state::sp_realigned.
8797 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
8798 (ix86_nsaved_sseregs): Likewise.
8799 (ix86_emit_save_regs): Likewise.
8800 (ix86_emit_save_regs_using_mov): Likewise.
8801 (ix86_emit_save_sse_regs_using_mov): Likewise.
8802 (get_scratch_register_on_entry): Likewise.
8803 (gen_frame_set): New function.
8804 (gen_frame_load): Likewise.
8805 (gen_frame_store): Likewise.
8806 (emit_outlined_ms2sysv_save): Likewise.
8807 (emit_outlined_ms2sysv_restore): Likewise.
8808 (ix86_expand_prologue): Modify stack re-alignment code and call
8809 emit_outlined_ms2sysv_save when appropriate.
8810 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
8811 parameter rtx_insn *insn, which allows the function to be used to only
8813 (ix86_expand_epilogue): Modify validity checks of frame and stack
8814 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
8815 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
8816 * config/i386/predicates.md
8817 (save_multiple): New predicate.
8818 (restore_multiple): Likewise.
8819 * config/i386/sse.md
8820 (save_multiple<mode>): New pattern.
8821 (save_multiple_realign<mode>): Likewise.
8822 (restore_multiple<mode>): Likewise.
8823 (restore_multiple_and_return<mode>): Likewise.
8824 (restore_multiple_leave_return<mode>): Likewise.
8825 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
8827 2017-05-14 Julia Koval <julia.koval@intel.com>
8829 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
8830 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
8831 (__builtin_ia32_xsetbv): New builtins.
8832 * config/i386/i386.c (ix86_expand_special_args_builtin):
8834 (ix86_expand_builtin): Special expand for new intrinsics.
8835 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
8836 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
8837 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
8839 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8841 * cfganal.c (inverted_post_order_compute): Change argument type
8843 * cfganal.h (inverted_post_order_compute): Adjust prototype.
8844 * df-core.c (rest_of_handle_df_initialize): Adjust.
8845 (rest_of_handle_df_finish): Likewise.
8846 (df_analyze_1): Likewise.
8847 (df_analyze): Likewise.
8848 (loop_inverted_post_order_compute): Change argument to be a vec *.
8849 (df_analyze_loop): Adjust.
8850 (df_get_n_blocks): Likewise.
8851 (df_get_postorder): Likewise.
8852 * df.h (struct df_d): Change field to be a vec.
8853 * lcm.c (compute_laterin): Adjust.
8854 (compute_available): Likewise.
8855 * lra-lives.c (lra_create_live_ranges_1): Likewise.
8856 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
8857 * tree-ssa-pre.c (compute_antic): Likewise.
8859 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8861 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
8862 (depth_first_search::depth_first_search): Change structure init
8863 function to this constructor.
8864 (depth_first_search::add_bb): Rename function to this member.
8865 (depth_first_search::execute): Likewise.
8866 (flow_dfs_compute_reverse_finish): Adjust.
8868 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8870 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
8871 (longest_simple_path): Likewise.
8872 * shrink-wrap.c (spread_components): Likewise.
8873 (disqualify_problematic_components): Likewise.
8874 (emit_common_heads_for_components): Likewise.
8875 (emit_common_tails_for_components): Likewise.
8876 (insert_prologue_epilogue_for_components): Likewise.
8878 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8880 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
8883 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8885 * df-core.c (df_set_blocks): Start using auto_bitmap.
8886 (df_compact_blocks): Likewise.
8887 * df-problems.c (df_rd_confluence_n): Likewise.
8888 * df-scan.c (df_insn_rescan_all): Likewise.
8889 (df_process_deferred_rescans): Likewise.
8890 (df_update_entry_block_defs): Likewise.
8891 (df_update_exit_block_uses): Likewise.
8892 (df_entry_block_bitmap_verify): Likewise.
8893 (df_exit_block_bitmap_verify): Likewise.
8894 (df_scan_verify): Likewise.
8895 * lra-constraints.c (lra_constraints): Likewise.
8896 (undo_optional_reloads): Likewise.
8897 (lra_undo_inheritance): Likewise.
8898 * lra-remat.c (calculate_gen_cands): Likewise.
8899 (do_remat): Likewise.
8900 * lra-spills.c (assign_spill_hard_regs): Likewise.
8901 (spill_pseudos): Likewise.
8902 * tree-ssa-pre.c (bitmap_set_and): Likewise.
8903 (bitmap_set_subtract_values): Likewise.
8905 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8907 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
8908 management with auto_bitmap.
8909 (fix_inter_tick): Likewise.
8910 (fix_recovery_deps): Likewise.
8911 * ira.c (add_store_equivs): Likewise.
8912 (find_moveable_pseudos): Likewise.
8913 (split_live_ranges_for_shrink_wrap): Likewise.
8914 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
8915 (rtx_reuse_manager::seen_def_p): Likewise.
8916 (rtx_reuse_manager::set_seen_def): Likewise.
8917 * print-rtl.h (class rtx_reuse_manager): Likewise.
8919 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8921 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
8923 (migrate_btr_def): Likewise.
8924 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
8925 * df-core.c (loop_post_order_compute): Likewise.
8926 (loop_inverted_post_order_compute): Likewise.
8927 * hsa-common.h: Likewise.
8928 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
8929 * init-regs.c (initialize_uninitialized_regs): Likewise.
8930 * ipa-inline.c (resolve_noninline_speculation): Likewise.
8931 (inline_small_functions): Likewise.
8932 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
8933 * ira.c (combine_and_move_insns): Likewise.
8934 (build_insn_chain): Likewise.
8935 * loop-invariant.c (find_invariants): Likewise.
8936 * lower-subreg.c (propagate_pseudo_copies): Likewise.
8937 * predict.c (tree_predict_by_opcode): Likewise.
8938 (predict_paths_leading_to): Likewise.
8939 (predict_paths_leading_to_edge): Likewise.
8940 (estimate_loops_at_level): Likewise.
8941 (estimate_loops): Likewise.
8942 * shrink-wrap.c (try_shrink_wrapping): Likewise.
8943 (spread_components): Likewise.
8944 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
8945 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
8946 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
8947 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
8948 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
8949 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
8950 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
8951 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
8952 (mark_threaded_blocks): Likewise.
8953 (thread_through_all_blocks): Likewise.
8954 * tree-ssa.c (verify_ssa): Likewise.
8955 (execute_update_addresses_taken): Likewise.
8956 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
8958 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8960 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
8962 (post_order_compute): Likewise.
8963 (inverted_post_order_compute): Likewise.
8964 (pre_and_rev_post_order_compute_fn): Likewise.
8966 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8968 * genrecog.c (int_set::int_set): Explicitly construct our
8969 auto_vec base class.
8970 * vec.h (auto_vec::auto_vec): New constructor.
8972 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8974 * bitmap.h (class auto_bitmap): New constructor taking
8975 bitmap_obstack * argument.
8977 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8979 * bitmap.h (class auto_bitmap): Change type of m_bits to
8980 bitmap_head, and adjust ctor / dtor and member operators.
8982 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
8984 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
8985 when returned register mode doesn't match original mode.
8987 2017-05-12 Jeff Law <law@redhat.com>
8988 Jakub Jelinek <jakub@redhat.com>
8990 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
8991 we look for cc setter after the compare-elim changes.
8992 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
8993 within the vector to match what compare-elim now expects.
8994 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
8995 (xorsi3_flags, one_cmplsi2_flags): Likewise.
8997 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
8998 after the compare-elim changes.
8999 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
9000 the vector to match what compare-elim now expects.
9001 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
9002 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
9003 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
9004 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
9005 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
9007 * config/visium/visium.c (single_set_and_flags): Fix where
9008 we look for cc setter after the compare-elim changes.
9009 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
9010 with the vector to match what compare-elim now expects.
9011 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
9012 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
9013 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
9014 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
9015 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
9016 (neg<mode>2_insn_set_overflow): Likewise.
9018 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
9021 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
9022 maybe_expand_insn call, set ops[0].target. If still set after call,
9023 set alt_rtl. Add extra arg to recursive calls.
9024 (extract_bit_field): Add alt_rtl argument. Pass to
9026 * expmed.h (extract_bit_field): Fix prototype.
9027 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
9028 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
9029 to extract_bit_field_calls.
9030 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
9031 Pass alt_rtl to extract_bit_field calls.
9032 * calls.c (store_unaligned_arguments_into_psuedos)
9033 load_register_parameters): Pass extra NULL to extract_bit_field calls.
9034 * optabs.c (maybe_legitimize_operand): Clear op->target when call
9036 * optabs.h (struct expand_operand): Add target bitfield.
9038 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
9040 * compare-elim.c (try_eliminate_compare): Canonicalize
9041 operation with embedded compare to
9042 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
9043 (set (reg) (operation)].
9045 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
9047 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
9050 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
9051 cost of adding a carry flag for ADC instruction.
9052 [case MINUS]: Ignore the cost of subtracting a carry flag
9053 for SBB instruction.
9055 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
9057 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
9059 * config/rs6000/bmiintrin.h: New file.
9060 * config/rs6000/bmi2intrin.h: New file.
9061 * config/rs6000/x86intrin.h: New file.
9063 2017-05-12 Jeff Law <law@redhat.com>
9065 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
9068 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
9071 * tree-cfg.c: Remove cfg edges of unreachable case statements.
9073 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9075 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
9076 early expansion of vector divide builtins.
9077 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
9078 builtins identified as having unsigned arguments.
9080 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9082 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
9083 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
9084 expansion of vector logical operations (and, andc, or, xor,
9087 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9089 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
9090 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
9092 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9094 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
9095 early GIMPLE expansion of vector multiplies.
9097 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9099 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
9100 TARGET_HAVE_MOVT conditional.
9101 (movt splitter): Likewise.
9103 2017-05-12 Richard Biener <rguenther@suse.de>
9105 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
9106 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9107 Fold all stmts not inplace.
9109 2017-05-12 Richard Biener <rguenther@suse.de>
9111 PR tree-optimization/80713
9112 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
9113 inserted_exprs bit for not removed stmts.
9115 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
9118 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
9119 parallelized" attribute for parallelized OpenACC kernels.
9120 * omp-offload.c (execute_oacc_device_lower): Use it.
9122 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
9123 Set "oacc kernels" attribute.
9124 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
9125 parameter. Adjust all users.
9126 (oacc_fn_attrib_kernels_p): Remove function.
9127 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
9128 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
9129 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
9130 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
9131 assert "oacc kernels" attribute is set.
9133 2017-05-11 Carl Love <cel@us.ibm.com>
9135 * config/rs6000/rs6000-c: Add support for built-in functions
9136 vector unsigned char vec_popcnt (vector signed char)
9137 vector unsigned char vec_popcnt (vector unsigned char)
9138 vector unsigned short vec_popcnt (vector signed short)
9139 vector unsigned short vec_popcnt (vector unsigned short)
9140 vector unsigned int vec_popcnt (vector signed int)
9141 vector unsigned int vec_popcnt (vector unsigned int)
9142 vector unsigned long long vec_popcnt (vector signed long long)
9143 vector unsigned long long vec_popcnt (vector unsigned long long)
9144 vector signed long long vec_slo (vector signed long long,
9146 vector signed long long vec_slo (vector signed long long,
9147 vector unsigned char)
9148 vector unsigned long long vec_slo (vector unsigned long long,
9150 vector unsigned long long vec_slo (vector unsigned long long,
9151 vector unsigned char)
9152 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
9153 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
9154 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
9155 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
9156 * doc/extend.texi: Update the built-in documentation file for the
9157 new built-in functions.
9159 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
9161 * attribs.h (sorted_attr_string): Move machine independent
9162 functions for target clone support from the i386 port to common
9163 code. Rename ix86_function_versions to common_function_versions.
9164 Rename make_name to make_unique_name.
9165 (common_function_versions): Likewise.
9166 (make_unique_name): Likewise.
9167 (make_dispatcher_decl): Likewise.
9168 (is_function_default_version): Likewise.
9169 * attribs.c (attr_strcmp): Likewise.
9170 (sorted_attr_string): Likewise.
9171 (common_function_versions): Likewise.
9172 (make_unique_name): Likewise.
9173 (make_dispatcher_decl): Likewise.
9174 (is_function_default_version): Likewise.
9175 * config/i386/i386.c (attr_strcmp): Likewise.
9176 (sorted_attr_string): Likewise.
9177 (ix86_function_versions): Likewise.
9178 (make_name): Likewise.
9179 (make_dispatcher_decl): Likewise.
9180 (is_function_default_version): Likewise.
9181 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
9183 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9186 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
9187 Account for direct move costs for vec_construct of integer
9190 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
9193 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
9194 (UNSPEC_STX_ATOMIC): Ditto.
9195 (loaddi_via_sse): New insn.
9196 (storedi_via_sse): Ditto.
9197 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
9198 Update corresponding peephole2 patterns.
9199 (atomic_storedi_fpu): Ditto.
9201 2017-05-11 Julia Koval <julia.koval@intel.com>
9203 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
9204 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
9206 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
9207 (__builtin_ia32_rsqrt14ss_mask): New builtins.
9208 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
9210 2017-05-11 Nathan Sidwell <nathan@acm.org>
9212 * graphite-poly.c: Include dumpfile.h.
9214 * dumpfle.h (dump_function): Declare here ...
9215 * tree-dump.h (dump_function): ... not here.
9216 * dumpfile.c: #include tree-cfg.h.
9217 (dump_function): Move here from ...
9218 * tree-dump.c (dump_function): ... here.
9219 * gimplify.c: #include splay-tree.h, not tree-dump.h.
9220 * graphite-poly.c: Don't include tree-dump.h.
9221 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
9222 * print-tree.c: Likewise.
9223 * stor-layout.c: Likewise.
9224 * tree-nested.c: Likewise.
9226 * dumpfile.c (dump_start): Use TDF_FLAGS.
9227 (dump_enable_all): Fix TDF_KIND check thinko.
9229 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
9231 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9232 array entries to represent two legal parameterizations of the
9233 overloaded __builtin_cmpb function, as represented by the
9234 P6_OV_BUILTIN_CMPB constant.
9235 (altivec_resolve_overloaded_builtin): Add special case handling
9236 for the __builtin_cmpb function, as represented by the
9237 P6_OV_BUILTIN_CMPB constant.
9238 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
9239 (BU_P6_64BIT_2): New macro.
9240 (BU_P6_OVERLOAD_2): New macro
9241 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
9242 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
9243 (CMPB): Add overload support to represent both 32-bit and 64-bit
9244 compare-bytes function.
9245 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
9246 support for TARGET_CMPB.
9247 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
9248 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
9249 documentation of the __builtin_cmpb overloaded built-in function.
9251 2017-05-11 Richard Biener <rguenther@suse.de>
9253 PR tree-optimization/80705
9254 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
9255 bases are not vectorizable.
9257 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9259 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
9260 when counting register pressure.
9262 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9264 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
9265 (iv_ca_more_deps): Renamed to ...
9266 (iv_ca_compare_deps): ... this.
9267 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
9269 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9271 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
9273 (determine_group_iv_costs): ... here.
9274 (find_inv_vars_cb): Record inv var if it's not recorded before.
9276 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9278 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
9279 (get_shiftadd_cost): Ditto.
9281 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9283 * tree-ssa-address.c: Include header file.
9284 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
9286 (add_to_parts): Refactor.
9287 (addr_to_parts): New parameter. Update use of move_hint_to_base.
9288 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
9291 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9293 PR tree-optimization/53090
9294 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
9296 (extract_cond_operands): Detect condition with IV on both sides
9297 and return COMP_IV_EXPR_2.
9298 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
9299 (rewrite_use_compare): Simplify by removing call to function
9300 extract_cond_operands.
9302 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9304 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
9305 (extract_cond_operands): Detect condition comparing against non-
9306 invariant bound and return appropriate enum value.
9307 (find_interesting_uses_cond): Update use of extract_cond_operands.
9308 Handle its return value accordingly.
9309 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
9311 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9313 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
9314 nonlinear iv_use computation in loop invariant sensitive way.
9316 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9318 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
9319 (find_iv_candidates): Call relate_compare_use_with_all_cands.
9321 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9323 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
9324 (dump_cand): Support iv_cand.inv_exprs.
9325 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
9327 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
9330 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9332 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
9334 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
9335 as local function. Include necessary header files.
9336 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
9338 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9340 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
9342 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9344 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
9345 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
9346 RSHIFT_EXPR and BIT_NOT_EXPR.
9348 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9350 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
9351 (adjust_setup_cost): New parameter supporting round up adjustment.
9352 (struct address_cost_data): Delete.
9353 (force_expr_to_var_cost): Don't bound cost with spill_cost.
9354 (split_address_cost, ptr_difference_cost): Delete.
9355 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
9356 (struct ainc_cost_data): New struct.
9357 (get_address_cost_ainc): New function.
9358 (get_address_cost, get_computation_cost): Reimplement.
9359 (determine_group_iv_cost_address): Record inv_expr for all uses of
9361 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
9362 (iv_ca_has_deps): Reimplemented to ...
9363 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
9365 (iv_ca_extend): Call iv_ca_more_deps.
9367 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9369 * tree-ssa-address.c (struct mem_address): Move to header file.
9370 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
9371 * tree-ssa-address.h (struct mem_address): Move from C file.
9372 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
9374 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9376 * tree-affine.h (aff_combination_type): New interface.
9377 (aff_combination_zero_p): Remove static.
9378 (aff_combination_const_p): New interface.
9379 (aff_combination_singleton_var_p): New interfaces.
9381 2017-05-11 Richard Biener <rguenther@suse.de>
9383 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9384 Skip unreachable blocks and destinations.
9385 (eliminate): Move stmt removal and fixup ...
9386 (fini_eliminate): ... here. Skip inserted exprs.
9387 (pass_pre::execute): Move fini_pre after fini_eliminate.
9388 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
9389 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
9390 PRE to get rid of dead code that has invalid SSA form and
9391 split critical edges again.
9393 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9395 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
9397 2017-05-11 Richard Biener <rguenther@suse.de>
9399 * passes.c (execute_function_todo): Verify loops if they are
9400 said to be up-to-date.
9401 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
9402 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
9404 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
9407 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
9408 handle calling assemble_external ourself.
9411 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
9412 modes with zero size. Enhance comment.
9414 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9416 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
9417 built-ins for vec_xl and vec_xst with short and char pointer
9420 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
9422 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
9423 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
9424 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
9425 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
9426 (_mm_maskz_min_round_ss): New intrinsics.
9427 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
9428 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
9429 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
9430 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
9431 (__builtin_ia32_minss_mask_round): New builtins.
9432 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9433 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9434 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
9436 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
9437 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
9439 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
9442 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
9444 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
9445 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
9446 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
9447 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
9448 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
9449 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
9450 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
9451 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9452 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
9453 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
9454 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
9455 (__builtin_ia32_mulss_mask_round): New builtins.
9456 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9457 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9458 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
9460 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
9461 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
9463 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
9466 2017-05-10 Julia Koval <julia.koval@intel.com>
9468 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
9469 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
9470 (_mm256_setr_m128i): New intrinsics.
9472 2017-05-10 Julia Koval <julia.koval@intel.com>
9474 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
9475 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
9476 (_mm_maskz_rcp14_ss): New intrinsics.
9477 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
9478 (__builtin_ia32_rcp14ss_mask): New builtins.
9479 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
9481 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
9483 PR tree-optimization/51513
9484 * tree-cfg.c (gimple_seq_unreachable_p): New function.
9485 (assert_unreachable_fallthru_edge_p): Use it.
9486 (group_case_labels_stmt): Likewise.
9487 * tree-cfg.h: Prototype it.
9488 * stmt.c: Include cfghooks.h and tree-cfg.h.
9489 (emit_case_dispatch_table) <gap_label>: New local variable.
9490 Use it to fill dispatch table gaps.
9491 Test for default_label before updating probabilities.
9492 (expand_case) <default_label>: Remove unneeded initialization.
9493 Test for unreachable default case statement and remove its edge.
9494 Set default_label accordingly.
9495 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
9497 2017-05-10 Carl Love <cel@us.ibm.com>
9499 * config/rs6000/rs6000-c: Add support for built-in functions
9500 vector signed char vec_neg (vector signed char)
9501 vector signed short int vec_neg (vector short int)
9502 vector signed int vec_neg (vector signed int)
9503 vector signed long long vec_neg (vector signed long long)
9504 vector float vec_neg (vector float)
9505 vector double vec_neg (vector double)
9506 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
9508 * config/rs6000/altivec.h: Add define for vec_neg
9509 * doc/extend.texi: Update the built-in documentation for the
9510 new built-in functions.
9512 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9514 PR tree-optimization/77644
9515 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
9517 2017-05-10 Nathan Sidwell <nathan@acm.org>
9519 * dumpfile.h (TDI_lang_all): New.
9520 (TDF_KIND): New. Renumber others
9521 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
9523 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
9525 (get_dump_file_name): Adjust suffix generation.
9526 (dump_enable_all): Use TDF_KIND.
9527 * doc/invoke.texi (-fdump-lang-all): Document.
9529 * dumpfile.h: Tabify.
9531 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
9534 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
9535 Move member access before delete.
9537 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
9539 * tree-inline.c (expand_call_inline): Split block at stmt
9542 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9545 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
9546 are now unused after splitting mov{sf,sd}_hardfloat.
9547 (f32_lr2): Likewise.
9549 (f32_lm2): Likewise.
9551 (f32_li2): Likewise.
9554 (f32_sr2): Likewise.
9556 (f32_sm2): Likewise.
9558 (f32_si2): Likewise.
9561 (f32_vsx): Likewise.
9563 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
9564 For movsf, order stores so the VSX stores occur before the GPR
9565 store which encourages the register allocator to use a traditional
9566 FPR instead of a GPR. For movsd, order the stores so that the GPR
9567 store comes before the VSX stores to allow the power6 to work.
9568 This is due to the power6 not having a 32-bit integer store
9569 instruction from a FPR.
9570 (movsf_hardfloat): Likewise.
9571 (movsd_hardfloat): Likewise.
9573 2017-05-09 Martin Sebor <msebor@redhat.com>
9575 PR translation/80280
9576 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
9579 PR translation/80280
9580 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
9581 data member added in r247778.
9582 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
9584 2017-05-09 Nathan Sidwell <nathan@acm.org>
9586 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
9588 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
9591 2017-05-09 Marek Polacek <polacek@redhat.com>
9593 * doc/invoke.texi: Fix typo.
9595 2017-05-09 Richard Biener <rguenther@suse.de>
9597 * tree-vrp.c (vrp_val_is_max): Adjust comment.
9598 (vrp_val_is_min): Likewise.
9599 (set_value_range_to_value): Likewise.
9600 (set_value_range_to_nonnegative): Likewise.
9601 (gimple_assign_nonzero_p): Likewise.
9602 (gimple_stmt_nonzero_p): Likewise.
9603 (vrp_int_const_binop): Likewise. Remove unreachable case.
9604 (adjust_range_with_scev): Adjust comments.
9605 (compare_range_with_value): Likewise.
9606 (extract_range_from_phi_node): Likewise.
9607 (test_for_singularity): Likewise.
9609 2017-05-09 Richard Biener <rguenther@suse.de>
9611 * tree-vrp.c (get_single_symbol): Add assert that we don't
9612 get overflowed constants as invariant part.
9613 (compare_values_warnv): Add comment before the TREE_NO_WARNING
9614 checks. Use wi::cmp instead of recursing for integer constants.
9615 (compare_values): Just ignore whether we assumed undefined
9616 overflow instead of failing the compare.
9617 (extract_range_for_var_from_comparison_expr): Add comment before the
9618 TREE_NO_WARNING sets.
9619 (test_for_singularity): Likewise.
9620 (extract_range_from_comparison): Do not disable optimization
9621 when we assumed undefined overflow.
9622 (extract_range_basic): Remove init of unused var.
9624 2017-05-09 Richard Biener <rguenther@suse.de>
9626 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
9627 (extract_range_from_multiplicative_op_1): Adjust.
9628 (extract_range_from_binary_expr_1): Use int_const_binop.
9630 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
9633 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
9634 rs6000_store_data_bypass_p in seven define_bypass directives and
9635 in several comments.
9636 * config/rs6000/rs6000-protos.h: Add prototype for
9637 rs6000_store_data_bypass_p function.
9638 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
9639 function implements slightly different (rs6000-specific) semantics
9640 than store_data_bypass_p, returning false rather than aborting
9641 with assertion error when arguments do not satisfy the
9642 requirements of store data bypass.
9643 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
9644 rs6000_store_data_bypass_p.
9646 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
9648 * config/xtensa/xtensa-protos.h
9649 (xtensa_initial_elimination_offset): New declaration.
9650 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
9651 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
9652 macro definition, add case for FRAME_POINTER_REGNUM when
9653 FRAME_GROWS_DOWNWARD.
9654 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
9655 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
9656 xtensa_initial_elimination_offset.
9658 2017-05-08 Nathan Sidwell <nathan@acm.org>
9660 * doc/invoke.texi: Alphabetize -fdump options.
9662 2017-05-08 Martin Sebor <msebor@redhat.com>
9664 PR translation/80280
9665 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
9667 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
9669 * target.def (compute_frame_layout): New optional target hook.
9670 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
9671 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
9672 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
9674 * reload1.c (verify_initial_elim_offsets): Likewise.
9675 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
9676 (use_simple_return_p): Call arm_compute_frame_layout if needed.
9677 (arm_get_frame_offsets): Split up into this ...
9678 (arm_compute_frame_layout): ... and this function.
9680 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
9682 * config/aarch64/constraints.md (Usa): New constraint.
9683 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
9685 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
9687 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
9688 with_multilib_list after it has been checked.
9690 2017-05-08 Richard Biener <rguenther@suse.de>
9692 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
9693 (bitmap_set_subtract_values): Likewise.
9695 2017-05-08 Richard Biener <rguenther@suse.de>
9697 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
9698 (gimple_assign_nonzero): ... this and remove strict_overflow_p
9700 (gimple_stmt_nonzero_warnv_p): Rename to ...
9701 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
9703 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
9704 (extract_range_basic): Adjust, do not disable propagation on
9705 strict overflow sensitive simplification.
9706 (vrp_visit_cond_stmt): Likewise.
9708 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
9710 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
9711 body size unconditionally.
9713 2017-05-07 Jeff Law <law@redhat.com>
9716 2017-05-06 Jeff Law <law@redhat.com>
9717 PR tree-optimization/78496
9718 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9721 PR tree-optimization/78496
9722 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9723 (simplify_stmt_using_ranges): Call it.
9724 (vrp_dom_walker::before_dom_children): Extract equivalences
9725 from an ASSERT_EXPR with an equality comparison against a
9728 2017-05-06 Jeff Law <law@redhat.com>
9730 PR tree-optimization/78496
9731 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9734 PR tree-optimization/78496
9735 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9736 (simplify_stmt_using_ranges): Call it.
9737 (vrp_dom_walker::before_dom_children): Extract equivalences
9738 from an ASSERT_EXPR with an equality comparison against a
9741 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9743 * lra-constraints.c (lra_copy_reg_equiv): New function.
9744 (split_reg): Use it to copy equivalence information from the
9745 original register to the spill register.
9747 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9749 PR rtl-optimization/75964
9750 * simplify-rtx.c (simplify_const_relational_operation): Remove
9751 invalid handling of comparisons of integer ABS.
9753 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
9755 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
9757 (init_regs): Remove declaration.
9758 (function_arg_advance_32): Initialize error_p as boolean variable.
9760 2017-05-05 Nathan Sidwell <nathan@acm.org>
9762 * store-motion.c (remove_reachable_equiv_notes): Reformat long
9763 lines. Use for (;;).
9765 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9767 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
9768 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
9769 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
9770 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
9771 VF=2 that require versioning.
9773 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9775 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
9778 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9780 * diagnostic.h (diagnostic_override_option_index): Convert from
9781 macro to inline function.
9783 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9785 * diagnostic.c (last_module_changed_p): New function.
9786 (set_last_module): New function.
9787 (diagnostic_report_current_module): Convert macro usage to
9788 the above functions.
9789 * diagnostic.h (diagnostic_context::last_module): Strengthen
9790 from const line_map * to const line_map_ordinary *.
9791 (diagnostic_last_module_changed): Delete macro.
9792 (diagnostic_set_last_module): Delete macro.
9794 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9796 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
9797 with diagnostic_report_diagnostic.
9798 (diagnostic_n_impl_richloc): Likewise.
9799 * diagnostic.h (report_diagnostic): Delete macro.
9800 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
9801 with diagnostic_report_diagnostic.
9802 * substring-locations.c (format_warning_va): Likewise.
9804 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9806 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
9807 save/restor of format_spec. Move option-printing code to...
9808 (print_option_information): ...this new function, and
9809 reimplement by simply printing to the pretty_printer,
9810 rather than appending to the format string.
9812 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9814 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
9815 handling logic into...
9816 (update_effective_level_from_pragmas): ...this new function.
9818 2017-05-04 Andrew Waterman <andrew@sifive.com>
9820 * config/riscv/riscv.opt (mstrict-align): New option.
9821 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
9822 (SLOW_UNALIGNED_ACCESS): Define.
9823 (riscv_slow_unaligned_access): Declare.
9824 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
9826 (riscv_slow_unaligned_access): New variable.
9827 (rocket_tune_info): Set slow_unaligned_access to true.
9828 (optimize_size_tune_info): Set slow_unaligned_access to false.
9829 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
9830 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
9831 (riscv_option_override): Set riscv_slow_unaligned_access.
9832 * doc/invoke.texi: Add -mstrict-align to RISC-V.
9834 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
9836 * config/riscv/riscv.md: Unify indentation.
9838 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
9843 * config/rs6000/rs6000.md (u code attribute): Add FIX and
9845 (extendsi<mode>2): Add support for doing sign extension via
9846 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
9847 don't have ISA 3.0 instructions.
9848 (extendsi<mode>2 splitter): Likewise.
9849 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
9850 generate the normal insns since SImode can now go in vector
9851 registers. Disallow the special UNSPECs needed for previous
9852 machines to hide SImode being used. Add new insns
9853 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
9854 (fix_trunc<mode>si2_stfiwx): Likewise.
9855 (fix_trunc<mode>si2_internal): Likewise.
9856 (fixuns_trunc<mode>si2): Likewise.
9857 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9858 (fctiw<u>z_<mode>_smallint): Likewise.
9859 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
9860 of floating point to 32-bit integer from doing a direct move to
9861 the GPR registers to do a store.
9862 (fctiwz_<mode>): Break long line.
9864 2017-05-05 Bin Cheng <bin.cheng@arm.com>
9866 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
9867 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
9868 (addr_list, addr_offset_valid_p): New.
9869 (split_address_groups): Check offset validity with above function.
9870 (gt-tree-ssa-loop-ivopts.h): Include header file.
9872 2017-05-05 Nathan Sidwell <nathan@acm.org>
9874 * config.gcc (arm*-*-*): Add missing 'fi'.
9876 2017-05-05 Steve Ellcey <sellcey@cavium.com>
9878 * doc/invoke.texi (-fopt-info): Explicitly say order of options
9879 included in -fopt-info does not matter.
9880 * doc/optinfo.texi (-fopt-info): Fix description of default
9881 behavour. Explicitly say order of options included in -fopt-info
9884 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
9886 * config.gcc: Allow combinations of aprofile and rmprofile values for
9887 --with-multilib-list.
9888 * config/arm/t-multilib: New file.
9889 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
9890 variables. Remove setting of ISA and floating-point ABI in
9891 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
9892 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
9893 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
9894 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
9896 * config/arm/t-rmprofile: Likewise except for the matches changes.
9897 * doc/install.texi (--with-multilib-list): Document the combination of
9898 aprofile and rmprofile values and warn about pitfalls in doing that.
9900 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9902 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
9903 (movdi_aarch64): Likewise.
9905 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9907 PR tree-optimization/80632
9908 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
9910 (build_arrays): Initialize it for virtual phis.
9911 (fix_phi_nodes): Use it for virtual phis.
9913 PR tree-optimization/80558
9914 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
9915 [x, y] op z into [x op, y op z] for op & or | if conditions
9918 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
9919 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9922 * config/arm/arm.md (use_literal_pool): Remove.
9923 (64-bit immediate split): No longer takes cost into consideration
9924 if arm_disable_literal_pool is enabled.
9925 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
9926 used when arm_disable_literal_pool is enabled.
9927 (arm_max_const_double_inline_cost): Remove use of
9928 arm_disable_literal_pool.
9929 (push_minipool_fix): Add assert.
9930 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
9931 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
9932 (no_literal_pool_sf_immediate): New.
9934 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9936 PR tree-optimization/80613
9937 * tree-ssa-dce.c (propagate_necessity): Remove cases for
9938 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
9940 2017-05-05 Richard Biener <rguenther@suse.de>
9942 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
9944 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
9946 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
9947 of this flag from insn conditions due to removal from r247495.
9949 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9951 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
9953 (arm_early_store_addr_dep_ptr): Likewise.
9954 * config/arm/aarch-common-protos.h
9955 (arm_early_load_addr_dep_ptr): Add prototype.
9956 (arm_early_store_addr_dep_ptr): Likewise.
9957 * config/arm/cortex-a53.md: Add new bypasses.
9959 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9961 * tree.c (next_type_uid): Change type to unsigned.
9962 (type_hash_canon): Decrement back next_type_uid if
9963 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
9964 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
9967 2017-05-04 Martin Sebor <msebor@redhat.com>
9969 * builtins.c: Fix a trivial typo in a comment.
9972 * builtins.c (check_sizes): Adjust to handle reading past the end.
9973 Avoid printing excessive upper bound of ranges. Use %E to print
9974 tree nodes instead of converting them to %wu.
9975 (expand_builtin_memchr): New function.
9976 (compute_dest_size): Rename...
9977 (compute_objsize): ...to this.
9978 (expand_builtin_memcpy): Adjust.
9979 (expand_builtin_mempcpy): Adjust.
9980 (expand_builtin_strcat): Adjust.
9981 (expand_builtin_strcpy): Adjust.
9982 (check_strncat_sizes): Adjust.
9983 (expand_builtin_strncat): Adjust.
9984 (expand_builtin_strncpy): Adjust and simplify.
9985 (expand_builtin_memset): Adjust.
9986 (expand_builtin_bzero): Adjust.
9987 (expand_builtin_memcmp): Adjust.
9988 (expand_builtin): Handle memcmp.
9989 (maybe_emit_chk_warning): Check strncat just once.
9991 2017-05-04 Martin Sebor <msebor@redhat.com>
9993 PR preprocessor/79214
9996 * builtins.c (check_sizes): Add inlining context and issue
9997 warnings even when -Wno-system-headers is set.
9998 (check_strncat_sizes): Same.
9999 (expand_builtin_strncat): Same.
10000 (expand_builtin_memmove): New function.
10001 (expand_builtin_stpncpy): Same.
10002 (expand_builtin): Handle memmove and stpncpy.
10004 2017-05-04 Bin Cheng <bin.cheng@arm.com>
10006 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
10007 which is not used any more.
10009 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
10011 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
10013 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
10015 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
10016 (cortexa53_tunings): Likewise.
10017 (cortexa57_tunings): Likewise.
10018 (cortexa72_tunings): Likewise.
10019 (cortexa73_tunings): Likewise.
10021 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
10023 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
10024 Set loop alignment to 8.
10026 2017-05-04 Martin Sebor <msebor@redhat.com>
10028 PR translation/80280
10029 * builtins.c (expand_builtin_object_size): Add missing quoting to
10030 %D and like directives.
10031 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
10032 (hsa_type_for_tree_type): Same.
10033 (verify_function_arguments): Same.
10034 * symtab.c (symbol_table::change_decl_assembler_name): Same.
10035 * varasm.c (get_section): Same.
10038 2017-05-04 Martin Sebor <msebor@redhat.com>
10040 PR translation/80280
10041 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
10043 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
10045 * config/aarch64/aarch64.c (generic_addrcost_table):
10046 Change HI/TI mode setting.
10048 2017-05-04 Martin Jambor <mjambor@suse.cz>
10050 PR tree-optimization/80622
10051 * tree-sra.c (comes_initialized_p): New function.
10052 (build_accesses_from_assign): Only set write lazily when
10053 comes_initialized_p is false.
10054 (analyze_access_subtree): Use comes_initialized_p.
10055 (propagate_subaccesses_across_link): Assert !comes_initialized_p
10056 instead of testing for PARM_DECL.
10058 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10060 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
10061 constraint on operand 0 to allow more general addressing modes.
10062 Adjust output template.
10063 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
10065 * config/aarch64/aarch64-protos.h
10066 (aarch64_address_valid_for_prefetch_p): Declare prototype.
10067 * config/aarch64/constraints.md (Dp): New address constraint.
10068 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
10071 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
10073 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
10074 update use of estimate_ipcp_clone_size_and_time.
10075 (estimate_local_effects): Update use of
10076 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
10077 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
10078 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
10079 Return nonspecialized time.
10081 2017-05-04 Richard Biener <rguenther@suse.de>
10083 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
10084 for the last VUSE which def dominates the PHI. Directly call
10086 (get_continuation_for_phi_1): Remove.
10088 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
10090 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
10091 to explain the use of truncating division. Cap the number of
10092 iterations to the maximum given by nb_iterations_upper_bound,
10095 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
10097 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
10098 * configure: Regenerate.
10099 * config.in: Regenerate.
10100 * config/i386/driver-mingw32.c: new file.
10101 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
10102 * config.host: Link driver-mingw32.o on MinGW host.
10103 * doc/install.texi: Document new --enable-mingw-wildcard configure
10106 2017-05-04 Marek Polacek <polacek@redhat.com>
10108 PR tree-optimization/80612
10109 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
10111 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10112 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
10114 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
10115 (movt splitter): Likewise.
10116 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
10117 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
10118 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
10119 block for Thumb-1 with MOVT.
10120 (thumb2_legitimate_address_p): Move code block ...
10121 (can_avoid_literal_pool_for_label_p): ... into this new function.
10122 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
10124 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
10125 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
10126 "M-profile targets with the MOVT instruction".
10128 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10130 * config/arm/arm-builtins.c (arm_init_builtins): Rename
10131 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
10132 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
10134 2017-05-04 Martin Liska <mliska@suse.cz>
10136 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
10137 variable cond_code.
10139 2017-05-04 Richard Biener <rguenther@suse.de>
10141 * tree.c (array_at_struct_end_p): Handle arrays at struct
10142 end with flexarrays more conservatively. Refactor and treat
10143 arrays of arrays or aggregates more strict. Fix
10144 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
10145 * tree.c (array_at_struct_end_p): Adjust prototype.
10146 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
10147 * gimple-fold.c (get_range_strlen): Likewise.
10148 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
10150 2017-05-04 Richard Biener <rguenther@suse.de>
10152 PR tree-optimization/31130
10153 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
10155 (supports_overflow_infinity): Likewise.
10156 (is_negative_overflow_infinity): Likewise.
10157 (is_positive_overflow_infinity): Likewise.
10158 (is_overflow_infinity): Likewise.
10159 (stmt_overflow_infinity): Likewise.
10160 (overflow_infinity_range_p): Likewise.
10161 (usable_range_p): Remove as always returning true.
10162 (make_overflow_infinity): Remove.
10163 (negative_overflow_infinity): Likewise.
10164 (positive_overflow_infinity): Likewise.
10165 (avoid_overflow_infinity): Likewise.
10166 (set_value_range): Adjust accordingly.
10167 (set_value_range_to_nonnegative): Likewise, remove now unused
10168 overflow_infinity arg.
10169 (vrp_operand_equal_p): Adjust.
10170 (update_value_range): Likewise.
10171 (range_int_cst_singleton_p): Likewise.
10172 (operand_less_p): Likewise.
10173 (compare_values_warnv): Likewise.
10174 (extract_range_for_var_from_comparison_expr): Likewise.
10175 (vrp_int_const_binop): Likewise.
10176 (zero_nonzero_bits_from_vr): Likewise.
10177 (extract_range_from_multiplicative_op_1): Likewise.
10178 (extract_range_from_binary_expr_1): Likewise.
10179 (extract_range_from_unary_expr): Likewise.
10180 (extract_range_from_comparison): Likewise.
10181 (extract_range_basic): Likewise.
10182 (adjust_range_with_scev): Likewise.
10183 (compare_ranges): Likewise.
10184 (compare_range_with_value): Likewise.
10185 (dump_value_range): Likewise.
10186 (test_for_singularity): Likewise, remove strict_overflow_p parameter
10188 (simplify_cond_using_ranges): Adjust.
10190 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
10192 * brig-builtins.def: Added a builtin for class_f64.
10193 * builtin-types.def: Added a builtin type needed by class_f64.
10195 2017-05-03 Jason Merrill <jason@redhat.com>
10197 * timevar.def: Add TV_CONSTEXPR.
10199 2017-05-03 David Malcolm <dmalcolm@redhat.com>
10201 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
10203 2017-05-03 Martin Jambor <mjambor@suse.cz>
10205 * ipa-prop.c (ipa_update_after_lto_read): Removed.
10206 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
10207 * ipa-cp.c (ipcp_propagate_stage): Do not call
10208 ipa_update_after_lto_read.
10209 * ipa-inline.c (ipa_inline): Likewise.
10211 2017-05-03 Martin Jambor <mjambor@suse.cz>
10213 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
10214 tag. Added a default constructor and a destructor.
10215 (ipa_edge_args_sum_t): New class;
10216 (ipa_edge_args_sum): Declare.
10217 (ipa_edge_args_vector): Remove declaration.
10218 (IPA_EDGE_REF): Use ipa_edge_args_sum.
10219 (ipa_free_edge_args_substructures): Remove declaration.
10220 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
10221 (ipa_edge_args_info_available_for_edge_p): Likewise.
10222 * ipa-prop.c (ipa_edge_args_vector): Removed.
10223 (edge_removal_hook_holder): Likewise.
10224 (edge_duplication_hook_holder): Likewise.
10225 (ipa_edge_args_sum): New variable.
10226 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
10227 ipa_edge_args_vector.
10228 (ipa_free_edge_args_substructures): Likewise.
10229 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
10230 ipa_edge_args_vector.
10231 (ipa_edge_removal_hook): Turned into method
10232 ipa_edge_args_sum_t::remove.
10233 (ipa_edge_duplication_hook): Turned into method
10234 ipa_edge_args_sum_t::duplicate.
10235 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
10236 registering edge hooks.
10237 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
10238 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
10239 ipa_edge_args_sum instead of ipa_edge_args_vector.
10240 * ipa-profile.c (ipa_profile): Likewise.
10242 2017-05-03 Martin Jambor <mjambor@suse.cz>
10244 * symbol-summary.h (function_summary): New method exists.
10245 (function_summary::symtab_removal): Deallocate through release.
10246 (call_summary): New class.
10247 (gt_ggc_mx): New overload.
10248 (gt_pch_nx): Likewise.
10249 (gt_pch_nx): Likewise.
10251 2017-05-03 Jeff Law <law@redhat.com>
10253 PR tree-optimization/78496
10254 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
10255 from simplify_cond_using_ranges. Split off code to walk
10256 backwards through casts into ...
10257 (simplify_cond_using_ranges_2): New function.
10258 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
10259 (execute_vrp): After identifying jump threads, call
10260 simplify_cond_using_ranges_2.
10262 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
10265 * ipa-inline.h (inline_summary): Add ctor.
10266 (create_ggc): Do not use ggc_cleared_alloc.
10268 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
10269 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10271 * gcc.c (handle_braces): Support escaping in switch matching
10273 * doc/invoke.texi (Spec Files): Document it.
10274 Remove superfluous @code markup in items.
10276 2017-05-03 David Malcolm <dmalcolm@redhat.com>
10278 * diagnostic-show-locus.c (struct column_range): New struct.
10279 (get_affected_columns): New function.
10280 (get_printed_columns): New function.
10281 (struct correction): New struct.
10282 (correction::ensure_capacity): New function.
10283 (correction::ensure_terminated): New function.
10284 (struct line_corrections): New struct.
10285 (line_corrections::~line_corrections): New dtor.
10286 (line_corrections::add_hint): New function.
10287 (layout::print_trailing_fixits): Reimplement in terms of the new
10289 (selftest::test_overlapped_fixit_printing): New function.
10290 (selftest::diagnostic_show_locus_c_tests): Call it.
10292 2017-05-03 Nathan Sidwell <nathan@acm.org>
10294 Canonicalize canonical type hashing
10295 * tree.h (type_hash_canon_hash): Declare.
10296 * tree.c (type_hash_list, attribute_hash_list): Move into
10297 type_hash_canon_hash.
10298 (build_type_attribute_qual_variant): Break out hash code calc into
10299 type_hash_canon_hash.
10300 (type_hash_canon_hash): New. Generic type hash computation.
10301 (build_range_type_1, build_array_type_1, build_function_type,
10302 build_method_type_directly, build_offset_type, build_complex_type,
10303 make_vector_type): Call it.
10305 2017-05-03 Richard Biener <rguenther@suse.de>
10307 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
10308 When all DRs have unknown misaligned do not always peel
10309 when there is a store but apply the same costing model as if
10310 there were only loads.
10312 2017-05-03 Richard Biener <rguenther@suse.de>
10315 PR tree-optimization/80492
10316 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
10317 compare_base_decls returning dont-know properly.
10319 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
10321 * config/arm/iterators.md (CCSI): New mode iterator.
10322 (arch): New mode attribute.
10323 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
10324 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
10325 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
10326 code iterator for success result mode.
10327 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
10328 the corresponding new insn generators.
10330 2017-05-03 Bin Cheng <bin.cheng@arm.com>
10333 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10334 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
10336 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
10338 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
10339 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
10340 (DDR_A): Wrap DDR argument in brackets.
10341 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
10342 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
10343 (DDR_REVERSED_P): Likewise.
10345 2017-05-03 Jakub Jelinek <jakub@redhat.com>
10347 PR tree-optimization/79472
10348 * tree-switch-conversion.c (struct switch_conv_info): Add
10349 contiguous_range and default_case_nonstandard fields.
10350 (collect_switch_conv_info): Compute contiguous_range and
10351 default_case_nonstandard fields, don't clear final_bb if
10352 contiguous_range and only the default case doesn't have the required
10354 (check_all_empty_except_final): Set default_case_nonstandard instead
10355 of failing if contiguous_range and the default case doesn't have empty
10357 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
10358 and only the default case doesn't have the required constants. Skip
10360 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
10361 if default_case_nonstandard.
10362 (build_constructors): Build constant 1 just once. Assert that default
10363 values aren't inserted in between cases if contiguous_range. Skip
10365 (build_arrays): Skip virtual phis.
10366 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
10367 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
10368 Handle virtual phis.
10369 (gen_inbound_check): Handle default_case_nonstandard case.
10370 (process_switch): Adjust check_final_bb caller. Call
10371 gather_default_values with the first non-default case instead of
10372 default case if default_case_nonstandard.
10374 2017-05-02 Nathan Sidwell <nathan@acm.org>
10376 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
10377 check. Fix formatting.
10379 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
10381 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
10382 errors when comparing specialized and unspecialized times.
10384 2017-05-02 David Malcolm <dmalcolm@redhat.com>
10386 * diagnostic-show-locus.c
10387 (layout::should_print_annotation_line_p): Make private.
10388 (layout::print_annotation_line): Make private.
10389 (layout::annotation_line_showed_range_p): Make private.
10390 (layout::show_ruler): Make private.
10391 (layout::print_source_line): Make private. Pass in line and
10392 line_width, rather than calling location_get_source_line. Drop
10394 (layout::print_leading_fixits): New method.
10395 (layout::print_any_fixits): Rename to...
10396 (layout::print_trailing_fixits): ...this, and make private.
10397 Don't print newline fixits.
10398 (diagnostic_show_locus): Move logic for printing one row into...
10399 (layout::print_line): ...this new function. Move the
10400 location_get_source_line call and error-handling from
10401 print_source_line to here. Call print_leading_fixits, and rename
10402 print_any_fixits to print_trailing_fixits.
10403 (selftest::test_fixit_insert_containing_newline): Update now that
10404 newlines are partially supported.
10405 (selftest::test_fixit_insert_containing_newline_2): New test.
10406 (selftest::test_fixit_replace_containing_newline): Update comments.
10407 (selftest::diagnostic_show_locus_c_tests): Call the new test.
10408 * edit-context.c (class added_line): New class.
10409 (class edited_line): Describe newline handling in comment.
10410 (edited_line::actually_edited_p): New method.
10411 (edited_line::print_content): Delete redundant decl.
10412 (edited_line::m_predecessors): New field.
10413 (edited_file::print_content): Call edited_line::print_content.
10414 (edited_file::print_diff): Update to support newlines.
10415 (edited_file::print_diff_hunk): Likewise.
10416 (edited_file::print_run_of_changed_lines): New function.
10417 (edited_file::print_diff_line): Convert to...
10418 (print_diff_line): ...this.
10419 (edited_file::get_effective_line_count): New function.
10420 (edited_line::edited_line): Initialize new field m_predecessors.
10421 (edited_line::~edited_line): Clean up m_predecessors.
10422 (edited_line::apply_fixit): Handle newlines.
10423 (edited_line::get_effective_line_count): New function.
10424 (edited_line::print_content): New function.
10425 (edited_line::print_diff_lines): New function.
10426 (selftest::test_applying_fixits_insert_containing_newline): New
10428 (selftest::test_applying_fixits_replace_containing_newline): New
10430 (selftest::insert_line): New function.
10431 (selftest::test_applying_fixits_multiple_lines): Add example of
10433 (selftest::edit_context_c_tests): Call the new tests.
10435 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10437 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
10438 parameter cand. Update dump information.
10439 (get_computation_cost): Update uses.
10441 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10443 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
10444 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
10445 (get_computation_at, rewrite_use_address): Update use of
10446 get_computation_aff.
10448 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10450 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
10451 (get_computation): Delete.
10452 (get_computation_cost): Implement like get_computation_cost_at.
10453 Use get_computation_at.
10454 (get_computation_cost_at): Delete.
10455 (rewrite_use_nonlinear_expr): Use get_computation_at.
10456 (rewrite_use_compare, remove_unused_ivs): Ditto.
10458 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10460 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
10462 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10464 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
10465 (ivopts_global_cost_for_size): Rename parameter and update uses.
10466 (iv_ca_recount_cost): Update uses.
10467 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
10468 candidates seperately in n_invs and n_cands.
10469 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
10471 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10473 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
10474 (find_inv_vars_cb): New.
10475 (find_depends): Renamed to ...
10476 (find_inv_vars): ... this.
10477 (add_candidate_1, force_var_cost): Call find_inv_vars.
10478 (split_address_cost, determine_group_iv_cost_cond): Ditto.
10480 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10482 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
10483 inv_vars. Add inv_exprs.
10484 (struct iv_cand): Rename depends_on to inv_vars.
10485 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
10486 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
10487 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
10488 (dump_cand): Dump inv_vars.
10489 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
10490 (record_invariant, find_depends, add_candidate_1): Ditto.
10491 (set_group_iv_cost, force_var_cost): Ditto.
10492 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
10493 (get_computation_cost_at, get_computation_cost): Ditto.
10494 (determine_group_iv_cost_generic): Ditto.
10495 (determine_group_iv_cost_address): Ditto.
10496 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
10497 (determine_group_iv_costs): Ditto.
10498 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
10499 (iv_ca_set_remove_invariants): Renamed to ...
10500 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
10501 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
10502 (iv_ca_set_add_invariants): Renamed to ...
10503 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
10504 (iv_ca_set_cp): Use iv_ca_set_add_invs.
10505 (iv_ca_has_deps): Support inv_vars and inv_exprs.
10506 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
10507 (create_new_ivs): Remove useless dump.
10509 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10511 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
10513 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
10514 (iv_ca_set_no_cp, create_new_iv): Ditto.
10516 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10518 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
10520 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10522 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
10523 function tree_check2.
10525 2017-05-02 Martin Liska <mliska@suse.cz>
10527 * doc/gcov.texi: Add missing preposition.
10528 * gcov.c (function_info::function_info): Properly fill up
10529 all member variables.
10531 2017-05-02 Tamar Christina <tamar.christina@arm.com>
10533 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
10535 2017-05-02 Tamar Christina <tamar.christina@arm.com>
10537 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
10539 2017-05-02 Martin Liska <mliska@suse.cz>
10542 * lto-streamer-in.c (lto_read_tree_1): Remove
10543 LTO_STREAMER_DEBUG.
10544 * lto-streamer.c (struct tree_hash_entry): Likewise.
10545 (struct tree_entry_hasher): Likewise.
10546 (tree_entry_hasher::hash): Likewise.
10547 (tree_entry_hasher::equal): Likewise.
10548 (lto_streamer_init): Likewise.
10549 (lto_orig_address_map): Likewise.
10550 (lto_orig_address_get): Likewise.
10551 (lto_orig_address_remove): Likewise.
10552 * lto-streamer.h: Likewise.
10553 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
10554 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
10556 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
10558 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
10559 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
10560 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
10561 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
10562 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
10563 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
10564 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
10565 (mm_maskz_sub_ss): New intrinsics.
10566 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
10567 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
10568 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
10569 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
10570 (__builtin_ia32_subss_mask_round): New builtins.
10571 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
10572 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
10573 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
10575 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
10576 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
10578 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
10581 2017-05-02 Martin Jambor <mjambor@suse.cz>
10583 PR tree-optimization/78687
10584 * tree-sra.c (access): New field parent.
10585 (process_subtree_disqualification): New function.
10586 (disqualify_candidate): Call it.
10587 (build_accesses_from_assign): Reset write flag if creating an
10589 (build_access_subtree): Fill in parent field and also prpagate
10590 down grp_write flag.
10591 (create_artificial_child_access): New parameter set_grp_write, set
10592 grp_write to its value.
10593 (propagate_subaccesses_across_link): Also propagate grp_write flag
10595 (propagate_all_subaccesses): Push the closest parent back to work
10596 queue if add_access_to_work_queue returned true.
10598 2017-05-02 Richard Biener <rguenther@suse.de>
10600 * common.opt (fstrict-overflow): Alias negative to fwrapv.
10601 * doc/invoke.texi (fstrict-overflow): Remove all traces of
10602 -fstrict-overflow documentation.
10603 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
10604 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
10605 flag_strict_overflow.
10606 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
10607 * lto-opts.c (lto_write_options): Do not stream it.
10608 * lto-wrapper.c (merge_and_complain): Do not handle it.
10609 * opts.c (default_options_table): Do not set -fstrict-overflow.
10610 (finish_options): Likewise do not clear it when sanitizing.
10611 * simplify-rtx.c (simplify_const_relational_operation): Do not
10612 test flag_strict_overflow.
10614 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10616 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
10617 using enabled attribute.
10618 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
10619 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
10620 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
10621 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
10622 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
10623 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
10624 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
10625 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
10626 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
10627 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
10629 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10631 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
10633 2017-05-02 Richard Biener <rguenther@suse.de>
10635 PR tree-optimization/80591
10637 2017-04-10 Richard Biener <rguenther@suse.de>
10639 * tree-ssa-structalias.c (find_func_aliases): Properly handle
10642 2017-05-02 Richard Biener <rguenther@suse.de>
10644 PR tree-optimization/80549
10645 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
10646 (cleanup_tree_cfg_noloop): Create forwarders to known loop
10647 headers if they do not have a preheader.
10649 2017-05-02 Martin Liska <mliska@suse.cz>
10652 * common.opt: Fix typo.
10653 * doc/invoke.texi: Likewise.
10655 2017-05-01 Jan Beulich <jbeulich@suse.com>
10657 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
10658 swapping, add (x,x,m,x,n) alternative.
10660 2017-05-01 Nathan Sidwell <nathan@acm.org>
10662 * calls.c (combine_pending_stack_adjustment_and_call): Remove
10663 unnecessary unadjusted_alignment check.
10665 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10668 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
10670 * gimplify.c (gimplify_cilk_detach): New function.
10671 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
10672 * tree-core.h: Document EXPR_CILK_SPAWN.
10673 * tree.h (EXPR_CILK_SPAWN): Define.
10675 2017-05-01 David Malcolm <dmalcolm@redhat.com>
10677 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
10678 to use new fixit_hint representation, using the "replace" logic.
10679 (get_line_span_for_fixit_hint): Likewise.
10680 (layout::print_any_fixits): Likewise.
10681 (selftest::test_one_liner_many_fixits): Rename to...
10682 (selftest::test_one_liner_many_fixits_1): ...this, and update
10683 comment and expected output to reflect that the multiple fix-it
10684 hints are now consolidated into one insertion.
10685 (selftest::test_one_liner_many_fixits_2): New test.
10686 (selftest::test_diagnostic_show_locus_one_liner): Update for
10688 (selftest::test_fixit_consolidation): Update for fix-it API
10690 * diagnostic.c (print_parseable_fixits): Likewise.
10691 * edit-context.c (edited_line::m_line_events): Convert from
10692 auto_vec <line_event *> to auto_vec <line_event>.
10693 (class line_event): Convert from abstract base class to a concrete
10694 class, taking over the role of replace_event.
10695 (class insert_event): Delete.
10696 (class replace_event): Rename to class line_event. Convert to
10698 (edit_context::add_fixits): Reimplement.
10699 (edit_context::apply_insert): Delete.
10700 (edit_context::apply_replace): Rename to...
10701 (edit_context::apply_fixit): ...this. Convert to half-open range.
10702 (edited_file::apply_insert): Delete.
10703 (edited_file::apply_replace): Rename to...
10704 (edited_file::apply_fixit): ...this.
10705 (edited_line::~edited_line): Drop deletion of events.
10706 (edited_line::apply_insert): Delete.
10707 (edited_line::apply_replace): Rename to...
10708 (edited_line::apply_fixit): ...this. Convert to half-open range.
10709 Update for change to type of m_line_events.
10710 * edit-context.h (edit_context::apply_insert): Delete.
10711 (edit_context::apply_replace): Rename to...
10712 (edit_context::apply_fixit): ...this.
10714 2017-05-01 Martin Sebor <msebor@redhat.com>
10716 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
10719 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
10722 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
10723 __get_cpuid_max returns 0.
10724 (__get_cpuid_count): Ditto.
10726 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
10728 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
10729 replacement expression is another instance of one of its arguments.
10731 2017-05-01 Jakub Jelinek <jakub@redhat.com>
10734 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
10735 check for stack push/pop autoinc.
10736 * config/i386/i386.c (ix86_agi_dependent): Return false
10737 if the only reason why modified_in_p returned true is that
10738 addr is SP based and set_insn is a push or pop.
10740 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10742 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
10745 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10748 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
10749 (account_size_time): Use two predicates - exec_pred and
10751 (evaluate_conditions_for_known_args): Compute both clause and
10753 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
10754 (inline_summary_t::duplicate): Update.
10755 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
10757 (compute_inline_parameters): Likewise.
10758 (estimate_edge_size_and_time): Update caluclation of time.
10759 (estimate_node_size_and_time): Compute both time and nonspecialized
10761 (estimate_ipcp_clone_size_and_time): Update.
10762 (inline_merge_summary): Update.
10763 (do_estimate_edge_time): Update.
10764 (do_estimate_edge_size): Update.
10765 (do_estimate_edge_hints): Update.
10766 (inline_read_section, inline_write_summary): Stream both new predicates.
10767 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
10769 (compute_inlined_call_time): Cleanup.
10770 (big_speedup_p): Update.
10771 (edge_badness): Update.
10772 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
10773 (size_time_entry): Replace predicate by exec_predicate and
10774 nonconst_predicate.
10775 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
10776 (estimate_edge_time): Return also nonspec_time.
10777 (reset_edge_growth_cache): Update.
10779 2017-04-29 Jakub Jelinek <jakub@redhat.com>
10781 PR rtl-optimization/80491
10782 * ifcvt.c (noce_process_if_block): When looking for x setter
10783 with missing else_bb, don't check only the insn right before
10784 cond_earliest, but look for the last insn that x is modified in
10785 within the same bb.
10787 PR rtl-optimization/80491
10788 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
10790 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
10792 PR tree-optimization/80487
10793 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
10795 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10797 PR tree-optimization/79697
10798 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
10799 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
10800 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
10802 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
10803 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
10805 2017-04-28 Martin Sebor <msebor@redhat.com>
10807 PR tree-optimization/80523
10808 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
10809 (init_target_to_host_charmap, target_to_host, target_strtol10): New
10811 (maybe_warn, format_directive, parse_directive): Use new functions.
10812 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
10814 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
10816 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
10818 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
10820 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
10821 target_header_dir): Set correctly.
10822 * configure: Regenerated.
10823 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
10824 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
10825 instead of SYSTEM_HEADER_DIR.
10827 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
10829 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
10830 (estimate_local_effects): Likewise.
10831 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
10832 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
10833 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
10834 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
10835 do_estimate_edge_time, estimate_edge_time): Likewise.
10836 * ipa-inline-analysis.c (estimate_node_size_and_time,
10837 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
10838 (estimate_time_after_inlining): Remove.
10840 2017-04-28 Martin Liska <mliska@suse.cz>
10842 * doc/gcov.texi: Enhance documentation of gcov.
10844 2017-04-28 Martin Liska <mliska@suse.cz>
10846 * doc/gcov.texi: Sort options in alphabetic order.
10847 * doc/gcov-dump.texi: Likewise.
10848 * doc/gcov-tool.texi: Likewise.
10849 * gcov.c (print_usage): Likewise.
10850 * gcov-dump.c (print_usage): Likewise.
10851 * gcov-tool.c (print_merge_usage_message): Likewise.
10852 (print_rewrite_usage_message): Likewise.
10853 (print_overlap_usage_message): Likewise.
10855 2017-04-28 Martin Liska <mliska@suse.cz>
10857 PR gcov-profile/53915
10858 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
10860 2017-04-28 Martin Liska <mliska@suse.cz>
10862 PR gcov-profile/79891
10863 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
10864 is marked by compiler as living on a line.
10865 (get_cycles_count): Remove usage of the union.
10866 (output_intermediate_file): Likewise.
10867 (find_source): Fix GNU coding style.
10868 (accumulate_line_counts): Remove old non-all block mode.
10869 (output_lines): Remove usage of the union.
10870 * profile.c (output_location): Include all BBs, even if
10871 belonging to a same line (and file) as a previous BB.
10873 2017-04-28 Martin Liska <mliska@suse.cz>
10875 * gcov.c (process_args): Handle new argument 'w'.
10876 (read_graph_file): Assign ID to BBs.
10877 (output_branch_count): Display BB # if verbose flag is set.
10878 (output_lines): Likewise for arcs.
10879 (print_usage): Add '--verbose' option help.
10880 * doc/gcov.texi: Document --verbose (-w) option.
10882 2017-04-28 Martin Liska <mliska@suse.cz>
10884 * gcov.c (struct block_location_info): New struct.
10885 (process_file): Fill up the new structure.
10886 (read_graph_file): Replace usage of encoding by the newly added
10888 (add_line_counts): Likewise.
10889 (accumulate_line_counts): Remove usage of the union.
10890 (function_info::function_info): New function.
10891 (function_info::~function_info): Likewise.
10892 (process_file): Call delete instead of release_function.
10893 (release_function): Release the function.
10894 (release_structures): Call delete instead of release_function.
10895 (solve_flow_graph): Replace usage of num_blocks.
10896 (find_exception_blocks): Likewise.
10897 (output_lines): Fix GNU coding style.
10899 2017-04-28 Martin Liska <mliska@suse.cz>
10902 * coverage.c (coverage_remove_note_file): New function.
10903 * coverage.h: Declare the function.
10904 * toplev.c (finalize): Clean if an error has been seen.
10906 2017-04-28 Martin Liska <mliska@suse.cz>
10908 PR gcov-profile/80031
10909 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
10910 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
10911 * gcov.c (read_graph_file): Read just number of blocks.
10912 * profile.c (branch_prob): Do not stream 0 flags per a basic
10915 2017-04-28 Martin Liska <mliska@suse.cz>
10917 * gcov-dump.c (tag_*): Add new argument to declarations.
10918 (dump_gcov_file): Likewise.
10919 (tag_blocks): Add and use new argument depth.
10920 (tag_arcs): Likewise.
10921 (tag_lines): Likewise.
10922 (tag_counters): Likewise.
10923 (tag_summary): Likewise.
10924 (dump_working_sets): Use depth to do a proper indentation.
10926 2017-04-28 Jakub Jelinek <jakub@redhat.com>
10929 * cgraph.h (symtab_node::debug_symtab): No longer inline.
10930 * symtab.c (symtab_node::debug_symtab): Move definition here.
10932 2017-04-28 Richard Biener <rguenther@suse.de>
10934 * lto-streamer.h (LTO_major_version): Bump to 7.
10936 2017-04-28 Richard Biener <rguenther@suse.de>
10938 * tree-vrp.c (assert_info): New struct.
10939 (add_assert_info): New helper.
10940 (register_edge_assert_for_2): Refactor to add asserts to a vector
10942 (register_edge_assert_for_1): Likewise.
10943 (register_edge_assert_for): Likewise.
10944 (finish_register_edge_assert_for): New helper actually registering
10945 asserts where live on edge.
10946 (find_conditional_asserts): Adjust.
10947 (find_switch_asserts): Likewise.
10948 (evrp_dom_walker::try_find_new_range): Generalize.
10949 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
10951 2017-04-27 Marek Polacek <polacek@redhat.com>
10954 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
10955 arg10 and arg11 to itype.
10957 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
10959 * doc/extend.texi (Object Size Checking): Improve grammar.
10961 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
10964 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
10965 that the logic for permitting reciprocal estimates matches that
10968 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10971 * tree.c (type_cache_hasher::equal): Only compare
10972 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
10973 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
10974 non-aggregate element types.
10975 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
10976 about the flag on ARRAY_TYPEs in the comment, formatting fix.
10978 2017-04-27 Richard Biener <rguenther@suse.de>
10980 PR middle-end/80533
10981 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
10982 stripping ARRAY_REFs from MEM_EXPR make sure we're not
10983 keeping a reference to a trailing array.
10985 2017-04-27 Richard Biener <rguenther@suse.de>
10987 PR middle-end/80539
10988 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
10989 being in loop-closed SSA form conservatively.
10990 (chrec_fold_multiply_poly_poly): Likewise.
10992 2017-04-27 Tamar Christina <tamar.christina@arm.com>
10994 PR middle-end/79665
10995 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
10996 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
10998 2017-04-27 Jakub Jelinek <jakub@redhat.com>
11001 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
11002 (aarch64_function_arg_alignment): Return unsigned int again, but still
11003 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
11004 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
11005 Don't emit -Wpsabi note.
11006 (aarch64_function_arg_boundary): Likewise.
11007 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
11010 2017-04-26 Nathan Sidwell <nathan@acm.org>
11012 * tree.h (crc32_unsigned_n): Declare.
11013 (crc32_unsigned, crc32_unsigned): Make inline.
11014 * tree.c (crc32_unsigned_bits): Replace with ...
11015 (crc32_unsigned_n): ... this.
11016 (crc32_unsigned, crc32_byte): Remove.
11017 (crc32_string): Remove unnecessary braces.
11019 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
11021 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
11022 * ipa-inline-analysis.c (MAX_TIME): Remove.
11023 (account_size_time): Use sreal for time.
11024 (dump_inline_summary): Update.
11025 (estimate_function_body_sizes): Update.
11026 (estimate_edge_size_and_time): Update.
11027 (estimate_calls_size_and_time): Update.
11028 (estimate_node_size_and_time): Update.
11029 (inline_merge_summary): Update.
11030 (inline_update_overall_summary): Update.
11031 (estimate_time_after_inlining): Update.
11032 (inline_read_section): Update.
11033 (inline_write_summary): Update.
11034 * ipa-inline.c (compute_uninlined_call_time): Update.
11035 (compute_inlined_call_time): Update.
11036 (recursive_inlining): Update.
11037 (inline_small_functions): Update.
11038 (dump_overall_stats): Update.
11039 * ipa-inline.h: Include sreal.h.
11040 (size_time_entry): Turn time to sreal.
11041 (inline_summary): Turn self_time nad time to sreal.
11043 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
11045 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
11047 (sreal::stream_out, sreal::stream_in): New.
11048 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
11050 2017-04-25 Jakub Jelinek <jakub@redhat.com>
11052 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
11055 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
11058 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
11060 (dimode_scalar_chain::compute_convert_gain): Ditto.
11061 (dimode_scalar_chain::make_vector_copies): Ditto.
11062 (dimode_scalar_chain::convert_reg): Ditto.
11063 (dimode_scalar_chain::convert_insn): Ditto.
11064 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
11065 (VI248_AVX512BW_1): New mode iterator.
11066 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
11067 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
11070 2017-04-25 Martin Sebor <msebor@redhat.com>
11072 PR tree-optimization/80497
11073 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
11074 constants are representable in HOST_WIDE_INT.
11075 (parse_directive): Ditto.
11077 2017-04-25 Martin Sebor <msebor@redhat.com>
11080 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
11081 (new_zero_array): Adjust signature.
11082 (dom_info::dom_init): Used unsigned rather that size_t.
11083 (dom_info::dom_info): Same.
11085 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11086 Jakub Jelinek <jakub@redhat.com>
11089 * config/arm/arm.c: Include gimple.h.
11090 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
11091 returns negative, increment ncrn only if it returned positive.
11092 (arm_needs_doubleword_align): Return int instead of bool,
11093 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
11094 members, but if there is any such non-FIELD_DECL
11095 > PARM_BOUNDARY aligned decl, return -1 instead of false.
11096 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
11097 returns negative, increment nregs only if it returned positive.
11098 (arm_setup_incoming_varargs): Likewise.
11099 (arm_function_arg_boundary): Emit -Wpsabi note if
11100 arm_needs_doubleword_align returns negative, return
11101 DOUBLEWORD_ALIGNMENT only if it returned positive.
11103 2017-04-25 Marek Polacek <polacek@redhat.com>
11106 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
11107 first argument to type.
11109 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
11112 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
11113 type checks to test for compatibility instead of equality.
11115 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11116 Jakub Jelinek <jakub@redhat.com>
11119 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
11121 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
11122 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
11123 the alignment computation, but return their maximum in warn_alignment.
11124 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
11125 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
11127 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
11128 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
11131 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11133 * config/arc/simdext.md (dmpyh): Fix typo.
11135 2017-04-25 Richard Biener <rguenther@suse.de>
11137 PR tree-optimization/80492
11138 * alias.c (compare_base_decls): Handle registers with asm
11139 specification conservatively.
11140 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
11141 compare_base_decls returning dont-know properly.
11143 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11145 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
11146 (legitimate_offset_address_p): New function.
11147 (arc_legitimate_address_p): Use above function.
11149 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11151 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
11153 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11155 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
11156 ACCH registers whenever they are available.
11158 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11160 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
11161 double regs fix when not used.
11163 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11165 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
11167 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
11168 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
11170 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11172 * config/arc/arc.c (arc_output_addsi): Check for h-register class
11173 when emitting short ADD instructions.
11175 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11177 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
11179 (cmpsi_cc_c_insn): Likewise.
11180 (cbranchsi4_scratch): Compute proper instruction length using
11181 compact_hreg_operand.
11182 * config/arc/predicates.md (compact_hreg_operand): New predicate.
11184 2017-04-25 Richard Biener <rguenther@suse.de>
11186 PR middle-end/80509
11187 * passes.c (pass_manager::pass_manager): Initialize
11188 m_name_to_pass_map.
11190 2017-04-25 Richard Biener <rguenther@suse.de>
11192 PR tree-optimization/79201
11193 * tree-ssa-sink.c (statement_sink_location): Handle calls.
11195 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11198 * config/s390/vector.md: Split MEM->GPR vector moves for
11199 non-s_operand addresses.
11201 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11204 * config/s390/predicates.md (reload_const_wide_int_operand): New
11206 * config/s390/s390.md ("movti"): Remove d/P alternative.
11207 ("movti_bigconst"): New pattern definition.
11209 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11212 * s390-protos.h (s390_expand_cs_hqi): Removed.
11213 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
11214 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
11215 modes as well as CCZ1mode and CCZmode.
11216 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
11217 signature of s390_emit_compare_and_swap.
11218 (s390_expand_cs_hqi): Likewise, make static.
11219 (s390_expand_cs_tdsi): Generate an explicit compare before trying
11220 compare-and-swap, in some cases.
11221 (s390_expand_cs): Wrapper function.
11222 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
11224 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
11225 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
11226 patterns for small and large integers. Forbid symref memory operands.
11227 Move expander to s390.c. Require cc register.
11228 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
11229 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
11230 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
11231 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
11232 symref memory operands. Remove CC mode and call s390_match_ccmode
11234 ("atomic_exchange<mode>"): Allow and implement all integer modes.
11236 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11238 * config/s390/s390.md (define_peephole2): New peephole to help
11239 combining the load-and-test pattern with volatile memory.
11241 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11243 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
11244 with CCZmode for TARGET_Z196.
11246 2017-04-25 Jakub Jelinek <jakub@redhat.com>
11248 PR rtl-optimization/80501
11249 * combine.c (make_compound_operation_int): Set subreg_code to SET
11250 even for AND with mask of the sign bit of mode.
11252 PR rtl-optimization/80500
11253 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
11254 sum's initial value.
11256 2017-04-25 Julian Brown <julian@codesourcery.com>
11257 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11259 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
11261 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
11263 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
11265 2017-04-25 Julian Brown <julian@codesourcery.com>
11266 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11268 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
11269 (thunderx2t99_sha): New Reservation.
11271 2017-04-25 Julian Brown <julian@codesourcery.com>
11272 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11274 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
11275 type for 1-element load.
11277 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
11279 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
11281 2017-04-24 Martin Jambor <mjambor@suse.cz>
11283 PR tree-optimization/80293
11284 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
11285 char arrays not totally scalarizable if it is false.
11286 (analyze_all_variable_accesses): Pass correct value in the new
11287 parameter. Add a statistics counter.
11289 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
11291 PR middle-end/79931
11292 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
11294 2017-04-24 Richard Biener <rguenther@suse.de>
11296 PR tree-optimization/80494
11297 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
11298 out for complex types.
11300 2017-04-24 Richard Biener <rguenther@suse.de>
11302 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
11303 * tree-ssa-sccvn.c (print_scc): Print SCC size.
11304 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
11305 (DFS): Adjust and never fail.
11306 (sccvn_dom_walker::fail): Remove.
11307 (sccvn_dom_walker::before_dom_children): Adjust.
11308 (run_scc_vn): Likewise and never fail.
11309 * tree-ssa-pre.c (pass_pre::execute): Adjust.
11310 (pass_fre::execute): Likewise.
11312 2017-04-24 Richard Biener <rguenther@suse.de>
11314 PR tree-optimization/79725
11315 * tree-ssa-sink.c (statement_sink_location): Return whether
11316 failure reason was zero uses. Move that check later.
11317 (sink_code_in_bb): Deal with zero uses by removing the stmt
11320 2017-04-24 Richard Biener <rguenther@suse.de>
11323 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
11324 pointer-based references.
11326 2017-04-24 Richard Biener <rguenther@suse.de>
11329 * pass_manager.h (pass_manager::operator new): Remove.
11330 (pass_manager::operator delete): Likewise.
11331 * passes.c (pass_manager::operator new): Remove.
11332 (pass_manager::operator delete): Likewise.
11333 (pass_manager::pass_manager): Zero individual pass members.
11335 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
11338 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
11339 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
11340 Check "XEXP (src, 1)" operand here.
11341 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
11342 Check "XEXP (src, 1)" operand here.
11343 (dimode_scalar_chain::make_vector_copies): Detect count register
11344 of a shift instruction. Zero extend count register from QImode
11345 to DImode to satisfy vector shift pattern count operand predicate.
11346 Substitute vector shift count operand with a DImode copy.
11347 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
11350 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
11352 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
11353 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
11354 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
11355 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
11356 (UNSPEC_NOREX_MEM): Remove definition.
11358 2017-04-21 Richard Biener <rguenther@suse.de>
11360 PR tree-optimization/79547
11361 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
11362 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
11363 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
11364 without any constraints.
11366 2017-04-21 Richard Biener <rguenther@suse.de>
11368 PR tree-optimization/78847
11369 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
11371 2017-04-21 Richard Biener <rguenther@suse.de>
11373 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
11374 (build_distinct_type_copy): Likewise.
11375 (build_variant_type_copy): Likewise.
11376 * tree.c (build_qualified_type): Pass down mem-stat info.
11377 (build_distinct_type_copy): Likewise.
11378 (build_variant_type_copy): Likewise.
11380 2017-04-21 Richard Biener <rguenther@suse.de>
11382 PR tree-optimization/80237
11383 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
11385 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
11386 for a simplified result.
11388 2016-04-21 Richard Biener <rguenther@suse.de>
11390 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
11391 sth as strict as a simple_iv but a chrec without symbols and an
11392 operand defined in the loop we are peeling (and not some subloop).
11393 (propagate_constants_for_unrolling): Propagate all constants.
11395 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
11398 * config/i386/i386.c (print_reg): Remove assert for disalowed
11399 regno values, call output_operand_lossage instead.
11401 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
11404 * config/i386/constraints.md (Yc): New register constraint.
11405 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
11406 Use Yc constraint for alternative 2 of operand 0. Remove
11407 preferred_for_speed attribute.
11409 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
11411 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
11412 lastprivate clauses in SIMT case.
11414 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
11416 * doc/invoke.texi (-Wextra-semi): Document new warning option.
11418 2017-04-20 Richard Biener <rguenther@suse.de>
11420 PR tree-optimization/57796
11421 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
11422 as N scalar stores.
11423 (vect_model_load_cost): Cost gathers as N scalar loads.
11425 2017-04-20 Richard Biener <rguenther@suse.de>
11427 * ggc-page.c (ggc_allocated_p): Rename to ...
11428 (safe_lookup_page_table_entry): ... this and return the lookup
11430 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
11432 2017-04-20 Richard Biener <rguenther@suse.de>
11434 PR tree-optimization/80453
11435 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
11436 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
11437 from the conditions.
11438 (vn_phi_eq): Pass them down.
11439 (vn_phi_lookup): Record them.
11440 (vn_phi_insert): Likewise.
11442 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
11444 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
11445 uninitialized variable warning to avoid buffer overrun.
11447 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
11450 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
11451 is suppressed for '{ 0 }' in C.
11453 2017-04-20 Jakub Jelinek <jakub@redhat.com>
11455 * BASE-VER: Set to 8.0.0.
11457 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
11459 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
11460 priority .init_array and .fini_array section with SECTION_NOTYPE
11463 2017-04-20 Jakub Jelinek <jakub@redhat.com>
11465 PR middle-end/80423
11466 * tree.h (build_array_type): Add typeless_storage default argument.
11467 * tree.c (type_cache_hasher::equal): Also compare
11468 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
11469 (build_array_type): Add typeless_storage argument, set
11470 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
11472 (build_nonshared_array_type): Adjust build_array_type_1 caller.
11473 (build_array_type): Likewise. Add typeless_storage argument.
11475 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
11476 Jakub Jelinek <jakub@redhat.com>
11478 PR tree-optimization/80426
11479 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
11480 operation on symbolic operands, also compute the overflow for the
11481 invariant part when the operation degenerates into a negation.
11483 2017-04-19 Jakub Jelinek <jakub@redhat.com>
11486 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
11487 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
11490 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
11492 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
11495 * config/avr/avr.c (tree.h): Include it.
11496 (cgraph.h): Include it.
11497 (avr_encode_section_info): Don't warn for uninitialized progmem
11498 variable if it's just an alias.
11500 2017-04-19 Richard Biener <rguenther@suse.de>
11503 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
11504 when needed by AutoPGO.
11506 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
11509 * doc/lto.texi: Remove an extra 'that'.
11511 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
11513 PR rtl-optimization/80429
11514 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
11515 are only used in debug insns.
11517 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
11518 Vladimir Makarov <vmakarov@redhat.com>
11520 * config/sparc/predicates.md (input_operand): Add comment. Return
11521 true for any memory operand when LRA is in progress.
11522 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
11524 2017-04-18 Jeff Law <law@redhat.com>
11527 * mips.md ({return,simple_return}_internal): Do not overwrite
11530 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11532 PR tree-optimization/80443
11533 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
11534 instead of adding 1, subtract -1 and similarly instead of subtracting
11537 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
11539 PR rtl-optimization/80357
11540 * haifa-sched.c (tmp_bitmap): New variable.
11541 (model_recompute): Handle duplicate use records.
11542 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
11543 (free_global_sched_pressure_data): Free it.
11545 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
11548 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
11549 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
11550 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
11551 instead of SYSTEM_HEADER_DIR.
11553 2017-04-18 Jeff Law <law@redhat.com>
11555 PR middle-end/80422
11556 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
11557 predecessors after walking up the insn chain.
11559 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11562 * dwarf2out.c (modified_type_die): Try harder not to emit internal
11563 sizetype type into debug info.
11565 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
11568 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
11569 unneeded test for TARGET_UPPER_REGS_SF.
11570 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
11572 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11575 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
11576 instead of gsi_after_labels.
11578 2017-04-18 Jeff Law <law@redhat.com>
11580 * regcprop.c (maybe_mode_change): Avoid creating copies of the
11584 2017-04-13 Jeff Law <law@redhat.com>
11585 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11586 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11588 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
11591 * config/avr/avr.c (intl.h): Include it.
11592 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
11594 2017-04-18 Martin Liska <mliska@suse.cz>
11596 PR gcov-profile/78783
11597 * gcov-tool.c (gcov_output_files): Validate that destination
11598 file is either removed by the tool or by a user.
11600 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
11601 Guy Benyei <guybe@mellanox.com>
11603 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
11604 block, and do not negate it, the stored id is already negative.
11606 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
11608 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
11610 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
11613 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
11614 masks of options that should be turned off if the VSX vector
11615 options are turned off.
11616 (OTHER_P8_VECTOR_MASKS): Likewise.
11617 (OTHER_VSX_VECTOR_MASKS): Likewise.
11618 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
11619 rs6000_disable_incompatible_switches to validate no type switches
11621 (rs6000_incompatible_switch): New function to disallow turning on
11622 other vector options if -mno-vsx, -mno-power8-vector, or
11623 -mno-power9-vector are specified.
11625 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11627 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
11629 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11631 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
11632 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
11633 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
11634 (ARG_POINTER_CFA_OFFSET): Likewise.
11636 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11638 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
11639 conditions to take advantage of various optimizations.
11641 2017-04-13 Jeff Law <law@redhat.com>
11643 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11644 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11645 (zero_extendsidi2_dext): Likewise.
11647 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11650 * fold-const.c (fold_ternary_loc): Revert
11651 use op0 instead of fold_convert_loc (loc, type, arg0) part of
11654 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
11656 PR rtl-optimization/80343
11657 * lra-remat.c (update_scratch_ops): Assign original hard reg to
11658 new scratch pseudo.
11660 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
11663 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
11664 to ubsan_encode_value.
11666 2017-04-13 Jeff Law <law@redhat.com>
11668 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
11669 appearing in DEBUG_INSNs.
11671 2017-04-13 Martin Liska <mliska@suse.cz>
11673 PR gcov-profile/80413
11674 * gcov-io.c (gcov_write_string): Copy to buffer just when
11675 allocated size is greater than zero.
11677 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11680 * dwarf2out.c (decls_for_scope): Ignore declarations of
11681 current_function_decl in BLOCK_NONLOCALIZED_VARS.
11683 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
11686 * ipa-visibility.c (non_local_p): Fix typos.
11687 (localize_node): When localizing symbol in same comdat group,
11688 dissolve the group only when we know external symbols are going
11690 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
11692 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11694 PR tree-optimization/79390
11695 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
11696 order does not result in usable sequence, retry with reversed operand
11702 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
11703 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
11704 op0 instead of fold_convert_loc (loc, type, arg0).
11706 2017-04-12 Jeff Law <law@redhat.com>
11708 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
11709 has a delay slot in the generated code.
11711 * config/cris/cris.md (cris_preferred_reload_class): Return
11712 GENNONACR_REGS rather than GENERAL_REGS.
11714 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11717 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
11718 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
11719 signedness of the result type.
11721 2017-04-12 Richard Biener <rguenther@suse.de>
11722 Jeff Law <law@redhat.com>
11724 PR tree-optimization/80359
11725 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
11726 trim stores to TARGET_MEM_REFs.
11728 2017-04-12 Richard Biener <rguenther@suse.de>
11730 PR tree-optimization/79390
11731 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
11732 threading case even more.
11734 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
11737 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
11738 for quad_address_p for TImode, instead of just not indexed_address.
11740 2017-04-12 Richard Biener <rguenther@suse.de>
11741 Bernd Edlinger <bernd.edlinger@hotmail.de>
11743 PR middle-end/79671
11744 * alias.c (component_uses_parent_alias_set_from): Handle
11745 TYPE_TYPELESS_STORAGE.
11746 (get_alias_set): Likewise.
11747 * tree-core.h (tree_type_common): Add typeless_storage flag.
11748 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
11749 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
11750 for types containing members with TYPE_TYPELESS_STORAGE.
11751 (place_field): Likewise.
11752 (layout_type): Likewise for ARRAY_TYPE.
11753 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
11754 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
11755 TYPE_TYPELESS_STORAGE.
11756 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
11758 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11761 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
11762 first argument to type.
11764 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11768 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
11769 CONST0_RTX (mode) rather than const0_rtx where appropriate.
11770 (rs6000_expand_binop_builtin): Likewise.
11771 (rs6000_expand_ternop_builtin): Likewise; also add missing
11772 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
11773 vshasigma built-ins.
11774 * doc/extend.texi: Document that vec_xxpermdi's third argument
11775 must be a constant.
11777 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
11779 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
11780 Use shift_const cost parameter when calculating gain of STV shifts.
11782 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
11784 PR rtl-optimization/70478
11785 * lra-constraints.c (process_alt_operands): Check memory for
11786 disfavoring memory insn operand.
11788 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11790 PR middle-end/80100
11791 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
11792 left shift in unsigned HOST_WIDE_INT type.
11794 PR rtl-optimization/80385
11795 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
11796 (not (neg X)) into (plus X -1) for complex or non-integral modes.
11799 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
11800 if they have any depend clauses.
11802 2017-04-11 Martin Liska <mliska@suse.cz>
11805 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
11806 * ipa-split.c (split_function): Create a local comdat symbol
11807 if caller is in a comdat group.
11809 2017-04-11 Martin Liska <mliska@suse.cz>
11812 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
11815 2017-04-11 Martin Sebor <msebor@redhat.com>
11817 PR middle-end/80364
11818 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
11819 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
11821 (directive::set_width, directive::set_precision, format_character):
11823 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
11826 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
11829 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
11830 conflict, set target->arch_name instead of target->cpu_name.
11832 2017-04-11 Richard Biener <rguenther@suse.de>
11834 PR tree-optimization/80374
11835 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
11836 build_zero_cst, remove fold_convertible_p check again.
11838 2017-04-11 Martin Liska <mliska@suse.cz>
11841 * ubsan.c (instrument_object_size): Do not instrument register
11844 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11847 * config/i386/i386-builtin-types.def
11848 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
11849 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
11850 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
11851 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
11852 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
11853 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
11854 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
11855 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
11856 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
11857 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
11858 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
11859 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
11860 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
11861 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
11862 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
11863 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
11864 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
11865 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
11866 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
11867 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
11868 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
11869 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
11870 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
11871 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
11872 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
11873 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
11874 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
11875 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
11876 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
11877 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
11878 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
11879 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
11880 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
11881 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
11882 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
11883 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
11884 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
11885 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
11886 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
11887 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
11888 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
11889 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
11890 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
11891 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
11892 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
11893 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
11895 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
11896 flag to second_arg_count, handle 4 argument function type _COUNT
11897 aliases, handle second_arg_count on second argument rather than last.
11899 2017-04-10 Jeff Law <law@redhat.com>
11901 PR tree-optimization/80374
11902 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
11903 record anything if we can not convert integer_zero_node to the
11906 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
11909 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11910 Enhance special handling given to the TARGET_P9_MINMAX option in
11911 relation to certain other options.
11913 2017-04-10 Bin Cheng <bin.cheng@arm.com>
11915 PR tree-optimization/80153
11916 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
11917 remove POINTER_PLUS_EXPR's base part directly, rather than through
11920 2017-04-10 Richard Biener <rguenther@suse.de>
11921 Bin Cheng <bin.cheng@arm.com>
11923 PR tree-optimization/80153
11924 * tree-affine.c (aff_combination_to_tree): Get base pointer from
11925 the first element of pointer type aff_tree. Build result expr in
11927 (add_elt_to_tree): Convert to type unconditionally. Remove other
11928 fold_convert calls.
11929 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
11930 (rewrite_use_nonlinear_expr): Check invariant using iv information.
11932 2017-04-10 Richard Biener <rguenther@suse.de>
11934 * tree-ssa-structalias.c (find_func_aliases): Properly handle
11937 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
11939 PR rtl-optimization/70478
11940 * lra-constraints.c (curr_small_class_check): New.
11941 (update_and_check_small_class_inputs): New.
11942 (process_alt_operands): Update curr_small_class_check. Disfavor
11943 alternative insn memory operands. Check available regs for small
11946 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
11949 * config/mips/mips.opt (-mvirt): Update description.
11950 * doc/invoke.texi (-mvirt): Likewise.
11952 2017-04-10 Richard Biener <rguenther@suse.de>
11954 PR middle-end/80362
11955 * fold-const.c (fold_binary_loc): Look at unstripped ops when
11956 looking for NEGATE_EXPR in -A / -B to A / B folding.
11958 2017-04-10 Martin Liska <mliska@suse.cz>
11960 PR gcov-profile/80224
11961 * gcov.c (print_usage): Fix usage string.
11962 (get_gcov_intermediate_filename): Remove.
11963 (output_gcov_file): Use both for normal and intermediate format.
11964 (generate_results): Do not initialize special file for
11965 intermediate format.
11967 2017-04-10 Richard Biener <rguenther@suse.de>
11969 PR tree-optimization/80304
11970 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
11973 2017-04-10 Nathan Sidwell <nathan@acm.org>
11976 * config/rs6000/rs6000.c (rs6000_vector_type): New.
11977 (rs6000_init_builtins): Use it.
11979 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11981 * config/arm/arm.md (<mrc>): Add mode to SET source.
11982 (<mrrc>): Likewise.
11984 2017-04-10 Richard Biener <rguenther@suse.de>
11986 PR middle-end/80344
11987 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
11989 2017-04-10 Jakub Jelinek <jakub@redhat.com>
11992 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
11993 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
11994 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
11995 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
11996 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
11997 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
11998 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
11999 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
12000 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
12001 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
12002 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
12003 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
12004 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
12005 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
12006 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
12007 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
12008 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
12009 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
12010 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
12011 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
12012 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
12013 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
12014 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
12016 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
12018 PR rtl-optimization/70478
12019 * lra-constraints.c: Reverse the last patch.
12021 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
12023 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
12024 Add comment for WCHAR_T.
12026 2017-04-08 Martin Liska <mliska@suse.cz>
12029 2017-04-07 Martin Liska <mliska@suse.cz>
12032 * ipa-split.c (split_function): Add function part to a same comdat
12035 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12038 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
12040 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
12042 * rs6000/rs6000.c (vec_load_pendulum): Rename...
12043 (vec_pairing): ...to this.
12044 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
12045 (rs6000_sched_init): Adjust for name change.
12046 (struct rs6000_sched_context): Likewise.
12047 (rs6000_init_sched_context): Likewise.
12048 (rs6000_set_sched_context): Likewise.
12050 2017-04-07 Jakub Jelinek <jakub@redhat.com>
12056 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
12058 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
12059 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
12060 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
12062 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
12064 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
12066 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
12068 PR rtl-optimization/70703
12069 * ira-color.c (update_conflict_hard_regno_costs): Use
12070 int64_t instead of HOST_WIDE_INT.
12072 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
12074 PR rtl-optimization/70478
12075 * lra-constraints.c (process_alt_operands): Disfavor alternative
12076 insn memory operands.
12078 2017-04-07 Jeff Law <law@redhat.com>
12080 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
12081 CALL and NOTE_INSN_CALL_ARG_LOCATION.
12083 2017-04-07 Martin Liska <mliska@suse.cz>
12086 * config/aarch64/aarch64.c (aarch64_process_target_attr):
12087 Show error message instead of an ICE.
12089 2017-04-07 Martin Liska <mliska@suse.cz>
12092 * ipa-split.c (split_function): Add function part to a same comdat
12095 2017-04-07 Richard Biener <rguenther@suse.de>
12097 PR middle-end/80341
12098 * tree.c (get_unwidened): Also handle ! for_type case for
12100 * convert.c (do_narrow): Split out from ...
12101 (convert_to_integer_1): ... here. Do not pass final truncation
12102 type to get_unwidened for TRUNC_DIV_EXPR.
12104 2017-04-07 Richard Biener <rguenther@suse.de>
12106 * tree-affine.c (wide_int_ext_for_comb): Take type rather
12108 (aff_combination_const): Adjust.
12109 (aff_combination_scale): Likewise.
12110 (aff_combination_add_elt): Likewise.
12111 (aff_combination_add_cst): Likewise.
12112 (aff_combination_convert): Likewise.
12113 (add_elt_to_tree): Likewise. Remove unused argument.
12114 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
12116 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
12118 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
12120 * config/arm/arm.c (arm_default_short_enums): Use
12121 ARM_DEFAULT_SHORT_ENUMS.
12122 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
12124 2017-04-06 Jakub Jelinek <jakub@redhat.com>
12127 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
12128 members with redundant out-of-class redeclaration.
12130 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
12133 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
12134 * config/i386/i386.md (*zero_extendsidi2):
12135 Add (?*x,*x) and (?*v,*v) alternatives.
12137 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
12140 * config/i386/i386.c (ix86_expand_builtin)
12141 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
12142 mode from insn data. Convert operands to insn operand mode.
12143 Copy operands that don't satisfy insn predicate to a register.
12145 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
12147 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
12150 2017-04-06 Richard Biener <rguenther@suse.de>
12152 PR tree-optimization/80334
12153 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
12154 preserve alignment of accesses.
12156 2017-04-06 Richard Biener <rguenther@suse.de>
12158 PR tree-optimization/80262
12159 * tree-sra.c (build_ref_for_offset): Preserve address-space
12161 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
12162 Drop useless address-space information on MEM_REF offsets.
12164 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
12166 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
12168 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
12170 PR rtl-optimization/70703
12171 * ira-color.c (update_conflict_hard_regno_costs): Use
12172 HOST_WIDE_INT instead of long.
12174 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
12177 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
12178 not defined for x86_64 target. Add -mmmx target option when __SSE2__
12180 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
12181 for x86_64 target. Handle -m3dnowa option.
12183 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
12185 PR rtl-optimization/70703
12186 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
12187 (update_conflict_hard_regno_costs): Use long instead of unsigned
12188 arithmetic for cost calculation.
12190 2017-04-05 Jakub Jelinek <jakub@redhat.com>
12191 Bernd Edlinger <bernd.edlinger@hotmail.de>
12194 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
12197 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
12200 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
12201 ptr_mode with Pmode throughout.
12202 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
12203 into probe_stack_range and use DImode.
12205 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
12208 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
12209 call_eh_return is true.
12211 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12213 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
12214 Initialize last_match_fntype_index.
12216 2017-04-05 Jakub Jelinek <jakub@redhat.com>
12219 * tree-nvr.c: Include internal-fn.h.
12220 (pass_return_slot::execute): Ignore internal calls without
12223 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12224 Richard Biener <rguenther@suse.de>
12227 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
12228 captures used multiple times, except for the last use.
12229 * generic-match-head.c: Include gimplify.h.
12231 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12233 PR tree-optimization/79390
12234 * target.h (struct noce_if_info): Declare.
12235 * targhooks.h (default_noce_conversion_profitable_p): Declare.
12236 * target.def (noce_conversion_profitable_p): New target hook.
12237 * ifcvt.h (struct noce_if_info): New type, moved from ...
12238 * ifcvt.c (struct noce_if_info): ... here.
12239 (noce_conversion_profitable_p): Renamed to ...
12240 (default_noce_conversion_profitable_p): ... this. No longer
12242 (noce_try_store_flag_constants, noce_try_addcc,
12243 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
12244 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
12245 instead of noce_conversion_profitable_p.
12246 * config/i386/i386.c: Include ifcvt.h.
12247 (ix86_option_override_internal): Don't override
12248 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
12249 (ix86_noce_conversion_profitable_p): New function.
12250 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
12251 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
12252 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
12253 * doc/tm.texi: Regenerated.
12255 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12257 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
12260 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
12263 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
12264 instructions for small multiply cores.
12266 2017-04-04 Jeff Law <law@redhat.com>
12268 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
12270 (mips_expand_vec_perm_const): Initialize elements in orig_perm
12271 that are not set by the loop over the elements.
12273 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12276 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
12277 int mode, convert_modes it to mode as unsigned, otherwise use
12278 lowpart_subreg to mode rather than SImode.
12279 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
12280 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
12281 Use DImode instead of SImode for the shift count operand.
12282 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
12285 2017-04-04 Richard Biener <rguenther@suse.de>
12287 PR middle-end/80281
12288 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
12289 arithmetic done for the negate or the plus. Simplify.
12290 (A - (-B) -> A + B): Likewise.
12291 * fold-const.c (split_tree): Make sure to not negate pointers.
12293 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
12295 PR rtl-optimization/60818
12296 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
12297 a compare of comparisons with the thing compared if this results
12298 in a different machine mode.
12300 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
12302 * alias.c (base_alias_check): Fix typo in comment.
12303 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
12304 * cgraphunit.c (symbol_table::compile): Likewise.
12305 * collect2.c (maybe_run_lto_and_relink): Likewise.
12306 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
12307 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
12308 * config/avr/avr.c (avr_map_op_t): Likewise.
12309 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
12310 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
12311 * config/epiphany/epiphany.md (movcc): Likewise.
12312 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
12313 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
12315 * config/mips/mips.c (mips_save_restore_reg): Likewise.
12316 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
12317 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
12318 * config/sh/sh.c (sh_rtx_costs): Likewise.
12319 * fold-const.c (fold_truth_andor): Likewise.
12320 * genautomata.c (collapse_flag): Likewise.
12321 * gengtype.h (struct type::u::s): Likewise.
12322 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
12323 * input.c (FORMAT_AMOUNT): Likewise.
12324 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
12325 (known_aggs_to_agg_replacement_list): Likewise.
12326 * ipa-inline-analysis.c: Likewise.
12327 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
12328 * ipa-polymorphic-call.c
12329 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
12330 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
12331 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
12333 * modulo-sched.c (apply_reg_moves): Likewise.
12334 * omp-expand.c (build_omp_regions_1): Likewise.
12335 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
12336 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
12337 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
12338 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
12339 * value-prof.c: Likewise.
12340 * var-tracking.c (val_reset): Likewise.
12342 2017-04-03 Richard Biener <rguenther@suse.de>
12344 PR tree-optimization/80275
12345 * fold-const.c (split_address_to_core_and_offset): Handle
12348 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
12350 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
12351 descriptors is at least equal to that of functions.
12353 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
12355 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
12357 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
12360 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
12361 (mov<IMOD4:mode>): New expander.
12362 (*mov<IMOD4:mode>_internal): New insn and split pattern.
12364 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
12366 PR rtl-optimization/79405
12367 * fwprop.c (propagations_left): New variable.
12368 (forward_propagate_into): Decrement it.
12369 (fwprop_init): Initialize it.
12370 (fw_prop): If the variable has reached zero, stop propagating.
12371 (fwprop_addr): Ditto.
12373 2017-03-31 Jakub Jelinek <jakub@redhat.com>
12376 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
12377 a FUNCTION_DECL, pass it as decl instead of origin to
12380 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
12382 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
12385 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
12388 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
12389 TARGET_VSX_SMALL_INTEGER.
12391 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12393 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
12394 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
12396 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
12398 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
12399 extraction from odd-numbered MSA register.
12401 2017-03-31 Jakub Jelinek <jakub@redhat.com>
12403 PR middle-end/80173
12404 * expmed.c (store_bit_field_1): Don't attempt to create
12405 a word subreg out of hard registers wider than word if they
12406 have HARD_REGNO_NREGS of 1 for their mode.
12408 PR middle-end/80163
12409 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
12410 conversions to integer types wider than word and pointer.
12413 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
12414 (rtx_equal_for_cselib_p): Pass 0 to it.
12415 * cselib.c (cselib_hasher::equal): Likewise.
12416 (rtx_equal_for_cselib_1): Add depth argument. If depth
12417 is 128, don't look up VALUE locs and punt. Increment
12418 depth in recursive calls when walking VALUE locs.
12420 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
12422 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
12423 (make_gcov_file_name): Use the canonical path name for generating
12425 (read_line): Fix handling of files with ascii null bytes.
12427 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
12429 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
12430 to initialise a vector register instead
12431 of using a const_int.
12433 2017-03-30 Jakub Jelinek <jakub@redhat.com>
12435 PR translation/80189
12436 * gimplify.c (omp_default_clause): Use %qs instead of %s in
12437 diagnostic messages.
12439 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
12442 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
12443 (dfp_diex_<mode>): Update mode of operand 1.
12444 * doc/extend.texi (dxex, dxexq): Document change to return type.
12445 (diex, diexq): Document change to argument type.
12447 2017-03-30 Martin Jambor <mjambor@suse.cz>
12450 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
12451 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
12452 it reflects the signature changes performed at the callee side.
12453 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
12454 to cgraph_build_function_type_skip_args.
12455 (build_function_decl_skip_args): Adjust call to the above function.
12457 2017-03-30 Jakub Jelinek <jakub@redhat.com>
12460 * config/i386/sse.md
12461 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
12462 register as dest whenever it is a MEM not rtx_equal_p to the
12463 corresponding dup operand, and when forcing into reg move the
12464 reg into the memory afterwards.
12465 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
12466 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
12467 for the force_reg mode.
12468 (avx512vl_vextractf128<mode>): Use register as dest either
12469 always when a MEM, or when it is a MEM not rtx_equal_p to the
12470 corresponding dup operand, or even not when it is a CONST_VECTOR
12471 depending on the mode and lo vs. hi.
12472 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
12474 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
12475 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
12476 Likewise. Require that operands[2] is even.
12477 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
12478 Remove extraneous parens. Require that operands[2] is a multiple
12480 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
12481 operands[0] is a MEM if <mask_applied>, the predicates/constraints
12482 disallow memory then.
12484 2017-03-30 Richard Biener <rguenther@suse.de>
12486 PR tree-optimization/77498
12487 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
12488 to non-constants over backedges.
12490 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
12492 PR rtl-optimization/80233
12493 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
12494 as last_combined_insn. Do not test for BARRIER_P separately.
12496 2017-03-29 Andreas Schwab <schwab@suse.de>
12499 * calls.c (prepare_call_address): Convert funexp to Pmode before
12500 copying to temp reg.
12502 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12504 PR tree-optimization/80158
12505 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
12506 Handle possible future case of more than one alternate
12508 (replace_rhs_if_not_dup): Likewise.
12509 (replace_one_candidate): Likewise.
12511 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
12513 PR rtl-optimization/80193
12514 * ira.c (ira): Do not check allocation for LRA.
12516 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
12518 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
12519 (nvptx_output_simt_exit): Declare.
12520 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
12521 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
12522 (init_softstack_frame): Move initialization of crtl->is_leaf to...
12523 (nvptx_declare_function_name): ...here. Emit declaration of local
12524 memory space buffer for omp_simt_enter insn.
12525 (nvptx_output_unisimt_switch): New.
12526 (nvptx_output_softstack_switch): New.
12527 (nvptx_output_simt_enter): New.
12528 (nvptx_output_simt_exit): New.
12529 * config/nvptx/nvptx.h (struct machine_function): New fields
12530 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
12531 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
12532 (UNSPECV_SIMT_EXIT): Ditto.
12533 (omp_simt_enter_insn): New insn.
12534 (omp_simt_enter): New expansion.
12535 (omp_simt_exit): New insn.
12536 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
12538 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
12539 (expand_GOMP_SIMT_ENTER_ALLOC): New.
12540 (expand_GOMP_SIMT_EXIT): New.
12541 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
12542 (GOMP_SIMT_ENTER_ALLOC): Ditto.
12543 (GOMP_SIMT_EXIT): Ditto.
12544 * target-insns.def (omp_simt_enter): New insn.
12545 (omp_simt_exit): Ditto.
12546 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
12548 (lower_rec_simd_input_clauses): Implement SIMT privatization.
12549 (lower_rec_input_clauses): Likewise.
12550 (lower_lastprivate_clauses): Handle SIMT privatization.
12552 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
12553 (ompdevlow_adjust_simt_enter): New.
12554 (find_simtpriv_var_op): New.
12555 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
12556 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
12558 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
12559 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
12560 (copy_decl_for_dup_finish): Ditto.
12562 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
12564 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
12567 * config/i386/i386.c (ix86_option_override_internal): Always
12568 allow -mpreferred-stack-boundary=3 for 64-bit targets.
12570 2017-03-28 Bin Cheng <bin.cheng@arm.com>
12572 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
12574 2017-03-28 Bin Cheng <bin.cheng@arm.com>
12576 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
12577 mark new edge's irreducible flag accordign to it.
12578 (vect_do_peeling): Check loop preheader edge's irreducible flag
12579 and pass it to function slpeel_add_loop_guard.
12581 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
12583 PR tree-optimization/80218
12584 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
12585 Update block frequencies and counts.
12587 2017-03-28 Richard Biener <rguenther@suse.de>
12589 PR tree-optimization/78644
12590 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
12591 of a simplification result we may not use it at all.
12593 2017-03-28 Richard Biener <rguenther@suse.de>
12596 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
12597 without arguments, generate default definition of a SSA name.
12599 2017-03-28 Richard Biener <rguenther@suse.de>
12601 PR middle-end/80222
12602 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
12603 TYPE_REF_CAN_ALIAS_ALL references.
12604 * fold-const.c (fold_indirect_ref_1): Likewise.
12606 2017-03-28 Martin Liska <mliska@suse.cz>
12609 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
12610 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
12612 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12613 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
12615 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
12616 (EXTRA_SPECS): Define.
12617 (SUBTARGET_EXTRA_SPECS): Likewise.
12618 (SUBTARGET_CPP_SPEC): Likewise.
12619 * config/arc/elf.h (EXTRA_SPECS): Renamed to
12620 SUBTARGET_EXTRA_SPECS.
12621 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
12623 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12625 * config/arc/simdext.md (vst64_insn): Update pattern.
12626 (vld32wh_insn): Likewise.
12627 (vld32wl_insn): Likewise.
12628 (vld64_insn): Likewise.
12629 (vld32_insn): Likewise.
12631 2017-03-28 Marek Polacek <polacek@redhat.com>
12634 * fold-const.c (fold_comparison): Use protected_set_expr_location
12635 instead of SET_EXPR_LOCATION.
12637 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
12639 * tree.c (add_expr): Avoid name lookup warning.
12641 2017-03-27 Jeff Law <law@redhat.com>
12643 PR tree-optimization/80216
12644 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
12645 function name. Limit recursion depth.
12646 (record_temporary_equivalences): Corresponding changes.
12648 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
12650 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
12653 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12656 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
12658 * cfgcleanup.c (reg_note_cfa_p): New array.
12659 (insns_have_identical_cfa_notes): New function.
12660 (old_insns_match_p): Don't cross-jump in between /f
12661 and non-/f instructions. If both i1 and i2 are frame related,
12662 verify all CFA notes, their order and content.
12664 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
12667 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
12668 HImode and SImode with zero extend to DImode to one insn.
12669 (bswap<mode>2_extenddi): Likewise.
12670 (bswapsi2_extenddi): Likewise.
12671 (bswaphi2_extendsi): Likewise.
12672 (bswaphi2): Combine bswap HImode and SImode into one insn.
12673 Separate memory insns from swapping register.
12674 (bswapsi2): Likewise.
12675 (bswap<mode>2): Likewise.
12676 (bswaphi2_internal): Delete, no longer used.
12677 (bswapsi2_internal): Likewise.
12678 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
12679 store, and gpr<-gpr swap insns.
12680 (bswap<mode>2_store): Likewise.
12681 (bswaphi2_reg): Register only splitter, combine with the splitter.
12682 (bswaphi2 splitter): Likewise.
12683 (bswapsi2_reg): Likewise.
12684 (bswapsi2 splitter): Likewise.
12685 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
12686 the insns into load, store, and register/register insns.
12687 (bswapdi2_ldbrx): Likewise.
12688 (bswapdi2_load): Likewise.
12689 (bswapdi2_store): Likewise.
12690 (bswapdi2_reg): Likewise.
12692 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
12694 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
12695 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
12697 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12700 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
12702 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
12703 special handling for target option conflicts between dform
12704 options (-mpower9-dform, -mpower9-dform-vector,
12705 -mpower9-dform-scalar) and -mno-direct-move.
12707 2017-03-27 Richard Biener <rguenther@suse.de>
12709 PR tree-optimization/80181
12710 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
12712 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12714 * config/arc/predicates.md (move_double_src_operand): Replace the
12715 call to move_double_src_operand with a call to address_operand.
12717 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12719 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
12720 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
12721 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
12723 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12725 * config/arc/predicates.md (long_immediate_loadstore_operand):
12726 Consider scaled addresses cases.
12728 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12730 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
12731 restored when in interrupt.
12732 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
12733 doesn't have delay slot.
12735 2017-03-27 Richard Biener <rguenther@suse.de>
12738 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
12739 inlined thunk clones.
12741 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12744 * asan.c (instrument_derefs): Copy over last operand from
12745 original COMPONENT_REF to the new COMPONENT_REF with
12746 DECL_BIT_FIELD_REPRESENTATIVE.
12747 * ubsan.c (instrument_object_size): Likewise.
12749 2017-03-27 Richard Biener <rguenther@suse.de>
12751 PR tree-optimization/80170
12752 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
12753 sure DR/SCEV didnt fold in constants we do not see when looking
12754 at the reference base alignment.
12756 2017-03-27 Richard Biener <rguenther@suse.de>
12758 PR middle-end/80171
12759 * gimple-fold.c (fold_ctor_reference): Properly guard against
12760 NULL return value from canonicalize_constructor_val.
12762 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
12765 * config/i386/i386.c (ix86_expand_builtin)
12766 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
12767 flags reg setting and flags reg using instructions.
12768 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
12769 clobbering instructions to zero extend op2.
12771 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
12773 * doc/install.texi (Configuration) <--with-aix-soname>:
12774 Update link to AIX ld.
12776 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
12778 PR rtl-optimization/80160
12779 PR rtl-optimization/80159
12780 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
12781 reg_alternate_class into account.
12783 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
12786 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
12787 to consider in curr_insn_transform.
12789 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12791 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
12792 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
12793 and emit_mode_inner.
12795 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12797 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
12798 argument to the overloaded builtin variants. Use the new flag to
12799 deprecate certain builtin variants.
12800 * config/s390/s390-builtin-types.def: Add new builtin types.
12801 * config/s390/s390-builtins.h: Support new flags field for
12802 overloaded builtins.
12803 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
12804 (s390_macro_to_expand): Enable vector float data type.
12805 (s390_cpu_cpp_builtins_internal): Indicate support of the new
12806 builtins by incrementing the __VEC__ version number.
12807 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
12809 (s390_resolve_overloaded_builtin): Emit error messages depending
12810 on the builtin flags.
12811 * config/s390/s390.c (s390_expand_builtin): Support additional
12812 flags argument. Change error message to match the messages
12813 emitted in s390-c.c.
12814 * config/s390/s390.md: New UNSPEC_* constants.
12815 (op_type): Add new instruction types.
12816 * config/s390/vecintrin.h: Add new builtins and test data class
12818 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
12819 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
12820 (VEC_INEXACT, VEC_NOINEXACT): New constants.
12821 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
12822 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
12823 ("vec_mergel<mode>"): V_HW -> VEC_HW.
12825 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
12826 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
12827 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
12828 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
12830 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
12831 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
12832 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
12833 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
12835 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
12836 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
12837 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
12838 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
12839 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
12840 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
12841 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
12843 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
12844 ("vec_scatter_element<V_HW_4:mode>_DI")
12845 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
12846 ("vec_fpint<mode>", "vflls")
12847 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
12848 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
12849 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
12850 ("*vec_cmphe<mode>_cc"): ... these.
12852 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
12853 mode constant instead of magic value.
12855 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12857 * config/s390/s390.c (s390_expand_vec_compare): Support other
12858 vector floating point modes than just V2DF.
12859 (s390_expand_vcond): Likewise.
12860 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
12861 (s390_cannot_change_mode_class): Prevent mode changes between TF
12862 and V1TF in vector registers.
12863 * config/s390/s390.md (DF, SF): New mode attributes.
12864 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
12865 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
12866 SFmode support for VRs.
12867 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
12869 (VFT, VF_HW): New mode iterators.
12870 (vw, sdx): New mode attributes.
12871 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
12872 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
12873 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
12874 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
12875 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
12876 also the new vector floating point modes. Renaming to ...
12878 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
12879 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
12880 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
12881 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
12882 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
12883 ("vec_unordered<mode>"): ... these.
12885 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
12886 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
12887 ("*vec_extendv2df"): New insn definitions.
12889 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12891 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
12892 ("mulditi3_2", "*muldi3_sign"): New patterns.
12893 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
12894 rename the pattern definition.
12896 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12898 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
12900 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
12902 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12904 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
12905 instruction if possible.
12906 * config/s390/vector.md (vec_halfnumelts): New mode
12908 ("*vec_vllezlf<mode>"): New pattern.
12910 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12912 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
12913 ("popcountv4si2", "popcountv2di2"): Rename to ...
12914 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
12915 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
12917 ("popcount<mode>2_vxe"): New pattern.
12919 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12921 * common/config/s390/s390-common.c (processor_flags_table): Add
12923 * config.gcc: Add arch12.
12924 * config/s390/driver-native.c (s390_host_detect_local_cpu):
12925 Default to arch12 for unknown CPU model numbers.
12926 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
12927 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
12928 PROCESSOR_max sanity check.
12929 * config/s390/s390-opts.h (enum processor_type): Add
12931 * config/s390/s390.c (processor_table): Add arch12.
12932 (s390_expand_builtin): Add check for B_VXE flag.
12933 (s390_issue_rate): Add PROCESSOR_ARCH12.
12934 (s390_get_sched_attrmask): Likewise.
12935 (s390_get_unit_mask): Likewise.
12936 (s390_sched_score): Enable z13 scheduling for arch12.
12937 (s390_sched_reorder): Likewise.
12938 (s390_sched_variable_issue): Likewise.
12939 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
12941 (s390_tune_attr): Use z13 scheduling also for arch12.
12942 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
12943 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
12944 (TARGET_VXE_P): New macros.
12945 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
12946 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
12947 * config/s390/s390.opt: Add arch12 as processor_type.
12949 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12951 * config/s390/s390.md
12952 ("fixuns_truncdddi2", "fixuns_trunctddi2")
12953 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
12954 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
12956 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
12957 Rename expanders to ...
12959 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
12960 ("fixuns_truncdddi2_emu"): ... these.
12962 ("fixuns_trunc<mode>si2_emu"): New expander.
12964 ("*fixuns_truncdfdi2_z13"): Rename to ...
12965 ("*fixuns_truncdfdi2_vx"): ... this.
12967 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12969 * config/s390/2964.md: Remove the single element vector compare
12970 instructions which are no longer used.
12971 * config/s390/s390.c (s390_select_ccmode): Remove handling of
12973 (s390_canonicalize_comparison): Remove handling of DFmode
12975 (s390_expand_vec_compare_scalar): Remove function.
12976 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
12977 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
12979 ("*cmp<mode>_ccs"): Add wfcdb instruction.
12981 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12983 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
12985 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
12986 will anyway by matched by mov<mode>_64dfp.
12988 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12990 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
12991 vlef/vstef. Add missing operand to vleif.
12993 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12995 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
12996 pair for all vector types with 64 bit elements.
12997 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
12998 * config/s390/vector.md (V_HW_64): ... here.
12999 (V_128_NOSINGLE): New mode iterator.
13000 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
13001 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
13002 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
13003 ("*vec_load_pairv2di"): Change to ...
13004 ("*vec_load_pair<mode>"): ... this one.
13006 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13008 * config/s390/constraints.md: Add comments.
13009 (jKK): Reject element sizes > 8 bytes.
13010 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
13012 * config/s390/s390.md: Add the s_operand checks formerly in
13013 s390_split_ok_p to various splitters where they are still
13015 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
13016 for 128 bit vectors. Plus two splitters.
13018 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13020 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
13023 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13026 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
13027 error if the boundary argument is not constant.
13029 2017-03-24 Jakub Jelinek <jakub@redhat.com>
13031 PR rtl-optimization/80112
13032 * loop-doloop.c (doloop_condition_get): Don't check condition
13033 if cmp isn't SET with IF_THEN_ELSE src.
13035 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13037 PR tree-optimization/80158
13038 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
13039 replacing a candidate statement, also replace it for the
13040 candidate's alternate interpretation.
13041 (replace_rhs_if_not_dup): Likewise.
13042 (replace_one_candidate): Likewise.
13044 2017-03-24 Richard Biener <rguenther@suse.de>
13046 PR tree-optimization/80167
13047 * graphite-isl-ast-to-gimple.c
13048 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
13050 (translate_isl_ast_to_gimple::get_rename): Likewise.
13052 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
13054 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
13055 handling of certain combinations of target options, including the
13056 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
13057 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
13059 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13062 * config/arm/arm.md (*load_multiple): Add reload_completed to
13063 matching condition.
13065 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13066 Richard Biener <rguenth@suse.de>
13068 PR tree-optimization/79908
13069 PR tree-optimization/80136
13070 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
13071 been cast away, gimplify_and_add suffices.
13073 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
13075 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
13077 2017-03-23 Richard Biener <rguenther@suse.de>
13079 PR tree-optimization/80032
13080 * gimplify.c (gimple_push_cleanup): Forced unconditional
13081 cleanups still have to go to the conditional_cleanups
13084 2017-03-22 Jakub Jelinek <jakub@redhat.com>
13086 PR tree-optimization/80072
13087 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
13089 (next_operand_entry_id): Change type to unsigned int.
13090 (sort_by_operand_rank): Make sure to return the right return value
13091 even if unsigned fields are bigger than INT_MAX.
13092 (struct oecount): Change cnt and id type to unsigned int.
13093 (oecount_hasher::equal): Formatting fix.
13094 (oecount_cmp): Make sure to return the right return value
13095 even if unsigned fields are bigger than INT_MAX.
13096 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
13099 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
13100 TREE_READONLY on result if writing it more than once.
13103 * doc/invoke.texi (-fsanitize=thread): Document that with
13104 -fnon-call-exceptions atomics are not able to throw
13108 * tsan.c: Include tree-eh.h.
13109 (instrument_builtin_call): Call maybe_clean_eh_stmt or
13110 maybe_clean_or_replace_eh_stmt where needed.
13111 (instrument_memory_accesses): Add cfg_changed argument.
13112 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
13113 if it returned true.
13114 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
13116 PR rtl-optimization/63191
13117 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
13118 wrapper function, moved the whole old content into ...
13119 (ix86_delegitimize_address_1): ... this. New inline function.
13120 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
13121 true as last argument instead of ix86_delegitimize_address.
13123 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
13125 * config/aarch64/aarch64.c (generic_branch_cost): Copy
13126 cortexa57_branch_cost.
13128 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
13130 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
13132 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13135 * doc/md.texi (Constraints): Document wA constraint.
13136 * config/rs6000/constraints.md (wA): New.
13137 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
13138 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
13139 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
13140 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
13142 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
13145 * gimplify.c (is_oacc_declared): New function.
13146 (oacc_default_clause): Use it to set default flags for acc declared
13147 variables inside parallel regions.
13148 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
13149 declared variables.
13150 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
13151 declare attribute to any decl as necessary.
13153 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
13156 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
13157 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
13158 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
13159 (arm_arch_lpae): This.
13160 * config/arm/arm.c (arm_arch7ve): Rename into ...
13161 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
13162 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
13165 2017-03-22 Martin Liska <mliska@suse.cz>
13168 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
13169 error message instead of an ICE.
13171 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13173 * doc/extend.texi (6.11 Additional Floating Types): Revise.
13175 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
13177 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
13179 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
13182 2017-03-21 Martin Sebor <msebor@redhat.com>
13184 * doc/extend.texi: Use "cannot" instead of "can't."
13185 * doc/hostconfig.texi: Same.
13186 * doc/install.texi: Same.
13187 * doc/invoke.texi: Same.
13188 * doc/loop.texi: Same.
13189 * doc/md.texi: Same.
13190 * doc/objc.texi: Same.
13191 * doc/rtl.texi: Same.
13192 * doc/tm.texi: Same.
13193 * doc/tm.texi.in: Same.
13194 * doc/trouble.texi: Same.
13196 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
13199 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
13200 (collect_checksum_attributes): Set it.
13201 (die_checksum_ordered): Use it.
13203 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13205 PR tree-optimization/79908
13206 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
13207 change: For a VA_ARG whose LHS has been cast away, use
13208 force_gimple_operand to construct the side effects.
13210 2017-03-21 David Malcolm <dmalcolm@redhat.com>
13212 PR translation/80001
13213 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
13214 more amenable to translation.
13215 (oacc_loop_auto_partitions): Likewise.
13217 2017-03-21 Marek Polacek <polacek@redhat.com>
13218 Martin Sebor <msebor@redhat.com>
13220 PR tree-optimization/80109
13221 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
13222 on INTEGRAL_TYPE_P.
13224 2017-03-21 Jakub Jelinek <jakub@redhat.com>
13225 Segher Boessenkool <segher@kernel.crashing.org>
13228 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
13229 check reg_used_between_p between insn and one of succ or succ2
13230 depending on if succ is artificial insn not inserted into insn
13233 2017-03-21 Martin Liska <mliska@suse.cz>
13235 PR gcov-profile/80081
13236 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
13237 * doc/gcc.texi: Include gcov-dump stuff.
13238 * doc/gcov-dump.texi: New file.
13240 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
13242 PR rtl-optimization/79150
13243 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
13244 conditional jump, if the jump is the last insn of the loop.
13246 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13247 Richard Biener <rguenth@suse.de>
13249 PR tree-optimization/79908
13250 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
13251 been cast away, use force_gimple_operand to construct the side
13254 2017-03-21 Martin Liska <mliska@suse.cz>
13256 PR libfortran/79956
13257 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
13260 2017-03-21 Brad Spengler <spender@grsecurity.net>
13263 * plugin.c (htab_hash_plugin): New function.
13264 (add_new_plugin): Use it and adjust.
13265 (parse_plugin_arg_opt): Adjust.
13266 (init_one_plugin): Likewise.
13268 2017-03-21 Richard Biener <rguenther@suse.de>
13270 PR tree-optimization/80032
13271 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
13272 if set force the cleanup to happen unconditionally.
13273 (gimplify_target_expr): Push inserted clobbers with force_uncond
13274 to avoid them being removed by control-dependent DCE.
13276 2017-03-21 Richard Biener <rguenther@suse.de>
13278 PR tree-optimization/80122
13279 * tree-inline.c (copy_bb): Do not expans va-arg packs or
13280 va_arg_pack_len when the inlined call stmt requires pack
13282 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
13284 2017-03-21 Jakub Jelinek <jakub@redhat.com>
13287 * tsan.c (instrument_builtin_call): If the memory model argument
13288 is not a constant, assume it is valid.
13291 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
13294 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
13296 PR rtl-optimization/79910
13297 * combine.c (can_combine_p): Do not allow combining an I0 or I1
13298 if its dest is used by an insn before I2 (other than the combined
13299 insns themselves, which are properly handled already).
13301 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
13304 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
13306 * combine.c (record_used_regs): New static function.
13307 (try_combine): Handle situations where there is an additional
13308 instruction between I2 and I3 which needs to have a LOG_LINK
13312 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
13314 * combine.c (try_combine): Delete redundant i1 test. Call
13315 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
13317 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13320 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
13321 alternatives 13/14.
13323 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13325 PR tree-optimization/80054
13326 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
13327 the optimization if a PHI or any of its arguments is not dominated
13328 by the candidate's basis. Use gphi* rather than gimple* as
13330 (replace_profitable_candidates): Clean up a gimple* variable that
13331 should be a gphi* variable.
13333 2017-03-20 Martin Sebor <msebor@redhat.com>
13336 * doc/extend.texi (attribute constructor): Document present limitation.
13338 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
13341 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
13342 __POWER9_VECTOR__ #ifdef control, change template definition to
13343 use Power9-specific built-in function.
13344 (vec_any_eq): Likewise.
13345 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
13346 to control outcomes from this test.
13347 (vector_ae_<mode>p): For VEC_F modes, likewise.
13349 2017-03-20 Ian Lance Taylor <iant@google.com>
13351 * config/i386/i386.c (ix86_function_regparm): Save an extra
13352 register for -fsplit-stack with DECL_STATIC_CHAIN.
13354 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
13357 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
13358 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
13360 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
13362 * config/riscv/riscv.c (riscv_print_operand): Use "fence
13364 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
13367 2017-03-20 Marek Polacek <polacek@redhat.com>
13370 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
13372 2017-03-20 Richard Biener <rguenther@suse.de>
13374 PR tree-optimization/80113
13375 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
13376 allocate extra SSA name for PHI def.
13377 (add_close_phis_to_outer_loops): Likewise.
13378 (add_close_phis_to_merge_points): Likewise.
13379 (copy_loop_close_phi_args): Likewise.
13380 (copy_cond_phi_nodes): Likewise.
13382 2017-03-20 Martin Liska <mliska@suse.cz>
13384 PR middle-end/79753
13385 * tree-chkp.c (chkp_build_returned_bound): Do not build
13386 returned bounds for a LHS that's not a BOUNDED_P type.
13388 2017-03-20 Martin Liska <mliska@suse.cz>
13392 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
13393 COMPLEX_CST and VECTOR_CST.
13395 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13398 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
13399 target operand. A new splitter adds the clobber statement in case
13400 the target operand is dead anyway.
13402 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
13404 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
13405 to age-old versions of binutils and glibc.
13407 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
13409 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
13411 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
13413 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
13415 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
13417 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
13418 requirement for binutils 2.13.
13420 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
13422 * combine.c (try_combine): Delete redundant i1 test. Call
13423 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
13425 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
13427 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
13428 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
13430 <riscv64-*-elf>: Re-arrange section
13431 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
13432 <riscv32-*-linux>: Likewise.
13433 <riscv64-*-elf>: Likewise
13434 <riscv64-*-linux>: Likewise.
13436 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
13439 * aarch64.opt(verbose-cost-dump): Fix typo.
13441 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
13444 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
13445 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
13447 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
13449 * reload.c (find_reloads): When reloading a nonoffsettable address,
13450 use RELOAD_OTHER for it and its address reloads.
13452 PR rtl-optimization/79910
13453 * combine.c (record_used_regs): New static function.
13454 (try_combine): Handle situations where there is an additional
13455 instruction between I2 and I3 which needs to have a LOG_LINK
13458 2017-03-17 Jeff Law <law@redhat.com>
13460 PR tree-optimization/71437
13461 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
13462 conditional in the hash table first.
13463 (vrp_dom_walker::before_dom_children): Extract condition from
13464 ASSERT_EXPR. Record condition, its inverion and any implied
13465 conditions as well.
13467 2017-03-17 Marek Polacek <polacek@redhat.com>
13468 Markus Trippelsdorf <markus@trippelsdorf.de>
13470 PR tree-optimization/80079
13471 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
13474 2017-03-17 Richard Biener <rguenther@suse.de>
13476 PR middle-end/80075
13477 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
13478 Properly verify the LHS before the RHS possibly claims to be
13480 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
13483 2017-03-17 Martin Jambor <mjambor@suse.cz>
13485 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
13486 (List of -O2 options): Likewise.
13487 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
13490 2017-03-17 Tom de Vries <tom@codesourcery.com>
13492 * gcov-dump.c (print_usage): Print bug_report_url.
13494 2017-03-17 Richard Biener <rguenther@suse.de>
13496 PR middle-end/80050
13497 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
13498 (parser::peek): Likewise.
13500 2017-03-17 Richard Biener <rguenther@suse.de>
13502 PR tree-optimization/80048
13503 * sese.c (free_sese_info): Properly release rename_map and
13504 copied_bb_map elements.
13506 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
13508 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
13509 Add linked-list forward and backlinks. Insert on
13510 construction, remove on destruction.
13511 (class pass_store_merging): Add m_stores_head field.
13512 (pass_store_merging::terminate_and_process_all_chains):
13513 Iterate over m_stores_head list.
13514 (pass_store_merging::terminate_all_aliasing_chains):
13516 (pass_store_merging::execute): Check for debug stmts first.
13517 Push new chains onto the m_stores_head stack.
13519 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
13522 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
13523 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
13524 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
13526 2017-03-16 Jeff Law <law@redhat.com>
13528 PR tree-optimization/71437
13529 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
13530 member function. Implementation moved into after_dom_children
13531 member function and into the threader's thread_outgoing_edges
13533 (dom_opt_dom_walker::after_dom_children): Simplify by moving
13534 some code into new thread_outgoing_edges.
13535 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
13536 definition. Simplify marker handling (do it here). Assume we always
13537 have the available expression and the const/copies tables.
13538 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
13540 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
13541 * tree-vrp.c (equiv_stack): No longer file scoped.
13542 (vrp_dom_walker): New class.
13543 (vrp_dom_walker::before_dom_children): New member function.
13544 (vrp_dom_walker::after_dom_children): Likewise.
13545 (identify_jump_threads): Setup domwalker. Use it rather than
13546 walking edges in a random order by hand. Simplify setup/finalization.
13547 (finalize_jump_threads): Remove.
13548 (vrp_finalize): Do not call identify_jump_threads here.
13549 (execute_vrp): Do it here instead and call thread_through_all_blocks
13552 PR tree-optimization/71437
13553 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
13555 (simplify_stmt_for_jump_threading): Add basic_block argument. All
13557 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
13558 (dom_opt_dom_walker::thread_across_edge): Remove
13559 handle_dominating_asserts argument. All callers changed.
13560 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
13561 changes. Remove calls to lhs_of_dominating_assert. Other
13562 uses of handle_dominating_asserts turn into unconditional code
13563 (simplify_control_stmt_condition_1): Likewise.
13564 (simplify_control_stmt_condition): Likewise.
13565 (thread_through_normal_block, thread_across_edge): Likewise.
13566 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
13567 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
13568 object if it is not an SSA_NAME.
13569 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
13570 before calling into the VRP specific simplifiers.
13571 (identify_jump_threads): Remove handle_dominating_asserts
13574 2017-03-16 Jakub Jelinek <jakub@redhat.com>
13577 * tree-diagnostic.c (default_tree_printer): No longer static.
13578 * tree-diagnostic.h (default_tree_printer): New prototype.
13580 2017-03-16 Tamar Christina <tamar.christina@arm.com>
13582 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
13583 Change ins into fmov.
13585 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13587 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
13588 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
13589 Use h_con constraint for operand 1.
13590 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
13591 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
13593 2017-03-15 Jeff Law <law@redhat.com>
13595 PR tree-optimization/71437
13596 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
13597 (record_temporary_equivalences): Use it.
13599 PR tree-optimization/71437
13600 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
13601 tree-ssa-scopedtables.
13602 (lookup_avail_expr, build_and_record_new_cond): Likewise.
13603 (record_conditions, record_cond, vuse_eq): Likewise.
13604 (record_edge_info): Adjust to API tweak of record_conditions.
13605 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
13606 (record_temporary_equivalences, optimize_stmt): Likewise.
13607 (eliminate_redundant_computations): Likewise.
13608 (record_equivalences_from_stmt): Likewise.
13609 * tree-ssa-scopedtables.c: Include options.h and params.h.
13610 (vuse_eq): New function, moved from tree-ssa-dom.c
13611 (build_and_record_new_cond): Likewise.
13612 (record_conditions): Likewise. Accept vector of conditions rather
13613 than edge_equivalence structure for first argument.
13614 for the first argument.
13615 (avail_exprs_stack::lookup_avail_expr): New member function, moved
13616 from tree-ssa-dom.c.
13617 (avail_exprs_stack::record_cond): Likewise.
13618 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
13619 from tree-ssa-dom.c.
13620 (avail_exprs_stack): Add new member functions lookup_avail_expr
13622 (record_conditions): Declare.
13624 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
13627 * lra-constraints.c (process_alt_operands): Increase reject for
13628 reloading an input/output operand.
13630 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
13633 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
13634 insns to convert from signed/unsigned char/short to IEEE 128-bit
13636 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
13638 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
13641 * config/i386/i386.c (ix86_vector_duplicate_value): Create
13642 subreg of inner mode for values already in registers.
13644 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
13646 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
13647 iteration reg is used after the loop.
13649 2017-03-14 Martin Sebor <msebor@redhat.com>
13651 PR tree-optimization/79800
13652 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
13653 precision in negative-positive range.
13654 (format_floating): Call non-const overload with adjusted precision.
13656 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
13659 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
13662 2017-03-14 Martin Sebor <msebor@redhat.com>
13664 PR middle-end/80020
13665 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
13666 * builtins.def (aligned_alloc): Use it.
13669 * Makefile.in (GTFILES): Add calls.c.
13670 * calls.c: Include "gt-calls.h".
13672 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
13674 PR rtl-optimization/79728
13675 * regs.h (struct target_regs): New field
13676 x_contains_allocatable_regs_of_mode.
13677 (contains_allocatable_regs_of_mode): New macro.
13678 * reginfo.c (init_reg_sets_1): Initialize it, and change
13679 contains_reg_of_mode so it includes global regs as well.
13680 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
13681 rather than contains_regs_of_mode.
13683 2017-03-14 Martin Liska <mliska@suse.cz>
13685 * doc/invoke.texi: Document options that can't be combined with
13686 -fcheck-pointer-bounds.
13688 2017-03-14 Martin Liska <mliska@suse.cz>
13690 PR middle-end/79831
13691 * doc/invoke.texi (-Wchkp): Document the option.
13693 2017-03-14 Martin Liska <mliska@suse.cz>
13695 * Makefile.in: Install gcov-dump.
13697 2017-03-14 Martin Liska <mliska@suse.cz>
13699 * multiple_target.c (expand_target_clones): Bail out for
13700 an invalid attribute.
13702 2017-03-14 Richard Biener <rguenther@suse.de>
13704 * alias.c (struct alias_set_entry): Pack properly.
13705 * cfgloop.h (struct loop): Likewise.
13706 * cse.c (struct set): Likewise.
13707 * ipa-utils.c (struct searchc_env): Likewise.
13708 * loop-invariant.c (struct invariant): Likewise.
13709 * lra-remat.c (struct cand): Likewise.
13710 * recog.c (struct change_t): Likewise.
13711 * rtl.h (struct address_info): Likewise.
13712 * symbol-summary.h (function_summary): Likewise.
13713 * tree-loop-distribution.c (struct partition): Likewise.
13714 * tree-object-size.c (struct object_size_info): Likewise.
13715 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
13716 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
13717 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
13718 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
13719 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
13720 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
13721 (struct _stmt_vec_info): Likewise.
13723 2017-03-14 Martin Liska <mliska@suse.cz>
13726 * multiple_target.c (create_dispatcher_calls): Check that
13727 a target can create a function dispatcher.
13729 2017-03-14 Martin Liska <mliska@suse.cz>
13732 * multiple_target.c (expand_target_clones): Drop local.local
13733 flag for default implementation.
13735 2017-03-14 Richard Biener <rguenther@suse.de>
13737 PR tree-optimization/80030
13738 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
13740 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
13742 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
13743 gcc_fallthrough() instead of __attribute__((fallthrough));
13745 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13747 * doc/gcc.texi: Remove "up" link to (DIR).
13748 * doc/gccint.texi: Ditto.
13750 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13752 * doc/install.texi (Specific) <avr>: Remove reference to
13755 2017-03-13 Jeff Law <law@redhat.com>
13757 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
13758 attribute rather than comments.
13760 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
13761 match_scratch operand is highest.
13763 2017-03-13 Martin Liska <mliska@suse.cz>
13765 PR middle-end/78339
13766 * ipa-pure-const.c (warn_function_noreturn): If the declarations
13767 is a CHKP clone, use original declaration.
13769 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13771 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
13772 (arc_conditional_register_usage): Use a different allocation order
13773 when optimizing for size.
13774 * common/config/arc/arc-common.c (arc_option_optimization_table):
13775 Section anchors default on when optimizing for size.
13777 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13779 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
13781 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13783 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
13784 * config/arc/arc.md (cpu_facility): Add cd variant.
13785 (*movqi_insn): Add code density variant.
13786 (*movhi_insn): Likewise.
13787 (*movqi_insn): Likewise.
13788 (*addsi3_mixed): Likewise.
13789 (subsi3_insn): Likewise.
13791 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13793 * config/arc/arc.md (movsi_cond_exec): Update constraint.
13795 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13797 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
13798 expressions with MINUS and UNARY ops.
13800 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13803 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
13805 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
13806 between vec_select and vector argument.
13807 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
13808 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
13809 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
13810 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
13811 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
13812 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
13814 2017-03-13 Richard Biener <rguenther@suse.de>
13817 * params.def (vect-max-peeling-for-alignment): Fix typo.
13819 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13821 * doc/install.texi (Specific) <mips-*-*>: Remove description of
13822 issue that only occurred with binutils below 2.18.
13824 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13826 * doc/install.texi (Specific) <cris-axis-elf>: No longer
13827 refer to binutils 2.11/2.12 minimum.
13829 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13831 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
13832 ftp.kernel.org and simplify binutils requirement.
13834 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
13836 * doc/invoke.texi (Warning Options): Fix spelling of link-time
13838 (Optimize Options): Ditto. Also remove redundancy.
13840 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13842 PR translation/79848
13843 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
13845 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
13846 to G_ to avoid double translation.
13848 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13850 PR translation/79923
13851 * auto-profile.c (get_combined_location): Convert leading
13852 character of diagnostics to lower case and remove trailing period.
13853 (read_profile): Likewise for various diagnostics.
13854 * config/arm/arm.c (arm_option_override): Remove trailing period
13855 from various diagnostics.
13856 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
13857 (msp430_expand_delay_cycles): Likewise.
13859 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13862 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
13863 full command-line argument, rather than just "str".
13864 (aarch64_validate_march): Likewise.
13865 (aarch64_validate_mtune): Likewise.
13867 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
13869 PR rtl-optimization/78911
13870 * lra-assigns.c (must_not_spill_p): New function.
13871 (spill_for): Use it.
13873 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13875 PR tree-optimization/79981
13876 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
13877 ATOMIC_COMPARE_EXCHANGE ifn result.
13878 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
13879 IFN_ATOMIC_COMPARE_EXCHANGE.
13881 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13884 * opts.c (parse_sanitizer_options): Add missing question mark to
13885 "did you mean" message.
13887 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13889 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
13891 (VMULEUH_UNS): Likewise.
13892 (VMULOUB_UNS): Likewise.
13893 (VMULOUH_UNS): Likewise.
13894 * config/rs6000/rs6000.c (builtin_function_type): Remove
13895 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
13897 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13900 * read-rtl-function.c (function_reader::read_rtx_operand): Update
13901 x with result of extra_parsing_for_operand_code_0.
13902 (function_reader::extra_parsing_for_operand_code_0): Convert
13903 return type from void to rtx, returning x. When reading
13904 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
13905 larger size containing struct block_symbol.
13907 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
13909 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
13910 -mfloat128-hardware without -m64.
13912 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
13915 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
13916 entries to the case statement that marks unsigned arguments to
13917 overloaded functions.
13919 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
13921 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
13922 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
13924 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
13927 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
13928 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
13930 2017-03-10 Martin Liska <mliska@suse.cz>
13934 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
13935 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
13938 2017-03-10 Olivier Hainque <hainque@adacore.com>
13940 * tree-switch-conversion (array_value_type): Start by resetting
13941 candidate type to it's main variant.
13943 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13945 PR rtl-optimization/79909
13946 * combine.c (try_combine): Use simplify_replace_rtx on individual
13947 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
13948 of the whole CALL_INSN_FUNCTION_USAGE.
13950 PR tree-optimization/79972
13951 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
13952 get_range_info on SSA_NAMEs. Formatting fixes.
13954 2017-03-10 Richard Biener <rguenther@suse.de>
13955 Jakub Jelinek <jakub@redhat.com>
13957 PR tree-optimization/77975
13958 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
13959 edge to be constant.
13960 (get_val_for): For constant x return it. Formatting fix.
13961 (loop_niter_by_eval): Avoid pointless looping if the next iteration
13962 would use the same bases as the current one.
13964 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13966 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
13967 instead of vec_select for V1TImode.
13968 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
13970 (VSX_LE_128): Add V1TI to this mode iterator.
13971 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
13972 (*vsx_le_perm_store_<mode>): Likewise.
13973 (pre-reload splitter for VSX stores): Likewise.
13974 (post-reload splitter for VSX stores): Likewise.
13975 (*vsx_xxpermdi2_le_<mode>): Likewise.
13976 (*vsx_lxvd2x2_le_<mode>): Likewise.
13977 (*vsx_stxvd2x2_le_<mode>): Likewise.
13979 2017-03-09 Michael Eager <eager@eagercon.com>
13981 Correct failures with --enable-checking=yes,rtl.
13983 * config/microblaze/microblaze.c (microblaze_expand_shift):
13984 Replace GET_CODE test with CONST_INT_P and INTVAL test with
13985 test for const0_rtx.
13986 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
13987 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
13989 2017-03-09 Richard Biener <rguenther@suse.de>
13991 PR tree-optimization/79977
13992 * graphite-scop-detection.c (scop_detection::merge_sese):
13993 Handle the case of extra exits to blocks dominating the entry.
13995 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
13997 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
14000 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
14002 PR rtl-optimization/79949
14003 * lra-constraints.c (process_alt_operands): Check memory when
14004 trying to predict a cycle. Print about the overall increase.
14006 2017-03-09 Richard Biener <rguenther@suse.de>
14008 PR middle-end/79971
14009 * gimple-expr.c (useless_type_conversion_p): Preserve
14010 TYPE_SATURATING for fixed-point types.
14012 2017-03-09 Richard Biener <rguenther@suse.de>
14015 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
14016 alignment of BLKmode params.
14018 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14021 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
14022 (VALL_NO_V2Q): Likewise.
14024 * config/aarch64/aarch64-simd.md
14025 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
14027 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
14028 VALL_NO_V2Q mode iterator.
14029 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
14031 2017-03-09 Martin Liska <mliska@suse.cz>
14033 PR tree-optimization/79631
14034 * tree-chkp-opt.c (chkp_is_constant_addr): Call
14035 tree_int_cst_sign_bit just for INTEGER constants.
14037 2017-03-09 Martin Liska <mliska@suse.cz>
14041 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
14044 2017-03-09 Marek Polacek <polacek@redhat.com>
14047 * tree.c (inchash::add_expr): Handle TREE_VEC.
14049 2017-03-09 Martin Liska <mliska@suse.cz>
14052 (chkp_narrow_size_and_offset): New function.
14053 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
14054 (void chkp_parse_bit_field_ref): New function.
14055 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
14056 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
14058 2017-03-09 Martin Liska <mliska@suse.cz>
14061 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
14062 (chkp_find_bounds_1): Remove gcc_unreachable.
14064 2017-03-09 Jakub Jelinek <jakub@redhat.com>
14067 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
14068 BUILT_IN_SYNC*, determine the access type from the size suffix and
14069 always build a MEM_REF with that type. Handle forgotten
14070 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
14073 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
14074 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
14075 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
14076 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
14077 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
14078 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
14079 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
14080 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
14081 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
14082 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
14083 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
14084 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
14085 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
14086 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
14087 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
14088 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
14089 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
14090 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
14091 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
14092 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
14093 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
14094 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
14095 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
14096 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
14097 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
14098 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
14099 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
14100 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
14101 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
14102 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
14103 definitions outside of __OPTIMIZE__ guarded section.
14106 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
14107 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
14108 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
14109 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
14112 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14114 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
14115 ("vfenez<mode>"): Add missing constraints.
14117 2017-03-08 Martin Sebor <msebor@redhat.com>
14120 * config/nds32/nds32.c (nds32_option_override):
14121 Fix misspelled diagnostic.
14123 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14126 * gimplify.c (gimplify_omp_for): Replace index var in outer
14127 taskloop statement with an artificial variable and add
14128 OMP_CLAUSE_PRIVATE clause for it.
14130 2017-03-08 Richard Biener <rguenther@suse.de>
14132 PR tree-optimization/79955
14133 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
14134 for accesses that are completely outside of the variable.
14136 2017-03-08 Andrew Haley <aph@redhat.com>
14138 PR tree-optimization/79943
14139 * tree-ssa-loop-split.c (compute_new_first_bound): When
14140 calculating the new upper bound, (END-BEG) should be added, not
14143 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14145 * config/avr/avr.md (setmemhi): Make sure match_dup
14146 operand number comes before match_scratch.
14148 2017-03-08 Richard Biener <rguenther@suse.de>
14150 PR tree-optimization/79920
14151 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
14152 with ncopies == 1 to ...
14153 (vect_transform_slp_perm_load): ... here. Properly compute
14154 all element loads by iterating VF times over the group. Do
14155 not handle ncopies (computed in a broken way) in
14156 vect_create_mask_and_perm.
14158 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14161 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
14162 is a uniform vector, use uniform_vector_p return value instead of
14163 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
14165 2017-03-07 Marek Polacek <polacek@redhat.com>
14167 PR middle-end/79809
14168 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
14169 (alloca_call_type): Likewise.
14171 2017-03-07 Martin Liska <mliska@suse.cz>
14173 * gcov.c (process_args): Put comment to correct location.
14175 2017-03-07 Martin Liska <mliska@suse.cz>
14177 PR middle-end/68270
14178 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
14179 Use array_at_struct_end_p instead of DECL_CHAIN (field).
14180 (chkp_narrow_bounds_for_field): Likewise.
14181 (chkp_parse_array_and_component_ref): Pass one more argument to
14184 2017-03-07 Richard Biener <rguenther@suse.de>
14186 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
14189 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
14191 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
14192 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
14194 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14197 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
14198 to end of description.
14199 (PARAM_MAX_STORES_TO_MERGE): Likewise.
14201 2017-03-07 Jakub Jelinek <jakub@redhat.com>
14203 PR rtl-optimization/79901
14204 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
14206 (*avx512f_<code><mode>3<mask_name>): ... this.
14207 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
14208 iterator instead of VI8_AVX2_AVX512BW.
14210 PR rtl-optimization/79901
14211 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
14212 min/max expander, expand it using expand_vec_cond_expr.
14215 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
14218 2017-03-06 Jakub Jelinek <jakub@redhat.com>
14221 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
14222 to void * for PCH reasons.
14223 * dwarf2out.c (output_loc_operands, output_die): Cast
14224 v.val_vec.array to unsigned char *.
14226 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
14229 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
14232 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
14234 PR rtl-optimization/79571
14235 * lra-constraints.c (process_alt_operands): Calculate static
14236 reject and subtract it from overall when only addresses will be
14239 2017-03-06 Julia Koval <julia.koval@intel.com>
14242 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
14243 incoming stack boundary to 128 for 64-bit targets.
14245 2017-03-06 Richard Biener <rguenther@suse.de>
14247 PR tree-optimization/79894
14248 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
14249 to NULL after folding it.
14251 2017-03-06 Richard Biener <rguenther@suse.de>
14253 PR tree-optimization/79824
14254 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
14255 check disabling peeling for gaps.
14257 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
14259 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
14260 attributes): Document gettimeofday.
14262 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
14264 * config/s390/s390.c (s390_option_override_internal): Set
14265 PARAM_MIN_VECT_LOOP_BOUND
14267 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
14269 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
14270 * config/s390/s390.md: Likewise.
14272 2017-03-06 Jakub Jelinek <jakub@redhat.com>
14275 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
14276 (<avx2_avx512>_perm<mode>): Rename to ...
14277 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
14279 (<avx512>_perm<mode>_mask): Rename to ...
14280 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
14282 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
14283 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
14284 instead of VI8F_256_512.
14285 (avx512f_perm<mode>): New define_expand.
14286 (avx512f_perm<mode>_mask): Likewise.
14287 (avx512f_perm<mode>_1<mask_name>): New define_insn.
14288 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
14290 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14292 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
14293 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
14295 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
14297 2017-03-06 Martin Liska <mliska@suse.cz>
14300 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
14301 when having a SSA NAME w/o VAR_DECL assigned to it.
14303 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14305 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
14306 msa_dpsub_<su>_d): Fix MODE for vec_select.
14308 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14310 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
14312 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
14314 2017-03-06 Richard Biener <rguenther@suse.de>
14316 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
14317 * plugin.c (register_plugin_info): Likewise.
14318 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
14320 2017-03-05 Jakub Jelinek <jakub@redhat.com>
14322 * config/i386/sse.md (sse_storehps, sse_storelps,
14323 avx_<castmode><avxsizesuffix>_<castmode>,
14324 avx512f_<castmode><avxsizesuffix>_<castmode>,
14325 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
14326 in condition that at least one operand is not a MEM.
14328 2017-03-03 Jakub Jelinek <jakub@redhat.com>
14330 PR middle-end/79805
14331 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
14332 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
14334 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
14335 gimple_call_nothrow_p flag based on whether original builtin can throw.
14336 If it can, emit following stmts on the fallthrough edge.
14337 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
14338 don't create new bb if inserting just debug stmts on the edge, try to
14339 insert them on the fallthru bb or just reset debug stmts.
14341 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
14344 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
14345 restore recog_data (including the operand rtxes inside it) around
14346 the call to get_insn_template.
14348 2017-03-03 Martin Sebor <msebor@redhat.com>
14350 PR tree-optimization/79699
14351 * context.c (context::~context): Free MPFR caches to avoid
14352 a memory leak on program exit.
14354 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14356 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
14357 Use wide_int::ulow () instead of .elt (0).
14359 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
14361 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
14362 (*pushxf): Limit oF constraint to 32bit targets and add oC
14363 constraint for 64bit targets.
14364 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
14365 (*pushdf): Change rmF constraint to rmC.
14367 2017-03-03 Martin Liska <mliska@suse.cz>
14369 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
14370 Remove unused variable.
14372 2017-03-03 Jakub Jelinek <jakub@redhat.com>
14375 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
14376 is a memory operand, increase num_memory.
14377 (ix86_expand_args_builtin): Likewise.
14379 2017-03-03 Jan Hubicka <jh@suse.cz>
14382 * ipa-devirt.c (maybe_record_node): Properly handle
14383 __cxa_pure_virtual visibility.
14385 2017-03-03 Martin Liska <mliska@suse.cz>
14387 PR tree-optimization/79803
14388 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
14390 (pass_loop_prefetch::execute): Disabled optimization if an
14391 assumption about L1 cache size is not met.
14393 2017-03-03 Martin Liska <mliska@suse.cz>
14395 PR rtl-optimization/79574
14396 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
14397 (hash_scan_set): Likewise.
14398 (dump_hash_table): Likewise.
14399 (hoist_code): Likewise.
14401 2017-03-03 Richard Biener <rguenther@suse.de>
14403 * fixed-value.c (fixed_from_string): Restore use of elt (1)
14404 in place of uhigh ().
14405 (fixed_convert_from_real): Likewise.
14407 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
14410 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
14412 2017-03-03 Richard Biener <rguenther@suse.de>
14414 PR middle-end/79818
14415 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
14416 TYPE_OVERFLOW_UNDEFINED check.
14418 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14420 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
14422 (vector_ae_<mode>_p): Likewise.
14423 (vector_nez_<mode>_p): Likewise.
14424 (vector_ne_v2di_p): Likewise.
14425 (vector_ae_v2di_p): Likewise.
14426 (vector_ne_<mode>_p): Likewise.
14427 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
14429 (vsx_tsqrt<mode>2_fe): Likewise.
14431 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
14434 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
14436 2017-03-02 Jakub Jelinek <jakub@redhat.com>
14438 PR rtl-optimization/79780
14439 * cprop.c (one_cprop_pass): When second and further conditional trap
14440 in a single basic block is turned into an unconditional trap, turn it
14441 into a deleted note to avoid RTL verification failures.
14443 2017-03-02 Richard Biener <rguenther@suse.de>
14445 * fold-const.c (const_binop): Use ulow () instead of elt (0).
14447 2017-03-02 Richard Biener <rguenther@suse.de>
14449 PR tree-optimization/79345
14451 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
14452 param and abort the walk, returning -1 if it is hit.
14453 (walk_aliased_vdefs): Take a limit param and pass it on.
14454 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
14455 defaulting to 0 and return a signed int.
14456 * tree-ssa-uninit.c (struct check_defs_data): New struct.
14457 (check_defs): New helper.
14458 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
14459 about uninitialized memory.
14460 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
14461 bogus uninitialized warning.
14462 (fixed_convert_from_real): Likewise.
14464 2017-03-02 Bin Cheng <bin.cheng@arm.com>
14466 PR tree-optimization/66768
14467 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
14468 iv_use if base object can't be determined.
14470 2017-03-02 Jakub Jelinek <jakub@redhat.com>
14472 PR tree-optimization/79345
14473 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
14474 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
14475 (get_pattern_stats): Initialize it.
14476 * genemit.c (gen_expand): Verify match_scratch numbers come after
14477 match_operand/match_dup numbers.
14478 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
14479 match_scratch numbers.
14480 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
14482 * config/s390/s390.md (trunctdsd2): Likewise.
14484 2017-03-02 Richard Biener <rguenther@suse.de>
14486 * wide-int.h (wide_int_storage::operator=): Implement in terms
14489 2017-03-02 Richard Biener <rguenther@suse.de>
14491 PR tree-optimization/79777
14492 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
14493 the to insert expression to sth existing.
14495 2017-03-01 Martin Sebor <msebor@redhat.com>
14497 PR middle-end/79692
14498 * gimple-ssa-sprintf.c
14499 (directive::known_width_and_precision): New function.
14500 (format_integer): Use it.
14501 (get_mpfr_format_length): Consider the full range of precision
14502 when computing %g output with the # flag. Set the likely byte
14503 count to 3 rather than 1 when precision is indeterminate.
14504 (format_floating): Correct the lower bound of precision.
14506 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14508 * doc/invoke.texi: Document default code model for 64-bit Linux.
14510 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14513 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
14514 udiv rather than div since input pattern is unsigned.
14516 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
14518 * config/i386/i386.c (print_reg): Warn for values of
14519 unsupported size in integer register.
14521 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
14524 * config/rs6000/predicates.md (current_file_function_operand): Do
14525 not allow self calls to be local if the function is replaceable.
14527 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
14530 * config/rs6000/altivec.h (vec_ctz and others): Change the
14531 preprocessor macro that controls conditional compilation from
14532 _ARCH_PWR9 to __POWER9_VECTOR__.
14533 (vec_all_ne): Change parameterization of __altivec_scalar_pred
14534 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
14535 control (instead of _ARCH_PWR9 control) so that template
14536 definition uses power9-specific function.
14537 (vec_any_eq): Likewise.
14538 (vec_all_ne): Change macro definition to use a power9-specific
14539 expansion under #ifdef __POWER9_VECTOR__ control (instead of
14540 _ARCH_PWR9 control).
14541 (vec_any_eq) Likewise.
14542 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
14543 expansion for CMPNEF to remove support for xvcmpnesp instruction.
14544 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
14545 support for xvcmpnedp instruction.
14546 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
14547 macro expansion so that Power9 implementation of vec_all_ne does
14548 not use the AltiVec predicate framework.
14549 (VCMPNEH_P): Likewise.
14550 (VCMPNEW_P): Likewise.
14551 (VCMPNED_P): Likewise.
14552 (VCMPNEFP_P): Likewise.
14553 (VCMPNEDP_P): Likewise.
14554 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
14555 implementation of vec_any_eq to not use AltiVec predicate
14557 (VCMPAEH_P): Likewise.
14558 (VCMPAEW_P): Likewise.
14559 (VCMPAED_P): Likewise.
14560 (VCMPAEFP_P): Likewise.
14561 (VCMPAEDP_P): Likewise.
14562 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
14563 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
14564 not use the AltiVec predicate framework.
14565 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
14566 of vec_any_eq to not use AltiVec predicate framework.
14567 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
14568 support for predefined __POWER9_VECTOR__ macro to indicate that
14569 Power9 instruction selection is enabled.
14570 (altivec_overloaded_builtins): Remove extraneous
14571 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
14572 function argument types RS6000_BTI_bool_V16QI and
14573 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
14574 entry for overloaded function argument types RS6000_BTI_bool_V4SI
14575 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
14576 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
14577 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
14578 Power9 for implementations of vec_cmpne. Change the signature for
14579 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
14580 (representing vec_all_ne) to remove the previously described first
14581 argument of type RS6000_BTI_INTSI, as this was an artifact of
14582 reliance on the AltiVec predicate framework, which is no longer
14583 used in the implementation of these functions. Add
14584 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
14585 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
14586 since, unlike the AltiVec predicate framework implementation, we
14587 do not share function descriptors between vec_alle and vec_anyeq.
14588 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
14589 set of modes that receive special treatment even when
14590 TARGET_P9_VECTOR is true. The special treatment emits code that
14591 does not depend on Power9 instructions.
14592 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
14593 define_expand to not rely on AltiVec predicate framework.
14594 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
14596 (vector_ne_v2di_p): Change this define_expand to not rely on
14597 AltiVec predicate framework.
14598 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
14600 (vector_ne_<mode>_p): Change this define_expand to not rely on
14601 AltiVec predicate framework.
14602 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
14604 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
14605 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
14606 define_insn pattern.
14607 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
14608 define_insn pattern because the xvcmpne<VSs>. instruction is not
14610 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
14611 instruction is not supported.
14613 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14615 * config/nvptx/nvptx.c: Include intl.h.
14617 2017-03-01 Martin Jambor <mjambor@suse.cz>
14620 * ipa-prop.h (ipa_bits): Removed field known.
14621 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
14622 to pointers. Adjusted their comments to warn about their sharing.
14623 (ipcp_transformation_summary): Change bits to a vector of pointers.
14624 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
14625 (ipa_get_ipa_bits_for_value): Declare.
14626 * tree-vrp.h (value_range): Mark as GTY((for_user)).
14627 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
14628 (ipa_bits_hash_table): Likewise.
14629 (ipa_vr_ggc_hash_traits): Likewise.
14630 (ipa_vr_hash_table): Likewise.
14631 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
14632 being pointers and vr_known being removed.
14633 (ipa_set_jf_unknown): Likewise.
14634 (ipa_get_ipa_bits_for_value): New function.
14635 (ipa_set_jfunc_bits): Likewise.
14636 (ipa_get_value_range): New overloaded functions.
14637 (ipa_set_jfunc_vr): Likewise.
14638 (ipa_compute_jump_functions_for_edge): Use the above functions to
14639 construct bits and vr parts of jump functions.
14640 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
14641 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14643 (ipcp_grow_transformations_if_necessary): Also allocate
14644 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14646 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
14647 them. Fix too long lines.
14648 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
14649 vr_known being removed.
14650 (ipa_read_jump_function): Use new setter functions to construct bits
14651 and vr parts of jump functions or set them to NULL.
14652 (write_ipcp_transformation_info): Adjust for bits being pointers.
14653 (read_ipcp_transformation_info): Likewise.
14654 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
14656 Include gt-ipa-prop.h.
14657 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
14659 (ipcp_store_bits_results): Likewise.
14660 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
14661 Do not write to existing jump functions but use a temporary instead.
14663 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14666 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
14667 attempt to use its first operand as BIT_FIELD_REF base.
14669 2017-03-01 Richard Biener <rguenther@suse.de>
14671 PR middle-end/79721
14672 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
14673 interpolating formula in wrapping arithmetic.
14674 (chrec_apply): Convert chrec_evaluate return value to wanted type.
14676 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14678 PR tree-optimization/79734
14679 * tree-vect-generic.c (expand_vector_condition): Optimize
14680 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
14681 Handle VEC_COND_EXPR where comparison has different inner width from
14682 type's inner width.
14684 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
14686 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
14687 markup, and similar issues. Remove @opindex entries for things
14688 that aren't options. Add missing -mmpy-option entries.
14690 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14692 PR tree-optimization/79737
14693 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
14694 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
14695 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
14696 instead of byte_size. Formatting fix.
14697 (shift_bytes_in_array_right): Formatting fix.
14699 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
14702 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
14703 condition on optimize for the leaf function test.
14705 2017-02-28 Martin Liska <mliska@suse.cz>
14708 * read-rtl-function.c (function_reader::handle_unknown_directive):
14709 Bail out when one uses -flto.
14711 2017-02-28 Martin Liska <mliska@suse.cz>
14713 * common.opt: Replace space with tabular for options of <number>
14715 * config/i386/i386.opt: Show <number> value for
14716 -mlarge-data-threshold.
14717 * opts.c (print_filtered_help): Do not display number in hexadecimal
14720 2017-02-28 Martin Liska <mliska@suse.cz>
14722 * common.opt: Fix --help=option -Q for options which are of
14725 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
14727 * config/i386/i386.c (print_reg): Error out for values
14728 of 8-bit size in invalid integer register.
14730 2017-02-28 Martin Sebor <msebor@redhat.com>
14732 PR tree-optimization/79691
14733 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
14735 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14738 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
14739 gcc_unreachable with output_operand_lossage.
14741 2017-02-28 Richard Biener <rguenther@suse.de>
14743 PR tree-optimization/79740
14744 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
14746 (visit_nary_op): Insert the nary into the hashtable if we
14747 pattern-matched sth.
14748 * tree-ssa-pre.c (eliminate_insert): Robustify.
14750 2017-02-28 Richard Biener <rguenther@suse.de>
14752 PR middle-end/79731
14753 * fold-const.c (decode_field_reference): Reject out-of-bound
14756 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14758 * config/i386/i386.c: Include intl.h.
14759 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
14760 instead of just cond ? "..." : "...".
14761 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
14762 * coverage.c (read_counts_file): Likewise.
14763 * omp-offload.c: Include intl.h.
14764 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
14765 of just cond ? "..." : "...".
14766 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
14767 of just cond ? "..." : "...".
14769 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
14772 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
14774 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
14775 'tune for' CPU name.
14776 * config/arm/arm-cpu-data.h: Regenerated.
14778 2017-02-28 Richard Biener <rguenther@suse.de>
14780 PR tree-optimization/79732
14781 * tree-inline.c (expand_call_inline): Do not shadow var.
14783 2017-02-28 Richard Biener <rguenther@suse.de>
14785 PR tree-optimization/79723
14786 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
14787 address-space properly.
14789 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
14791 * doc/optinfo.texi (Optimization groups): Fix option used for
14793 * doc/invoke.texi (-fopt-info): Document "omp".
14794 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
14795 (OPTGROUP_ALL): Add OPTGROUP_OMP.
14796 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
14797 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
14798 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
14800 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
14802 * dumpfile.c (optgroup_options): Instead of "openmp", associate
14803 OPTGROUP_OMP with "omp".
14805 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
14808 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
14809 for arithmetic shift of unsigned V2DI.
14811 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
14813 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
14814 arc/linux.h headers.
14815 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
14816 (LINK_SPEC): Likewise.
14817 (ARC_TLS_EXTRA_START_SPEC): Likewise.
14818 (EXTRA_SPECS): Likewise.
14819 (STARTFILE_SPEC): Likewise.
14820 (ENDFILE_SPEC): Likewise.
14821 (LIB_SPEC): Likewise.
14822 (TARGET_SDATA_DEFAULT): Likewise.
14823 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
14824 (MULTILIB_DEFAULTS): Likewise.
14825 (DWARF2_UNWIND_INFO): Likewise.
14826 * config/arc/big.h: New file.
14827 * config/arc/elf.h: Likewise.
14828 * config/arc/linux.h: Likewise.
14829 * config/arc/t-uClibc: Remove.
14831 2017-02-27 Bin Cheng <bin.cheng@arm.com>
14833 PR tree-optimization/77536
14834 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
14835 (tree_transform_and_unroll_loop): Use above function to compute the
14836 estimated niter of unrolled loop and use it when scaling profile.
14837 Also use count info rather than frequency if it's non-zero.
14838 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
14839 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
14840 (vect_transform_loop): Call above function.
14842 2017-02-27 Richard Biener <rguenther@suse.de>
14844 PR tree-optimization/45397
14845 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
14846 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
14847 (visit_nary_op): Add pattern matching for CSEing sign-changed
14848 or truncated operations with wider ones.
14850 2017-02-27 Richard Biener <rguenther@suse.de>
14852 PR tree-optimization/79690
14853 * tree-vect-stmts.c (vectorizable_store): Use vector type
14854 built from the DR with address-space.
14856 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
14858 * doc/invoke.texi (Optimize Options): Refine the description
14859 of asan-use-after-return.
14861 2017-02-25 Alan Modra <amodra@gmail.com>
14863 PR rtl-optimization/79584
14864 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
14865 base, not ad->base_term, the reg within base. Remove assertion
14866 that ad->base == ad->base_term. Replace gen_int_mode using
14867 bogus mode with const0_rtx.
14869 2017-02-25 Jakub Jelinek <jakub@redhat.com>
14871 PR middle-end/79396
14872 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
14873 FMA_EXPR like tcc_binary or tcc_unary.
14875 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
14878 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
14880 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
14881 (output_loc_operands): Handle DW_OP_call_ref and
14882 DW_OP_GNU_variable_value.
14883 (struct variable_value_struct): New type.
14884 (struct variable_value_hasher): Likewise.
14885 (variable_value_hash): New variable.
14886 (string_types): Remove.
14887 (copy_loc_descr): New function.
14888 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
14889 (prepend_loc_descr_to_each): New function.
14890 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
14891 instead of add_loc_descr_to_each if the first argument is single
14892 location list and the second has multiple.
14893 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
14894 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
14895 when looking for variable value which doesn't have other location info.
14896 (loc_list_from_tree): Formatting fix.
14897 (gen_array_type_die): Simplify DW_AT_string_length handling.
14898 (adjust_string_types): Remove.
14899 (gen_subprogram_die): Don't call adjust_string_types nor test/set
14900 string_types. Call resolve_variable_values.
14901 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
14902 (resolve_addr_in_expr): Likewise. Add A argument.
14903 (copy_deref_exprloc): Remove deref argument. Adjust for the
14904 original expression being DW_OP_GNU_variable_value with optionally
14905 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
14906 optionally after it.
14907 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
14908 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
14909 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
14910 (variable_value_hasher::hash, variable_value_hasher::equal): New
14912 (resolve_variable_value_in_expr, resolve_variable_value,
14913 resolve_variable_values, note_variable_value_in_expr,
14914 note_variable_value): New functions.
14915 (dwarf2out_early_finish): Call note_variable_value on all toplevel
14918 2017-02-24 Jakub Jelinek <jakub@redhat.com>
14921 * opts.h (handle_generated_option): Add GENERATED_P argument.
14922 * opts-common.c (handle_option): Adjust function comment.
14923 (handle_generated_option): Add GENERATED_P argument, pass it to
14925 (control_warning_option): Pass false to handle_generated_option
14927 * opts.c (maybe_default_option): Pass true to handle_generated_option
14929 * optc-gen.awk: Likewise.
14931 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14933 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
14934 a REG, look at the REG it is a SUBREG of.
14935 (splitter for cmpeqsi_t): Ditto.
14937 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14939 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
14940 the special USEs with the pattern of the insn, not the insn itself.
14942 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
14945 * doc/invoke.texi: Document -mload-store-pairs.
14947 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14948 Sandra Loosemore <sandra@codesourcery.com>
14950 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
14951 argument isn't a CONST_INT.
14952 (nios2_alternate_compare_const): Assert op is a CONST_INT.
14953 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
14954 (nios2_validate_compare): Bypass alternate compare logic if *op2
14955 is not a CONST_INT.
14956 (ldstwm_operation_p): Return false if first_base is not a REG or
14957 if first_offset is not a CONST_INT.
14959 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14961 * config/cris/cris.md: Use correct operand in a define_peephole2.
14963 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14965 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
14967 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14969 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
14970 this_insn if it is an INSN or JUMP_INSN.
14971 (force_offsettable): Look at base, not at addr.
14972 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
14973 on things that aren't necessarily CONST_INTs.
14975 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
14977 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
14978 -mfpmath=sse is the default also for x86-32 targets with SSE2
14979 instruction set when @option{-ffast-math} is enabled
14981 2017-02-24 Jeff Law <law@redhat.com>
14983 PR rtl-optimizatoin/79286
14984 * ira.c (update_equiv_regs): Drop may_trap_p exception to
14987 2017-02-24 Richard Biener <rguenther@suse.de>
14989 PR tree-optimization/79389
14990 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
14993 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
14995 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
14996 function comment to reflect reality.
14997 (loop_exits_before_overflow): Fix typo in function description.
14999 2017-02-24 Richard Biener <rguenther@suse.de>
15001 PR tree-optimization/79389
15002 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
15003 properly that a threading opportunity exists. Detect conditional
15004 copy/constant propagation opportunities.
15006 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
15008 * config/visium/visium.md (type): Add trap.
15009 (b): New mode attribute.
15010 (*btst): Rename into...
15011 (*btst<mode>): ...this and adjust.
15012 (*cbranchsi4_btst_insn): Rename into...
15013 (*cbranch<mode>4_btst_insn): ...this and adjust.
15014 (trap): New define_insn.
15016 2017-02-23 Jakub Jelinek <jakub@redhat.com>
15018 PR tree-optimization/79389
15019 * ifcvt.c (struct noce_if_info): Add rev_cond field.
15020 (noce_reversed_cond_code): New function.
15021 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
15022 reversed_comparison_code. Formatting fix.
15023 (noce_try_store_flag): Test rev_cond != NULL in addition to
15024 reversed_comparison_code.
15025 (noce_try_store_flag_constants): Likewise.
15026 (noce_try_store_flag_mask): Likewise.
15027 (noce_try_addcc): Use rev_cond if non-NULL instead of
15028 reversed_comparison_code.
15029 (noce_try_cmove_arith): Likewise. Formatting fixes.
15030 (noce_try_minmax, noce_try_abs): Clear rev_cond.
15031 (noce_find_if_block): Initialize rev_cond.
15032 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
15033 instead of false as last argument never attempt to reverse it
15036 2017-02-23 Bin Cheng <bin.cheng@arm.com>
15038 PR tree-optimization/79663
15039 * tree-predcom.c (combine_chains): Process refs in reverse order
15040 only for ZERO length chains, and add explaining comment.
15042 2017-02-23 Jeff Law <law@redhat.com>
15044 PR tree-optimization/79578
15045 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
15046 in call to operand_equal_p.
15048 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
15051 * config/i386/cpuid.h: Fix another undefined behavior.
15053 2017-02-23 Richard Biener <rguenther@suse.de>
15055 PR tree-optimization/79683
15056 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
15057 vector types for data-refs.
15059 2017-02-23 Martin Liska <mliska@suse.cz>
15061 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
15063 2017-02-23 Jakub Jelinek <jakub@redhat.com>
15065 PR middle-end/79665
15066 * internal-fn.c (get_range_pos_neg): Moved to ...
15067 * tree.c (get_range_pos_neg): ... here. No longer static.
15068 * tree.h (get_range_pos_neg): New prototype.
15069 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
15070 are known to be in between 0 and signed maximum inclusive, try to
15071 expand both unsigned and signed divmod and use the cheaper one from
15074 2017-02-22 Jeff Law <law@redhat.com>
15076 PR tree-optimization/79578
15077 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
15078 to compare base operands.
15080 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
15083 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
15084 gpc_reg_operand instead of fpr_reg_operand.
15086 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
15088 * config/mips/mips.c (mips_return_in_memory): Force FP
15089 vector types to be returned in memory for o32 ABI.
15091 2017-02-22 Jakub Jelinek <jakub@redhat.com>
15093 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
15094 instead of DW_TAG_member for static data member declarations and don't
15095 set no_linkage_name for static inline data members.
15096 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
15099 2017-02-22 Martin Liska <mliska@suse.cz>
15101 * doc/invoke.texi: Replace inequality signs with square brackets
15104 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15106 PR tree-optimization/68644
15107 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
15109 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
15112 * lra-constraints.c (simplify_operand_subreg): Handle
15113 WORD_REGISTER_OPERATIONS targets.
15115 2017-02-22 Jakub Jelinek <jakub@redhat.com>
15118 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
15119 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
15120 elimination by swapping fld*.
15122 2017-02-22 Richard Biener <rguenther@suse.de>
15124 PR tree-optimization/79673
15125 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
15126 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
15127 irrelevant address-space qualifiers and avoiding a
15128 ADDR_SPACE_CONVERT_EXPR from fold_convert.
15130 2017-02-22 Richard Biener <rguenther@suse.de>
15132 PR tree-optimization/79666
15133 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
15134 to not symbolically negate if that may introduce undefined
15137 2017-02-22 Martin Liska <mliska@suse.cz>
15140 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
15141 * data-streamer-out.c (streamer_write_gcov_count_stream):
15143 * value-prof.c (stream_out_histogram_value): Make assert more
15144 precise based on type of counter.
15146 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
15149 * config/i386/i386.md (standard_x87sse_constant_load splitter):
15150 Use nonimmediate_operand instead of memory_operand for operand 1.
15151 (float-extend standard_x87sse_constant_load splitter): Ditto.
15153 2017-02-21 Jeff Law <law@redhat.com>
15155 PR tree-optimization/79621
15156 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
15157 blocks with edges to themselves.
15159 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15162 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
15163 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
15164 Use gimple_call_builtin_p.
15167 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
15168 on temporarily removed DEBUG_INSNs.
15170 PR tree-optimization/79649
15171 * tree-loop-distribution.c (classify_partition): Give up on
15172 non-generic address space loads/stores.
15174 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
15176 * doc/loop.texi (Loop manipulation): Remove nonexistent
15177 tree_ssa_loop_version from the documentation.
15178 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
15180 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15183 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
15184 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
15185 * config/rs6000/rs6000.c: Include except.h.
15186 (rs6000_expand_split_stack_prologue): Call
15187 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
15189 2017-02-21 Martin Jambor <mjambor@suse.cz>
15192 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
15193 have been analyzed.
15195 2017-02-21 Martin Jambor <mjambor@suse.cz>
15197 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
15198 for backward compatibility only.
15199 * doc/invoke.texi (Option Summary): Remove all references to
15200 -fipa-cp-alignment.
15202 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
15206 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15208 * lra-constraints.c (curr_insn_transform): Handle
15209 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
15211 2017-02-21 Martin Liska <mliska@suse.cz>
15213 * config/i386/i386.opt: Replace -masm-dialect with -masm.
15215 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
15217 PR translation/79638
15218 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
15220 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
15223 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
15224 (arm_function_ok_for_sibcall): Return false for an indirect call by
15225 descriptor if all the argument registers are used.
15226 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
15227 alignment of the function.
15229 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15231 PR tree-optimization/61441
15232 * simplify-rtx.c (simplify_const_unary_operation): For
15233 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
15234 the sNaN unmodified.
15236 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
15238 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
15239 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
15240 instead of SYSTEM_HEADER_DIR.
15242 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
15243 Martin Liška <mliska@suse.cz>
15245 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
15246 Fix typos and grammar, use active voice, and clarify.
15248 2017-02-20 Marek Polacek <polacek@redhat.com>
15250 PR middle-end/79537
15251 * gimplify.c (gimplify_expr): Handle unused *&&L;.
15254 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
15256 2017-02-20 Jakub Jelinek <jakub@redhat.com>
15259 * config/i386/i386.c (ix86_expand_builtin): Handle
15260 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
15261 ix86_builtins_isa[fcode].isa as a requirement of those
15262 flags and any other flag in the bitmask.
15263 (ix86_init_mmx_sse_builtins): Use 0 instead of
15264 ~OPTION_MASK_ISA_64BIT as mask.
15265 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
15266 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
15267 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
15268 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
15270 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15273 * lra-constraints.c (split_reg): Check requested split mode
15274 is supported by the register.
15276 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15278 * lra-constraints.c (simplify_operand_subreg): Remove early
15281 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15284 * lra-constraints.c (curr_insn_transform): Tighten condition
15285 for converting SUBREG reloads from OP_OUT to OP_INOUT.
15287 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15290 * lra-constraints.c (curr_insn_transform): Handle
15291 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
15293 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
15296 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
15298 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
15300 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
15303 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
15306 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15308 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
15311 2017-02-18 Jakub Jelinek <jakub@redhat.com>
15313 * final.c (last_columnnum, override_columnnum): New variables.
15314 (final_start_function): Set last_columnnum, pass it to begin_prologue
15315 hook and pass 0 to dwarf2out_begin_prologue.
15316 (final_scan_insn): Update override_columnnum. Pass last_columnnum
15317 to source_line debug hook.
15318 (notice_source_line): Compute last_columnnum and for debug_column_info
15319 return true on column changes.
15320 * debug.h (struct gcc_debug_hooks): Add column argument to
15321 source_line and begin_prologue hooks.
15322 (debug_nothing_int_charstar_int_bool): Remove prototype.
15323 (debug_nothing_int_int_charstar,
15324 debug_nothing_int_int_charstar_int_bool): New prototypes.
15325 (dwarf2out_begin_prologue): Add column argument.
15326 * debug.c (do_nothing_debug_hooks): Adjust source_line and
15327 begin_prologue hooks.
15328 (debug_nothing_int_charstar_int_bool): Remove.
15329 (debug_nothing_int_int_charstar,
15330 debug_nothing_int_int_charstar_int_bool): New functions.
15331 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
15332 through to dwarf2out_source_line.
15333 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
15334 (dwarf2out_source_line): Add column argument, emit it if requested.
15335 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
15337 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
15338 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
15339 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
15340 through to dwarf2out_begin_prologue.
15341 (vmsdbgout_source_line): Add column argument, pass it through to
15342 dwarf2out_source_line.
15343 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
15344 dbxout_source_line caller.
15345 (dbxout_source_line): Add column argument.
15347 * common.opt (gno-column-info, gcolumn-info): New options.
15348 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
15349 (check_die): Also test for multiple DW_AT_decl_column attributes.
15350 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
15351 DW_AT_decl_column if requested.
15352 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
15354 (gen_variable_die): Likewise.
15355 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
15356 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
15359 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
15360 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
15361 (ix86_handle_option): Handle OPT_m3dnowa.
15362 * doc/invoke.texi (-m3dnowa): Document.
15363 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
15364 -m3dnowa instead of -m3dnow -march=athlon.
15367 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
15368 instead of gcc_assert for K, r and R code checks. Formatting fixes.
15370 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15373 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
15374 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
15375 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
15376 generator for vsx_xxpermdi_<mode>_be.
15377 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
15378 force big-endian semantics.
15379 (vsx_xxpermdi_<mode>_be): New define_expand with same
15380 implementation as previous version of vsx_xxpermdi_<mode>.
15382 2017-02-17 Jakub Jelinek <jakub@redhat.com>
15384 PR tree-optimization/79327
15385 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
15386 variable, its initialization and use.
15388 2017-02-17 Julia Koval <julia.koval@intel.com>
15390 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
15391 (OPTION_MASK_ISA_PKU_UNSET): New.
15392 (ix86_handle_option): Handle -mrdpid.
15393 * config/i386/cpuid.h (bit_RDPID): New.
15394 * config/i386/driver-i386.c (host_detect_local_cpu):
15395 Detect RDPID feature.
15396 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
15397 * config/i386/i386-c.c (ix86_target_macros_internal):
15399 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
15400 (ix86_valid_target_attribute_inner_p): Add "rdpid".
15401 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
15402 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
15403 * config/i386/i386.md (define_insn "rdpid"): New.
15404 * config/i386/i386.opt Add -mrdpid.
15405 * config/i386/immintrin.h (_rdpid_u32): New.
15407 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
15409 PR rtl-optimization/79541
15410 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
15411 instead of transforming it into USE.
15413 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
15415 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
15416 If HONOR_SNANS (SFmode) force the input to a register.
15417 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
15418 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
15419 an frsp or similar insn.
15421 2017-02-17 Martin Liska <mliska@suse.cz>
15423 PR rtl-optimization/79577
15424 * params.def (selsched-max-sched-times): Increase minimum to 1.
15426 2017-02-17 Martin Liska <mliska@suse.cz>
15428 PR rtl-optimization/79574
15429 * gcse.c (want_to_gcse_p): Prevent integer overflow.
15431 2017-02-17 Martin Liska <mliska@suse.cz>
15433 PR tree-optimization/79529
15434 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
15435 ssa_defined_default_def_p to handle cases which are implicitly
15437 * tree-ssa.c (ssa_defined_default_def_p): New function.
15438 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
15439 which are implicitly defined.
15440 * tree-ssa.h (ssa_defined_default_def_p): Declare.
15442 2017-02-17 Richard Biener <rguenther@suse.de>
15444 PR middle-end/79576
15445 * params.def (max-ssa-name-query-depth): Limit to 10.
15447 2017-02-17 Richard Biener <rguenther@suse.de>
15449 PR tree-optimization/79552
15450 * tree-ssa-structalias.c (visit_loadstore): Properly verify
15453 2017-02-17 Richard Biener <rguenther@suse.de>
15456 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
15458 2017-02-17 Marek Polacek <polacek@redhat.com>
15460 PR middle-end/79536
15461 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
15462 (fold_negate_expr): New wrapper.
15464 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
15466 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
15467 Correct terminology and de-emphasize pre-standard behavior.
15469 2017-02-16 Alan Modra <amodra@gmail.com>
15471 PR rtl-optimization/79286
15472 * ira.c (def_dominates_uses): New function.
15473 (update_equiv_regs): Don't create an equivalence for insns that
15474 may trap where the register def does not dominate the use.
15476 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
15478 PR rtl-optimization/78127
15479 * lra.c (lra): Call lra_eliminate before finish the loop after
15482 2017-02-16 Richard Biener <rguenther@suse.de>
15484 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
15486 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
15487 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
15488 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
15489 (isl_val_int_from_wi): New function.
15490 (extract_affine_gmp): Rename to ...
15491 (extract_affine_wi): ... this, take a widest_int.
15492 (extract_affine_int): Just wrap extract_affine_wi.
15493 (add_param_constraints): Use isl_val_int_from_wi.
15494 (add_loop_constraints): Likewise, and extract_affine_wi.
15496 2017-02-15 Jeff Law <law@redhat.com>
15498 PR middle-end/79521
15499 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
15500 ira_init_register_move_cost_if_necessary.
15502 2017-02-15 Martin Sebor <msebor@redhat.com>
15504 PR middle-end/32003
15505 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
15506 removed in a prior commit.
15508 2017-02-15 Bin Cheng <bin.cheng@arm.com>
15510 PR tree-optimization/79347
15511 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
15512 counters during peeling.
15514 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
15516 * Makefile.in (site.exp): Remove "set ISLVER".
15518 2017-02-15 Jakub Jelinek <jakub@redhat.com>
15521 * real.c (real_from_integer): Call real_convert even for decimal.
15523 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
15526 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
15528 2017-02-14 Andrew Pinski <apinski@cavium.com>
15530 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
15531 cores and change the partno/implementer to be correct.
15532 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
15533 the 'B" as the implementer.
15534 * config/aarch64/aarch64-tune.md: Regenerate.
15536 2017-02-14 Carl Love <cel@us.ibm.com>
15538 * config/rs6000/rs6000.c: Add case statement entry to make the
15539 xvcvuxdsp built-in argument unsigned.
15540 * config/rs6000/vsx.md: Fix the source and return operand types so they
15541 match the instruction definitions from the ISA document. Fix typo
15542 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
15545 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
15548 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
15549 member early_clobber_alts.
15550 * lra-lives.c (reg_early_clobber_p): New.
15551 (process_bb_lives): Use it.
15552 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
15553 (debug_operand_data): Initialize early_clobber_alts.
15554 (setup_operand_alternative): Set up early_clobber_alts.
15555 (collect_non_operand_hard_regs): Ditto. Pass early clobber
15556 alternatives to new_insn_reg.
15557 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
15559 (lra_update_insn_regno_info): Pass the new arg.
15561 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15563 PR middle-end/79505
15564 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
15565 (new_oacc_loop_raw): Don't clear already cleared fields.
15568 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
15569 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
15570 _mm512_prefetch_i64gather_ps): New inline functions and macros.
15572 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
15575 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
15577 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
15580 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
15581 the extra instruction to the right place to store 128-bit constant
15584 2017-02-14 Martin Sebor <msebor@redhat.com>
15586 PR middle-end/79448
15587 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
15588 warning for strings of unknown length.
15590 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
15592 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
15594 2017-02-14 Jeff Law <law@redhat.com>
15597 * ira-costs.c (scan_one_insn): Initialize register move costs
15598 for pseudos seen in USE/CLOBBER insns.
15600 PR tree-optimization/79095
15601 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
15602 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
15603 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
15604 if the operands are known to be not equal, then the resulting range
15606 (intersect_ranges): If the new range is ~[0,0] and the old range is
15607 wide, then prefer ~[0,0].
15608 * tree-vrp.c (overflow_comparison_p_1): New function.
15609 (overflow_comparison_p): New function.
15610 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
15611 if NAME is used in an overflow test.
15612 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
15613 overflow check that can be expressed as an equality test, then adjust
15614 ops to be that equality test.
15616 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15618 * config/s390/s390-builtin-types.def: Remove flags argument.
15619 * config/s390/s390.c (s390_init_builtins): Likewise.
15621 2017-02-14 Martin Liska <mliska@suse.cz>
15623 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
15624 vector. Fix trailing white spaces.
15626 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
15628 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
15631 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15633 PR rtl-optimization/68664
15634 * config/arm/arm.c (arm_sched_can_speculate_insn):
15635 New function. Declare prototype.
15636 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15638 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15640 PR rtl-optimization/68664
15641 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
15643 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15645 2017-02-14 Amit Pawar <amit.pawar@amd.com>
15647 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
15648 max skip bytes for function, loop and jump.
15650 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15652 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
15653 ABS_EXPR for gimple dump.
15655 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15658 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
15660 PR tree-optimization/79408
15661 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
15662 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
15663 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
15664 also if rhs1 is INTEGER_CST.
15666 2017-02-14 Richard Biener <rguenther@suse.de>
15668 PR middle-end/79432
15669 * tree-into-ssa.c (insert_phi_nodes): When the function can
15670 have abnormal edges rewrite SSA names with broken use-def
15671 dominance out of SSA and register them for PHI insertion.
15673 2017-02-13 Martin Sebor <msebor@redhat.com>
15675 PR middle-end/79496
15676 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
15677 clearing info.nowrite flag when snprintf size argument is a range.
15679 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15681 * cprop.c (cprop_jump): Add missing space in string literal.
15682 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
15683 (get_constraint_for_component_ref): Likewise.
15684 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
15685 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
15686 * lra-constraints.c (process_alt_operands): Likewise.
15687 * ipa-inline.c (inline_small_functions): Likewise.
15688 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
15689 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15690 * trans-mem.c (diagnose_tm_1_op): Likewise.
15691 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
15692 (grid_parallel_clauses_gridifiable): Likewise.
15694 * config/nvptx/mkoffload.c (process): Add space in between
15697 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
15698 "MOD4_SSE_REGS" and "ALL_REGS".
15700 * spellcheck.c (test_data): Add , in between "foo" and "food".
15702 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15705 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
15706 boundary crossing check and subsequent code generation agree.
15708 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15710 * config/aarch64/aarch64.c (has_memory_op): Delete.
15711 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
15714 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15716 PR rtl-optimization/79388
15717 PR rtl-optimization/79450
15718 * combine.c (distribute_notes): When removing TEM_INSN for which
15719 corresponding dest has last value recorded, invalidate that last
15722 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15724 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
15725 of explicit '@'. Add missing assembly comment marker on branch costs
15728 2017-02-13 Nathan Sidwell <nathan@acm.org>
15730 * gengtype-lex.l (<in_struct>): Add '/'.
15732 2017-02-13 Martin Liska <mliska@suse.cz>
15735 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
15737 2017-02-13 Richard Biener <rguenther@suse.de>
15739 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
15741 * configure: Re-generate.
15742 * config.in: Likewise.
15743 * graphite-dependences.c: Simplify as if
15744 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
15745 * graphite-isl-ast-to-gimple.c: Likewise.
15746 * graphite-optimize-isl.c: Likewise.
15747 * graphite-poly.c: Likewise.
15748 * graphite-sese-to-poly.c: Likewise.
15749 * graphite.h: Likewise.
15750 * toplev.c: Include isl/version.h and use isl_version () for
15751 printing the ISL version.
15752 * doc/install.texi: Update ISL requirement.
15754 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15756 * doc/standards.texi (Standards): Update reference to
15759 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15761 * doc/extend.texi (Named Address Spaces): sourceware.org now
15763 * doc/install.texi (Binaries): Ditto.
15766 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15768 * doc/cpp.texi: Replace "stringify"/"stringification" with C
15769 standard terminology "stringize"/"stringizing" throughout.
15770 * doc/cppinternals.texi: Likewise.
15772 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15774 * doc/extend.texi: Fix some spelling mistakes and typos.
15775 * doc/invoke.texi: Likewise.
15777 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15780 * params.def (inline-min-speedup) Change from 10 to 8.
15782 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15784 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
15787 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15790 * ipa-inline-analysis.c (get_minimal_bb): New function.
15791 (record_modified): Use it.
15792 (remap_edge_change_prob): Handle also ancestor functions.
15794 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
15796 * doc/contrib.texi (Contributors): Remove broken link into
15797 the Mauve CVS repository.
15799 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15801 PR middle-end/79454
15802 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
15803 result computation whenever lhs doesn't have vector mode, not
15804 just when it has BLKmode.
15806 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15808 * doc/makefile.texi (profiledbootstrap): Refer to the
15809 installation instructions only in textual form.
15811 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15814 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
15816 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15818 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
15819 (Specific): Update mingw-w64 reference.
15821 (Specific): Remove broken link to Renesas RX processor.
15823 2017-02-10 Richard Biener <rguenther@suse.de>
15825 * toplev.c (process_options): Do not mention obsolete graphite
15826 options when printing sorry message about missing graphite support.
15827 Mention -floop-nest-optimize.
15829 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
15831 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
15832 (vtst_p16): Likewise.
15833 (vtstq_p8): Likewise.
15834 (vtstq_p16): Likewise.
15836 (vtstq_p64): Likewise.
15837 * config/arm/arm_neon.h (vgetq_lane_p64): New.
15838 (vset_lane_p64): New.
15839 (vsetq_lane_p64): New.
15841 2017-02-10 Jakub Jelinek <jakub@redhat.com>
15843 PR tree-optimization/79411
15844 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
15845 stmt operands are SSA_NAMEs used in abnormal phis.
15846 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
15849 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15852 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
15855 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15857 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
15859 2017-02-09 Jakub Jelinek <jakub@redhat.com>
15861 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
15865 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
15866 not arbitrary TREE_CONSTANT.
15869 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
15870 "omp declare target link" attribute unless is_global_var.
15871 * omp-offload.c (find_link_var_op): Likewise.
15873 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
15874 Chung-Lin Tang <cltang@codesourcery.com>
15876 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
15878 (gimplify_adjust_omp_clauses): Don't delete TILE.
15879 (gimplify_omp_for): Deal with TILE.
15880 * internal-fn.c (expand_GOACC_TILE): New function.
15881 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
15883 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
15884 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
15886 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
15887 avoid DIV for outermost collapse var.
15888 (expand_oacc_for): Insert tile element loop as needed. Adjust.
15889 Remove out of date comments, fix whitespace.
15890 * omp-general.c (omp_extract_for_data): Deal with tiling.
15891 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
15892 adjust OLF_DIM_BASE value.
15893 (struct omp_for_data): Add tiling field.
15894 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
15895 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
15896 for auto loops. Remove default auto determining, moved to
15897 oacc_loop_fixed_partitions.
15898 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
15899 stmts, add e_mask field.
15900 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
15901 (oacc_thread_numbers): Use oacc_dim_call.
15902 (oacc_xform_tile): New.
15903 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
15904 (finish_oacc_loop): Adjust for ifns vector.
15905 (oacc_loop_discover_walk): Append loop abstraction sites to list,
15906 add case for GOACC_TILE fns.
15907 (oacc_loop_xform_loop): Delete.
15908 (oacc_loop_process): Iterate over call list directly, and add
15909 handling for GOACC_TILE fns.
15910 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
15912 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
15913 vector partitioning to outer loops. Assign 2 partitions to loops
15914 when available. Add TILE handling.
15915 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
15916 (execite_oacc_device_lower): Process GOACC_TILE fns,
15917 ignore unknown specs.
15918 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
15919 * tree.c (omp_clause_num_ops): Adjust TILE ops.
15920 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
15922 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
15924 * configure.ac (ACX_BUGURL): Update.
15925 * configure: Regenerate.
15927 2017-02-09 Richard Biener <rguenther@suse.de>
15929 PR tree-optimization/69823
15930 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
15931 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
15933 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15935 * config/arc/arc-c.def: Add __NPS400__ definition.
15936 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
15937 (TARGET_NPS400): Define.
15939 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15941 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
15943 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
15944 pointer, arch_info.
15945 (arc_cpu_types): Fill the arch_info field with a pointer into the
15946 arc_arch_types table.
15947 (arc_selected_cpu): Declare.
15948 * config/arc/arc.c (arc_selected_cpu): Make global.
15949 (arc_selected_arch): Delete.
15950 (arc_base_cpu): Delete.
15951 (arc_override_options): Remove references to deleted variables,
15952 update access to arch information.
15953 (ARC_OPT): Update access to arch information.
15954 (ARC_OPTX): Likewise.
15955 * config/arc/arc.h (arc_base_cpu): Remove declaration.
15956 (TARGET_ARC600): Update access to arch information.
15957 (TARGET_ARC601): Likewise.
15958 (TARGET_ARC700): Likewise.
15959 (TARGET_EM): Likewise.
15960 (TARGET_HS): Likewise.
15961 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
15964 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
15967 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
15968 condition/operands for integer GE/LE/GEU/LEU operations.
15970 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
15972 PR translation/79397
15973 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
15976 2017-02-08 Martin Jambor <mjambor@suse.cz>
15979 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
15980 whether allocation happened.
15981 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
15982 nothing was allocated.
15984 2017-02-08 Jakub Jelinek <jakub@redhat.com>
15986 PR tree-optimization/79408
15987 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
15988 constant, but SSA_NAME with a known integer range, use the minimum
15989 of that range instead of op1 to determine if modulo can be replaced
15990 with its first operand.
15992 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15994 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
15996 2017-02-08 Richard Biener <rguenther@suse.de>
15998 PR tree-optimization/71824
15999 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
16000 Check all loops contained in the merged region.
16002 2017-02-07 Andrew Pinski <apinski@cavium.com>
16004 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
16006 2017-02-07 Andrew Pinski <apinski@cavium.com>
16008 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
16009 (thunderxt88): Likewise.
16010 (thunderxt81): Disable LSE and change v8.1 to v8.
16011 (thunderxt83): Likewise.
16013 2017-02-07 Jakub Jelinek <jakub@redhat.com>
16014 Richard Biener <rguenther@suse.de>
16016 PR middle-end/79399
16017 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
16018 type from int to size_t.
16019 * ira-costs.c (struct_costs_size): Change type from int to size_t.
16021 2017-02-07 Jakub Jelinek <jakub@redhat.com>
16023 PR rtl-optimization/79386
16024 * cprop.c (bypass_conditional_jumps): Initialize
16025 bypass_last_basic_block already before splitting bbs after
16026 unconditional traps...
16027 (bypass_conditional_jumps): ... rather than here.
16030 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
16031 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
16032 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
16033 fix -masm=intel patterns.
16035 2017-02-07 Richard Biener <rguenther@suse.de>
16037 PR tree-optimization/79256
16038 PR middle-end/79278
16039 * builtins.c (get_object_alignment_2): Use min_align_of_type
16040 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
16041 and ADJUST_FIELD_ALIGN.
16043 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
16045 * doc/tm.texi: Regenerate.
16046 * stor-layout.c (layout_decl): Adjust.
16047 (update_alignment_for_field): Likewise.
16048 (place_field): Likewise.
16049 (min_align_of_type): Likewise.
16050 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
16051 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
16052 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
16053 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
16054 * config/frv/frv.c (frv_adjust_field_align): Likewise.
16055 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
16056 * config/i386/i386.c (x86_field_alignment): Likewise.
16057 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
16058 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
16059 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
16060 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
16061 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
16062 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
16066 2017-01-30 Richard Biener <rguenther@suse.de>
16068 PR tree-optimization/79256
16069 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
16070 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
16073 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
16075 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
16076 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
16077 builtins to SImode and emit a zero-extend, if necessary.
16079 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
16081 * docs/invoke.texi (RISC-V Options): Alphabetize.
16083 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
16085 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
16088 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
16090 * config/riscv/riscv.c: New file.
16091 * common/config/riscv/riscv-common.c: Likewise.
16092 * config.gcc: Likewise.
16093 * config/riscv/constraints.md: Likewise.
16094 * config/riscv/elf.h: Likewise.
16095 * config/riscv/generic.md: Likewise.
16096 * config/riscv/linux.h: Likewise.
16097 * config/riscv/multilib-generator: Likewise.
16098 * config/riscv/peephole.md: Likewise.
16099 * config/riscv/pic.md: Likewise.
16100 * config/riscv/predicates.md: Likewise.
16101 * config/riscv/riscv-builtins.c: Likewise.
16102 * config/riscv/riscv-c.c: Likewise.
16103 * config/riscv/riscv-ftypes.def: Likewise.
16104 * config/riscv/riscv-modes.def: Likewise.
16105 * config/riscv/riscv-opts.h: Likewise.
16106 * config/riscv/riscv-protos.h: Likewise.
16107 * config/riscv/riscv.h: Likewise.
16108 * config/riscv/riscv.md: Likewise.
16109 * config/riscv/riscv.opt: Likewise.
16110 * config/riscv/sync.md: Likewise.
16111 * config/riscv/t-elf-multilib: Likewise.
16112 * config/riscv/t-linux: Likewise.
16113 * config/riscv/t-linux-multilib: Likewise.
16114 * config/riscv/t-riscv: Likewise.
16115 * configure.ac: Likewise.
16116 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
16117 Waterman as RISC-V maintainers.
16118 * doc/install.texi: Add RISC-V entries.
16119 * doc/invoke.texi: Add RISC-V options section.
16120 * doc/md.texi: Add RISC-V constraints section.
16121 * configure: Regenerated.
16123 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
16126 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
16127 false values to be constant vectors with all 0 or all 1 bits set.
16128 (vcondu<mode><mode>): Likewise.
16129 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
16131 (fpmask_comparison_operator): Update comment.
16132 (vecint_comparison_operator): New predicate.
16133 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
16134 vector conditionals when the true and false values are constant
16135 vectors with all 0 bits or all 1 bits set.
16137 2017-02-06 Martin Sebor <msebor@redhat.com>
16139 PR tree-optimization/79376
16140 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
16142 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
16144 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
16145 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
16146 to simplify split condition.
16148 2017-02-06 Jakub Jelinek <jakub@redhat.com>
16150 * omp-expand.c (oxpand_omp_atomic_fetch_op,
16151 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
16154 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
16156 PR rtl-optimization/68664
16157 * target.def (can_speculate_insn): New hook.
16158 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
16159 * doc/tm.texi: Regenerate.
16160 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
16161 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
16162 (rs6000_sched_can_speculate_insn): New function.
16164 2017-02-06 Jakub Jelinek <jakub@redhat.com>
16166 PR tree-optimization/79284
16167 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
16168 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
16169 vectorizable_mask_load_store, vectorizable_operation,
16170 vect_is_simple_cond, get_same_sized_vectype): Use it instead
16171 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
16172 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
16173 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
16174 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
16175 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
16176 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
16177 is_gimple_assign (stmt). Replace another such test with
16178 is_gimple_assign (stmt).
16180 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
16183 * config/avr/avr.c (rtl-iter.h): Include it.
16184 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
16185 (avr_legitimate_combined_insn): ...and implementation.
16187 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
16189 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
16190 * config/s390/s390.c (s390_const_operand_ok)
16191 (s390_canonicalize_comparison, s390_extract_part)
16192 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
16193 (s390_contiguous_bitmask_p, s390_rtx_costs)
16194 (legitimize_pic_address): Likewise.
16195 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
16196 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
16197 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
16198 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
16199 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
16201 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
16203 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
16204 REGNO($0) == REGNO($1).
16206 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16208 * config/s390/linux.h(SIZE_TYPE): Add comment.
16210 2017-02-06 Julian Brown <julian@codesourcery.com>
16211 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16212 Virendra Pathak <virendra.pathak@broadcom.com>
16214 * config/aarch64/aarch64-cores.def: Change the scheduler
16216 * config/aarch64/aarch64.md: Include thunderx2t99.md.
16217 * config/aarch64/thunderx2t99.md: New file.
16219 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16221 * doc/standards.texi (Go Language): Update link to language
16224 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
16226 * tree-eh.c (lower_resx): Sanitize profile.
16227 (cleanup_empty_eh_move_lp): Likewise.
16229 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
16232 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
16234 * cfgloopmanip.h (loop_version): Update prototype.
16235 * modulo-sched.c (sms_schedule): Update call of loop_version.
16236 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
16237 * tree-parloops.c (gen_parallel_loop): Likewise.
16238 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
16239 * tree-ssa-loop-split.c (split_loop): Likewise.
16240 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
16241 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
16243 2017-02-05 Martin Liska <mliska@suse.cz>
16246 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
16248 (print_operand_address): Initialize a struct to zero.
16250 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16252 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
16253 garbage collector only in textual form.
16255 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16257 * doc/extend.texi (x86 specific memory model extensions for
16258 transactional memory): Simplify a phrase.
16260 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
16263 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
16264 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
16265 (atomic_storedi_1): Likewise.
16267 2017-02-04 Jakub Jelinek <jakub@redhat.com>
16269 PR tree-optimization/79338
16270 * tree-parloops.c (gather_scalar_reductions): Don't call
16271 vect_analyze_loop_form for loop->inner before destroying loop's
16274 2017-02-03 Martin Sebor <msebor@redhat.com>
16276 PR tree-optimization/79327
16277 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
16278 when precision has resulted in leading zeros.
16279 (format_integer): Adjust the likely counter to assume an unknown
16280 argument that may be zero is non-zero.
16282 2017-02-03 Jason Merrill <jason@redhat.com>
16285 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
16286 avoid copying non-taken branch.
16288 2017-02-03 Jakub Jelinek <jakub@redhat.com>
16290 PR tree-optimization/79340
16291 * tree-vect-loop.c (vectorizable_reduction): Release
16292 vec_defs elements after safe_splicing them into other vectors.
16295 PR tree-optimization/79327
16296 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
16297 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
16299 (format_integer): Use wide_int_to_tree instead of build_int_cst
16300 + to_?hwi. If argmin is NULL, just set argmin and argmax to
16301 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
16302 of shortest and longest sequence.
16304 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
16306 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
16307 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
16309 2017-02-03 Walter Lee <walt@tilera.com>
16312 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
16313 after initial stackframe link reg save.
16314 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
16316 2017-02-03 Jakub Jelinek <jakub@redhat.com>
16319 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
16320 wu for stxssp alternative.
16322 2017-02-03 Martin Sebor <msebor@redhat.com>
16324 PR tree-optimization/79352
16325 * gimple-fold.c (get_range_strlen): Add argument.
16326 (get_range_strlen): Change return type to bool.
16327 (get_maxval_strlen): Pass in a dummy argument.
16328 * gimple-fold.h (get_range_strlen): Change return type to bool.
16329 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
16330 * tree.h (array_at_struct_end_p): Add argument.
16331 * tree.c (array_at_struct_end_p): Handle it.
16333 2017-02-03 Martin Liska <mliska@suse.cz>
16336 * multiple_target.c (create_dispatcher_calls): Redirect edge
16337 from a caller of a dispatcher.
16338 (expand_target_clones): Make the clones local.
16339 (ipa_target_clone): Do both target clones and resolvers.
16340 (ipa_dispatcher_calls): Remove the pass.
16341 (pass_dispatcher_calls::gate): Likewise.
16342 (make_pass_dispatcher_calls): Likewise.
16343 * passes.def (pass_target_clone): Put as very first IPA early
16346 2017-02-03 Martin Liska <mliska@suse.cz>
16348 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
16349 in case of a function with ifunc attribute.
16351 2017-02-03 Martin Liska <mliska@suse.cz>
16353 * cgraph.c (cgraph_node::dump): Dump function version info.
16354 * symtab.c (symtab_node::dump_base): Add missing new line.
16356 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
16358 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
16359 (ifcombine_ifandif): Use it.
16361 2017-02-03 Martin Liska <mliska@suse.cz>
16363 * doc/invoke.texi: Document default value for
16364 use-after-scope-direct-emission-threshold.
16366 2017-02-03 Martin Liska <mliska@suse.cz>
16368 PR tree-optimization/79339
16369 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
16370 (format_floating): Likewise.
16372 2017-02-03 Martin Liska <mliska@suse.cz>
16375 * ipa-prop.c (ipa_node_params_t::insert): Remove current
16377 (ipa_node_params_t::remove): Likewise.
16378 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
16379 initialization from removed ipa_node_params_t::insert.
16380 (ipa_node_params::~ipa_node_params): Move from removed
16381 ipa_node_params_t::release.
16382 * symbol-summary.h (symbol_summary::m_released): New member.
16383 Do not release a summary twice. Do not allow to call finalizer
16384 for types of a summary that live in GGC memory.
16386 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16388 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
16391 2017-02-02 Martin Sebor <msebor@redhat.com>
16393 PR middle-end/79275
16394 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
16395 (format_string): Tighten up the range of output for non-constant
16396 strings and correct the expected range for wide non-constant strings.
16398 2017-02-02 Martin Sebor <msebor@redhat.com>
16400 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
16402 PR middle-end/32003
16403 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
16405 (-fdump-tree-@var): Add to index and document how to come up
16406 with pass-specific option and dump file names.
16407 (-fdump-passes): Clarify where to look for output.
16409 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
16411 PR middle-end/77445
16412 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
16413 statistics of the analyzed path; allow threading for speed when
16414 any of BBs along the path are optimized for speed.
16416 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
16418 PR middle-end/78468
16419 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
16420 settings of the virtual registers.
16423 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
16425 * explow.c (get_dynamic_stack_size): Take known alignment of stack
16426 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
16429 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16431 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
16432 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
16434 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16436 * config/s390/s390.md: Add missing comments with the expanded
16438 * config/s390/vector.md: Likewise.
16439 * config/s390/vx-builtins.md: Likewise.
16441 2017-02-02 Jakub Jelinek <jakub@redhat.com>
16444 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
16445 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
16446 conditions on a single line.
16448 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16450 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
16451 __S390_VX__ to __VX__.
16453 2017-02-01 Andrew Pinski <apinski@cavium.com>
16455 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
16456 stmt_info to record_stmt_cost.
16457 (vect_get_known_peeling_cost): Pass stmt_info if known to
16459 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
16460 cpu_vector_cost field into
16461 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
16462 field into vec_int_stmt_cost and vec_fp_stmt_cost.
16463 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
16464 splitting of scalar_stmt_cost and vec_stmt_cost.
16465 (thunderx_vector_cost): Likewise.
16466 (cortexa57_vector_cost): LIkewise.
16467 (exynosm1_vector_cost): Likewise.
16468 (xgene1_vector_cost): Likewise.
16469 (thunderx2t99_vector_cost): Improve after the splitting of the two
16471 (aarch64_builtin_vectorization_cost): Update for the splitting of
16472 scalar_stmt_cost and vec_stmt_cost.
16474 2017-02-01 Torvald Riegel <triegel@redhat.com>
16475 Richard Henderson <rth@redhat.com>
16477 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
16478 conditional on existance of a fast atomic load.
16479 * optabs-query.c (can_atomic_load_p): New function.
16480 * optabs-query.h (can_atomic_load_p): Declare it.
16481 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
16482 no fast atomic load is available for the particular size of access.
16483 (expand_atomic_compare_and_swap): Likewise.
16484 (expand_atomic_load): Likewise.
16485 (expand_atomic_store): Likewise.
16486 (expand_atomic_fetch_op): Likewise.
16487 * testsuite/lib/target-supports.exp
16488 (check_effective_target_sync_int_128): Remove x86 because it provides
16489 no fast atomic load.
16490 (check_effective_target_sync_int_128_runtime): Likewise.
16492 2017-02-01 Richard Biener <rguenther@suse.de>
16494 * graphite.c: Include tree-vectorizer.h for find_loop_location.
16495 (graphite_transform_loops): Provide opt-info for optimized nests.
16496 * tree-parloop.c (parallelize_loops): Provide opt-info for
16497 parallelized loops.
16499 2017-02-01 Richard Biener <rguenther@suse.de>
16501 PR middle-end/79315
16502 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
16503 was not set before.
16505 2017-02-01 Richard Biener <rguenther@suse.de>
16507 PR tree-optimization/71824
16508 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
16509 Verify the loops are valid in the merged SESE region.
16510 (scop_detection::can_represent_loop_1): Check analyzing the
16511 evolution of the number of iterations in the region succeeds.
16513 2017-01-31 Ian Lance Taylor <iant@golang.org>
16515 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
16516 REG_ARGS_SIZE note to 32-bit push insns and call insn.
16518 2017-01-31 David Malcolm <dmalcolm@redhat.com>
16520 PR preprocessor/79210
16521 * input.c (get_substring_ranges_for_loc): Replace line_width
16522 assertion with error-handling.
16524 2017-01-31 Richard Biener <rguenther@suse.de>
16526 PR tree-optimization/77318
16527 * graphite-sese-to-poly.c (extract_affine): Fix assert.
16528 (create_pw_aff_from_tree): Take loop parameter.
16529 (add_condition_to_pbb): Pass loop of the condition to
16530 create_pw_aff_from_tree.
16532 2017-01-31 Jakub Jelinek <jakub@redhat.com>
16534 * config/s390/s390.c (s390_asan_shadow_offset): New function.
16535 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
16537 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
16541 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
16543 (convert_int_to_float128): Likewise.
16544 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
16545 (convert_int_to_float128): Likewise.
16546 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
16547 (UNSPEC_IEEE128_CONVERT): Likewise.
16548 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
16549 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
16550 Use local variables for IBM extended format.
16551 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
16552 (fix_trunc<mode>si2_fprs): Likewise.
16553 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
16554 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
16555 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
16556 to know that we can now have integers of all sizes in vector
16558 (fix<uns>_<mode>di2_hw): Likewise.
16559 (float<uns>_<mode>si2_hw): Likewise.
16560 (fix_<mode>si2_hw): Likewise.
16561 (fixuns_<mode>si2_hw): Likewise.
16562 (float<uns>_<mode>di2_hw): Likewise.
16563 (float_<mode>di2_hw): Likewise.
16564 (float_<mode>si2_hw): Likewise.
16565 (floatuns_<mode>di2_hw): Likewise.
16566 (floatuns_<mode>si2_hw): Likewise.
16567 (xscvqp<su>wz_<mode>): Delete, no longer used.
16568 (xscvqp<su>dz_<mode>): Likewise.
16569 (xscv<su>dqp_<mode>): Likewise.
16570 (ieee128_mfvsrd_64bit): Likewise.
16571 (ieee128_mfvsrd_32bit): Likewise.
16572 (ieee128_mfvsrwz): Likewise.
16573 (ieee128_mtvsrw): Likewise.
16574 (ieee128_mtvsrd_64bit): Likewise.
16575 (ieee128_mtvsrd_32bit): Likewise.
16577 2017-01-31 Martin Liska <mliska@suse.cz>
16580 * ipa-prop.c (ipa_free_all_node_params): Call release method
16581 instead of ~sumbol_summary to not to trigger double times
16584 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
16586 PR tree-optimization/71691
16587 * bitmap.h (class auto_bitmap): New.
16588 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
16589 is_maybe_undefined instead of ssa_undefined_value_p.
16591 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16593 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
16594 __S390_ARCH_LEVEL__ to __ARCH__.
16596 2017-01-31 Jakub Jelinek <jakub@redhat.com>
16598 PR tree-optimization/79267
16599 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
16600 if should_remove_lhs_p is true.
16602 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
16605 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
16606 (add_alignment_attribute): New.
16607 (base_type_die): Add alignment attribute.
16608 (subrange_type_die): Likewise.
16609 (modified_type_die): Likewise.
16610 (gen_array_type_die): Likewise.
16611 (gen_descr_array_type_die: Likewise.
16612 (gen_enumeration_type_die): Likewise.
16613 (gen_subprogram_die): Likewise.
16614 (gen_variable_die): Likewise.
16615 (gen_field_die): Likewise.
16616 (gen_ptr_to_mbr_type_die): Likewise.
16617 (gen_struct_or_union_type_die): Likewise.
16618 (gen_subroutine_type_die): Likewise.
16619 (gen_typedef_die): Likewise.
16620 (base_type_cmp): Compare alignment attribute.
16622 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16625 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
16626 (setb_unsigned) New pattern for setb with CCUNS.
16627 * config/rs6000/rs6000.c (expand_block_compare): Use a different
16628 subfc./subfe sequence to avoid overflow problems. Generate a
16629 shorter sequence with cmpld/setb for power9.
16630 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
16631 for generating subfc. instruction.
16632 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
16633 now uses this instruction.
16635 2017-01-30 Ian Lance Taylor <iant@google.com>
16638 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
16639 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
16641 2017-01-30 Martin Sebor <msebor@redhat.com>
16643 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
16644 Move constant to the right of a relational operator.
16645 (get_mpfr_format_length, format_character, format_string): Ditto.
16646 (should_warn_p, maybe_warn): Same.
16648 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
16650 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
16653 * asan.c (get_translation_unit_decl): Remove function.
16654 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
16656 2017-01-30 Martin Liska <mliska@suse.cz>
16658 PR gcov-profile/79259
16659 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
16660 -fprofile-generate.
16662 2017-01-30 Martin Liska <mliska@suse.cz>
16665 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
16666 Initialize variables with NULL value.
16668 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
16671 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
16673 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
16675 2017-01-30 Richard Biener <rguenther@suse.de>
16677 PR tree-optimization/79276
16678 * tree-vrp.c (process_assert_insertions): Properly adjust common
16679 when removing a duplicate.
16681 * gcc.dg/torture/pr79276.c: New testcase.
16683 2017-01-30 Richard Biener <rguenther@suse.de>
16685 PR tree-optimization/79256
16686 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
16687 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
16689 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
16691 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
16694 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
16695 ("*r<noxa>sbg_<mode>_sll_bitmask")
16696 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
16697 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
16698 Use contiguous_bitmask_nowrap_operand.
16700 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16703 * config/rs6000/altivec.h (vec_xl): Revise #define.
16704 (vec_xst): Likewise.
16706 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
16708 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
16710 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
16712 PR rtl-optimization/79194
16713 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
16714 traps before call to bypass_conditional_jumps.
16716 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16718 PR tree-optimization/71374
16719 * lra-constraints.c (check_conflict_input_operands): New.
16720 (match_reload): Use it.
16722 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16725 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
16726 account to calculate conflict_set.
16728 2017-01-27 Bin Cheng <bin.cheng@arm.com>
16730 PR rtl-optimization/78559
16731 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
16732 other_insn in combine.
16734 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16736 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
16737 uint16_type_node for BT_UINT16.
16739 2017-01-27 David Malcolm <dmalcolm@redhat.com>
16741 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
16742 "RTL Tests" to menu.
16743 (GIMPLE Tests): New node.
16744 (RTL Tests): New node.
16746 2017-01-27 Richard Biener <rguenther@suse.de>
16748 PR tree-optimization/79245
16749 * tree-loop-distribution.c (distribute_loop): Apply cost
16750 modeling also to detected patterns.
16752 2017-01-27 Richard Biener <rguenther@suse.de>
16754 PR tree-optimization/71433
16755 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
16756 (compare_assert_loc): New function.
16757 (process_assert_insertions): Sort and optimize assert locations
16758 to remove duplicates and push down identical assertions on
16759 edges to their destination block.
16761 2017-01-27 Richard Biener <rguenther@suse.de>
16763 PR tree-optimization/79244
16764 * tree-vrp.c (remove_range_assertions): Forcefully propagate
16765 out SSA names even if abnormal.
16767 2017-01-27 Jakub Jelinek <jakub@redhat.com>
16769 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
16770 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
16771 instead of MPFR_RNDN.
16773 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
16776 * arm.c (arm_option_override): Don't call build_target_option_node
16777 until after doing all option overrides.
16778 (arm_valid_target_attribute_tree): Likewise.
16780 2017-01-27 Martin Liska <mliska@suse.cz>
16782 * doc/invoke.texi (-fprofile-arcs): Document profiling support
16783 for {cd}tors and C++ {cd}tors.
16785 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
16787 * config/s390/s390.md ("*setmem_long_and")
16788 ("*setmem_long_and_31z"): Use zero_extend instead of and.
16790 2017-01-26 Martin Sebor <msebor@redhat.com>
16792 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
16795 2017-01-26 Martin Sebor <msebor@redhat.com>
16797 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
16798 HAVE_DFmode before using XFmode or DFmode.
16799 (parse_directive): Avoid using the z length modifier to avoid
16800 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
16802 PR middle-end/78703
16803 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
16804 to accept adjustment as an array.
16805 (get_int_range): New function.
16806 (struct directive): Make width and prec arrays.
16807 (directive::set_width, directive::set_precision): Call get_int_range.
16808 (format_integer, format_floating): Handle width and precision ranges.
16809 (format_string, parse_directive): Same.
16811 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16814 * dwarf2out.c (generate_skeleton_bottom_up): For children with
16815 comdat_type_p set, just clone them, but keep the children in the
16819 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
16820 which have direct callers with -fvar-tracking-assignments enabled
16822 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
16823 inside of type units.
16825 2017-01-26 Martin Sebor <msebor@redhat.com>
16827 PR middle-end/78703
16828 * gimple-ssa-sprintf.c (struct result_range): Add likely and
16830 (struct format_result): Replace number_chars, number_chars_min,
16831 and number_chars_max with a single member of struct result_range.
16833 (format_result::operator+=): Adjust.
16834 (struct fmtresult): Remove bounded. Handle likely and unlikely
16836 (fmtresult::adjust_for_width_or_precision): New function.
16837 (fmtresult:type_max_digits): New function.
16838 (bytes_remaining): Handle likely and unlikely counters.
16839 (min_bytes_remaining): Remove.
16840 (format_percent): Simplify.
16841 (format_integer, format_floating): Set likely and unlikely counters.
16842 (get_string_length, format_character, format_string): Same.
16843 (format_plain, should_warn_p): New function.
16844 (maybe_warn): Call should_warn_p. Update diagnostic messages
16845 and handle those for all directives, including plain strings.
16846 (format_directive): Handle likely and unlikely counters.
16847 Remove unnecessary quoting from diagnostics. Add an informational
16849 (add_bytes): Remove.
16850 (pass_sprintf_length::compute_format_length): Simplify.
16851 (try_substitute_return_value): Handle likely and unlikely counters.
16853 2017-01-26 Carl Love <cel@us.ibm.com>
16855 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
16856 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
16858 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
16861 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
16862 endianess for subregs into account.
16863 * lra-constraints.c (lra_constraints): Do risky transformations
16864 always on the first iteration.
16865 * lra-lives.c (check_pseudos_live_through_calls): Add arg
16866 last_call_used_reg_set.
16867 (process_bb_lives): Define and use last_call_used_reg_set.
16868 * lra.c (lra): Always continue after lra_constraints on the first
16871 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
16873 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
16875 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
16877 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16879 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
16880 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
16881 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
16882 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
16883 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
16884 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
16885 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
16886 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
16887 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
16889 2017-01-26 Marek Polacek <polacek@redhat.com>
16892 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
16893 for the third operand.
16895 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16897 PR middle-end/79236
16898 * omp-low.c (struct omp_context): Add simt_stmt field.
16899 (scan_omp_for): Return omp_context *.
16900 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
16901 context to the _simt_ SIMD stmt.
16902 (lower_omp_for): For combined SIMD with sibling _simt_
16903 SIMD, make sure to use the same decls in _looptemp_
16904 clauses as in the sibling.
16906 2017-01-26 David Sherwood <david.sherwood@arm.com>
16908 PR middle-end/79212
16909 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
16912 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16915 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
16916 emit fld b; fld a; if possible.
16918 * brig-builtins.def: Update copyright years.
16919 * config/arm/arm_acle_builtins.def: Update copyright years.
16921 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16924 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
16925 constraint instead of o for the stxsd instruction.
16927 2017-01-25 Carl Love <cel@us.ibm.com>
16929 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
16930 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
16932 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
16934 * doc/invoke.texi (C++ Dialect Options): Fix typo.
16936 2017-01-25 Richard Biener <rguenther@suse.de>
16938 PR tree-optimization/69264
16939 * target.def (vector_alignment_reachable): Improve documentation.
16940 * doc/tm.texi: Regenerate.
16941 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
16943 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
16944 earlier changes with respect to TYPE_USER_ALIGN.
16945 (vector_alignment_reachable_p): Likewise. Improve dumping.
16947 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16950 * config/arm/arm.md (xordi3): Force constant operand into a register
16953 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16955 * doc/invoke.texi (-fstore-merging): Correct default optimization
16956 levels at which it is enabled.
16957 (-O): Move -fstore-merging from list to...
16960 2017-01-25 Richard Biener <rguenther@suse.de>
16963 * omp-expand.c: Include debug.h.
16964 (expand_omp_taskreg): Make sure to generate early debug before
16965 outlining anything from a function.
16966 (expand_omp_target): Likewise.
16967 (grid_expand_target_grid_body): Likewise.
16969 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
16972 * asan.c (get_translation_unit_decl): New function.
16973 (asan_add_global): Extract modules file name from globals
16974 TRANSLATION_UNIT_DECL name.
16976 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
16979 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
16980 for long calls with APCS frame and VFP.
16982 2017-01-24 David Malcolm <dmalcolm@redhat.com>
16984 * cfg.c (original_copy_tables_initialized_p): New function.
16985 * cfg.h (original_copy_tables_initialized_p): New decl.
16986 * cfgrtl.c (relink_block_chain): Guard the call to
16987 free_original_copy_tables with a call to
16988 original_copy_tables_initialized_p.
16989 * cgraph.h (symtab_node::native_rtl_p): New decl.
16990 * cgraphunit.c (symtab_node::native_rtl_p): New function.
16991 (symtab_node::needed_p): Don't assert for early assembly output
16992 for __RTL functions.
16993 (cgraph_node::finalize_function): Set "force_output" for __RTL
16995 (cgraph_node::analyze): Bail out early for __RTL functions.
16996 (analyze_functions): Update assertion to support __RTL functions.
16997 (cgraph_node::expand): Bail out early for __RTL functions.
16998 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
17000 * function.h (struct function): Update comment for field
17002 * gimple-expr.c: Include "tree-pass.h".
17003 (gimple_has_body_p): Return false for __RTL functions.
17004 * Makefile.in (OBJS): Add run-rtl-passes.o.
17005 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
17007 (gcc::pass_manager::get_clean_slate): New accessor.
17008 * passes.c: Include "insn-addr.h".
17009 (should_skip_pass_p): Add logging. Update logic for running
17010 "expand" to be compatible with both __GIMPLE and __RTL. Guard
17011 property-provider override so it is only done for gimple passes.
17013 (skip_pass): New function.
17014 (execute_one_pass): Call skip_pass when skipping passes.
17015 * read-md.c (md_reader::read_char): Support filtering
17016 the input to a subset of line numbers.
17017 (md_reader::md_reader): Initialize fields
17018 m_first_line and m_last_line.
17019 (md_reader::read_file_fragment): New function.
17020 * read-md.h (md_reader::read_file_fragment): New decl.
17021 (md_reader::m_first_line): New field.
17022 (md_reader::m_last_line): New field.
17023 * read-rtl-function.c (function_reader::create_function): Only
17024 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
17025 curr_properties. Set DECL_INITIAL to a dummy block.
17026 (read_rtl_function_body_from_file_range): New function.
17027 * read-rtl-function.h (read_rtl_function_body_from_file_range):
17029 * run-rtl-passes.c: New file.
17030 * run-rtl-passes.h: New file.
17032 2017-01-24 Jeff Law <law@redhat.com>
17034 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
17037 2017-01-24 Bin Cheng <bin.cheng@arm.com>
17039 PR tree-optimization/79159
17040 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
17041 (record_nonwrapping_iv): Improve boundary using above function if no
17042 value range information.
17044 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
17045 Martin Jambor <mjambor@suse.cz>
17047 * brig-builtins.def: New file.
17048 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
17049 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
17050 (DEF_HSAIL_SAT_BUILTIN): Likewise.
17051 (DEF_HSAIL_INTR_BUILTIN): Likewise.
17052 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
17053 * builtin-types.def (BT_INT8): New.
17054 (BT_INT16): Likewise.
17055 (BT_UINT8): Likewise.
17056 (BT_UINT16): Likewise.
17057 (BT_FN_ULONG): Likewise.
17058 (BT_FN_UINT_INT): Likewise.
17059 (BT_FN_UINT_ULONG): Likewise.
17060 (BT_FN_UINT_LONG): Likewise.
17061 (BT_FN_UINT_PTR): Likewise.
17062 (BT_FN_ULONG_PTR): Likewise.
17063 (BT_FN_INT8_FLOAT): Likewise.
17064 (BT_FN_INT16_FLOAT): Likewise.
17065 (BT_FN_UINT32_FLOAT): Likewise.
17066 (BT_FN_UINT16_FLOAT): Likewise.
17067 (BT_FN_UINT8_FLOAT): Likewise.
17068 (BT_FN_UINT64_FLOAT): Likewise.
17069 (BT_FN_UINT16_UINT32): Likewise.
17070 (BT_FN_UINT32_UINT16): Likewise.
17071 (BT_FN_UINT16_UINT16_UINT16): Likewise.
17072 (BT_FN_INT_PTR_INT): Likewise.
17073 (BT_FN_UINT_PTR_UINT): Likewise.
17074 (BT_FN_LONG_PTR_LONG): Likewise.
17075 (BT_FN_ULONG_PTR_ULONG): Likewise.
17076 (BT_FN_VOID_UINT64_UINT64): Likewise.
17077 (BT_FN_UINT8_UINT8_UINT8): Likewise.
17078 (BT_FN_INT8_INT8_INT8): Likewise.
17079 (BT_FN_INT16_INT16_INT16): Likewise.
17080 (BT_FN_INT_INT_INT): Likewise.
17081 (BT_FN_UINT_FLOAT_UINT): Likewise.
17082 (BT_FN_FLOAT_UINT_UINT): Likewise.
17083 (BT_FN_ULONG_UINT_UINT): Likewise.
17084 (BT_FN_ULONG_UINT_PTR): Likewise.
17085 (BT_FN_ULONG_ULONG_ULONG): Likewise.
17086 (BT_FN_UINT_UINT_UINT): Likewise.
17087 (BT_FN_VOID_UINT_PTR): Likewise.
17088 (BT_FN_UINT_UINT_PTR: Likewise.
17089 (BT_FN_UINT32_UINT64_PTR): Likewise.
17090 (BT_FN_INT_INT_UINT_UINT): Likewise.
17091 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
17092 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
17093 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
17094 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
17095 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
17096 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
17097 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
17098 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
17099 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
17100 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
17101 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
17102 * doc/frontends.texi: List BRIG FE.
17103 * doc/install.texi (Testing): Add BRIG tesring requirements.
17104 * doc/invoke.texi (Overall Options): Mention BRIG.
17105 * doc/standards.texi (Standards): Doucment BRIG HSA version.
17107 2017-01-24 Richard Biener <rguenther@suse.de>
17109 PR translation/79208
17110 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
17112 2017-01-24 Martin Jambor <mjambor@suse.cz>
17115 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
17116 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
17117 and known_contexts.
17119 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
17121 PR middle-end/79123
17122 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
17123 casts from signed to unsigned really don't have a range.
17125 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
17127 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
17128 GMP_RNDx for compatiblity.
17130 2017-01-24 Martin Liska <mliska@suse.cz>
17133 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
17134 that would prevent us to call alloca with -1 as argument.
17136 2017-01-24 Jakub Jelinek <jakub@redhat.com>
17138 * dwarf2out.c (output_compilation_unit_header, output_file_names):
17139 Avoid -Wformat-security warning.
17141 2017-01-23 Andrew Pinski <apinski@cavium.com>
17143 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
17146 2017-01-23 Martin Sebor <msebor@redhat.com>
17148 PR middle-end/78703
17149 * gimple-ssa-sprintf.c (warn_level): New global.
17150 (format_integer): Use it here and throughout the rest of the file.
17151 Use the same switch to compute sign as base.
17152 (maybe_warn): New function.
17153 (format_directive): Factor out warnings into maybe_warn.
17154 Add debugging output. Use warn_level.
17155 (add_bytes): Use warn_level.
17156 (pass_sprintf_length::compute_format_length): Add debugging output.
17157 (try_substitute_return_value): Same.
17158 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
17160 PR middle-end/78703
17161 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
17162 (struct fmtresult, format_integer, format_floating): Adjust.
17163 (fmtresult::fmtresult): Set max correctly in two argument ctor.
17164 (get_string_length, format_string,format_directive): Same.
17165 (pass_sprintf_length::compute_format_length): Same.
17166 (try_substitute_return_value): Simplify slightly.
17168 PR middle-end/78703
17169 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
17170 (fmtresult::operator+=): Outlined.
17171 (struct fmtresult): Add ctors.
17172 (struct conversion_spec): Rename...
17173 (struct directive): ...to this. Add and remove data members.
17174 (directive::set_width, directive::set_precision): New functions.
17175 (format_percent): Use fmtresult ctor.
17176 (get_width_and_precision): Remove.
17177 (format_integer): Make naming changes. Avoid computing width and
17179 (format_floating): Same. Adjust indentation.
17180 (format_character, format_none): New functions.
17181 (format_string): Moved character handling to format_character.
17182 (format_directive): Remove arguments, change return type.
17183 (parse_directive): New function.
17184 (pass_sprintf_length::compute_format_length): Move directive
17185 parsing to parse_directive.
17187 2017-01-23 Jakub Jelinek <jakub@redhat.com>
17189 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
17190 (assign_assembler_name_if_needed): ... this.
17191 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
17192 (assign_assembler_name_if_needed): ... this.
17193 (free_lang_data_in_cgraph): Adjust callers.
17194 * cgraphunit.c (cgraph_node::analyze): Likewise.
17195 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
17198 2017-01-23 Richard Biener <rguenther@suse.de>
17200 PR tree-optimization/79088
17201 PR tree-optimization/79188
17202 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
17203 resetting loop bounds after last path deletion. Reset loop
17204 bounds of the target loop, make code match the comments.
17205 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
17206 Make sure loops need no fixups.
17208 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
17210 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
17211 exponent support with double type for first argument.
17212 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
17213 type returned by __builtin_vec_extract_sig,
17214 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
17215 functions from "vector int" to "vector unsigned int" or from
17216 "vector long long int" to "vector unsigned long long int".
17217 Changed type returned by __builtin_vec_extract_exp,
17218 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
17219 functions from "vector int" to "vector unsigned int" or from
17220 "vector long long int" to "vector unsigned long long int".
17221 Changed return type of __builtin_vec_test_data_class,
17222 __builtin_vec_test_data_class_sp, and
17223 __builtin_vec_test_data_class_dp from "vector int" to
17224 "vector bool int" or from "vector long long int" to "vector bool
17225 long long int" and changed second argument type from "unsigned
17226 int" to "int". Added new overloaded function forms "vector float
17227 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
17228 "vector float __builtin_vec_insert_exp_sp (vector float, vector
17229 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
17230 double, vector unsigned long long int)" and "vector double
17231 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
17232 long int)". Changed return type of
17233 __builtin_scalar_test_data_class and
17234 __builtin_scalar_test_data_class_sp and
17235 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
17236 int" and changed second argument from "unsigned int" to "int".
17237 Changed type returned by __builtin_scalar_test_neg,
17238 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
17239 from "int" to "bool int". Added new overloaded function form
17240 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
17241 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
17242 exponent double-precision with floating point first argument.
17243 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
17244 documentation of scalar_test_data_class, scalar_test_neg,
17245 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
17246 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
17247 vec_test_data_class built-in functions to reflect refinements in
17248 their type signatures.
17250 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
17252 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
17254 (aarch64_elf_asm_destructor): Likewise.
17256 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
17258 PR rtl-optimization/78634
17259 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
17260 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
17261 * ifcvt.c (noce_try_cmove): Add missing cost check.
17263 PR rtl-optimization/71724
17264 * combine.c (if_then_else_cond): Look for situations where it is
17265 beneficial to undo the work of one of the recursive calls.
17267 2017-01-23 Bin Cheng <bin.cheng@arm.com>
17269 PR tree-optimization/70754
17270 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
17271 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
17272 combined stmt before it if not NULL.
17273 (combine_chains): Process refs reversely and compute dominance point
17276 2017-01-23 Martin Liska <mliska@suse.cz>
17278 PR tree-optimization/79196
17279 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
17280 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
17282 (strlen_optimize_stmt): Call the renamed function.
17284 2017-01-23 Michael Matz <matz@suse.de>
17286 PR tree-optimization/78384
17287 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
17289 2017-01-23 Richard Biener <rguenther@suse.de>
17291 PR tree-optimization/79186
17292 * tree-vrp.c (register_new_assert_for): Make sure we've seen
17293 both incoming edges before moving an assert.
17295 2017-01-23 Martin Jambor <mjambor@suse.cz>
17297 * ipa-prop.c (load_from_param_1): Removed.
17298 (load_from_unmodified_param): Bits from load_from_param_1 put back
17300 (load_from_param): Removed.
17301 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
17302 with stmt. Reverted back to use of load_from_unmodified_param.
17304 2017-01-23 Martin Jambor <mjambor@suse.cz>
17307 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
17308 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
17309 field a pointer to garbage collected vector, mark lattices and
17310 ipcp_orig_node with GTY((skip)).
17311 (ipa_get_param_count): Adjust to descriptors being a pointer.
17312 (ipa_get_param): Likewise.
17313 (ipa_get_type): Likewise.
17314 (ipa_get_param_move_cost): Likewise.
17315 (ipa_set_param_used): Likewise.
17316 (ipa_get_controlled_uses): Likewise.
17317 (ipa_set_controlled_uses): Likewise.
17318 (ipa_is_param_used): Likewise.
17319 (ipa_node_params_t): Move into garbage collector. New methods insert
17321 (ipa_node_params_sum): Annotate wth GTY(()).
17322 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
17324 (ipa_load_from_parm_agg): Adjust declaration.
17325 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
17326 * ipa-profile.c (ipa_profile): Likewise.
17327 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
17328 (ipa_populate_param_decls): Make descriptors parameter garbage
17330 (ipa_dump_param): Adjust to descriptors being a pointer.
17331 (ipa_alloc_node_params): Likewise.
17332 (ipa_initialize_node_params): Likewise.
17333 (load_from_param_1): Make descriptors parameter garbage collected.
17334 (load_from_unmodified_param): Likewise.
17335 (load_from_param): Likewise.
17336 (ipa_load_from_parm_agg): Likewise.
17337 (ipa_node_params::~ipa_node_params): Removed.
17338 (ipa_free_all_node_params): Remove call to delete operator.
17339 (ipa_node_params_t::insert): New.
17340 (ipa_node_params_t::remove): Likewise.
17341 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
17342 copy known_csts and known_contexts vectors.
17343 (ipa_read_node_info): Adjust to descriptors being a pointer.
17344 (ipcp_modif_dom_walker): Make m_descriptors field garbage
17346 (ipcp_transform_function): Make descriptors variable garbage
17349 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
17351 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
17352 * config/i386/avx512dqintrin.h: Ditto.
17353 * config/i386/avx512fintrin.h: Ditto.
17354 * config/i386/i386.c: Handle new builtins.
17355 * config/i386/i386-builtin.def: Add new builtins.
17356 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
17357 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
17359 2017-01-23 Jakub Jelinek <jakub@redhat.com>
17360 Martin Liska <mliska@suse.cz>
17362 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
17363 * asan.c (asan_expand_poison_ifn): Support stores and use
17364 appropriate ASAN report function.
17365 * internal-fn.c (expand_ASAN_POISON_USE): New function.
17366 * internal-fn.def (ASAN_POISON_USE): Declare.
17367 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
17368 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
17369 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
17370 ASAN_POISON calls w/o LHS.
17371 * tree-ssa.c (execute_update_addresses_taken): Create clobber
17372 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
17373 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
17374 * gimplify.c (asan_poison_variables): Add attribute
17375 use_after_scope_memory to variables that really needs to live
17377 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
17378 having the attribute.
17380 2017-01-23 Martin Liska <mliska@suse.cz>
17382 * asan.c (create_asan_shadow_var): New function.
17383 (asan_expand_poison_ifn): Likewise.
17384 * asan.h (asan_expand_poison_ifn): New declaration.
17385 * internal-fn.c (expand_ASAN_POISON): Likewise.
17386 * internal-fn.def (ASAN_POISON): New builtin.
17387 * sanopt.c (pass_sanopt::execute): Expand
17388 asan_expand_poison_ifn.
17389 * tree-inline.c (copy_decl_for_dup_finish): Make function
17391 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
17392 * tree-ssa.c (is_asan_mark_p): New function.
17393 (execute_update_addresses_taken): Rewrite local variables
17394 (identified just by use-after-scope as addressable) into SSA.
17396 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
17398 * doc/install.texi (Specific): opensource.apple.com uses https
17399 now. Remove trailing slash.
17401 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
17403 * README.Portability: Remove note on an Irix compatibility issue.
17405 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
17407 * gcov.c (INCLUDE_ALGORITHM): Define.
17408 (INCLUDE_VECTOR): Define.
17409 No longer include <vector> and <algorithm> directly.
17411 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
17413 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
17415 * doc/invoke.texi (Code Gen Options): Ditto.
17417 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
17420 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
17422 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
17424 rtl-optimization/79125
17425 * cprop.c (local_cprop_pass): Handle cases where we make an
17426 unconditional trap.
17428 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
17432 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
17433 read from, for big endian.
17435 2017-01-20 Jiong Wang <jiong.wang@arm.com>
17437 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
17438 register pauth builtins for LP64 only.
17440 2017-01-20 Marek Polacek <polacek@redhat.com>
17443 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
17446 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
17448 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
17450 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
17451 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
17452 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
17454 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17457 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
17458 in big-endian mode when they are not a single duplicated value.
17460 2017-01-20 Richard Biener <rguenther@suse.de>
17462 * BASE-VER: Bump to 7.0.1.
17464 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
17466 * omp-low.c (omplow_simd_context): New struct. Use it...
17467 (lower_rec_simd_input_clauses): ...here and...
17468 (lower_rec_input_clauses): ...here to hold common data. Adjust all
17469 references to idx, lane, max_vf, is_simt.
17471 2017-01-20 Graham Markall <graham.markall@embecosm.com>
17473 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
17476 2017-01-20 Martin Jambor <mjambor@suse.cz>
17478 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
17479 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
17481 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
17482 (GTFILES): Rename hsa.c to hsa-common.c.
17483 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
17484 * hsa-dump.c: Likewise.
17485 * hsa-gen.c: Likewise.
17486 * hsa-regalloc.c: Likewise.
17487 * ipa-hsa.c: Likewise.
17488 * omp-expand.c: Likewise.
17489 * omp-low.c: Likewise.
17490 * toplev.c: Likewise.
17492 2017-01-20 Marek Polacek <polacek@redhat.com>
17495 * doc/invoke.texi: Document -Wduplicated-branches.
17496 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
17497 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
17498 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
17499 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
17500 return 0 only when not OEP_LEXICOGRAPHIC.
17501 (fold_build_cleanup_point_expr): Use the expression
17502 location when building CLEANUP_POINT_EXPR.
17503 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
17504 * tree.c (add_expr): Handle error_mark_node.
17506 2017-01-20 Martin Liska <mliska@suse.cz>
17509 * tree-profile.c (init_ic_make_global_vars): Do not call
17511 (gimple_init_gcov_profiler): Likewise.
17513 2017-01-20 Martin Liska <mliska@suse.cz>
17516 * cgraph.h (maybe_create_reference): Remove argument and
17518 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
17520 * ipa-cp.c (create_specialized_node): Likewise.
17521 * symtab.c (symtab_node::maybe_create_reference): Handle
17522 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
17524 2017-01-20 Martin Liska <mliska@suse.cz>
17526 * read-rtl-function.c (function_reader::create_function): Use
17527 build_decl instread of build_decl_stat.
17529 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
17531 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
17532 * config/i386/avx512dqintrin.h: Ditto.
17533 * config/i386/avx512fintrin.h: Ditto.
17534 * config/i386/i386-builtin-types.def: Add new types.
17535 * config/i386/i386.c: Handle new types.
17536 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
17537 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
17538 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
17539 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
17540 (__builtin_ia32_kshiftridi): New.
17541 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
17543 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
17547 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
17548 define to rs6000_init_stack_protect_guard.
17549 (rs6000_init_stack_protect_guard): New function.
17551 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
17552 Yunqiang Su <yunqiang.su@imgtec.com>
17554 * config.gcc (supported_defaults): Add madd4.
17555 (with_madd4): Add validation.
17556 (all_defaults): Add madd4.
17557 * config/mips/mips.opt (mmadd4): New option.
17558 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17560 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17562 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
17563 (ISA_HAS_FUSED_MADD4): Likewise.
17564 * doc/invoke.texi (-mmadd4): Document the new option.
17565 * doc/install.texi (--with-madd4): Document the new option.
17567 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17569 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
17570 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
17571 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
17572 (aarch64_init_pauth_hint_builtins): New.
17573 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
17574 (aarch64_expand_builtin): Expand new builtins.
17576 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17578 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
17579 * combine-stack-adj.c (no_unhandled_cfa): Handle
17580 REG_CFA_TOGGLE_RA_MANGLE.
17581 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
17582 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
17583 info for return address signing.
17584 (aarch64_expand_epilogue): Likewise.
17586 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17588 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
17589 * config/aarch64/aarch64-protos.h
17590 (aarch64_return_address_signing_enabled): New declaration.
17591 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
17593 (aarch64_expand_prologue): Sign return address before it's pushed onto
17595 (aarch64_expand_epilogue): Authenticate return address fetched from
17597 (aarch64_override_options): Sanity check for ILP32 and ISA level.
17598 (aarch64_attributes): New function attributes for "sign-return-address".
17599 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
17600 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
17601 ("*do_return"): Generate combined instructions according to key index.
17602 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
17603 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
17605 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
17606 * config/aarch64/aarch64.opt (msign-return-address=): New.
17607 * doc/extend.texi (AArch64 Function Attributes): Documents
17608 "sign-return-address=".
17609 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
17611 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
17613 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
17614 overall option summary.
17616 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17618 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
17619 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
17620 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
17621 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
17623 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
17625 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
17626 -mpower9-minmax by default for -mcpu=power9.
17627 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
17628 128-bit floating point.
17630 2017-01-20 Alan Modra <amodra@gmail.com>
17632 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
17633 optimizing for size.
17635 2017-01-20 Alan Modra <amodra@gmail.com>
17638 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
17639 for strcmp and strncmp from corresponding builtin decl.
17641 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17643 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
17644 instead of i386/rtems-64.h.
17645 * config/i386/rtems-64.h: Remove.
17647 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17651 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
17653 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
17655 2017-01-19 Tamar Christina <tamar.christina@arm.com>
17657 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
17658 Change int to HOST_WIDE_INT.
17659 * config/aarch64/aarch64-protos.h
17660 (aarch64_simd_gen_const_vector_dup): Likewise.
17661 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
17663 2017-01-19 David Malcolm <dmalcolm@redhat.com>
17665 * langhooks-def.h (lhd_type_for_size): New decl.
17666 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
17667 * langhooks.c (lhd_type_for_size): New function, taken from
17670 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
17672 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
17673 define_bypass for CR latency.
17674 (power9-cracked-alu): Update bypass latency and remove power9-branch.
17675 (power9-alu2): Add define_bypass for CR latency.
17677 (power9-mul): Update insn latency.
17678 (power9-mul-compare): Update insn latency, bypass latency and remove
17681 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17683 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
17685 * config/aarch64/aarch64.md
17686 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
17687 aarch64_nopcrelative_literal_loads.
17688 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
17690 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
17692 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
17693 TARGET_LOONGSON_3A.
17694 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
17696 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
17699 * config.gcc (supported_defaults): Add lxc1-sxc1.
17700 (with_lxc1_sxc1): Add validation.
17701 (all_defaults): Add lxc1-sxc1.
17702 * config/mips/mips.opt (mlxc1-sxc1): New option.
17703 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17705 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17706 __mips_no_lxc1_sxc1.
17707 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
17708 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
17709 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
17711 2017-01-19 Richard Biener <rguenther@suse.de>
17713 PR tree-optimization/72488
17714 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
17715 sure to restore SSA info.
17716 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
17718 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
17720 PR rtl-optimization/79121
17721 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
17722 of the inner type when shifting an extended value.
17724 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
17727 * symtab.c (symtab_node::equal_address_to): Fix comparing of
17728 interposable aliases.
17730 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
17733 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
17734 Use the evmergelohi instruction.
17735 (mov_si<mode>_e500_subreg4_2_le): Likewise.
17736 (mov_sitf_e500_subreg8_2_be): Likewise.
17737 (mov_sitf_e500_subreg12_2_le): Likewise.
17738 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
17739 (mov_si<mode>_e500_subreg4_2_be): Likewise.
17740 (mov_sitf_e500_subreg8_2_le): Likewise.
17741 (mov_sitf_e500_subreg12_2_be): Likewise.
17743 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17745 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
17746 attribute from vecsimple to vecperm.
17747 (altivec_vbpermq2): Likewise.
17749 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17752 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
17754 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17755 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
17756 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
17757 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
17758 case where N arg is SIZE_MAX.
17759 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
17760 (cmpstrsi): Add pattern.
17762 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17764 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17765 __builtin_vec_revb builtins.
17766 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
17767 built-in functions to support generation of the ISA 3.0 XXBR<x>
17768 vector byte reverse instructions.
17769 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
17770 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
17771 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
17772 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
17773 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
17774 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
17775 (P9V_BUILTIN_VEC_REVB): Likewise.
17776 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
17777 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
17778 (p9_xxbrq_v16qi): Likewise.
17779 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
17780 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
17781 (p9_xxbrh_v8hi): Likewise.
17782 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
17783 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
17784 vec_revb built-in functions.
17786 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
17788 PR rtl-optimization/78952
17789 * config/i386/i386.md (any_extract): New code iterator.
17790 (*insvqi_2): Use any_extract for source operand.
17791 (*insvqi_3): Use any_shiftrt for source operand.
17793 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
17795 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
17797 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
17799 2017-01-18 Matthias Klose <doko@ubuntu.com>
17801 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
17803 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17805 * config/rs6000/altivec.h (vec_bperm): Change #define.
17806 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
17807 (altivec_vbpermq2): New define_insn.
17808 (altivec_vbpermd): Likewise.
17809 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
17810 function interface.
17811 (VBPERMD): Likewise.
17812 (VBPERM): New polymorphic function interface.
17813 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
17814 Add entries for P9V_BUILTIN_VEC_VBPERM.
17815 * doc/extend.texi: Add interfaces for vec_bperm.
17817 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17819 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
17820 first letter of error messages.
17821 (s390_resolve_overloaded_builtin): Likewise.
17822 * config/s390/s390.c (s390_expand_builtin): Likewise.
17823 (s390_invalid_arg_for_unprototyped_fn): Likewise.
17824 (s390_valid_target_attribute_inner_p): Likewise.
17825 * config/s390/s390.md ("tabort"): Likewise.
17827 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
17829 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
17830 (ISA_AVOID_DIV_HILO): New macro.
17831 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
17832 (ISA_HAS_DDIV): Likewise.
17834 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17836 * doc/invoke.texi (fabi-version): Correct number of occurrences.
17838 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17840 * doc/invoke.texi (fabi-version): Spelling fix.
17842 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17845 * doc/invoke.texi (fabi-version): Mention mangling fix for
17848 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17851 * doc/invoke.texi (fabi-version): Document discriminator mangling.
17853 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
17856 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
17857 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
17859 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
17861 (stack_protect_test): Ditto.
17862 * config/rs6000/rs6000.opt (mstack-protector-guard=,
17863 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
17865 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
17866 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
17867 -mstack-protector-guard-offset=.
17868 (RS/6000 and PowerPC Options): Ditto.
17870 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
17872 * config/i386/i386.h (MASK_CLASS_P): New define.
17873 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
17874 there are no registers from different register sets also when
17875 mask registers are used. Update function comment.
17876 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
17877 to (*k/*r) and (*k/*km) alternatives.
17879 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
17881 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
17882 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
17883 (EH_RETURN_HANDLER_RTX): New define.
17884 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
17885 Force frame pointer in EH return functions.
17886 (aarch64_expand_epilogue): Add barrier for eh_return.
17887 (aarch64_final_eh_return_addr): Remove.
17888 (aarch64_eh_return_handler_rtx): New function.
17889 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
17891 (aarch64_eh_return_handler_rtx): New prototype.
17893 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17895 * config/rs6000/altivec.h (vec_rlmi): New #define.
17896 (vec_vrlnm): Likewise.
17897 (vec_rlnm): Likewise.
17898 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
17899 (UNSPEC_VRLNM): Likewise.
17900 (VIlong): New mode iterator.
17901 (altivec_vrl<VI_char>mi): New define_insn.
17902 (altivec_vrl<VI_char>nm): Likewise.
17903 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
17905 (VRLDNM): Likewise.
17906 (RLNM): New polymorphic function entry.
17907 (VRLWMI): New monomorphic function entry.
17908 (VRLDMI): Likewise.
17909 (RLMI): New polymorphic function entry.
17910 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
17911 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
17912 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
17915 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17918 * dwarf2out.c (field_byte_offset): Restore the
17919 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
17920 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
17921 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
17924 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17927 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
17929 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17932 * dwarf2out.c (add_data_member_location_attribute): For constant
17933 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
17934 instead of DW_AT_data_member_location, DW_AT_bit_offset and
17935 DW_AT_byte_size attributes.
17937 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17939 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
17940 after forcing to constant memory when the code model is medium.
17942 2017-01-17 Julia Koval <julia.koval@intel.com>
17945 * config/i386/avx512fintrin.h
17946 (_mm512_i32gather_ps): Change __addr type to void const*.
17947 (_mm512_mask_i32gather_ps): Ditto.
17948 (_mm512_i32gather_pd): Ditto.
17949 (_mm512_mask_i32gather_pd): Ditto.
17950 (_mm512_i64gather_ps): Ditto.
17951 (_mm512_mask_i64gather_ps): Ditto.
17952 (_mm512_i64gather_pd): Ditto.
17953 (_mm512_mask_i64gather_pd): Ditto.
17954 (_mm512_i32gather_epi32): Ditto.
17955 (_mm512_mask_i32gather_epi32): Ditto.
17956 (_mm512_i32gather_epi64): Ditto.
17957 (_mm512_mask_i32gather_epi64): Ditto.
17958 (_mm512_i64gather_epi32): Ditto.
17959 (_mm512_mask_i64gather_epi32): Ditto.
17960 (_mm512_i64gather_epi64): Ditto.
17961 (_mm512_mask_i64gather_epi64): Ditto.
17962 (_mm512_i32scatter_ps): Change __addr type to void*.
17963 (_mm512_mask_i32scatter_ps): Ditto.
17964 (_mm512_i32scatter_pd): Ditto.
17965 (_mm512_mask_i32scatter_pd): Ditto.
17966 (_mm512_i64scatter_ps): Ditto.
17967 (_mm512_mask_i64scatter_ps): Ditto.
17968 (_mm512_i64scatter_pd): Ditto.
17969 (_mm512_mask_i64scatter_pd): Ditto.
17970 (_mm512_i32scatter_epi32): Ditto.
17971 (_mm512_mask_i32scatter_epi32): Ditto.
17972 (_mm512_i32scatter_epi64): Ditto.
17973 (_mm512_mask_i32scatter_epi64): Ditto.
17974 (_mm512_i64scatter_epi32): Ditto.
17975 (_mm512_mask_i64scatter_epi32): Ditto.
17976 (_mm512_i64scatter_epi64): Ditto.
17977 (_mm512_mask_i64scatter_epi64): Ditto.
17978 * config/i386/avx512pfintrin.h
17979 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
17980 (_mm512_mask_prefetch_i32gather_ps): Ditto.
17981 (_mm512_mask_prefetch_i64gather_pd): Ditto.
17982 (_mm512_mask_prefetch_i64gather_ps): Ditto.
17983 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
17984 (_mm512_prefetch_i32scatter_ps): Ditto.
17985 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
17986 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
17987 (_mm512_prefetch_i64scatter_pd): Ditto.
17988 (_mm512_prefetch_i64scatter_ps): Ditto.
17989 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
17990 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
17991 * config/i386/avx512vlintrin.h
17992 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
17993 (_mm_mmask_i32gather_ps): Ditto.
17994 (_mm256_mmask_i32gather_pd): Ditto.
17995 (_mm_mmask_i32gather_pd): Ditto.
17996 (_mm256_mmask_i64gather_ps): Ditto.
17997 (_mm_mmask_i64gather_ps): Ditto.
17998 (_mm256_mmask_i64gather_pd): Ditto.
17999 (_mm_mmask_i64gather_pd): Ditto.
18000 (_mm256_mmask_i32gather_epi32): Ditto.
18001 (_mm_mmask_i32gather_epi32): Ditto.
18002 (_mm256_mmask_i32gather_epi64): Ditto.
18003 (_mm_mmask_i32gather_epi64): Ditto.
18004 (_mm256_mmask_i64gather_epi32): Ditto.
18005 (_mm_mmask_i64gather_epi32): Ditto.
18006 (_mm256_mmask_i64gather_epi64): Ditto.
18007 (_mm_mmask_i64gather_epi64): Ditto.
18008 (_mm256_i32scatter_ps): Change __addr type to void*.
18009 (_mm256_mask_i32scatter_ps): Ditto.
18010 (_mm_i32scatter_ps): Ditto.
18011 (_mm_mask_i32scatter_ps): Ditto.
18012 (_mm256_i32scatter_pd): Ditto.
18013 (_mm256_mask_i32scatter_pd): Ditto.
18014 (_mm_i32scatter_pd): Ditto.
18015 (_mm_mask_i32scatter_pd): Ditto.
18016 (_mm256_i64scatter_ps): Ditto.
18017 (_mm256_mask_i64scatter_ps): Ditto.
18018 (_mm_i64scatter_ps): Ditto.
18019 (_mm_mask_i64scatter_ps): Ditto.
18020 (_mm256_i64scatter_pd): Ditto.
18021 (_mm256_mask_i64scatter_pd): Ditto.
18022 (_mm_i64scatter_pd): Ditto.
18023 (_mm_mask_i64scatter_pd): Ditto.
18024 (_mm256_i32scatter_epi32): Ditto.
18025 (_mm256_mask_i32scatter_epi32): Ditto.
18026 (_mm_i32scatter_epi32): Ditto.
18027 (_mm_mask_i32scatter_epi32): Ditto.
18028 (_mm256_i32scatter_epi64): Ditto.
18029 (_mm256_mask_i32scatter_epi64): Ditto.
18030 (_mm_i32scatter_epi64): Ditto.
18031 (_mm_mask_i32scatter_epi64): Ditto.
18032 (_mm256_i64scatter_epi32): Ditto.
18033 (_mm256_mask_i64scatter_epi32): Ditto.
18034 (_mm_i64scatter_epi32): Ditto.
18035 (_mm_mask_i64scatter_epi32): Ditto.
18036 (_mm256_i64scatter_epi64): Ditto.
18037 (_mm256_mask_i64scatter_epi64): Ditto.
18038 (_mm_i64scatter_epi64): Ditto.
18039 (_mm_mask_i64scatter_epi64): Ditto.
18040 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
18041 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
18042 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
18043 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
18044 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
18045 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
18046 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
18047 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
18048 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
18049 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
18050 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
18051 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
18052 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
18053 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
18054 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
18055 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
18056 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
18057 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
18058 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
18059 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
18060 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
18061 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
18062 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
18063 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
18064 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
18065 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
18066 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
18067 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
18068 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
18069 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
18070 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
18071 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
18072 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
18073 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
18074 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
18075 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
18076 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
18077 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
18078 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
18079 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
18080 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
18081 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
18082 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
18083 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
18084 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
18085 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
18086 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
18087 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
18088 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
18089 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
18090 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
18091 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
18092 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
18093 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
18094 definitions accordingly.
18096 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
18097 Kuan-Lin Chen <kuanlinchentw@gmail.com>
18100 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
18103 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
18106 * ira-conflicts.c (ira_build_conflicts): Update total conflict
18107 hard regs for inner regno.
18109 2017-01-17 Martin Liska <mliska@suse.cz>
18112 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
18113 assumption and add comment.
18115 2017-01-17 Nathan Sidwell <nathan@acm.org>
18117 * ipa-visibility.c (localize_node): New function, broken out of ...
18118 (function_and_variable_visibility): ... here. Call it.
18120 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
18122 PR middle-end/77445
18123 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
18124 correctly set frequency of oudgoing edge.
18125 (duplicate_thread_path): Fix profile updating.
18127 2017-01-17 Jakub Jelinek <jakub@redhat.com>
18130 * configure.ac: Add GCC_BASE_VER.
18131 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
18132 version from BASE-VER file.
18133 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
18134 (gcc.o): Depend on $(BASEVER).
18135 * common.opt (dumpfullversion): New option.
18136 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
18137 * doc/invoke.texi: Document -dumpfullversion.
18138 * doc/install.texi: Document --with-gcc-major-version-only.
18139 * configure: Regenerated.
18141 2017-01-17 Richard Biener <rguenther@suse.de>
18143 PR tree-optimization/71433
18144 * tree-vrp.c (register_new_assert_for): Merge same asserts
18145 on all incoming edges.
18146 (process_assert_insertions_for): Handle insertions at the
18149 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
18151 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
18152 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
18154 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
18157 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
18160 2017-01-17 Alan Modra <amodra@gmail.com>
18163 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
18164 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
18165 symbolic stack limit when pic.
18167 2017-01-16 Martin Sebor <msebor@redhat.com>
18169 PR tree-optimization/78608
18170 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
18172 2017-01-16 Jeff Law <law@redhat.com>
18175 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
18176 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
18177 for several include directories that may be relative to sysroot.
18178 * config/i386/x-mingw32 (gplus_includedir): Define.
18179 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
18180 (native_system_includedir): Likewise.
18181 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
18182 override if TARGET_SYSTEM_ROOT is defined.
18183 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
18185 PR tree-optimization/79090
18186 PR tree-optimization/33562
18187 PR tree-optimization/61912
18188 PR tree-optimization/77485
18189 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
18190 and computed trims into the dump file.
18192 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
18194 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
18196 2017-01-16 Jakub Jelinek <jakub@redhat.com>
18199 * gimplify.c (gimplify_init_constructor): If want_value and
18200 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
18204 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
18205 sequence. Formatting fixes.
18206 (doloop_optimize): Formatting fixes.
18209 * gcc.c (debug_level_greater_than_spec_func): New function.
18210 (static_spec_functions): Add debug-level-gt spec function.
18211 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
18213 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
18214 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
18215 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
18216 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
18217 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
18218 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
18220 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
18222 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
18223 QImode fixups to general and mask registers only.
18225 2017-01-16 Carl Love <cel@us.ibm.com>
18227 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
18228 for built-in functions
18229 vector signed char vec_nabs (vector signed char)
18230 vector signed short vec_nabs (vector signed short)
18231 vector signed int vec_nabs (vector signed int)
18232 vector signed long long vec_nabs (vector signed long long)
18233 vector float vec_nabs (vector float)
18234 vector double vec_nabs (vector double)
18235 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
18237 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
18238 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
18239 * doc/extend.texi: Update the documentation file for the new built-in
18242 2017-01-16 Martin Sebor <msebor@redhat.com>
18244 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
18247 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18249 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
18250 UNSPEC_VSX__XXSPLTD to require special splat handling.
18252 2017-01-16 David Malcolm <dmalcolm@redhat.com>
18255 * system.h: Poison strndup.
18257 2017-01-16 Alan Modra <amodra@gmail.com>
18260 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
18263 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
18265 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
18267 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
18269 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
18270 call recog here. Assert that INSN_CODE (insn) is non-negative.
18272 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
18275 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
18277 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
18278 in the currently scheduled RTL fragment.
18280 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
18282 PR rtl-optimization/78751
18283 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
18286 2017-01-14 Jeff Law <law@redhat.com>
18288 PR tree-optimization/79090
18289 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
18290 variable length stores.
18291 (compute_trims): Delete dead assignment to *trim_tail.
18292 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
18295 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
18297 PR rtl-optimization/78626
18298 PR rtl-optimization/78727
18299 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
18300 of a block, and split such blocks after everything else is finished.
18302 2017-01-14 Alan Modra <amodra@gmail.com>
18305 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
18306 target legitimate_combined_insn.
18307 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
18308 (rs6000_legitimate_combined_insn): New function.
18309 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
18311 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
18312 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
18313 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
18315 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
18317 * doc/frontends.texi (G++ and GCC): Remove references to Java.
18319 2017-01-13 Jeff Law <law@redhat.com>
18321 PR tree-optimization/33562
18322 PR tree-optimization/61912
18323 PR tree-optimization/77485
18324 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
18326 (delete_dead_assignment): Likewise.
18327 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
18328 statement to delete_dead_call and delete_dead_assignment.
18330 2017-01-13 David Malcolm <dmalcolm@redhat.com>
18333 * substring-locations.c (format_warning_va): Strengthen case 1 so
18334 that both endpoints of the substring must be within the format
18335 range for just the substring to be printed.
18337 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
18339 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
18340 * config/i386/i386.c (ix86_target_string): Add missing options
18341 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
18342 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
18343 flags_other and ix86_target_other to flags2_other. Display unknown
18345 (ix86_valid_target_attribute_inner_p): Add missing options and
18346 reorder options by implied ISAs, as in ix86_target_string.
18348 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
18350 * hash-table.h (hash_table::too_empty_p): New function.
18351 (hash_table::expand): Use it.
18352 (hash_table::traverse): Likewise.
18353 (hash_table::empty_slot): Use sizeof (value_type) instead of
18354 sizeof (PTR) to convert bytes to elements. Shrink the table
18355 if the current size is excessive for the current number of
18358 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
18360 * ira-costs.c (record_reg_classes): Break from the inner loop
18361 early once alt_fail is known to be true. Update outer loop
18362 handling accordingly.
18364 2017-01-13 Jeff Law <law@redhat.com>
18366 * tree-ssa-dse.c (decrement_count): New function.
18367 (increment_start_addr, maybe_trim_memstar_call): Likewise.
18368 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
18369 when we know the partially dead statement is a mem* function.
18371 PR tree-optimization/61912
18372 PR tree-optimization/77485
18373 * tree-ssa-dse.c: Include expr.h.
18374 (maybe_trim_constructor_store): New function.
18375 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
18377 PR tree-optimization/33562
18378 PR tree-optimization/61912
18379 PR tree-optimization/77485
18380 * doc/invoke.texi: Document new dse-max-object-size param.
18381 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
18382 * tree-ssa-dse.c: Include params.h.
18383 (dse_store_status): New enum.
18384 (initialize_ao_ref_for_dse): New, partially extracted from
18386 (valid_ao_ref_for_dse, normalize_ref): New.
18387 (setup_live_bytes_from_ref, compute_trims): Likewise.
18388 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
18389 (maybe_trim_partially_dead_store): Likewise.
18390 (maybe_trim_complex_store): Likewise.
18391 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
18392 Track what bytes live from the original store. Return tri-state
18393 for dead, partially dead or live.
18394 (dse_dom_walker): Add constructor, destructor and new private members.
18395 (delete_dead_call, delete_dead_assignment): New extracted from
18397 (dse_optimize_stmt): Make a member of dse_dom_walker.
18398 Use initialize_ao_ref_for_dse.
18400 PR tree-optimization/33562
18401 PR tree-optimization/61912
18402 PR tree-optimization/77485
18403 * sbitmap.h (bitmap_count_bits): Prototype.
18404 (bitmap_clear_range, bitmap_set_range): Likewise.
18405 * sbitmap.c (bitmap_clear_range): New function.
18406 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
18408 2017-01-13 Martin Liska <mliska@suse.cz>
18411 * function.c (set_cfun): Add new argument force.
18412 * function.h (set_cfun): Likewise.
18413 * ipa-inline-transform.c (inline_call): Use the function when
18414 strict alising from is dropped for function we inline to.
18416 2017-01-13 Richard Biener <rguenther@suse.de>
18418 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
18419 for dumping GIMPLE INTEGER_CSTs.
18421 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18423 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
18424 to 201112L since C++17.
18426 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
18429 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
18430 if -fsanitize=kernel-address is present.
18432 2017-01-13 Richard Biener <rguenther@suse.de>
18434 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
18435 as _Literal ( type ) number in case usual suffixes do not
18436 preserve all information.
18438 2017-01-13 Richard Biener <rguenther@suse.de>
18440 PR tree-optimization/77283
18441 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
18442 and ssa-iterators.h.
18443 (is_feasible_trace): Implement a cost model based on joiner
18446 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
18449 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
18450 char or short to __float128/_Float128 directly.
18452 2017-01-12 Martin Sebor <msebor@redhat.com>
18454 to -Wformat-overflow.
18455 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
18456 (min_bytes_remaining): Same.
18457 (get_string_length): Same.
18458 (format_string): Same.
18459 (format_directive): Same.
18461 (pass_sprintf_length::handle_gimple_call): Same.
18463 2017-01-12 Jakub Jelinek <jakub@redhat.com>
18465 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
18466 info.nowrite calls with no lhs that can't throw. Return bool
18467 whether gsi_remove has been called or not.
18468 (pass_sprintf_length::handle_gimple_call): Return bool whether
18469 try_substitute_return_value called gsi_remove. Formatting fix.
18470 (pass_sprintf_length::execute): Don't use gsi_remove if
18471 handle_gimple_call returned true.
18474 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
18475 be removed due to side-effects, don't remove following barrier nor
18476 turn the successor edge into fallthru edge.
18478 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18481 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
18482 element-reversing loads and stores as not swappable.
18484 2017-01-12 Nathan Sidwell <nathan@acm.org>
18485 Nicolai Stange <nicstange@gmail.com>
18487 * combine.c (try_combine): Don't ignore result of overlap checking
18488 loop. Combine overlap & asm check into single loop.
18490 2017-01-12 Richard Biener <rguenther@suse.de>
18492 * tree-pretty-print.c (dump_generic_node): Provide -gimple
18493 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
18495 2017-01-12 Richard Biener <rguenther@suse.de>
18497 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
18498 and TS_TARGET_OPTION directly derive from TS_BASE.
18499 * tree-core.h (tree_optimization_option): Derive from tree_base.
18500 (tree_target_option): Likewise.
18502 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
18504 * config/i386/i386.c (memory_address_length): Increase len
18505 only when rip_relative_addr_p returns false.
18507 2017-01-11 Julia Koval <julia.koval@intel.com>
18509 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
18510 (OPTION_MASK_ISA_SGX_SET): New.
18511 (ix86_handle_option): Handle OPT_msgx.
18512 * config.gcc: Added sgxintrin.h.
18513 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
18514 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
18515 * config/i386/i386.c (ix86_target_string): Add -msgx.
18517 (ix86_option_override_internal): Handle new options.
18518 (ix86_valid_target_attribute_inner_p): Add sgx.
18519 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
18520 * config/i386/i386.opt: Add msgx.
18521 * config/i386/sgxintrin.h: New file.
18522 * config/i386/x86intrin.h: Add sgxintrin.h.
18524 2017-01-11 Jakub Jelinek <jakub@redhat.com>
18527 * fold-const.c (maybe_nonzero_address): Return 1 for function
18529 (tree_single_nonzero_warnv_p): Don't handle function local objects
18533 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
18536 2017-01-11 David Malcolm <dmalcolm@redhat.com>
18539 * opts.c: Include "spellcheck.h"
18540 (struct string_fragment): New struct.
18541 (struct edit_distance_traits<const string_fragment &>): New
18543 (get_closest_sanitizer_option): New function.
18544 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
18546 2017-01-11 Jakub Jelinek <jakub@redhat.com>
18548 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
18550 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
18551 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
18552 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
18553 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
18554 for initial die_offset if dwarf_split_debug_info.
18555 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
18556 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
18558 (output_skeleton_debug_sections): Formatting fix. Use
18559 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
18560 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
18562 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
18564 * config/arm/cortex-a53.md: Add bypasses for
18565 cortex_a53_r2f_cvt.
18566 (cortex_a53_r2f): Only use for transfers.
18567 (cortex_a53_f2r): Likewise.
18568 (cortex_a53_r2f_cvt): Add reservation for conversions.
18569 (cortex_a53_f2r_cvt): Likewise.
18571 2017-01-11 Tamar Christina <tamar.christina@arm.com>
18573 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
18574 to all inlined functions, change static to extern.
18576 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
18579 * config/arm/arm.c (legitimize_pic_address): Handle reference to
18581 (arm_assemble_integer): Likewise.
18583 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
18585 * config.gcc: Use new awk script to check CPU, FPU and architecture
18586 parameters for --with-... options.
18587 * config/arm/parsecpu.awk: New file
18588 * config/arm/arm-cpus.in: New file.
18589 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
18591 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
18593 * config/arm/t-arm: Update dependency rules.
18594 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
18595 of processing .def files.
18596 * config/arm/genopt.sh: Deleted.
18597 * config/arm/gentune.sh: Deleted.
18598 * config/arm/arm-cores.def: Deleted.
18599 * config/arm/arm-arches.def: Deleted.
18600 * config/arm/arm-fpus.def: Deleted.
18601 * config/arm/arm-tune.md: Regenerated.
18602 * config/arm/arm-tables.opt: Regenerated.
18603 * config/arm/arm-cpu.h: New generated file.
18604 * config/arm/arm-cpu-data.h: New generated file.
18605 * config/arm/arm-cpu-cdata.h: New generated file.
18607 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
18610 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
18612 (input_varpool_node): Unpack dynamically_initialized bit.
18614 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
18616 PR rtl-optimization/79032
18617 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
18618 the alignment of the adjusted memory reference against that of MODE,
18619 instead of the alignment of the original memory reference.
18621 2017-01-11 Martin Jambor <mjambor@suse.cz>
18623 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
18625 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
18626 decorated functions.
18628 2017-01-11 Richard Biener <rguenther@suse.de>
18630 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
18631 set range/nonnull info for PHI results. Do not set it on
18632 stmts marked for removal.
18634 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
18636 * expr.c (store_field): In the bitfield case, fetch the return value
18637 from the registers before applying a single big-endian adjustment.
18638 Always do a final load for a BLKmode value not larger than a word.
18640 2017-01-10 David Malcolm <dmalcolm@redhat.com>
18643 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18644 that we correctly handle column numbers greater than
18645 LINE_MAP_MAX_COLUMN_NUMBER.
18647 2017-01-10 Martin Sebor <msebor@redhat.com>
18649 PR middle-end/78245
18650 * gimple-ssa-sprintf.c (get_destination_size): Call
18651 {init,fini}object_sizes.
18652 * tree-object-size.c (addr_object_size): Adjust.
18653 (pass_through_call): Adjust.
18654 (pass_object_sizes::execute): Adjust.
18655 * tree-object-size.h (fini_object_sizes): Declare.
18657 2017-01-10 Martin Sebor <msebor@redhat.com>
18659 PR tree-optimization/78775
18660 * builtins.c (get_size_range): Move...
18661 * calls.c: ...to here.
18662 (alloc_max_size): Accept zero argument.
18663 (operand_signed_p): Remove.
18664 (maybe_warn_alloc_args_overflow): Call get_size_range.
18665 * calls.h (get_size_range): Declare.
18667 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
18669 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
18670 from TI's devices.csv file as of September 2016.
18671 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
18673 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
18675 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
18676 * doc/invoke.texi: Likewise.
18677 * doc/md.texi: Likewise.
18678 * doc/objc.texi: Likewise.
18680 2017-01-10 Joshua Conner <joshconner@google.com>
18682 * config/arm/fuchsia-elf.h: New file.
18683 * config/fuchsia.h: New file.
18684 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
18685 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
18687 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
18689 2016-01-10 Richard Biener <rguenther@suse.de>
18691 PR tree-optimization/79034
18692 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18693 Propagate out degenerate PHIs in the joiner.
18695 2017-01-10 Martin Liska <mliska@suse.cz>
18697 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
18698 (sort_congruence_classes_by_decl_uid): Likewise.
18699 (sort_congruence_class_groups_by_decl_uid): Likewise.
18700 (sem_item_optimizer::merge_classes): Sort class, groups in these
18701 classes and members in the groups by DECL_UID of declarations.
18702 This would make merge operations stable.
18704 2017-01-10 Martin Liska <mliska@suse.cz>
18706 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
18707 usage of m_classes_vec.
18708 (sem_item_optimizer::~sem_item_optimizer): Likewise.
18709 (sem_item_optimizer::get_group_by_hash): Likewise.
18710 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
18711 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
18712 (sem_item_optimizer::verify_classes): Likewise.
18713 (sem_item_optimizer::process_cong_reduction): Likewise.
18714 (sem_item_optimizer::dump_cong_classes): Likewise.
18715 (sem_item_optimizer::merge_classes): Likewise.
18716 * ipa-icf.h (congruence_class_hash): Rename from
18717 congruence_class_group_hash. Remove declaration of m_classes_vec.
18719 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
18721 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
18722 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
18723 * config.gcc: Add avx512vpopcntdqintrin.h.
18724 * config/i386/avx512vpopcntdqintrin.h: New.
18725 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
18726 * config/i386/i386-builtin-types.def: Add new types.
18727 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
18728 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
18729 __builtin_ia32_vpopcountq_v8di_mask): New.
18730 * config/i386/i386-c.c (ix86_target_macros_internal): Define
18731 __AVX512VPOPCNTDQ__.
18732 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
18733 (PTA_AVX512VPOPCNTDQ): Define.
18734 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
18735 TARGET_AVX512VPOPCNTDQ_P): Define.
18736 * config/i386/i386.opt: Add mavx512vpopcntdq.
18737 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
18738 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
18740 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18742 PR middle-end/77484
18743 * predict.def (PRED_CALL): Set to 67.
18745 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
18747 * expr.c (store_field): In the bitfield case, if the value comes from
18748 a function call and is of an aggregate type returned in registers, do
18749 not modify the field mode; extract the value in all cases if the mode
18750 is BLKmode and the size is not larger than a word.
18752 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
18755 * config/i386/cpuid.h: Fix undefined behavior.
18757 2017-01-04 Jeff Law <law@redhat.com>
18759 PR tree-optimization/79007
18760 PR tree-optimization/67955
18761 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
18762 conservative for pt.null when flag_non_call_exceptions is on.
18764 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18766 PR translation/79019
18767 PR translation/79020
18768 * params.def (PARAM_INLINE_MIN_SPEEDUP,
18769 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
18770 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
18772 * config/avr/avr.opt (maccumulate-args): Likewise.
18773 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
18774 * common.opt (freport-bug): Likewise.
18775 * cif-code.def (CIF_FINAL_ERROR): Likewise.
18776 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
18777 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
18778 translatable string.
18779 * config/i386/i386.c (function_value_32): Likewise.
18780 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
18781 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
18783 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
18784 * common/config/msp430/msp430-common.c (msp430_handle_option):
18786 * symtab.c (symtab_node::verify_base): Likewise.
18787 * opts.c (set_debug_level): Likewise.
18788 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
18789 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
18790 missing whitespace to translatable strings.
18791 * config/avr/avr.md (bswapsi2): Fix typo in comment.
18792 * config/sh/superh.h: Likewise.
18793 * config/i386/xopintrin.h: Likewise.
18794 * config/i386/znver1.md: Likewise.
18795 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
18796 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
18797 * double-int.h (struct double_int): Likewise.
18798 * double-int.c (div_and_round_double): Likewise.
18799 * wide-int.cc: Likewise.
18800 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
18801 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
18802 * cfgcleanup.c (crossjumps_occured): Renamed to ...
18803 (crossjumps_occurred): ... this.
18804 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
18807 PR tree-optimization/78899
18808 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
18809 returning bool return struct loop *, NULL for failure and the new
18811 (versionable_outer_loop_p): Don't version outer loop if it has
18812 dont_vectorized bit set.
18813 (tree_if_conversion): When versioning outer loop, ensure
18814 tree_if_conversion is performed also on the inner loop of the
18815 non-vectorizable outer loop copy.
18816 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
18817 LOOP_VECTORIZED in inner loop of the scalar outer loop and
18818 prevent vectorization of it.
18819 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
18820 the outer loop vectorization of the non-scalar version is attempted
18821 before vectorization of the inner loop in scalar version. If
18822 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
18823 vectorization of its inner loop.
18824 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
18825 has 2 inner loops, rename also on edges from bb whose single pred
18826 is outer_loop->header. Fix typo in function comment.
18828 2017-01-09 Martin Sebor <msebor@redhat.com>
18831 * asan.c (asan_emit_stack_protection): Increase local buffer size
18832 to avoid snprintf truncation warning.
18834 2017-01-09 Andrew Pinski <apinski@cavium.com>
18836 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
18837 to reference thunderx2t99 for the tuning structure
18838 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
18840 (thunderx2t99_extra_costs): This.
18841 * config/aarch64/aarch64-tune.md: Regenerate.
18842 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
18843 (vulcan_addrcost_table): This.
18844 (vulcan_regmove_cost): Rename to ...
18845 (thunderx2t99_regmove_cost): This.
18846 (vulcan_vector_cost): Rename to ...
18847 (thunderx2t99_vector_cost): this.
18848 (vulcan_branch_cost): Rename to ...
18849 (thunderx2t99_branch_cost): This.
18850 (vulcan_tunings): Rename to ...
18851 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
18852 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
18854 2017-01-09 Martin Jambor <mjambor@suse.cz>
18858 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
18859 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
18860 (propagate_vr_accross_jump_function): Use the above function for all
18861 value range computations for pass-through jump functions and type
18862 converasion from explicit value range values.
18863 (ipcp_propagate_stage): Do not attempt to deduce types of formal
18864 parameters from TYPE_ARG_TYPES.
18865 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
18866 (ipa_write_node_info): Stream type of the actual argument.
18867 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
18869 2017-01-09 Martin Liska <mliska@suse.cz>
18872 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
18873 (lookup_compiler): Do not show error message with have_E.
18875 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18877 PR tree-optimization/78938
18878 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
18879 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
18880 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
18881 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
18884 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18886 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
18889 2017-01-09 Richard Biener <rguenther@suse.de>
18891 PR tree-optimization/78997
18892 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
18893 name condition properly.
18895 2017-01-09 Richard Biener <rguenther@suse.de>
18898 * dwarf2out.c (is_cxx): New overload with context.
18899 (is_naming_typedef_decl): Use it.
18901 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18903 * invoke.texi (Option Summary): Correct spacing in option lists
18904 and add line breaks to fix over-long lines.
18906 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18908 PR middle-end/17660
18910 * extend.texi (Common Variable Attributes): Add xref to GCC
18911 Internals manual to explain mode attribute keywords.
18913 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18916 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
18917 and Preprocessor Options.
18918 (Options for Linking): Document -pthread here....
18919 (RS/6000 and PowerPC Options): ...not here.
18920 (Solaris 2 Options): ...or here.
18921 * doc/cppopts.texi: Document -pthread.
18923 2017-01-08 Martin Sebor <msebor@redhat.com>
18925 PR middle-end/77708
18926 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
18927 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
18928 New member functions.
18929 (format_directive): Used them.
18931 (pass_sprintf_length::handle_gimple_call): Same.
18932 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
18933 to avoid truncation for any argument.
18934 (extract_affine_mul): Same.
18935 * tree.c (get_file_function_name): Same.
18937 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18939 PR middle-end/77484
18940 * predict.def (PRED_INDIR_CALL): Set to 86.
18942 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18944 PR preprocessor/54124
18945 * doc/cppopts.texi: Reformat -d subtable to list the full name
18946 of the options. Add cross-reference to the docs for the general
18947 compiler -d options.
18948 * doc/invoke.texi (Developer Options): Add cross-reference to the
18949 preprocessor-specific -d option documentation.
18951 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18953 PR preprocessor/13498
18954 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
18955 redudant material, and reflect new command-line options.
18956 (System Headers): Likewise.
18958 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18960 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
18961 -isystem, and -idirafter. Copy-edit.
18962 * doc/cppopts.texi: Copy-edit. Remove contradiction about
18963 default for -ftrack-macro-expansion. Delete obsolete and
18964 badly-formatted implementation details about -fdebug-cpp output.
18965 * doc/cppwarnopts.texi: Copy-edit.
18967 2017-01-07 David Malcolm <dmalcolm@redhat.com>
18970 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18971 that the transition from a max line width >= 1<<10 to narrower
18972 lines works correctly.
18974 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
18976 * doc/options.texi (PerFunction): New.
18977 * opt-functions.awk (switch_flags): Map both Optimization and
18978 PerFunction to CL_OPTIMIZATION.
18979 * opth-gen.awk: Test for PerFunction flag along with
18981 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
18982 it only when the latter is present. Skip those that don't in
18983 the hash function generator.
18984 * common.opt (fvar-tracking): Mark as PerFunction instead of
18986 (fvar-tracking-assignments): Likewise.
18987 (fvar-tracking-assignments-toggle): Likewise.
18988 (fvar-tracking-uninit): Likewise.
18990 2017-01-07 Jakub Jelinek <jakub@redhat.com>
18992 PR translation/79018
18993 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
18996 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
18999 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
19000 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
19001 TARGET_LONG_JUMP_TABLE_OFFSETS.
19002 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
19003 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
19004 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
19005 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
19006 * config/m68k/m68k.md (tablejump expander): Likewise.
19007 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
19008 TARGET_LONG_JUMP_TABLE_OFFSETS.
19009 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
19010 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
19012 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
19013 David Holsgrove <david.holsgrove@xilinx.com>
19015 * common/config/microblaze/microblaze-common.c
19016 (TARGET_EXCEPT_UNWIND_INFO): Remove.
19017 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
19019 * config/microblaze/microblaze.c (microblaze_must_save_register)
19020 (microblaze_expand_epilogue, microblaze_return_addr): Handle
19022 (microblaze_eh_return): New function.
19023 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
19024 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
19025 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
19026 * config/microblaze/microblaze.md (eh_return): New pattern.
19028 2017-01-06 Jakub Jelinek <jakub@redhat.com>
19030 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
19031 GCC_DIAGNOSTIC_STRINGIFY): Define.
19033 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
19035 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19037 * config/arm/arm.md (<mcrr>): New.
19039 * config/arm/arm.c (arm_arch5te): New.
19040 (arm_option_override): Set arm_arch5te.
19041 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
19043 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
19044 (arm_mcrr_qualifiers): ... this. New.
19045 (MRRC_QUALIFIERS): Define to...
19046 (arm_mrrc_qualifiers): ... this. New.
19047 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
19049 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
19050 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
19051 (MRRCI, mrrc, MRRC): New.
19052 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
19053 VUNSPEC_MRRC2): New.
19055 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19057 * config/arm/arm.md (<mcr>): New.
19059 * config/arm/arm.c (arm_coproc_builtin_available): Add
19060 support for mcr, mrc, mcr2 and mrc2.
19061 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
19062 (arm_mcr_qualifiers): ... this. New.
19063 (MRC_QUALIFIERS): Define to ...
19064 (arm_mrc_qualifiers): ... this. New.
19065 (MCR_QUALIFIERS): Define to ...
19066 (arm_mcr_qualifiers): ... this. New.
19067 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
19069 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
19070 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
19071 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
19072 VUNSPEC_MRC2): New.
19074 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19076 * config/arm/arm.md (*ldc): New.
19080 * config/arm/arm.c (arm_coproc_builtin_available): Add
19081 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
19082 (arm_coproc_ldc_stc_legitimate_address): New.
19083 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
19084 'qualifier_const_pointer'.
19085 (LDC_QUALIFIERS): Define to...
19086 (arm_ldc_qualifiers): ... this. New.
19087 (STC_QUALIFIERS): Define to...
19088 (arm_stc_qualifiers): ... this. New.
19089 * config/arm/arm-protos.h
19090 (arm_coproc_ldc_stc_legitimate_address): New.
19091 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
19092 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
19093 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
19094 stc2, stcl, stc2l): New.
19095 * config/arm/constraints.md (Uz): New.
19096 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
19097 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
19098 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
19099 VUNSPEC_STC2L): New.
19101 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19103 * config/arm/arm.md (<cdp>): New.
19104 * config/arm/arm.c (neon_const_bounds): Rename this ...
19105 (arm_const_bounds): ... this.
19106 (arm_coproc_builtin_available): New.
19107 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
19108 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
19109 (CDP_QUALIFIERS): Define to...
19110 (arm_cdp_qualifiers): ... this. New.
19112 (arm_expand_builtin_args): Add case for 6 arguments.
19113 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
19114 (arm_const_bounds): ... this.
19115 (arm_coproc_builtin_available): New.
19116 * config/arm/arm_acle.h (__arm_cdp): New.
19118 * config/arm/arm_acle_builtins.def (cdp): New.
19120 * config/arm/iterators.md (CDPI,CDP,cdp): New.
19121 * config/arm/neon.md: Rename all 'neon_const_bounds' to
19122 'arm_const_bounds'.
19123 * config/arm/types.md (coproc): New.
19124 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
19125 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
19126 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
19127 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
19129 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19131 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
19132 (UBINOP_QUALIFIERS): New.
19134 (acle_builtin_data): New. Change comment.
19135 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
19136 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
19137 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
19138 arm_acle_builtins.def.
19139 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
19140 (arm_init_acle_builtins): New.
19141 (CRC32_BUILTIN): Remove.
19142 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
19143 crc32cb, crc32ch and crc32cw.
19144 (arm_init_crc32_builtins): Remove.
19145 (arm_init_builtins): Use arm_init_acle_builtins rather
19146 than arm_init_crc32_builtins.
19147 (arm_expand_acle_builtin): New.
19148 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
19149 * config/arm/arm_acle_builtins.def: New.
19151 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19153 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
19154 (arm_builtin_datum): ... this.
19155 (arm_init_neon_builtin): Rename to ...
19156 (arm_init_builtin): ... this. Add a new parameters PREFIX
19157 and USE_SIG_IN_NAME.
19158 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
19159 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
19160 'arm_builtin_datum'.
19161 (arm_init_vfp_builtins): Likewise.
19162 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
19163 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
19164 (arm_expand_neon_args): Rename to ...
19165 (arm_expand_builtin_args): ... this. Rename builtin_arg
19166 enum values and differentiate between ARG_BUILTIN_MEMORY
19167 and ARG_BUILTIN_NEON_MEMORY.
19168 (arm_expand_neon_builtin_1): Rename to ...
19169 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
19170 values, arm_expand_builtin_args and add bool parameter NEON.
19171 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
19172 (arm_expand_vfp_builtin): Likewise.
19173 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
19175 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19177 PR middle-end/77484
19178 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
19179 * predict.c (tree_estimate_probability_bb): Reverse direction of
19180 polymorphic call predictor.
19182 2017-01-06 David Malcolm <dmalcolm@redhat.com>
19184 * passes.c (execute_one_pass): Split out pass-skipping logic into...
19185 (determine_pass_name_match): ...this new function and...
19186 (should_skip_pass_p): ...this new function.
19188 2017-01-06 Nathan Sidwell <nathan@acm.org>
19190 * ipa-visibility.c (function_and_variable_visibility): Reformat
19191 comments and long lines. Remove extrneous if.
19192 * symtab.c (symtab_node::make_decl_local): Fix code format.
19193 (symtab_node::set_section_for_node): Fix comment typo.
19195 2017-01-06 Martin Liska <mliska@suse.cz>
19198 * lra-constraints.c: Rename invariant to lra_invariant.
19199 * predict.c (set_even_probabilities): Initialize e to NULL.
19201 2017-01-05 Martin Sebor <msebor@redhat.com>
19203 PR tree-optimization/78910
19204 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
19205 (format_integer): Correct off-by-one error in the handling
19206 of precision with negative numbers in signed conversions..
19208 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
19210 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
19212 2017-01-05 Jakub Jelinek <jakub@redhat.com>
19214 PR tree-optimization/71016
19215 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
19216 factor_out_conditional_conversion. Formatting fix.
19217 (factor_out_conditional_conversion): Add cond_stmt argument.
19218 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
19219 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
19222 2017-01-05 David Malcolm <dmalcolm@redhat.com>
19224 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
19225 read-rtl-function.o, and selftest-rtl.o.
19226 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
19227 (selftest::aarch64_test_loading_full_dump): New function.
19228 (selftest::aarch64_run_selftests): New function.
19229 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
19230 selftest::aarch64_run_selftests.
19231 * config/i386/i386.c
19232 (selftest::ix86_test_loading_dump_fragment_1): New function.
19233 (selftest::ix86_test_loading_call_insn): New function.
19234 (selftest::ix86_test_loading_full_dump): New function.
19235 (selftest::ix86_test_loading_unspec): New function.
19236 (selftest::ix86_run_selftests): Call the new functions.
19237 * emit-rtl.c (maybe_set_max_label_num): New function.
19238 * emit-rtl.h (maybe_set_max_label_num): New decl.
19239 * function.c (instantiate_decls): Guard call to
19240 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
19241 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
19243 * gensupport.c (gen_reader::gen_reader): Pass "false"
19244 for new "compact" param of rtx_reader.
19245 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
19246 rather than an empty string for NULL strings.
19247 * read-md.c: Potentially include config.h rather than bconfig.h.
19248 Wrap include of errors.h with #ifdef GENERATOR_FILE.
19249 (have_error): New global, copied from errors.c.
19250 (md_reader::read_name): Rename to...
19251 (md_reader::read_name_1): ...this, adding "out_loc" param,
19252 and converting "missing name or number" to returning false, rather
19254 (md_reader::read_name): Reimplement in terms of read_name_1.
19255 (md_reader::read_name_or_nil): New function.
19256 (md_reader::read_string): Handle "(nil)" by returning NULL.
19257 (md_reader::md_reader): Add new param "compact".
19258 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
19259 (md_reader::read_file): New method.
19260 * read-md.h (md_reader::md_reader): Add new param "compact".
19261 (md_reader::read_file): New method.
19262 (md_reader::is_compact): New accessor.
19263 (md_reader::read_name): Convert return type from void to file_location.
19264 (md_reader::read_name_or_nil): New decl.
19265 (md_reader::read_name_1): New decl.
19266 (md_reader::m_compact): New field.
19267 (noop_reader::noop_reader): Pass "false" for new "compact" param
19269 (rtx_reader::rtx_reader): Add new "compact" param.
19270 (rtx_reader::read_rtx_operand): Make virtual and convert return
19271 type from void to rtx.
19272 (rtx_reader::read_until): New decl.
19273 (rtx_reader::handle_any_trailing_information): New virtual function.
19274 (rtx_reader::postprocess): New virtual function.
19275 (rtx_reader::finalize_string): New virtual function.
19276 (rtx_reader::m_in_call_function_usage): New field.
19277 (rtx_reader::m_reuse_rtx_by_id): New field.
19278 * read-rtl-function.c: New file.
19279 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
19280 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
19281 (selftest::verify_three_block_rtl_cfg): New decl.
19282 * read-rtl-function.h: New file.
19283 * read-rtl.c: Potentially include config.h rather than bconfig.h.
19284 For host, include function.h, memmodel.h, and emit-rtl.h.
19285 (one_time_initialization): New function.
19286 (struct compact_insn_name): New struct.
19287 (compact_insn_names): New array.
19288 (find_code): Handle insn codes in compact dumps.
19289 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
19290 (bind_subst_iter_and_attr): Likewise.
19291 (add_condition_to_string): Likewise.
19292 (add_condition_to_rtx): Likewise.
19293 (apply_attribute_uses): Likewise.
19294 (add_current_iterators): Likewise.
19295 (apply_iterators): Likewise.
19296 (initialize_iterators): Guard usage of apply_subst_iterator with
19297 #ifdef GENERATOR_FILE.
19298 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
19299 (md_reader::read_mapping): Likewise.
19300 (add_define_attr_for_define_subst): Likewise.
19301 (add_define_subst_attr): Likewise.
19302 (read_subst_mapping): Likewise.
19303 (check_code_iterator): Likewise.
19304 (rtx_reader::read_rtx): Likewise. Move one-time initialization
19306 (one_time_initialization): New function.
19307 (rtx_reader::read_until): New method.
19308 (read_flags): New function.
19309 (parse_reg_note_name): New function.
19310 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
19311 Handle reuse_rtx ids.
19312 Wrap iterator lookup within #ifdef GENERATOR_FILE.
19313 Add parsing support for RTL dumps, mirroring the special-cases in
19314 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
19315 values, and calling handle_any_trailing_information.
19316 (rtx_reader::read_rtx_operand): Convert return type from void
19317 to rtx, returning return_rtx. Handle case 'e'. Call
19318 finalize_string on XSTR and XTMPL fields.
19319 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
19320 "(nil)" values were omitted. Call the postprocess vfunc on the
19322 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
19323 class ctor. Initialize m_in_call_function_usage. Call
19324 one_time_initialization.
19325 * rtl-tests.c (selftest::test_uncond_jump): Call
19326 set_new_first_and_last_insn.
19327 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
19328 * selftest-rtl.c: New file.
19329 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
19330 (selftest::get_insn_by_uid): New decl.
19331 * selftest-run-tests.c (selftest::run_tests): Call
19332 read_rtl_function_c_tests.
19333 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
19334 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
19337 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
19339 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
19340 operands in a special way. Assert that pos+len <= mode precision.
19342 2017-01-05 Jakub Jelinek <jakub@redhat.com>
19344 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
19345 3 argument Alias with unlimited for the negative form.
19346 (fno-vect-cost-model): Removed.
19348 2017-01-05 Martin Liska <mliska@suse.cz>
19350 * hsa-gen.c (gen_hsa_divmod): New function.
19351 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
19353 2017-01-05 Martin Liska <mliska@suse.cz>
19356 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
19359 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19361 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
19362 small constant length operands.
19364 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19366 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
19367 between loop iterations.
19369 2017-01-05 Martin Liska <mliska@suse.cz>
19372 * gimplify.c (gimplify_decl_expr): Compare to
19373 asan_poisoned_variables instread of checking flags.
19374 (gimplify_target_expr): Likewise.
19375 (gimplify_expr): Likewise.
19376 (gimplify_function_tree): Conditionally initialize
19377 asan_poisoned_variables.
19379 2017-01-04 Jeff Law <law@redhat.com>
19381 PR tree-optimizatin/78812
19382 * rtl.h (contains_mem_rtx_p): Prototype.
19383 * ifcvt.c (containts_mem_rtx_p): Move from here to...
19384 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
19385 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
19386 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
19387 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
19389 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19391 * input.c (assert_char_at_range): Default-initialize actual_range.
19393 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19395 * df-scan.c (df_ref_create_structure): Make regno unsigned,
19396 to match the caller.
19398 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19400 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
19401 insns after final jump in test to emit dummy move.
19403 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19405 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
19406 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
19408 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19410 * multiple_target.c (create_dispatcher_calls): Init e_next.
19411 * tree-ssa-loop-split.c (split_loop): Init border.
19412 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
19415 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
19420 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
19421 (altivec_register_operand): Do not return true if the operand
19422 contains a SUBREG mixing SImode and SFmode.
19423 (vsx_register_operand): Likewise.
19424 (vsx_reg_sfsubreg_ok): New predicate.
19425 (vfloat_operand): Do not return true if the operand contains a
19426 SUBREG mixing SImode and SFmode.
19427 (vint_operand): Likewise.
19428 (vlogical_operand): Likewise.
19429 (gpc_reg_operand): Likewise.
19430 (int_reg_operand): Likewise.
19431 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
19432 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
19433 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
19435 (rs6000_emit_move_si_sf_subreg): New helper function.
19436 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
19437 fixup SUBREGs involving SImode and SFmode.
19438 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
19439 numbers for the new peephole2 optimization.
19440 (peephole2 for SFmode unions): New peephole2 to optimize cases in
19441 the GLIBC math library that do AND/IOR/XOR operations on single
19442 precision floating point.
19443 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
19444 target macros to say whether we need to avoid SUBREGs mixing
19446 (TARGET_ALLOW_SF_SUBREG): Likewise.
19447 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
19448 (UNSPEC_SI_FROM_SF): Likewise.
19449 (iorxor): Change spacing.
19450 (and_ior_xor): New iterator for AND, IOR, and XOR.
19451 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
19452 (movdi_from_sf_zero_ext): Likewise.
19453 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
19454 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
19455 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
19456 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
19457 (fms<mode>4): Likewise.
19458 (fnma<mode>4): Likewise.
19459 (fnms<mode>4): Likewise.
19460 (nfma<mode>4): Likewise.
19461 (nfms<mode>4): Likewise.
19463 2017-01-04 Marek Polacek <polacek@redhat.com>
19466 * doc/invoke.texi: Document -Wpointer-compare.
19468 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19470 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
19473 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
19474 descriptions for -gdwarf-5 and emit them as uleb128 instead of
19477 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
19480 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
19481 documentation of the powerpc_popcntb_ok attribute.
19482 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19483 code to issue warning messages if a requested CPU configuration is
19484 not supported by the binary (assembler and loader) toolchain.
19485 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
19486 made to define a built-in function that has been disabled.
19487 (paired_init_builtins): Add assertion to prevent ICE if attempt is
19488 made to define a built-in function that has been disabled.
19489 (altivec_init_builtins): Add comment explaining why definition
19490 of the DST built-in functions is not preceded by an assertion
19491 check. Add assertions to prevent ICE if attempts are made to
19492 define an altivec predicate or an abs* built-in function that has
19494 (htm_init_builtins): Add comment explaining why definition of the
19495 htm built-in functions is not preceded by an assertion check.
19497 2017-01-04 Jeff Law <law@redhat.com>
19499 PR tree-optimizatin/67955
19500 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
19501 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
19502 the points-to solution does not include pt_null. Use DECL_PT_UID
19505 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
19507 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
19508 Use gen_int_mode instead of gen_lopwart for const_int operands.
19510 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19512 PR tree-optimization/71563
19513 * match.pd: Simplify X << Y into X if Y is known to be 0 or
19514 out of range value - has low bits known to be zero.
19516 2017-01-04 Alan Modra <amodra@gmail.com>
19518 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
19519 * configure: Regenerate.
19520 * config.in: Regenerate.
19522 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19525 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
19526 a substring of the message, but strcmp with the whole message. Ifdef
19527 ENABLE_NLS, translate the message first using dgettext.
19529 2017-01-03 Jeff Law <law@redhat.com>
19531 PR tree-optimizatin/78856
19532 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
19533 (mark_threaded_blocks): Remove code to truncate thread paths that
19534 cross multiple loop headers. Instead invalidate the cached loop
19535 iteration information and handle case of a thread path walking
19536 into an irreducible region.
19538 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
19541 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
19542 assertions. Add support for doing the signbit if the IEEE 128-bit
19543 floating point value is in a GPR.
19544 * config/rs6000/rs6000.md (Fsignbit): Delete.
19545 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
19546 Update the length attribute if the value is in a GPR.
19547 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
19548 the sign or zero extension instruction, since the value is always 0/1.
19549 (signbit<mode>2_dm2): Delete using <Fsignbit>.
19552 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
19553 extracting SImode to a GPR register so that we can generate a
19554 store, limit the vector to be in a traditional Altivec register
19555 for the vextuwrx instruction.
19557 2017-01-03 Ian Lance Taylor <iant@google.com>
19559 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
19561 2017-01-03 Martin Sebor <msebor@redhat.com>
19563 PR tree-optimization/78696
19564 * gimple-ssa-sprintf.c (format_floating): Correct handling of
19565 precision. Use MPFR for %f for greater fidelity. Correct handling
19567 (pass_sprintf_length::compute_format_length): Set width and precision
19568 specified by asrerisk to void_node for vararg functions.
19569 (try_substitute_return_value): Adjust dump output.
19571 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
19573 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
19575 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
19577 * doc/invoke.texi (SPARC options): Document -mlra as the default.
19578 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
19579 -mlra/-mno-lra was passed to the compiler.
19581 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
19583 PR rtl-optimization/65618
19584 * emit-rtl.c (try_split): Move initialization of "before" and
19585 "after" to just before the call to emit_insn_after_setloc.
19587 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
19589 * doc/md.texi (Standard Names): Remove reference to Java frontend.
19591 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
19593 * dwarf2out.c (gen_enumeration_type_die): When
19594 -gno-strict-dwarf, add a DW_AT_encoding attribute.
19596 2017-01-03 Jakub Jelinek <jakub@redhat.com>
19598 PR tree-optimization/78965
19599 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
19600 Change first argument from const call_info & to call_info &. For %n
19601 set info.nowrite to false.
19603 PR middle-end/78901
19604 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
19605 possibly throwing calls.
19607 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
19608 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
19609 and fns handling, rather than in a separate case SSA_NAME.
19611 2017-01-02 Jeff Law <law@redhat.com>
19613 * config/darwin-driver.c (darwin_driver_init): Const-correctness
19614 fixes for first_period and second_period variables.
19616 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
19619 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
19620 (*insvqi_1): New insn pattern.
19621 (*insvqi_1_mem_rex64): Ditto.
19622 (*insvqi_2): Ditto.
19623 (*insvqi_3): Rename from *insvqi.
19625 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
19627 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
19629 * doc/cfg.texi (Edges): Remove reference to Java.
19630 (Maintaining the CFG): Ditto.
19632 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19634 PR middle-end/77674
19635 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
19636 transparent aliases.
19638 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19640 PR middle-end/77484
19641 * predict.def (PRED_CALL): Update hitrate.
19642 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
19643 * predict.c (tree_estimate_probability_bb): Split CALL predictor
19644 into direct/indirect/polymorphic variants.
19646 2017-01-01 Jakub Jelinek <jakub@redhat.com>
19648 Update copyright years.
19650 * gcc.c (process_command): Update copyright notice dates.
19651 * gcov-dump.c (print_version): Ditto.
19652 * gcov.c (print_version): Ditto.
19653 * gcov-tool.c (print_version): Ditto.
19654 * gengtype.c (create_file): Ditto.
19655 * doc/cpp.texi: Bump @copying's copyright year.
19656 * doc/cppinternals.texi: Ditto.
19657 * doc/gcc.texi: Ditto.
19658 * doc/gccint.texi: Ditto.
19659 * doc/gcov.texi: Ditto.
19660 * doc/install.texi: Ditto.
19661 * doc/invoke.texi: Ditto.
19663 Copyright (C) 2017 Free Software Foundation, Inc.
19665 Copying and distribution of this file, with or without modification,
19666 are permitted in any medium without royalty provided the copyright
19667 notice and this notice are preserved.