1 2017-07-28 Tamar Christina <tamar.christina@arm.com>
3 * config/aarch64/aarch64.c
4 (aarch64_simd_container_mode): Add prototype.
5 (aarch64_expand_mov_immediate): Add HI support.
6 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
7 (aarch64_can_const_movi_rtx_p): New.
8 (aarch64_preferred_reload_class):
9 Remove restrictions of using FP registers for certain SIMD operations.
10 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
11 (aarch64_valid_floating_const): Add integer move validation.
12 (aarch64_simd_imm_scalar_p): Remove.
13 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
14 (aarch64_legitimate_constant_p): Expand list of supported cases.
15 * config/aarch64/aarch64-protos.h
16 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
17 (aarch64_reinterpret_float_as_int): New.
18 (aarch64_simd_imm_scalar_p): Remove.
19 * config/aarch64/constraints.md (Uvi): New.
20 (Dd): Split into Ds and new Dd.
21 * config/aarch64/aarch64.md (*movsi_aarch64):
23 (*movdi_aarch64): Add SIMD mov case.
25 2017-07-28 Bin Cheng <bin.cheng@arm.com>
27 * tree-predcom.c: (struct chain): Handle store-store chain in which
28 stores for elimination only store loop invariant values.
29 (execute_pred_commoning_chain): Ditto.
30 (prepare_initializers_chain_store_elim): Ditto.
31 (prepare_finalizers): Ditto.
32 (is_inv_store_elimination_chain): New function.
33 (initialize_root_vars_store_elim_1): New function.
35 2017-07-28 Bin Cheng <bin.cheng@arm.com>
37 * tree-predcom.c: Revise general description of the pass.
38 (enum chain_type): New enum type for store elimination.
39 (struct chain): New field supporting store elimination.
40 (struct component): Ditto.
41 (dump_chain): Dump store-stores chain.
42 (release_chain): Release resources.
43 (split_data_refs_to_components): Compute and create component
44 contains only stores for elimination.
45 (get_chain_last_ref_at): New function.
46 (make_invariant_chain): Initialization.
47 (make_rooted_chain): Specify chain type in parameter and record it.
48 (add_looparound_copies): Skip for store-stores chain.
49 (determine_roots_comp): Compute type of chain and pass it to
51 (initialize_root_vars_store_elim_2): New function.
52 (finalize_eliminated_stores): New function.
53 (remove_stmt): Handle store for elimination.
54 (execute_pred_commoning_chain): Execute predictive commoning on
56 (determine_unroll_factor): Skip unroll for store-stores chain.
57 (prepare_initializers_chain_store_elim): New function.
58 (prepare_initializers_chain): Hanlde store-store chain.
59 (prepare_finalizers_chain, prepare_finalizers): New function.
60 (tree_predictive_commoning_loop): Return integer value indicating
61 if loop is unrolled or lcssa form is corrupted.
62 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
64 2017-07-28 Bin Cheng <bin.cheng@arm.com>
66 * tree-predcom.c (initialize_root): Delete.
67 (execute_pred_commoning_chain): Initialize root vars and replace
68 reference of non-combined chain directly, rather than call above
71 2017-07-28 Bin Cheng <bin.cheng@arm.com>
73 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
74 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
76 2017-07-28 Bin Cheng <bin.cheng@arm.com>
78 * tree-predcom.c (struct chain): New field init_seq.
79 (release_chain): Release init_seq.
80 (prepare_initializers_chain): Record intialization stmts in above
82 (insert_init_seqs): New function.
83 (tree_predictive_commoning_loop): Call insert_init_seqs.
85 2017-07-28 Bin Cheng <bin.cheng@arm.com>
87 * tree-predcom.c (determine_roots_comp): Skip trivial components.
89 2017-07-28 Richard Biener <rguenther@suse.de>
91 * match.pd: Remove superfluous :c.
92 * genmatch.c (simplify::id): Add member.
93 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
95 (current_id): New global.
96 (dt_node::parent): Move from ...
97 (dt_operand::parent): ... here. Add for_id member.
98 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
99 (decision_tree::find_node): Relax order requirement when
100 merging DT_TRUE nodes to ones inbetween the current simplify
101 and the one we try to merge with. Add diagnostic whenever
102 we need to enforce pattern order by not merging.
103 (decision_tree::insert): Set current_id.
104 (decision_tree::print_node): Dump parent node and for_id.
105 (parser::last_id): Add member.
106 (parser::push_simplify): Assign unique id.
107 (parser::parser): Initialize last_id.
109 2017-07-28 Martin Liska <mliska@suse.cz>
112 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
113 gimple_build_debug_bind.
115 2017-07-28 Richard Biener <rguenther@suse.de>
117 PR tree-optimization/81502
118 * match.pd: Add pattern combining BIT_INSERT_EXPR with
120 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
122 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
123 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
124 for BIT_FIELD_REF args.
125 * fold-const.c (make_bit_field_ref): Likewise.
126 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
128 2017-07-28 Jakub Jelinek <jakub@redhat.com>
131 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
132 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
133 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
134 Or it into SANITIZER_UNDEFINED.
135 * ubsan.c: Include gimple-fold.h and varasm.h.
136 (ubsan_expand_ptr_ifn): New function.
137 (instrument_pointer_overflow): New function.
138 (maybe_instrument_pointer_overflow): New function.
139 (instrument_object_size): Formatting fix.
140 (pass_ubsan::execute): Call instrument_pointer_overflow
141 and maybe_instrument_pointer_overflow.
142 * internal-fn.c (expand_UBSAN_PTR): New function.
143 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
144 * sanitizer.def (__ubsan_handle_pointer_overflow,
145 __ubsan_handle_pointer_overflow_abort): New builtins.
146 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
147 * internal-fn.def (UBSAN_PTR): New internal function.
148 * opts.c (sanitizer_opts): Add pointer-overflow.
149 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
150 * fold-const.c (build_range_check): Compute pointer range check in
151 integral type if pointer arithmetics would be needed. Formatting
154 2017-07-28 Martin Liska <mliska@suse.cz>
157 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
158 parameters that are of a variable-length.
160 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
162 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
164 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
165 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
166 (CRT_CALL_STATIC_FUNCTION): Likewise.
167 (ASM_DEFAULT_SPEC): New define.
168 (ASM_SPEC32): Likewise.
169 (ASM_SPEC64): Likewise.
170 (ASM_SPEC_COMMON): Likewise.
171 (ASM_SPEC): Likewise.
172 (INVALID_64BIT): Likewise.
173 (LINK_OS_DEFAULT_SPEC): Likewise.
174 (LINK_OS_SPEC32): Likewise.
175 (LINK_OS_SPEC64): Likewise.
176 (POWERPC_LINUX): Likewise.
177 (PTRDIFF_TYPE): Likewise.
178 (RESTORE_FP_PREFIX): Likewise.
179 (RESTORE_FP_SUFFIX): Likewise.
180 (SAVE_FP_PREFIX): Likewise.
181 (SAVE_FP_SUFFIX): Likewise.
182 (SIZE_TYPE): Likewise.
183 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
184 (TARGET_64BIT): Likewise.
185 (TARGET_64BIT): Likewise.
186 (TARGET_AIX): Likewise.
187 (WCHAR_TYPE_SIZE): Likewise.
188 (WCHAR_TYPE): Undefine.
189 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
190 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
191 (CPP_OS_RTEMS_SPEC): Delete.
192 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
193 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
195 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
197 2017-07-28 Jakub Jelinek <jakub@redhat.com>
199 PR tree-optimization/81578
200 * tree-parloops.c (build_new_reduction): Bail out if
201 reduction_code isn't one of the standard OpenMP reductions.
202 Move the details printing after that decision.
204 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
206 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
207 related to reload_in_progress.
208 (splat_input_operand): Likewise.
209 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
211 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
213 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
214 (TARGET_INSTANTIATE_DECLS): Likewise.
215 (legitimate_indexed_address_p): Delete reload_in_progress code.
216 (rs6000_debug_legitimate_address_p): Likewise.
217 (rs6000_eliminate_indexed_memrefs): Likewise.
218 (rs6000_emit_le_vsx_store): Likewise.
219 (rs6000_emit_move_si_sf_subreg): Likewise.
220 (rs6000_emit_move): Likewise.
221 (register_to_reg_type): Likewise.
222 (rs6000_pre_atomic_barrier): Likewise.
223 (rs6000_machopic_legitimize_pic_address): Likewise.
224 (rs6000_allocate_stack_temp): Likewise.
225 (rs6000_address_for_fpconvert): Likewise.
226 (rs6000_address_for_altivec): Likewise.
227 (rs6000_secondary_memory_needed_rtx): Delete function.
228 (rs6000_check_sdmode): Likewise.
229 (rs6000_alloc_sdmode_stack_slot): Likewise.
230 (rs6000_instantiate_decls): Likewise.
231 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
232 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
233 Delete reload_in_progress.
234 (*vec_reload_and_plus_<mptrsize>): Likewise.
235 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
236 (vsx_div_v2di): Likewise.
237 (vsx_udiv_v2di): Likewise.
239 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
241 * config/rs6000/rs6000.opt (mlra): Replace with stub.
242 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
243 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
244 (rs6000_debug_reg_global): Delete print of LRA status.
245 (rs6000_option_override_internal): Delete dead LRA related code.
246 (rs6000_lra_p): Delete function.
247 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
249 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
251 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
252 * config/riscv/rtems.h: New file.
254 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
255 Sudakshina Das <sudi.das@arm.com>
257 * config/aarch64/aarch64.md
258 (define_split for and<mode>3nr_compare): Move
259 non aarch64_logical_operand to a register.
260 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
261 register immediate operand to a register.
262 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
264 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
267 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
269 2017-07-27 Richard Biener <rguenther@suse.de>
271 PR tree-optimization/81573
272 PR tree-optimization/81494
273 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
274 multi defuse cycle case.
276 2017-07-27 Richard Biener <rguenther@suse.de>
278 PR tree-optimization/81571
279 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
282 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
284 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
285 earlier and only if MASK_FPU is set. Adjust formatting.
287 2017-07-27 Martin Liska <mliska@suse.cz>
289 * opt-functions.awk: Add validation of value of Init.
290 * optc-gen.awk: Pass new argument.
292 2017-07-27 Martin Liska <mliska@suse.cz>
294 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
297 2017-07-27 Martin Liska <mliska@suse.cz>
299 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
300 BBs and edges seen by autoFDO.
302 2017-07-27 Richard Biener <rguenther@suse.de>
304 PR tree-optimization/81502
305 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
306 with incompatible but same sized type.
307 (execute_update_addresses_taken): Likewise.
309 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
311 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
312 flag_tree_loop_vectorize rather than flag_tree_vectorize.
314 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
317 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
318 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
319 Change s_operand to memory_operand.
321 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
323 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
324 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
325 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
326 Emit instructions rather than returning an expression. Handle TFmode
327 and KFmode by casting to TImode.
328 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
329 (rs6000_emit_le_vsx_store): Likewise.
330 * config/rs6000/vsx.md (VSX_TI): New iterator.
331 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
332 (*vsx_le_undo_permute_<mode>): Likewise.
333 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
334 emit the split sequence.
335 (*vsx_le_perm_store_<mode>): Likewise.
337 2017-07-27 Jakub Jelinek <jakub@redhat.com>
339 PR tree-optimization/81555
340 PR tree-optimization/81556
341 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
342 if true, force CHANGED for the recursive invocation.
343 (reassociate_bb): Remember original length of ops array, pass
344 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
346 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
347 attributes for noipa attribute. For naked attribute use
348 lookup_attribute first before lookup_attribute_spec.
349 * final.c (rest_of_handle_final): Disable IPA RA for functions with
351 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
352 for functions with noipa attribute.
353 (cgraph_externally_visible_p): Return true for functions with noipa
355 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
356 for functions with noipa attribute.
357 * doc/extend.texi: Document noipa function attribute.
358 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
359 also for functions with noipa attribute.
360 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
362 2017-07-26 Andrew Pinski <apinski@cavium.com>
364 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
365 vec_unalign_load_cost and vec_unalign_store_cost.
367 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
369 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
370 -mvsx-small-integer option.
371 (ISA_3_0_MASKS_IEEE): Likewise.
372 (OTHER_VSX_VECTOR_MASKS): Likewise.
373 (POWERPC_MASKS): Likewise.
374 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
375 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
376 code, only testing for DImode being allowed in non-VSX floating
378 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
379 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
381 (rs6000_option_override_internal): Delete -mvsx-small-integer.
382 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
383 TARGET_P8_VECTOR test.
384 (rs6000_secondary_reload_simple_move): Likewise.
385 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
386 since TARGET_P9_VECTOR was already tested.
387 (rs6000_opt_masks): Remove -mvsx-small-integer.
388 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
389 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
391 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
392 test for TARGET_VEXTRACTUB was used, and that uses
394 (p9 extract splitter): Likewise.
395 (vsx_extract_<mode>_di_p9): Likewise.
396 (vsx_extract_<mode>_store_p9): Likewise.
397 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
398 for TARGET_P9_VECTOR was used. Delete code that is now dead with
399 the elimination of TARGET_VSX_SMALL_INTEGER.
400 (vsx_extract_<mode>_p8): Likewise.
401 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
402 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
403 (vsx_set_<mode>_p9): Likewise.
404 (vsx_set_v4sf_p9): Likewise.
405 (vsx_set_v4sf_p9_zero): Likewise.
406 (vsx_insert_extract_v4sf_p9): Likewise.
407 (vsx_insert_extract_v4sf_p9_2): Likewise.
408 * config/rs6000/rs6000.md (sign extend splitter): Change
409 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
410 (floatsi<mode>2_lfiwax_mem): Likewise.
411 (floatunssi<mode>2_lfiwzx_mem): Likewise.
412 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
413 since a test for TARGET_P9_VECTOR was used.
414 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
415 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
416 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
417 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
418 TARGET_P8_VECTOR test.
419 (fix_trunc<mode>si2_stfiwx): Likewise.
420 (fix_trunc<mode>si2_internal): Likewise.
421 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
422 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
424 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
425 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
426 TARGET_P8_VECTOR test.
427 (fixuns_trunc<mode>si2_stfiwx): Likewise.
428 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
429 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
431 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
432 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
433 since a test for TARGET_P9_VECTOR was used.
434 (splitter for loading small constants): Likewise.
436 2017-07-26 Andrew Pinski <apinski@cavium.com>
438 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
441 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
444 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
445 (fp_valid_at): Likewise.
447 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
449 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
450 (qdf24xx_addrcost_table): Likewise.
451 (cortexa57_tunings): Update to use generic_branch_cost.
452 (cortexa72_tunings): Likewise.
453 (cortexa73_tunings): Likewise.
454 (qdf24xx_tunings): Likewise.
456 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
458 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
459 (thunderx2t99_branch_cost): Likewise.
460 (cortexa35_tunings): Update to use generic_branch_cost.
461 (cortexa53_tunings): Likewise.
462 (cortexa57_tunings): Likewise.
463 (cortexa72_tunings): Likewise.
464 (cortexa73_tunings): Likewise.
465 (thunderx2t99_tunings): Likewise.
467 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
469 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
470 (sparc_option_override): Honour MASK_FSMULD.
471 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
472 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
473 * config/sparc/sparc.opt (mfsmuld): New option.
474 * doc/invoke.texi (mfsmuld): Document option.
476 2017-07-26 Marek Polacek <polacek@redhat.com>
479 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
481 2017-07-26 Richard Biener <rguenther@suse.de>
483 * gimple-match-head.c (do_valueize): Return OP if valueize
485 (get_def): New helper to get at the def stmt of a SSA name
487 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
488 do_valueize to get at the def stmt.
489 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
491 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
494 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
496 2017-07-26 Martin Liska <mliska@suse.cz>
499 * function.c (expand_function_start): Make expansion of
500 nonlocal_goto_save_area after parm_birth_insn.
502 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
504 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
505 from all CPU target flags enable members.
507 2017-07-26 Richard Biener <rguenther@suse.de>
509 * genmatch.c (dt_simplify::gen): Make iterator vars const.
510 (decision_tree::gen): Make 'type' const.
511 (write_predicate): Likewise.
513 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
515 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
516 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
517 (rs6000_option_override_internal): Likewise.
518 (rs6000_expand_vector_set): Likewise.
519 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
520 (TARGET_UPPER_REGS_SF): Likewise.
521 (TARGET_UPPER_REGS_DI): Likewise.
522 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
523 (TARGET_DIRECT_MOVE_64BIT): Likewise.
524 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
525 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
526 (Splitters for DI constants in Altivec registers): Likewise.
527 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
528 (vsx_set_v4sf_p9): Likewise.
529 (vsx_set_v4sf_p9_zero): Likewise.
530 (vsx_insert_extract_v4sf_p9): Likewise.
531 (vsx_insert_extract_v4sf_p9_2): Likewise.
533 2017-07-25 Carl Love <cel@us.ibm.com>
535 * doc/extend.texi: Update the built-in documentation file for the
536 existing built-in functions
537 vector signed char vec_cnttz (vector signed char);
538 vector unsigned char vec_cnttz (vector unsigned char);
539 vector signed short vec_cnttz (vector signed short);
540 vector unsigned short vec_cnttz (vector unsigned short);
541 vector signed int vec_cnttz (vector signed int);
542 vector unsigned int vec_cnttz (vector unsigned int);
543 vector signed long long vec_cnttz (vector signed long long);
544 vector unsigned long long vec_cnttz (vector unsigned long long);
546 2017-07-25 Andrew Pinski <apinski@cavium.com>
548 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
549 accesses where the use is for the first operand of a BIT_INSERT.
551 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
554 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
555 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
557 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
559 * config/i386/gstabs.h: Delete.
560 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
562 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
564 * config/i386/i386.c (ix86_decompose_address): Do not check for
565 register RTX when looking at index_reg or base_reg.
566 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
568 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
570 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
571 to update EH info here.
573 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
575 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
577 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
579 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
581 2017-07-25 Torsten Duwe <duwe@suse.de>
583 * common.opt: Introduce -fpatchable-function-entry
584 command line option, and its variables function_entry_patch_area_size
585 and function_entry_patch_area_start.
586 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
587 including a two-value parser.
588 * target.def (print_patchable_function_entry): New target hook.
589 * targhooks.h (default_print_patchable_function_entry): New function.
590 * targhooks.c (default_print_patchable_function_entry): Likewise.
591 * toplev.c (process_options): Switch off IPA-RA if
592 patchable function entries are being generated.
593 * varasm.c (assemble_start_function): Look at the
594 patchable-function-entry command line switch and current
595 function attributes and maybe generate NOP instructions by
596 calling the print_patchable_function_entry hook.
597 * doc/extend.texi: Document patchable_function_entry attribute.
598 * doc/invoke.texi: Document -fpatchable_function_entry
600 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
602 * doc/tm.texi: Re-generate.
604 2017-07-25 Jakub Jelinek <jakub@redhat.com>
607 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
608 TARGET_AVX512DQ rather than TARGET_AVX512BW.
610 2017-07-25 Tamar Christina <tamar.christina@arm.com>
612 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
614 2017-07-25 Richard Biener <rguenther@suse.de>
616 PR tree-optimization/81455
617 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
618 not walk in cycles when looking for guards.
620 2017-07-25 Richard Biener <rguenther@suse.de>
622 PR tree-optimization/81529
623 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
624 when optimizing backedge uses.
626 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
628 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
630 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
631 to dl_section_ref. On AIX, append an expression to subtract
632 the size of the section length to dl_section_ref.
634 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
636 * configure.ac: If any of the config.* scripts fail, exit 1.
637 * configure: Regenerate.
639 2017-07-25 Richard Biener <rguenther@suse.de>
642 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
643 of immediate uses, be more verbose on errors.
645 2017-07-25 Richard Biener <rguenther@suse.de>
647 PR tree-optimization/81510
648 * tree-vect-loop.c (vect_is_simple_reduction): When the
649 reduction stmt is not inside the loop bail out.
651 2017-07-25 Richard Biener <rguenther@suse.de>
653 PR tree-optimization/81303
654 * tree-vect-loop-manip.c (vect_loop_versioning): Build
655 profitability check against LOOP_VINFO_NITERSM1.
657 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
659 * domwalk.c (cmp_bb_postorder): Simplify.
660 (sort_bbs_postorder): New function. Use it...
661 (dom_walker::walk): ...here to optimize common cases.
663 2017-07-25 Martin Liska <mliska@suse.cz>
666 * ipa-visibility.c (function_and_variable_visibility): Make the
667 redirection just on target that supports aliasing.
668 Fix GNU coding style.
670 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
673 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
675 * config/arm/rtems.h: Likewise.
676 * config/bfin/rtems.h: Likewise.
677 * config/i386/rtemself.h: Likewise.
678 * config/lm32/rtems.h: Likewise.
679 * config/m32c/rtems.h: Likewise.
680 * config/m68k/rtemself.h: Likewise.
681 * config/microblaze/rtems.h: Likewise.
682 * config/mips/rtems.h: Likewise.
683 * config/moxie/rtems.h: Likewise.
684 * config/nios2/rtems.h: Likewise.
685 * config/powerpcspe/rtems.h: Likewise.
686 * config/rs6000/rtems.h: Likewise.
687 * config/rtems.h: Likewise.
688 * config/sh/rtems.h: Likewise.
689 * config/sh/rtemself.h: Likewise.
690 * config/sparc/rtemself.h: Likewise.
692 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
695 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
696 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
697 * tree-ssa-structalias.c (alias_get_name): Same.
699 2017-07-25 Bin Cheng <bin.cheng@arm.com>
702 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
703 instructions if no du chain is found.
705 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
707 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
709 2017-07-25 Richard Biener <rguenther@suse.de>
712 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
715 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
717 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
719 (ISA_2_7_MASKS_SERVER): Likewise.
720 (ISA_3_0_MASKS_IEEE): Likewise.
721 (OTHER_P8_VECTOR_MASKS): Likewise.
722 (OTHER_VSX_VECTOR_MASKS): Likewise.
723 (POWERPC_MASKS): Likewise.
724 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
725 duplicate list of options.
726 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
727 explicit -mupper-regs options.
728 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
729 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
730 alias for -mupper-regs-df.
731 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
732 (rs6000_init_hard_regno_mode_ok): Likewise.
733 (rs6000_option_override_internal): Likewise.
734 (rs6000_opt_masks): Likewise.
735 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
736 options in terms of whether -mvsx or -mpower8-vector was used.
737 (TARGET_UPPER_REGS_DI): Likewise.
738 (TARGET_UPPER_REGS_SF): Likewise.
739 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
740 -mupper-regs-* options.
742 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
744 * passes.c (emergency_dump_function): Print some empty lines and a
745 header before the RTL dump.
747 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
749 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
751 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
754 * config/aarch64/aarch64.c (aarch64_classify_symbol):
755 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
757 2017-07-24 Carl Love <cel@us.ibm.com>
759 * config/rs6000/rs6000-c.c: Add support for built-in functions
760 vector float vec_extract_fp32_from_shorth (vector unsigned short);
761 vector float vec_extract_fp32_from_shortl (vector unsigned short);
762 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
763 vec_extract_fp_from_shortl): Add defines for the two builtins.
764 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
765 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
767 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
768 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
769 * doc/extend.texi: Update the built-in documentation file for the
770 new built-in function.
772 2017-07-24 Jakub Jelinek <jakub@redhat.com>
775 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
777 * doc/generic.texi: Likewise.
778 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
779 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
781 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
783 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
784 (aarch64_mls_elt_merge<mode>): Likewise.
786 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
788 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
791 2017-07-23 Michael Collison <michael.collison@arm.com>
793 * config/arm/arm.c (arm_option_override): Deprecate
794 use of -mstructure-size-boundary.
795 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
796 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
798 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
801 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
802 Reduce cost estimate for direct moves.
804 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
807 * config/i386/i386.c (ix86_option_override_internal): Disable
808 BMI, BMI2 and TBM instructions for -m16.
810 2017-07-21 Carl Love <cel@us.ibm.com>
812 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
813 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
814 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
815 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
816 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
817 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
818 VMULOSW): New enum "unspec" values.
819 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
820 altivec_vmulosw): New patterns.
821 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
822 VMULOSW): Add definitions.
824 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
826 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
828 * config/aarch64/aarch64-options-extensions.def (rdma); New.
829 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
830 (AARCH64_FL_V8_1): Renumber.
831 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
832 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
833 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
834 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
835 rdma to feature modifiers list.
837 2017-07-21 Yury Gribov <tetra2005@gmail.com>
840 * ipa-visibility (function_and_variable_visibility): Convert
841 recursive PLT call to direct call if appropriate.
843 2017-07-21 Andrew Pinski <apinski@cavium.com>
845 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
846 operand 1 to see if the types precision matches.
847 * fold-const.c (operand_equal_p): Likewise.
849 2017-07-21 Richard Biener <rguenther@suse.de>
851 PR tree-optimization/81303
852 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
853 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
854 (vect_peeling_hash_get_lowest_cost): Adjust.
855 (vect_enhance_data_refs_alignment): Likewise. Use
856 vect_get_peeling_costs_all_drs to compute the penalty for no
857 peeling to match up costs.
859 2017-07-21 Richard Biener <rguenther@suse.de>
861 PR tree-optimization/81500
862 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
863 we didn't identify a reduction path.
865 2017-07-21 Tom de Vries <tom@codesourcery.com>
866 Cesar Philippidis <cesar@codesourcery.com>
868 PR gcov-profile/81442
869 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
872 2017-07-21 Tom de Vries <tom@codesourcery.com>
875 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
877 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
878 nvptx_override_options_after_change.
880 2017-07-21 Ulrich Drepper <drepper@redhat.com>
882 * dwarf2out.c (output_file_names): Avoid double testing for
885 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
887 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
889 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
891 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
893 (try_crossjump_to_edge): Do not punt on partitioned functions.
895 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
897 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
898 Put all BBs reachable only via paths crossing cold region to cold
900 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
902 2016-07-21 Richard Biener <rguenther@suse.de>
904 PR tree-optimization/81303
905 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
906 into account prologue and epilogue iterations when raising
907 min_profitable_iters to sth at least covering one vector iteration.
909 2017-07-21 Tamar Christina <tamar.christina@arm.com>
911 * config/arm/arm.c (arm_test_cpu_arch_dat):
914 2017-07-20 Nathan Sidwell <nathan@acm.org>
917 * tree.h (TYPE_METHODS): Delete.
918 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
919 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
920 (dbxout_type_methods): Scan TYPE_FIELDS.
921 (dbxout_type): Don't check TYPE_METHODS here.
922 * function.c (use_register_for_decl): Always ignore register for
923 class types when not optimizing.
924 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
925 * tree.c (free_lang_data_in_type): Stitch out member functions and
926 templates from TYPE_FIELDS.
927 (build_distinct_type_copy, verify_type_variant,
928 verify_type): Member fns are on TYPE_FIELDS.
929 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
930 * tree-pretty-print.c (dump_generic_node): Likewise.
932 2017-07-20 Jakub Jelinek <jakub@redhat.com>
935 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
936 V2TImode and V4TImode.
937 (ix86_expand_vector_extract): Likewise.
938 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
939 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
940 (ssescalarmode): Handle V4TImode and V2TImode.
941 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
942 (*vec_extractv2ti, *vec_extractv4ti): New insns.
943 (VEXTRACTI128_MODE): New mode iterator.
944 (splitter for *vec_extractv?ti first element): New.
945 (VEC_INIT_MODE): New mode iterator.
946 (vec_init<mode>): Consolidate 3 expanders into one using
947 VEC_INIT_MODE mode iterator.
949 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
951 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
952 non_spilled_static_chain_regno_p.
954 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
956 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
958 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
960 * bb-reorder.c (connect_traces): Allow copying of blocks within
963 2017-07-20 Richard Biener <rguenther@suse.de>
965 * gimple.h (gimple_phi_result): Add gphi * overload.
966 (gimple_phi_result_ptr): Likewise.
967 (gimple_phi_arg): Likewise. Adjust index assert to only
968 allow actual argument accesses rather than all slots available
970 (gimple_phi_arg_def): Add gphi * overload.
971 * tree-phinodes.c (make_phi_node): Initialize only actual
973 (resize_phi_node): Clear memory not covered by old node,
974 do not initialize excess argument slots.
975 (reserve_phi_args_for_new_edge): Initialize new argument slot
978 2017-07-20 Bin Cheng <bin.cheng@arm.com>
980 PR tree-optimization/81388
982 2016-07-21 Bin Cheng <bin.cheng@arm.com>
984 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
985 by removing computation of may_be_zero.
987 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
988 Tom de Vries <tom@codesourcery.com>
991 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
992 when gimple level profile disagrees with what RTL expander did.
994 2017-07-20 Richard Biener <rguenther@suse.de>
996 PR tree-optimization/61171
997 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
998 (vect_analyze_stmt): Add slp instance parameter.
999 (vectorizable_reduction): Likewise.
1000 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
1001 (vect_is_simple_reduction): Deal with chains not detected
1002 as SLP reduction chain, specifically not properly associated
1003 chains containing a mix of plus/minus.
1004 (get_reduction_op): Remove.
1005 (get_initial_defs_for_reduction): Simplify, pass in whether
1006 this is a reduction chain, pass in the SLP node for the PHIs.
1007 (vect_create_epilog_for_reduction): Get the SLP instance as
1009 (vectorizable_reduction): Get the SLP instance as arg.
1010 During analysis remember the SLP node with the PHIs in the
1011 instance. Simplify getting at the vectorized reduction PHIs.
1012 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
1013 through SLP instance.
1014 (vect_slp_analyze_operations): Likewise.
1015 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
1016 (vect_transform_stmt): Likewise.
1018 2017-07-20 Tom de Vries <tom@codesourcery.com>
1020 PR tree-optimization/81489
1021 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
1022 read of phi arg location to before loop that modifies phi.
1024 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
1026 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
1029 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1032 * except.c (execute): Fix ordering issue.
1034 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
1036 PR rtl-optimization/81423
1037 * combine.c (make_compound_operation_int): Don't try to optimize
1038 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
1040 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
1042 PR rtl-optimization/81423
1043 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
1044 with a constant that is -1 in the truncated to mode.
1046 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1048 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
1049 (determine_unlikely_bbs): ... here.
1050 * predict.h (propagate_unlikely_bbs_forward): Declare.
1051 * cfgexpand.c (pass_expand::execute): Use it.
1052 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
1054 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
1055 propagate_unlikely_bbs_forward.
1057 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1060 * except.c (maybe_add_nop_after_section_switch): New function.
1063 2017-07-19 Tom de Vries <tom@codesourcery.com>
1065 * gimple.h (gimple_phi_set_arg): Make assert more strict.
1067 2017-07-19 Tom de Vries <tom@codesourcery.com>
1069 * gimple.h (gimple_phi_arg): Make assert more strict.
1071 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
1073 * config.gcc (powerpc*-*-*): Add mmintrin.h.
1074 * config/rs6000/mmintrin.h: New file.
1075 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
1077 2017-07-19 Jakub Jelinek <jakub@redhat.com>
1079 PR tree-optimization/81346
1080 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
1082 2017-07-19 Tom de Vries <tom@codesourcery.com>
1084 * config/nvptx/nvptx.md (VECIM): Add V2DI.
1086 2017-07-19 Tom de Vries <tom@codesourcery.com>
1088 * config/nvptx/nvptx-modes.def: Add V2DImode.
1089 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
1090 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
1091 (nvptx_output_mov_insn): Handle lack of mov.b128.
1092 (nvptx_print_operand): Handle 'H' and 'L' codes.
1093 (nvptx_vector_mode_supported): Allow V2DImode.
1094 (nvptx_preferred_simd_mode): New function.
1095 (nvptx_data_alignment): New function.
1096 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
1097 nvptx_preferred_simd_mode.
1098 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
1100 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
1102 2017-07-19 Tom de Vries <tom@codesourcery.com>
1104 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
1105 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
1106 (nvptx_vector_mode_supported): New function. Allow V2SImode.
1107 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
1108 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
1109 (mov<VECIM>_insn): New define_insn.
1110 (define_expand "mov<VECIM>): New define_expand.
1112 2017-07-19 Tom de Vries <tom@codesourcery.com>
1114 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
1116 2017-07-19 Jakub Jelinek <jakub@redhat.com>
1118 PR tree-optimization/81346
1119 * fold-const.h (fold_div_compare, range_check_type): Declare.
1120 * fold-const.c (range_check_type): New function.
1121 (build_range_check): Use range_check_type.
1122 (fold_div_compare): No longer static, rewritten into
1123 a match.pd helper function.
1124 (fold_comparison): Don't call fold_div_compare here.
1125 * match.pd (X / C1 op C2): New optimization using fold_div_compare
1128 2017-07-19 Nathan Sidwell <nathan@acm.org>
1130 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
1131 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
1132 * tree.c (find_decls_types_r, verify_type): Use
1133 TYPE_{MIN,MAX}_VALUE_RAW.
1134 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
1135 (hash_tree): Likewise.
1136 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
1138 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
1141 2017-07-18 Tom de Vries <tom@codesourcery.com>
1144 * omp-expand.c (expand_omp_for_static_chunk): Handle
1145 equal-argument loop exit phi.
1147 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
1150 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
1151 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
1152 operand 2 predicate.
1153 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
1154 operand 2 predicate.
1155 (ror,rol -> rorx splitters): Use const_int_operand as
1156 operand 2 predicate.
1158 2017-06-18 Richard Biener <rguenther@suse.de>
1160 PR tree-optimization/81410
1161 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
1162 the gap in the ! slp_perm SLP case after each group.
1164 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1167 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
1170 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1173 * predict.c (set_even_probabilities): Cleanup; do not affect
1174 probabilities that are already known.
1175 (combine_predictions_for_bb): Call even when count is set.
1177 2017-07-18 Nathan Sidwell <nathan@acm.org>
1179 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
1182 2017-07-18 Bin Cheng <bin.cheng@arm.com>
1185 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
1186 optimization for loop niter analysis.
1188 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
1191 * config/avr/avr.c (avr_optimize_casesi): Don't use
1192 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
1194 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1196 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
1197 body_cost_vec from _vect_peel_extended_info.
1198 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
1199 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
1202 2017-07-18 Bin Cheng <bin.cheng@arm.com>
1204 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
1206 2017-07-18 Richard Biener <rguenther@suse.de>
1208 PR tree-optimization/80620
1209 PR tree-optimization/81403
1210 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
1211 info when re-using a VN table entry.
1213 2017-07-18 Richard Biener <rguenther@suse.de>
1215 PR tree-optimization/81418
1216 * tree-vect-loop.c (vectorizable_reduction): Properly compute
1217 vectype_in. Verify that with lane-reducing reduction operations
1218 we have a single def-use cycle.
1220 2017-07-17 Carl Love <cel@us.ibm.com>
1222 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
1224 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1225 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
1226 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
1227 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
1228 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
1229 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
1230 VMULOSW): New enum "unspec" values.
1231 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
1232 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
1233 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
1234 altivec_vmulosw): New patterns.
1235 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
1236 VMULOSW): Add definitions.
1237 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
1239 * config/alpha/alpha.c: Include predict.h.
1241 2017-07-17 Yury Gribov <tetra2005@gmail.com>
1243 * tree-vrp.c (compare_assert_loc): Fix comparison function
1244 to return predictable results.
1246 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1248 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
1251 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
1252 * doc/invoke.texi (mexpand-adddi): Update text.
1254 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1256 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
1257 that also clobbers the CC register. The old expand code is moved
1259 (*arc_clzsi2): ... here.
1260 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
1261 the CC register. The old expand code is moved to ...
1262 (arc_ctzsi2): ... here.
1264 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1266 * config/arc/arc.opt (mindexed-loads): Use initial value
1267 TARGET_INDEXED_LOADS_DEFAULT.
1268 (mauto-modify-reg): Use initial value
1269 TARGET_AUTO_MODIFY_REG_DEFAULT.
1270 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
1271 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
1272 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
1273 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
1275 2017-07-17 Martin Liska <mliska@suse.cz>
1278 * opts.c (finish_options): Do not allow -fgnu-tm
1279 w/ -fsanitize={kernel-,}address. Say sorry.
1281 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1284 * tree-loop-distribution.c (classify_partition): Only assert on
1285 numer of iterations.
1286 (merge_dep_scc_partitions): Delete prameter. Update function call.
1287 (distribute_loop): Remove code handling loop with unknown niters.
1288 (pass_loop_distribution::execute): Skip loop with unknown niters.
1290 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1293 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
1294 function sort_partitions_by_post_order.
1296 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1298 PR tree-optimization/81374
1299 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
1300 the max index of basic blocks, rather than number of basic blocks.
1302 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1304 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
1306 (arc_legitimate_pic_operand_p): Likewise.
1307 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
1309 (arc_needs_pcl_p): Likewise.
1310 (arc_legitimate_pc_offset_p): Likewise.
1311 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
1312 function is also used in constrains.md.
1313 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
1314 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
1315 PLUS. Only return true/false in known cases, otherwise assert.
1316 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
1317 is already called in arc_legitimate_constant_p.
1318 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
1320 (LEGITIMATE_PIC_OPERAND_P): Use
1321 arc_raw_symbolic_reference_mentioned_p function.
1322 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
1327 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1328 Andrew Burgess <andrew.burgess@embecosm.com>
1330 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
1331 (arc_return_address_register): New function.
1332 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
1333 (arc_handle_fndecl_attribute): Add naked attribute.
1334 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
1335 (TARGET_WARN_FUNC_RETURN): Likewise.
1336 (arc_allocate_stack_slots_for_args): New function.
1337 (arc_warn_func_return): Likewise.
1338 (machine_function): Change type fn_type.
1339 (arc_compute_function_type): Consider new naked function type,
1340 change function return type.
1341 (arc_must_save_register): Adapt to handle new
1342 arc_compute_function_type's return type.
1343 (arc_expand_prologue): Likewise.
1344 (arc_expand_epilogue): Likewise.
1345 (arc_return_address_regs): Delete.
1346 (arc_return_address_register): New function.
1347 (arc_epilogue_uses): Use above function.
1348 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
1349 (arc_function_type): Change encoding, add naked type.
1350 (ARC_INTERRUPT_P): Change to handle the new encoding.
1351 (ARC_FAST_INTERRUPT_P): Likewise.
1352 (ARC_NORMAL_P): Define.
1353 (ARC_NAKED_P): Likewise.
1354 (arc_compute_function_type): Delete prototype.
1355 * config/arc/arc.md (in_ret_delay_slot): Use
1356 arc_return_address_register function.
1357 (simple_return): Likewise.
1358 (p_return_i): Likewise.
1360 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1362 PR tree-optimization/81428
1363 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
1364 can't be built for those types.
1366 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
1368 Remove stuff dead since r239246.
1370 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
1371 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
1372 (avr_inform_devices): Remove dead stuff.
1374 2017-07-17 Tamar Christina <tamar.christina@arm.com>
1376 * config/arm/arm_neon.h: Fix softp typo.
1378 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1380 PR tree-optimization/81365
1381 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
1382 aggregate moves onto bb predecessor edges, make sure there are no
1383 loads that could alias the lhs in between the start of bb and the
1386 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
1389 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
1390 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
1391 [LSHIFTRT, outer_code = TRUNCATE]: Same.
1393 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1395 PR tree-optimization/81396
1396 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
1397 (init_symbolic_number): Initialize it to 1.
1398 (perform_symbolic_merge): Add n_ops from both operands into the new
1400 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
1401 without base_addr as useless if they need more than one operation.
1402 (bswap_replace): Handle !bswap case for NULL base_addr.
1404 2017-07-17 Tom de Vries <tom@codesourcery.com>
1407 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
1410 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
1412 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
1413 conditional builtin define __FIX_LEON3FT_B2BST.
1415 2017-07-17 Daniel Cederman <cederman@gaisler.com>
1417 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
1418 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
1421 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1423 PR rtl-optimization/81424
1424 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
1425 to remove potential trapping from operands if -fnon-call-exceptions.
1427 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1429 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
1430 profile_proability for scalling.
1431 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
1433 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1435 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
1437 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1439 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
1440 fixpoint arithmetics.
1442 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1444 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
1445 fixpoint arithmetics.
1447 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1449 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
1450 fixpoint arithmetics.
1452 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1454 * profile-count.h (profile_probability::from_reg_br_prob_note,
1455 profile_probability::to_reg_br_prob_note): New functions.
1456 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
1457 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
1458 * predict.c (probability_reliable_p): Update.
1459 (edge_probability_reliable_p): Update.
1460 (br_prob_note_reliable_p): Update.
1461 (invert_br_probabilities): Update.
1462 (add_reg_br_prob_note): New function.
1463 (combine_predictions_for_insn): Update.
1464 * asan.c (asan_clear_shadow): Update.
1465 * cfgbuild.c (compute_outgoing_frequencies): Update.
1466 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
1467 (update_br_prob_note): Update.
1468 (rtl_verify_edges): Update.
1469 (purge_dead_edges): Update.
1470 (fixup_reorder_chain): Update.
1471 * emit-rtl.c (try_split): Update.
1472 * ifcvt.c (cond_exec_process_insns): Update.
1473 (cond_exec_process_if_block): Update.
1474 (dead_or_predicable): Update.
1475 * internal-fn.c (expand_addsub_overflow): Update.
1476 (expand_neg_overflow): Update.
1477 (expand_mul_overflow): Update.
1478 * loop-doloop.c (doloop_modify): Update.
1479 * loop-unroll.c (compare_and_jump_seq): Update.
1480 * optabs.c (emit_cmp_and_jump_insn_1): Update.
1481 * predict.h: Update.
1482 * reorg.c (mostly_true_jump): Update.
1484 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
1485 * config/alpha/alpha.c (emit_unlikely_jump): Update.
1486 * config/arc/arc.c: (emit_unlikely_jump): Update.
1487 * config/arm/arm.c: (emit_unlikely_jump): Update.
1488 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
1489 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
1490 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
1491 (ix86_print_operand): Update.
1492 (ix86_split_fp_branch): Update.
1493 (predict_jump): Update.
1494 * config/ia64/ia64.c (ia64_print_operand): Update.
1495 * config/mmix/mmix.c (mmix_print_operand): Update.
1496 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
1497 (rs6000_expand_split_stack_prologue): Update.
1498 * config/rs6000/rs6000.c: Update.
1499 * config/s390/s390.c (s390_expand_vec_strlen): Update.
1500 (s390_expand_vec_movstr): Update.
1501 (s390_expand_cs_tdsi): Update.
1502 (s390_expand_split_stack_prologue): Update.
1503 * config/sh/sh.c (sh_print_operand): Update.
1504 (expand_cbranchsi4): Update.
1505 (expand_cbranchdi4): Update.
1506 * config/sparc/sparc.c (output_v9branch): Update.
1507 * config/spu/spu.c (get_branch_target): Update.
1508 (ea_load_store_inline): Update.
1509 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
1510 * config/tilepro/tilepro.c: Update.
1512 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1514 * gimplify.c (mostly_copy_tree_r): Revert latest change.
1515 (gimplify_save_expr): Likewise.
1517 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1519 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
1521 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1523 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
1525 * timevar.def (TV_IPA_FNSUMMARY): Define.
1527 2017-07-16 Daniel Cederman <cederman@gaisler.com>
1529 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
1530 to back store errata sensitive sequence from being generated.
1531 (sqrtdf2_fix): Likewise.
1533 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1535 * tree-ssa-threadupdate.c (compute_path_counts,
1536 update_joiner_offpath_counts): Use profile_probability.
1538 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
1541 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1543 * config/arm/arm-c.c (arm_cpu_builtins): Define
1544 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1546 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
1548 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1549 array entries to represent __ieee128 versions of the
1550 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
1551 scalar_extract_sig, and scalar_insert_exp built-in functions.
1552 (altivec_resolve_overloaded_builtin): Add special case handling
1553 for the __builtin_scalar_insert_exp function, as represented by
1554 the P9V_BUILTIN_VEC_VSIEDP constant.
1555 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
1556 exponent support for __ieee128 argument.
1557 (VSESQP): Add scalar extract signature support for __ieee128
1559 (VSTDCNQP): Add scalar test negative support for __ieee128
1561 (VSIEQP): Add scalar insert exponent support for __int128 argument
1562 with __ieee128 result.
1563 (VSIEQPF): Add scalar insert exponent support for __ieee128
1564 argument with __ieee128 result.
1565 (VSTDCQP): Add scalar test data class support for __ieee128
1567 (VSTDCNQP): Add overload support for scalar test negative with
1569 (VSTDCQP): Add overload support for scalar test data class
1571 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
1573 (UNSPEC_VSX_SIEXPQP): New constant.
1574 (xsxexpqp): New insn for VSX scalar extract exponent quad
1576 (xsxsigqp): New insn for VSX scalar extract significand quad
1578 (xsiexpqpf): New insn for VSX scalar insert exponent quad
1579 precision with floating point argument.
1580 (xststdcqp): New expand for VSX scalar test data class quad
1582 (xststdcnegqp): New expand for VSX scalar test negative quad
1584 (xststdcqp): New insn to match expansions for VSX scalar test data
1585 class quad precision and VSX scalar test negative quad precision.
1586 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
1587 special case operand checking to enforce that second operand of
1588 VSX scalar test data class with quad precision argument is a 7-bit
1590 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
1591 prototypes and descriptions of __ieee128 versions of
1592 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
1593 scalar_test_data_class, and scalar_test_neg built-in functions.
1595 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1597 PR tree-optimization/81162
1598 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
1599 replace a negate with an add.
1601 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
1603 * doc/invoke.texi (arm/-mcpu): Document +crypto.
1605 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1607 * config/arm/arm-c.c (arm_cpu_builtins): Define
1608 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1610 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1612 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
1613 (armv8-r): Set ARM Cortex-R52 as default CPU.
1614 * config/arm/arm-tables.opt: Regenerate.
1615 * config/arm/arm-tune.md: Regenerate.
1616 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
1618 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
1619 extension for -mcpu=cortex-r52.
1621 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1623 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
1624 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
1625 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
1626 (fp-armv8): Define it as FP_ARMv8 only.
1627 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
1628 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
1630 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
1631 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
1632 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
1633 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
1634 than TARGET_FPU_ARMV8.
1635 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
1636 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
1637 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
1639 * config/arm/neon.md (neon_vrint): Likewise.
1640 (neon_vcvt): Likewise.
1641 (neon_<fmaxmin_op><mode>): Likewise.
1642 (<fmaxmin><mode>3): Likewise.
1643 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
1644 * config/arm/predicates.md (arm_cond_move_operator): Check against
1645 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
1647 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
1649 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
1652 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1654 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
1655 loop in comment with memset.
1657 2017-07-14 Martin Liska <mliska@suse.cz>
1659 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
1660 * dwarf2out.c (is_java): Remove the function.
1661 (output_pubname): Remove usage of the function.
1662 (lower_bound_default): Remove usage of DW_LANG_Java.
1663 (gen_compile_unit_die): Likewise.
1664 * gcc.c: Remove compiler defaults for .java and .zip files.
1665 * gimple-expr.c (remove_suffix): Change as there's no longer
1666 extension than 4-letter one.
1667 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
1668 (gimplify_save_expr): Likewise.
1669 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
1670 as it's possible even for other languages than Java.
1671 * langhooks.h (struct lang_hooks): Remove Java from a comment.
1672 * lto-opts.c (lto_write_options): Remove reference to Java.
1673 * opts.c (strip_off_ending): Update file extension handling.
1674 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
1675 * tree-eh.c (lower_resx): Likewise.
1676 * tree.c (free_lang_data_in_type): Remove dead code.
1677 (find_decls_types_r): Likewise.
1678 (build_common_builtin_nodes): Remove Java from a comment.
1679 (verify_type): Remove dead code.
1680 * varasm.c (assemble_external): Remove Java from a comment.
1682 2017-07-14 Martin Liska <mliska@suse.cz>
1684 * opts.c (finish_options): Add quotes.
1685 (common_handle_option): Likewise.
1687 2017-07-14 Martin Liska <mliska@suse.cz>
1689 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
1690 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
1692 * dwarf2out.c (lower_bound_default): Do not handle
1694 (gen_compile_unit_die): Likewise.
1695 * gcc.c: Remove default extension binding for GNU Pascal.
1696 * stmt.c: Remove Pascal language from a comment.
1697 * xcoffout.c: Likewise.
1699 2017-07-13 David Malcolm <dmalcolm@redhat.com>
1702 * diagnostic-show-locus.c (fixit_cmp): New function.
1703 (layout::layout): Sort m_fixit_hints.
1704 (column_range::column_range): Assert that the values are valid.
1705 (struct char_span): New struct.
1706 (correction::overwrite): New method.
1707 (struct source_line): New struct.
1708 (line_corrections::add_hint): Add assertions. Reimplement memcpy
1709 calls in terms of classes source_line and char_span, and
1710 correction::overwrite.
1711 (selftest::test_overlapped_fixit_printing_2): New function.
1712 (selftest::diagnostic_show_locus_c_tests): Call it.
1714 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
1716 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
1717 early if there is no lhs.
1719 2017-07-13 Martin Liska <mliska@suse.cz>
1721 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
1722 (gen_reference_type_die): Likewise.
1723 * stor-layout.c: Remove Pascal-related comment.
1725 2017-07-13 Martin Liska <mliska@suse.cz>
1727 * opts.c (finish_options): Add quotes to error messages.
1728 (parse_sanitizer_options): Likewise.
1730 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
1732 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
1734 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
1736 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
1738 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
1740 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
1742 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
1744 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1747 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
1748 provides the hardware capability bits, define the macro
1749 __BUILTIN_CPU_SUPPORTS__.
1750 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
1751 if GLIBC does not provide the hardware capability bits. Add a
1752 gcc_unreachable call if the built-in cpu function is neither
1753 __builtin_cpu_is nor __builtin_cpu_supports.
1754 (rs6000_get_function_versions_dispatcher): Change the warning
1755 that an old GLIBC is used which does not export the capability
1756 bits to be an error.
1757 * doc/extend.texi (target_clones attribute): Document the
1758 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
1759 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
1760 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
1761 the macros defined by GCC if the newer GLIBC is available.
1763 2017-07-12 Jeff Law <law@redhat.com>
1765 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
1766 remaining includes slightly.
1767 * config/riscv/riscv-builtins.c: Include profile-count.h.
1769 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1772 * config/avr/avr.c (avr_set_current_function): In diagnostic
1773 messages: Quote keywords and (parts of) identifiers.
1774 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
1777 2017-07-12 Carl Love <cel@us.ibm.com>
1779 * config/rs6000/rs6000-c.c: Add support for built-in functions
1780 vector bool char vec_revb (vector bool char);
1781 vector bool short vec_revb (vector short char);
1782 vector bool int vec_revb (vector bool int);
1783 vector bool long long vec_revb (vector bool long long);
1784 * doc/extend.texi: Update the built-in documentation file for the
1785 new built-in functions.
1787 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1789 * config/s390/s390.md: Remove movcc splitter.
1791 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1793 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
1794 load/store on condition.
1796 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1799 * config/avr/avr.c (avr_encode_section_info)
1800 [progmem && !TREE_READONLY]: Error if progmem object needs
1803 2017-07-11 Michael Collison <michael.collison@arm.com>
1805 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
1808 2017-07-11 Carl Love <cel@us.ibm.com>
1810 * config/rs6000/rs6000-c.c: Add support for builtins
1811 vector unsigned int vec_parity_lsbb (vector signed int);
1812 vector unsigned int vec_parity_lsbb (vector unsigned int);
1813 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
1814 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
1815 vector unsigned long long vec_parity_lsbb (vector signed long long);
1816 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
1817 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
1818 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
1819 * doc/extend.texi: Update the built-in documentation file for the
1820 new built-in functions.
1822 2017-07-11 David Malcolm <dmalcolm@redhat.com>
1824 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
1825 (layout::m_primary_loc): New field.
1826 (layout::layout): Initialize new field. Move location filtering
1827 logic from here to...
1828 (layout::maybe_add_location_range): ...this new method. Add
1829 support for filtering to just the lines already specified by other
1831 (layout::will_show_line_p): New method.
1832 (gcc_rich_location::add_location_if_nearby): New method.
1833 (selftest::test_add_location_if_nearby): New test function.
1834 (selftest::diagnostic_show_locus_c_tests): Call it.
1835 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
1838 2017-07-11 Tom de Vries <tom@codesourcery.com>
1840 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
1841 (bb_first_real_insn): New function.
1842 (nvptx_single): Add extra initialization of broadcasted condition
1845 2017-07-11 Nathan Sidwell <nathan@acm.org>
1847 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
1849 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
1851 * doc/extend.texi (AVR Function Attributes): Remove weblink to
1852 Binutils doc as TEXI will mess them up.
1853 * doc/invoke.texi (AVR Options): Same here.
1855 2017-07-11 Daniel Cederman <cederman@gaisler.com>
1857 * config/sparc/sparc.opt (mfix-ut700): New option.
1858 (mfix-gr712rc): Likewise.
1859 (sparc_fix_b2bst): New variable.
1860 * doc/invoke.texi (SPARC options): Document them.
1861 (ARM options): Fix warnings.
1862 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
1863 instructions to prevent sequences that can trigger the store-store
1864 errata for certain LEON3FT processors.
1865 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
1866 (sparc_option_override): Set sparc_fix_b2bst appropriately.
1867 * config/sparc/sparc.md (fix_b2bst): New attribute.
1868 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
1870 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
1873 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
1875 (*rsqrtsf2_sse): Ditto.
1877 (div<mode>3): Macroize insn from divdf3 and divsf3
1878 using MODEF mode iterator.
1880 2017-07-10 Martin Sebor <msebor@redhat.com>
1882 PR tree-optimization/80397
1883 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
1884 instead of testing for equality to INTEGER_TYPE.
1886 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
1888 * config.gcc: Remove uclibc from arc target spec.
1890 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
1892 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
1894 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1897 * lto-wrapper.c (remove_option): New function.
1898 (merge_and_complain): Merge PIC/PIE options more realistically.
1900 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
1902 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
1906 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
1907 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
1908 * config.in: Regenerate.
1909 * configure: Regenerate.
1910 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
1911 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
1912 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
1913 (TARGET_GASISR_PROLOGUES): ...target mask.
1914 * common/config/avr/avr-common.c
1915 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
1916 Set -mgas-isr-prologues.
1917 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
1918 INSERT_PASS_BEFORE for it.
1919 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
1920 * config/avr/avr.c (avr_option_override)
1921 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
1922 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
1923 (avr_attribute_table) <no_gccisr>: Add new function attribute.
1924 (avr_set_current_function) <is_no_gccisr>: Init machine field.
1925 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
1927 (make_avr_pass_pre_proep): New function.
1928 (emit_push_sfr) <treg>: Add argument to function and use it
1930 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
1931 and set machine->gasisr.yes.
1932 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
1933 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
1934 __gcc_isr.n_pushed to .L__stack_usage.
1935 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
1936 (avr_asm_final_postscan_insn): ...this new static function.
1937 * config/avr/avr.h (machine_function)
1938 <is_no_gccisr, use_L__stack_usage>: New fields.
1939 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
1940 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
1941 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
1942 (gasisr, *gasisr): New expander and insn.
1943 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
1944 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
1945 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
1947 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
1949 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
1952 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
1954 Move jump-tables out of .text again.
1957 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
1958 (ASM_OUTPUT_ADDR_VEC): New function.
1959 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
1960 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
1961 INSN_ADDRESSes as asm comment.
1962 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
1963 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
1964 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
1965 * config/avr/avr.md (*tablejump): Adjust comment.
1966 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
1967 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
1969 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
1970 (avr_output_addr_vec): New proto.
1971 (avr_log_t) <insn_addresses>: New field.
1973 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
1976 * config/i386/i386.c (ix86_function_arg_advance): Set
1977 outgoing_args_on_stack to true if there are outgoing arguments
1979 (ix86_function_arg): Likewise.
1980 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
1981 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
1982 * config/i386/i386.h (machine_function): Add
1983 outgoing_args_on_stack.
1985 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
1987 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
1989 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
1991 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
1993 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
1994 (REAL_H): Remove $(MACHMODE_H).
1995 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
1997 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
1998 $(MACHMODE_H) and double-int.h.
1999 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
2001 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
2004 2017-07-07 Andrew Pinski <apinski@cavium.com>
2006 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
2007 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
2009 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
2011 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
2012 Add warning if GCC was not configured to link against a GLIBC that
2013 exports the hardware capability bits.
2014 (make_resolver_func): Make resolver function private and not a
2015 COMDAT function. Create the name with clone_function_name instead
2016 of make_unique_name.
2019 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
2020 correct operand in doing the split.
2022 2017-07-07 Carl Love <cel@us.ibm.com>
2024 * config/rs6000/rs6000-c: Add support for built-in function
2025 vector unsigned short vec_pack_to_short_fp32 (vector float,
2027 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
2028 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
2029 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
2030 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
2031 (convert_4f32_8i16): Add define_expand.
2032 * doc/extend.texi: Update the built-in documentation file for the
2033 new built-in function.
2035 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2037 * config/sparc/m8.md: New file.
2038 * config/sparc/sparc.md: Include m8.md.
2040 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2042 * config/sparc/sparc.opt: New option -mvis4b.
2043 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
2044 (sparc_option_override): Handle VIS4B.
2045 (enum sparc_builtins): Define
2046 SPARC_BUILTIN_DICTUNPACK{8,16,32},
2047 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
2048 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
2049 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
2050 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
2051 (check_constant_argument): New function.
2052 (sparc_vis_init_builtins): Define builtins
2053 __builtin_vis_dictunpack{8,16,32},
2054 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
2055 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
2056 __builtin_vis_fpcmpde{8,16,32}shl and
2057 __builtin_vis_fpcmpur{8,16,32}shl.
2058 (sparc_expand_builtin): Check that the constant operands to
2059 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
2060 constant and in range.
2061 * config/sparc/sparc-c.c (sparc_target_macros): Handle
2063 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
2064 (SPARC_IMM5_P): Likewise.
2065 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
2066 (enabled): Handle vis4b.
2067 (UNSPEC_DICTUNPACK): New unspec.
2068 (UNSPEC_FPCMPSHL): Likewise.
2069 (UNSPEC_FPUCMPSHL): Likewise.
2070 (UNSPEC_FPCMPDESHL): Likewise.
2071 (UNSPEC_FPCMPURSHL): Likewise.
2072 (cpu_feature): New CPU feature `vis4b'.
2073 (dictunpack{8,16,32}): New insns.
2074 (FPCSMODE): New mode iterator.
2075 (fpcscond): New code iterator.
2076 (fpcsucond): Likewise.
2077 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
2078 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
2079 (fpcmpde{8,16,32}{si,di}shl): Likewise.
2080 (fpcmpur{8,16,32}{si,di}shl): Likewise.
2081 * config/sparc/constraints.md: Define constraints `q' for unsigned
2082 2-bit integer constants and `t' for unsigned 5-bit integer
2084 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
2086 (imm5_operand_dictunpack16): Likewise.
2087 (imm5_operand_dictunpack32): Likewise.
2088 (imm2_operand): Likewise.
2089 * doc/invoke.texi (SPARC Options): Document -mvis4b.
2090 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
2091 ditunpack* and fpcmp*shl builtins.
2093 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2095 * config.gcc: Handle m8 in --with-{cpu,tune} options.
2096 * config.in: Add HAVE_AS_SPARC6 define.
2097 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
2099 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
2101 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
2102 (CPP_CPU_SPEC): Handle m8.
2103 (ASM_CPU_SPEC): Likewise.
2104 * config/sparc/sparc-opts.h (enum processor_type): Add
2106 * config/sparc/sparc.c (m8_costs): New struct.
2107 (sparc_option_override): Handle TARGET_CPU_m8.
2108 (sparc32_initialize_trampoline): Likewise.
2109 (sparc64_initialize_trampoline): Likewise.
2110 (sparc_issue_rate): Likewise.
2111 (sparc_register_move_cost): Likewise.
2112 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
2113 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
2114 (ASM_CPU64_DEFAULT_SPEC): Likewise.
2115 (CPP_CPU_SPEC): Handle M8.
2116 (ASM_CPU_SPEC): Likewise.
2117 (AS_M8_FLAG): Define.
2118 * config/sparc/sparc.md: Add m8 to the cpu attribute.
2119 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
2120 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
2122 * configure: Regenerate.
2123 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
2126 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2128 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
2130 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
2131 ("*movdi_insn_sp32"): Do not set v3pipe.
2132 ("*movsi_insn"): Likewise.
2133 ("*movdi_insn_sp64"): Likewise.
2134 ("*movsf_insn"): Likewise.
2135 ("*movdf_insn_sp32"): Likewise.
2136 ("*movdf_insn_sp64"): Likewise.
2137 ("*zero_extendsidi2_insn_sp64"): Likewise.
2138 ("*sign_extendsidi2_insn"): Likewise.
2139 ("*mov<VM32:mode>_insn"): Likewise.
2140 ("*mov<VM64:mode>_insn_sp64"): Likewise.
2141 ("*mov<VM64:mode>_insn_sp32"): Likewise.
2142 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
2143 ("<vlop:code><VL:mode>3"): Likewise.
2144 ("*not_<vlop:code><VL:mode>3"): Likewise.
2145 ("*nand<VL:mode>_vis"): Likewise.
2146 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
2147 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
2148 ("one_cmpl<VL:mode>2"): Likewise.
2149 ("faligndata<VM64:mode>_vis"): Likewise.
2150 ("alignaddrsi_vis"): Likewise.
2151 ("alignaddrdi_vis"): Likweise.
2152 ("alignaddrlsi_vis"): Likewise.
2153 ("alignaddrldi_vis"): Likewise.
2154 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
2155 ("bmaskdi_vis"): Likewise.
2156 ("bmasksi_vis"): Likewise.
2157 ("bshuffle<VM64:mode>_vis"): Likewise.
2158 ("cmask8<P:mode>_vis"): Likewise.
2159 ("cmask16<P:mode>_vis"): Likewise.
2160 ("cmask32<P:mode>_vis"): Likewise.
2161 ("pdistn<P:mode>_vis"): Likewise.
2162 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
2164 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2166 * config/sparc/sparc.md ("subtype"): New insn attribute.
2167 ("*wrgsr_sp64"): Set insn subtype.
2168 ("*rdgsr_sp64"): Likewise.
2169 ("alignaddrsi_vis"): Likewise.
2170 ("alignaddrdi_vis"): Likewise.
2171 ("alignaddrlsi_vis"): Likewise.
2172 ("alignaddrldi_vis"): Likewise.
2173 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
2174 ("fexpand_vis"): Likewise.
2175 ("fpmerge_vis"): Likewise.
2176 ("faligndata<VM64:mode>_vis"): Likewise.
2177 ("bshuffle<VM64:mode>_vis"): Likewise.
2178 ("cmask8<P:mode>_vis"): Likewise.
2179 ("cmask16<P:mode>_vis"): Likewise.
2180 ("cmask32<P:mode>_vis"): Likewise.
2181 ("fchksm16_vis"): Likewise.
2182 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
2183 ("fmean16_vis"): Likewise.
2184 ("fp<plusminus_insn>64_vis"): Likewise.
2185 ("<plusminus_insn>v8qi3"): Likewise.
2186 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
2187 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
2188 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
2189 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
2190 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
2191 ("*movqi_insn"): Likewise.
2192 ("*movhi_insn"): Likewise.
2193 ("*movsi_insn"): Likewise.
2194 ("movsi_pic_gotdata_op"): Likewise.
2195 ("*movdi_insn_sp32"): Likewise.
2196 ("*movdi_insn_sp64"): Likewise.
2197 ("movdi_pic_gotdata_op"): Likewise.
2198 ("*movsf_insn"): Likewise.
2199 ("*movdf_insn_sp32"): Likewise.
2200 ("*movdf_insn_sp64"): Likewise.
2201 ("*zero_extendhisi2_insn"): Likewise.
2202 ("*zero_extendqihi2_insn"): Likewise.
2203 ("*zero_extendqisi2_insn"): Likewise.
2204 ("*zero_extendqidi2_insn"): Likewise.
2205 ("*zero_extendhidi2_insn"): Likewise.
2206 ("*zero_extendsidi2_insn_sp64"): Likewise.
2207 ("ldfsr"): Likewise.
2208 ("prefetch_64"): Likewise.
2209 ("prefetch_32"): Likewise.
2210 ("tie_ld32"): Likewise.
2211 ("tie_ld64"): Likewise.
2212 ("*tldo_ldub_sp32"): Likewise.
2213 ("*tldo_ldub1_sp32"): Likewise.
2214 ("*tldo_ldub2_sp32"): Likewise.
2215 ("*tldo_ldub_sp64"): Likewise.
2216 ("*tldo_ldub1_sp64"): Likewise.
2217 ("*tldo_ldub2_sp64"): Likewise.
2218 ("*tldo_ldub3_sp64"): Likewise.
2219 ("*tldo_lduh_sp32"): Likewise.
2220 ("*tldo_lduh1_sp32"): Likewise.
2221 ("*tldo_lduh_sp64"): Likewise.
2222 ("*tldo_lduh1_sp64"): Likewise.
2223 ("*tldo_lduh2_sp64"): Likewise.
2224 ("*tldo_lduw_sp32"): Likewise.
2225 ("*tldo_lduw_sp64"): Likewise.
2226 ("*tldo_lduw1_sp64"): Likewise.
2227 ("*tldo_ldx_sp64"): Likewise.
2228 ("*mov<VM32:mode>_insn"): Likewise.
2229 ("*mov<VM64:mode>_insn_sp64"): Likewise.
2230 ("*mov<VM64:mode>_insn_sp32"): Likewise.
2232 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2234 * config/sparc/sparc.md ("type"): New insn type viscmp.
2235 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
2237 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
2238 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
2239 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
2240 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
2242 ("n7_vis_logical_11cycle"): Likewise.
2243 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
2244 * config/sparc/niagara2.md ("niag3_vis": Likewise.
2245 * config/sparc/niagara.md ("niag_vis"): Likewise.
2246 * config/sparc/ultra3.md ("us3_fga"): Likewise.
2247 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
2249 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2251 * config/sparc/sparc.md: New instruction type `bmask'.
2252 (bmaskdi_vis): Use the `bmask' type.
2253 (bmasksi_vis): Likewise.
2254 * config/sparc/ultra3.md (us3_array): Likewise.
2255 * config/sparc/niagara7.md (n7_array): Likewise.
2256 * config/sparc/niagara4.md (n4_array): Likewise.
2257 * config/sparc/niagara2.md (niag2_vis): Likewise.
2258 (niag3_vis): Likewise.
2259 * config/sparc/niagara.md (niag_vis): Likewise.
2261 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2263 * ipa-comdats.c: Remove optimize check from gate.
2264 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
2265 for functions not optimized.
2266 (ipa_fn_summary_read): Skip optimize check.
2267 (ipa_fn_summary_write): Likewise.
2268 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
2270 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
2272 (can_inline_edge_p): Check flag_pcc_struct_return for match.
2273 (check_callers): Give up on caller which is not optimized.
2274 (inline_small_functions): Likewise.
2275 (ipa_inline): Do not give up when not optimizing.
2276 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
2277 away unoptimizes cdtors.
2278 (whole_program_function_and_variable_visibility): Do
2279 ipa_discover_readonly_nonaddressable_vars in LTO mode.
2280 * ipa.c (process_references): Do not check optimize.
2281 (symbol_table::remove_unreachable_nodes): Update optimize check.
2282 (set_writeonly_bit): Update optimize check.
2283 (pass_ipa_cdtor_merge::gate): Do not check optimize.
2284 (pass_ipa_single_use::gate): Remove.
2286 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2288 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
2289 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
2290 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
2291 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
2292 permute_load, permute_store, adjust_extract, adjust_splat,
2293 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
2294 replace_swap_with_copy, dump_swap_insn_table,
2295 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
2296 recombine_lvx_pattern, recombine_stvx_pattern,
2297 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
2298 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
2299 to file rs6000-p8swap.c.
2300 * config/rs6000/rs6000-p8swap.c: New file.
2301 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
2302 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
2303 and rs6000*-*-* targets.
2305 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2307 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
2309 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2311 * lto-wrapper.c (merge_and_complain): Do not merge
2312 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
2313 fsigned_zeros, ftrapping_math, fwrapv.
2314 (append_compiler_options): Do not track these options.
2315 (append_linker_options): Likewie
2317 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2319 * cgraphunit.c (cgraph_node::finalize_function): When
2320 !flag_toplevel_reorde set no_reorder flag.
2321 (varpool_node::finalize_decl): Likewise.
2322 (symbol_table::compile): Drop no toplevel reorder path.
2324 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2326 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
2327 edges; zero probability is not better than uninitialized.
2329 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
2331 * asan.h (asan_sanitize_allocas_p): Declare.
2332 * asan.c (asan_sanitize_allocas_p): New function.
2333 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
2334 (handle_builtin_alloca): Likewise.
2335 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
2336 if !asan_sanitize_allocas_p.
2337 * params.def (asan-instrument-allocas): Add new option.
2338 * params.h (ASAN_PROTECT_ALLOCAS): Define.
2339 * opts.c (common_handle_option): Disable allocas sanitization for
2342 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
2344 * asan.c: Include gimple-fold.h.
2345 (get_last_alloca_addr): New function.
2346 (handle_builtin_stackrestore): Likewise.
2347 (handle_builtin_alloca): Likewise.
2348 (asan_emit_allocas_unpoison): Likewise.
2349 (get_mem_refs_of_builtin_call): Add new parameter, remove const
2350 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
2351 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
2352 (instrument_builtin_call): Pass gimple iterator to
2353 get_mem_refs_of_builtin_call.
2354 (last_alloca_addr): New global.
2355 * asan.h (asan_emit_allocas_unpoison): Declare.
2356 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
2357 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
2358 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
2359 if function calls alloca.
2360 * gimple-fold.c (replace_call_with_value): Remove static keyword.
2361 * gimple-fold.h (replace_call_with_value): Declare.
2362 * internal-fn.c: Include asan.h.
2363 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
2364 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
2366 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2368 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
2369 (C_SELFTEST_FLAGS): New.
2370 (CPP_SELFTEST_FLAGS): New.
2371 (SELFTEST_DEPS): New, from deps of s-selftest.
2372 (C_SELFTEST_DEPS): New, from deps of s-selftest.
2373 (CPP_SELFTEST_DEPS): New.
2374 (selftest): Add dependency on s-selftest-c++.
2375 (s-selftest): Rename to...
2376 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
2377 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
2378 than SELFTEST_FLAGS.
2379 (selftest-gdb): Rename to...
2380 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
2382 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
2383 (selftest-valgrind): Rename to...
2384 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
2386 (selftest-valgrind): Reintroduce as an alias for
2387 selftest-c-valgrind.
2388 (s-selftest-c++): New.
2389 (selftest-c++-gdb): New.
2390 (selftest-c++-valgrind): New.
2392 2017-07-06 Olivier Hainque <hainque@adacore.com>
2394 * gcc.c (process_command): When deciding if undefined variables
2395 should be ignored when processing specs, accept "gcc -v" as well.
2397 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2399 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
2400 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
2402 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
2404 * config/arm/arm-cpus.in (armv8-r): Add new entry.
2405 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
2406 * config/arm/arm-tables.opt: Regenerate.
2407 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
2409 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
2411 2017-07-06 Carl Love <cel@us.ibm.com>
2413 * ChangeLog: Clean up from mid air collision
2415 2017-07-06 Carl Love <cel@us.ibm.com>
2417 * config/rs6000/rs6000-c.c: Add support for built-in functions
2418 vector signed int vec_subc (vector signed int, vector signed int);
2419 vector signed __int128 vec_subc (vector signed __int128,
2420 vector signed __int128);
2421 vector unsigned __int128 vec_subc (vector unsigned __int128,
2422 vector unsigned __int128);
2423 vector signed int vec_sube (vector signed int, vector signed int,
2425 vector unsigned int vec_sube (vector unsigned int,
2426 vector unsigned int,
2427 vector unsigned int);
2428 vector signed __int128 vec_sube (vector signed __int128,
2429 vector signed __int128,
2430 vector signed__int128);
2431 vector unsigned __int128 vec_sube (vector unsigned __int128,
2432 vector unsigned __int128,
2433 vector unsigned __int128);
2434 vector signed int vec_subec (vector signed int, vector signed int,
2436 vector unsigned int vec_subec (vector unsigned int,
2437 vector unsigned int,
2438 vector unsigned int);
2439 vector signed __int128 vec_subec (vector signed __int128,
2440 vector signed __int128,
2441 vector signed__int128);
2442 vector unsigned __int128 vec_subec (vector unsigned __int128,
2443 vector unsigned __int128,
2444 vector unsigned __int128);
2445 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
2446 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
2447 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
2448 BU_ALTIVEC_OVERLOAD_X definitions.
2449 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
2450 * doc/extend.texi: Update the built-in documentation file for the new
2453 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2456 * diagnostic-show-locus.c (layout::layout): Use start and finish
2457 spelling location for the start and finish of each range.
2458 * genmatch.c (linemap_client_expand_location_to_spelling_point):
2459 Add unused aspect param.
2460 * input.c (expand_location_1): Add "aspect" param, and use it
2461 to access the correct part of the location.
2462 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
2464 (expand_location_to_spelling_point): Likewise.
2465 (linemap_client_expand_location_to_spelling_point): Add "aspect"
2466 param, and pass it to expand_location_1.
2468 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
2470 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
2471 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
2472 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
2473 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
2474 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
2475 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
2476 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
2477 _mm_maskz_getmant_ss): New intrinsics.
2478 (__builtin_ia32_getexpss128_mask): Changed to ...
2479 __builtin_ia32_getexpss128_round ... this.
2480 (__builtin_ia32_getexpsd128_mask): Changed to ...
2481 __builtin_ia32_getexpsd128_round ... this.
2482 * config/i386/i386-builtin-types.def
2483 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
2484 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
2485 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
2486 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
2487 __builtin_ia32_getmantss_mask_round): New builtins.
2488 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
2489 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
2490 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
2491 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
2492 * config/i386/sse.md
2493 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
2494 avx512f_sgetexp<mode><mask_scalar_name>
2495 <round_saeonly_scalar_name> ... this.
2496 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
2497 %0, %1, %2<round_saeonly_op3>}): Changed to ...
2498 vgetexp<ssescalarmodesuffix>
2499 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2500 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
2501 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
2502 avx512f_vgetmant<mode><mask_scalar_name>
2503 <round_saeonly_scalar_name> ... this.
2504 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
2505 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
2506 vgetmant<ssescalarmodesuffix>
2507 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
2508 %0<mask_scalar_operand4>, %1, %2
2509 <round_saeonly_scalar_mask_op4>, %3} ... this.
2510 * config/i386/subst.md (mask_scalar_operand4,
2511 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
2512 round_saeonly_scalar_nimm_predicate): New subst attributes.
2514 2017-07-06 Julia Koval <julia.koval@intel.com>
2516 * config/i386/i386.c (ix86_erase_embedded_rounding):
2517 Remove code for old rounding pattern.
2519 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
2521 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
2523 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
2525 * doc/sourcebuild.texi (Test Directives, Variants of
2526 dg-require-support): Add documentation for dg-require-stack-check.
2528 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
2530 * config/i386/subst.md (mask_scalar, round_scalar,
2531 round_saeonly_scalar): New meta-templates.
2532 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
2533 round_scalar_mask_operand3, round_scalar_mask_op3,
2534 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
2535 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
2536 round_saeonly_scalar_constraint,
2537 round_saeonly_scalar_prefix): New subst attribute.
2538 * config/i386/sse.md
2539 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
2540 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
2541 <round_scalar_name> ... this.
2542 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
2543 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
2544 <round_scalar_name> ... this.
2545 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
2546 <sse>_vm<code><mode>3<mask_scalar_name>
2547 <round_saeonly_scalar_name> ... this.
2548 (v<plusminus_mnemonic><ssescalarmodesuffix>
2549 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2550 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2551 v<plusminus_mnemonic><ssescalarmodesuffix>
2552 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2553 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2554 (v<multdiv_mnemonic><ssescalarmodesuffix>
2555 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2556 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2557 v<multdiv_mnemonic><ssescalarmodesuffix>
2558 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2559 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2560 (v<maxmin_float><ssescalarmodesuffix>
2561 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
2562 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
2563 v<maxmin_float><ssescalarmodesuffix>
2564 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2565 %0<mask_scalar_operand3>, %1, %<iptr>2
2566 <round_saeonly_scalar_mask_op3>} ... this.
2568 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
2570 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
2571 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
2573 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
2574 Alan Hayward <alan.hayward@arm.com>
2575 David Sherwood <david.sherwood@arm.com>
2577 * combine.c (simplify_if_then_else): Remove "enum" before
2579 * compare-elim.c (can_eliminate_compare): Likewise.
2580 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
2582 (aarch64_lookup_simd_builtin_type): Likewise.
2583 (aarch64_simd_builtin_type): Likewise.
2584 (aarch64_init_simd_builtin_types): Likewise.
2585 (aarch64_simd_expand_args): Likewise.
2586 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
2588 (aarch64_reverse_mask): Likewise.
2589 (aarch64_simd_emit_reg_reg_move): Likewise.
2590 (aarch64_gen_adjusted_ldpstp): Likewise.
2591 (aarch64_ccmp_mode_to_code): Likewise.
2592 (aarch64_operands_ok_for_ldpstp): Likewise.
2593 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2594 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
2596 (aarch64_min_divisions_for_recip_mul): Likewise.
2597 (aarch64_reassociation_width): Likewise.
2598 (aarch64_get_condition_code_1): Likewise.
2599 (aarch64_simd_emit_reg_reg_move): Likewise.
2600 (aarch64_simd_attr_length_rglist): Likewise.
2601 (aarch64_reverse_mask): Likewise.
2602 (aarch64_operands_ok_for_ldpstp): Likewise.
2603 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2604 (aarch64_gen_adjusted_ldpstp): Likewise.
2605 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
2607 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
2608 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
2609 (arm_lookup_simd_builtin_type): Likewise.
2610 (arm_simd_builtin_type): Likewise.
2611 (arm_init_simd_builtin_types): Likewise.
2612 (arm_expand_builtin_args): Likewise.
2613 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
2614 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
2615 (ft32_setup_incoming_varargs): Likewise.
2616 (ft32_function_arg): Likewise.
2617 (ft32_function_arg_advance): Likewise.
2618 (ft32_pass_by_reference): Likewise.
2619 (ft32_arg_partial_bytes): Likewise.
2620 (ft32_valid_pointer_mode): Likewise.
2621 (ft32_addr_space_pointer_mode): Likewise.
2622 (ft32_addr_space_legitimate_address_p): Likewise.
2623 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
2625 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
2626 (ix86_emit_outlined_ms2sysv_restore): Likewise.
2627 (iamcu_alignment): Likewise.
2628 (canonicalize_vector_int_perm): Likewise.
2629 (ix86_noce_conversion_profitable_p): Likewise.
2630 (ix86_mpx_bound_mode): Likewise.
2631 (ix86_operands_ok_for_move_multiple): Likewise.
2632 * config/microblaze/microblaze-protos.h
2633 (microblaze_expand_conditional_branch_reg): Likewise.
2634 * config/microblaze/microblaze.c
2635 (microblaze_expand_conditional_branch_reg): Likewise.
2636 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
2638 (rs6000_reassociation_width): Likewise.
2639 (rs6000_invalid_binary_op): Likewise.
2640 (fusion_p9_p): Likewise.
2641 (emit_fusion_p9_load): Likewise.
2642 (emit_fusion_p9_store): Likewise.
2643 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
2645 (riscv_hard_regno_mode_ok_p): Likewise.
2646 (riscv_address_insns): Likewise.
2647 (riscv_split_symbol): Likewise.
2648 (riscv_legitimize_move): Likewise.
2649 (riscv_function_value): Likewise.
2650 (riscv_hard_regno_nregs): Likewise.
2651 (riscv_expand_builtin): Likewise.
2652 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
2653 (riscv_build_integer): Likewise.
2654 (riscv_split_integer): Likewise.
2655 (riscv_legitimate_constant_p): Likewise.
2656 (riscv_cannot_force_const_mem): Likewise.
2657 (riscv_regno_mode_ok_for_base_p): Likewise.
2658 (riscv_valid_base_register_p): Likewise.
2659 (riscv_valid_offset_p): Likewise.
2660 (riscv_valid_lo_sum_p): Likewise.
2661 (riscv_classify_address): Likewise.
2662 (riscv_legitimate_address_p): Likewise.
2663 (riscv_address_insns): Likewise.
2664 (riscv_load_store_insns): Likewise.
2665 (riscv_force_binary): Likewise.
2666 (riscv_split_symbol): Likewise.
2667 (riscv_force_address): Likewise.
2668 (riscv_legitimize_address): Likewise.
2669 (riscv_move_integer): Likewise.
2670 (riscv_legitimize_const_move): Likewise.
2671 (riscv_legitimize_move): Likewise.
2672 (riscv_address_cost): Likewise.
2673 (riscv_subword): Likewise.
2674 (riscv_output_move): Likewise.
2675 (riscv_canonicalize_int_order_test): Likewise.
2676 (riscv_emit_int_order_test): Likewise.
2677 (riscv_function_arg_boundary): Likewise.
2678 (riscv_pass_mode_in_fpr_p): Likewise.
2679 (riscv_pass_fpr_single): Likewise.
2680 (riscv_pass_fpr_pair): Likewise.
2681 (riscv_get_arg_info): Likewise.
2682 (riscv_function_arg): Likewise.
2683 (riscv_function_arg_advance): Likewise.
2684 (riscv_arg_partial_bytes): Likewise.
2685 (riscv_function_value): Likewise.
2686 (riscv_pass_by_reference): Likewise.
2687 (riscv_setup_incoming_varargs): Likewise.
2688 (riscv_print_operand): Likewise.
2689 (riscv_elf_select_rtx_section): Likewise.
2690 (riscv_save_restore_reg): Likewise.
2691 (riscv_for_each_saved_reg): Likewise.
2692 (riscv_register_move_cost): Likewise.
2693 (riscv_hard_regno_mode_ok_p): Likewise.
2694 (riscv_hard_regno_nregs): Likewise.
2695 (riscv_class_max_nregs): Likewise.
2696 (riscv_memory_move_cost): Likewise.
2697 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
2698 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
2699 (rl78_addr_space_address_mode): Likewise.
2700 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2702 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2703 (rs6000_reassociation_width): Likewise.
2704 (rs6000_invalid_binary_op): Likewise.
2705 (fusion_p9_p): Likewise.
2706 (emit_fusion_p9_load): Likewise.
2707 (emit_fusion_p9_store): Likewise.
2708 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
2709 (ok_for_simple_move_operands): Likewise.
2710 (ok_for_simple_move_strict_operands): Likewise.
2711 (ok_for_simple_arith_logic_operands): Likewise.
2712 (visium_legitimize_reload_address): Likewise.
2713 (visium_select_cc_mode): Likewise.
2714 (output_cbranch): Likewise.
2715 (visium_split_double_move): Likewise.
2716 (visium_expand_copysign): Likewise.
2717 (visium_expand_int_cstore): Likewise.
2718 (visium_expand_fp_cstore): Likewise.
2719 * config/visium/visium.c (visium_pass_by_reference): Likewise.
2720 (visium_function_arg): Likewise.
2721 (visium_function_arg_advance): Likewise.
2722 (visium_libcall_value): Likewise.
2723 (visium_setup_incoming_varargs): Likewise.
2724 (visium_legitimate_constant_p): Likewise.
2725 (visium_legitimate_address_p): Likewise.
2726 (visium_legitimize_address): Likewise.
2727 (visium_secondary_reload): Likewise.
2728 (visium_register_move_cost): Likewise.
2729 (visium_memory_move_cost): Likewise.
2730 (prepare_move_operands): Likewise.
2731 (ok_for_simple_move_operands): Likewise.
2732 (ok_for_simple_move_strict_operands): Likewise.
2733 (ok_for_simple_arith_logic_operands): Likewise.
2734 (visium_function_value_1): Likewise.
2735 (rtx_ok_for_offset_p): Likewise.
2736 (visium_legitimize_reload_address): Likewise.
2737 (visium_split_double_move): Likewise.
2738 (visium_expand_copysign): Likewise.
2739 (visium_expand_int_cstore): Likewise.
2740 (visium_expand_fp_cstore): Likewise.
2741 (visium_split_cstore): Likewise.
2742 (visium_select_cc_mode): Likewise.
2743 (visium_split_cbranch): Likewise.
2744 (output_cbranch): Likewise.
2745 (visium_print_operand_address): Likewise.
2746 * expmed.c (flip_storage_order): Likewise.
2747 * expmed.h (emit_cstore): Likewise.
2748 (flip_storage_order): Likewise.
2749 * genrecog.c (validate_pattern): Likewise.
2750 * hsa-gen.c (gen_hsa_addr): Likewise.
2751 * internal-fn.c (expand_arith_overflow): Likewise.
2752 * ira-color.c (allocno_copy_cost_saving): Likewise.
2753 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2754 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
2755 (process_invariant_for_inheritance): Likewise.
2756 * lra-eliminations.c (move_plus_up): Likewise.
2757 * omp-low.c (lower_oacc_reductions): Likewise.
2758 * simplify-rtx.c (simplify_subreg): Likewise.
2759 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
2760 (TARGET_CHKP_BOUND_MODE): Likewise..
2761 * targhooks.c (default_chkp_bound_mode): Likewise.
2762 (default_setup_incoming_vararg_bounds): Likewise.
2763 * targhooks.h (default_chkp_bound_mode): Likewise.
2764 (default_setup_incoming_vararg_bounds): Likewise.
2765 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
2766 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
2767 (have_whole_vector_shift): Likewise.
2768 * tree-vect-stmts.c (vectorizable_load): Likewise.
2769 * doc/tm.texi: Regenerate.
2771 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2773 Graceful degrade if Binutils PR21472 is not available.
2776 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
2777 .rodata in flash test fails.
2778 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
2779 * confgure: Regenerate.
2780 * config.in: Regenerate.
2781 * config/avr/avr.c (avr_asm_named_section)
2782 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
2783 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
2784 (avr_asm_init_sections): Same.
2786 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2788 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
2789 (fma<VH:mode>4_intrinsic): Likewise.
2790 (*fmsub<VCVTF:mode>4): Likewise.
2791 (*fmsub<VH:mode>4_intrinsic): Likewise.
2793 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2796 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
2797 Don't depend on "optimize > 0".
2798 (out_movhi_r_mr, out_movqi_mr_r): Same.
2799 (out_movhi_mr_r, out_movqi_r_mr): Same.
2800 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
2801 io_address_operand on "optimize > 0".
2803 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2805 * tree-loop-distribution.c: Add general explanantion on the pass.
2806 (generate_loops_for_partition): Mark distributed loop.
2807 (pg_add_dependence_edges): New parameter. Handle alias data
2808 dependence specially and record it in the parameter if asked.
2809 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
2810 (init_partition_graph_vertices, add_partition_graph_edge): New.
2811 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
2812 (free_partition_graph_vdata, build_partition_graph): New.
2813 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
2814 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
2815 (data_ref_segment_size, latch_dominated_by_data_ref): New.
2816 (compute_alias_check_pairs, version_loop_by_alias_check): New.
2817 (version_for_distribution_p, finalize_partitions): New.
2818 (distribute_loop): Handle alias data dependence specially. Factor
2819 out loop fusion code as functions and call these functions.
2821 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2823 * tree-loop-distribution.c (classify_partition): New parameter and
2824 better handle reduction statement.
2825 (rdg_build_partitions): Revise comment.
2826 (distribute_loop): Compute statements in all partitions and pass it
2827 to classify_partition.
2829 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2831 * tree-loop-distribution.c (enum partition_type): New.
2832 (struct partition): New field type.
2833 (partition_merge_into): Add parameter. Update partition type.
2834 (data_dep_in_cycle_p, update_type_for_merge): New functions.
2835 (build_rdg_partition_for_vertex): Compute partition type.
2836 (rdg_build_partitions): Dump partition type.
2837 (distribute_loop): Update calls to partition_merge_into.
2839 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2841 * tree-loop-distribution.c (struct ddr_hasher): New.
2842 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
2844 (classify_partition): Call get_data_dependence.
2845 (pg_add_dependence_edges): Ditto.
2846 (distribute_loop): Release data dependence hash table.
2848 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2850 * tree-loop-distribution.c (ref_base_address): Delete.
2851 (similar_memory_accesses): Rename ...
2852 (share_memory_accesses): ... to this. Check if partitions access
2853 the same memory reference.
2854 (distribute_loop): Call share_memory_accesses.
2856 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2858 * tree-loop-distribution.c (struct partition): New field recording
2860 (partition_alloc, partition_free): Init and release data refs.
2861 (partition_merge_into): Merge data refs.
2862 (build_rdg_partition_for_vertex): Collect data refs for partition.
2863 (pg_add_dependence_edges): Change parameters from vector to bitmap.
2865 (distribute_loop): Remve data refs from vertice data of partition
2868 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2870 * tree-loop-distribution.c (params.h): Include header file.
2871 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
2872 (datarefs_vec): New global var.
2873 (create_rdg_vertices): Use datarefs_vec directly.
2874 (free_rdg): Don't free data references.
2875 (build_rdg): Update use. Don't free data references.
2876 (distribute_loop): Compute global variable for data references.
2877 Bail out if there are too many data references.
2879 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2881 * tree-loop-distribution.c (loop_nest): New global var.
2882 (build_rdg): Use loop directly, rather than loop nest.
2883 (pg_add_dependence_edges): Remove loop nest parameter. Use global
2885 (distribute_loop): Compute global variable loop nest. Update use.
2887 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2889 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
2890 (partition_merge_into): New parameter. Dump reason for fusion.
2891 (distribute_loop): Update use of partition_merge_into.
2893 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2895 * tree-loop-distribution.c (bb_top_order_index): New.
2896 (bb_top_order_index_size, bb_top_order_cmp): New.
2897 (stmts_from_loop): Use topological order.
2898 (pass_loop_distribution::execute): Compute and release topological
2899 order for basic blocks.
2901 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2903 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
2906 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2908 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
2909 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
2910 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
2911 * internal-fn.def (LOOP_DIST_ALIAS): New.
2912 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
2913 (fold_loop_internal_call): ... this.
2914 (vect_loop_dist_alias_call): New function.
2915 (set_uid_loop_bbs): Call fold_loop_internal_call.
2916 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
2919 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
2922 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
2923 Require dead FLAGS_REG at the beginning of a peephole.
2925 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
2928 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
2929 arguments in the call to __builtin_ia32_sbb_u32.
2930 (_subborrow_u64): Swap _X and _Y arguments in the call to
2931 __builtin_ia32_sbb_u64.
2933 2017-07-04 Jakub Jelinek <jakub@redhat.com>
2936 * tree-vrp.c (compare_assert_loc): Turn into a function template
2937 with stable template parameter. Only test if a->e is NULL,
2938 !a->e == !b->e has been verified already. Use e == NULL or
2939 e != NULL instead of e or ! e tests. If stable is true, don't use
2940 iterative_hash_expr, on the other side allow a or b or both NULL
2941 and sort the NULLs last.
2942 (process_assert_insertions): Sort using compare_assert_loc<false>
2943 instead of compare_assert_loc, later sort using
2944 compare_assert_loc<true> before calling process_assert_insertions_for
2945 in a loop. Use break instead of continue once seen NULL pointer.
2947 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
2949 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
2950 Cortex-R7 and Cortex-R8 processors.
2952 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2954 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
2955 uninitialized while src is not.
2957 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
2959 * common/config/arm/arm-common.c: Adjust include path for
2961 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
2962 (arm-cpu.h): Create in build directory. Adjust dependency rules.
2963 (arm-cpu-data.h): Likewise.
2964 (arm-cpu-cdata.h): Likewise.
2965 * config/arm/arm-cpu.h: Delete.
2966 * config/arm/arm-cpu-cdata.h: Delete.
2967 * config/arm/arm-cpu-data.h: Delete.
2969 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
2971 * config/arm/arm-cpus.in (cortex-a55): New.
2972 (cortex-a75): Likewise.
2973 (cortex-a75.cortex-a55): Likewise.
2974 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
2976 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
2977 * config/arm/arm-cpu-cdata.h: Regenerate.
2978 * config/arm/arm-cpu-data.h: Regenerate.
2979 * config/arm/arm-cpu.h: Regenerate.
2980 * config/arm/arm-tables.opt: Regenerate.
2981 * config/arm/arm-tune.md: Regenerate.
2983 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2985 * haifa-sched.c (sched_create_recovery_edges): Update profile.
2987 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2989 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
2992 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
2994 PR tree-optimization/81292
2995 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
2996 full_string_p, also call adjust_related_strinfos if the adjustment
2997 is simple, otherwise invalidate related strinfos.
2999 2017-07-04 Martin Liska <mliska@suse.cz>
3002 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
3003 newly created variable as DECL_IGNORED_P.
3005 2017-07-04 Martin Liska <mliska@suse.cz>
3008 * ipa-inline.c (inline_small_functions):
3009 Use xstrdup_for_dump.
3011 2017-07-04 Tom de Vries <tom@codesourcery.com>
3013 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
3015 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
3018 * config/darwin.c (darwin_function_switched_text_sections):
3021 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
3023 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
3025 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
3027 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
3029 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3031 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
3032 min_profitable_iters, and th as inclusive lower bounds.
3033 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
3034 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
3035 for min_profitable_iters and min_profitable_estimate.
3036 (vect_transform_loop): Treat th as an inclusive lower bound.
3037 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
3039 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
3042 * config/darwin.c (darwin_function_switched_text_sections):
3043 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
3044 in two pieces, and suppress the use of buf.
3046 2017-07-03 Nathan Sidwell <nathan@acm.org>
3048 * hash-table.h (hash_table_mod1): Fix indentation.
3050 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3053 * predict.c (force_edge_cold): Be more careful about propagation
3055 * profile-count.h (profile_probability::guessed,
3056 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
3058 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
3060 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
3062 * doc/invoke.texi (rcpc architecture extension): Document it.
3064 2017-07-03 Richard Biener <rguenther@suse.de>
3066 PR tree-optimization/60510
3067 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
3068 the scalar reduction PHI and use it.
3069 (vectorizable_reduction): Properly guard the single_defuse_cycle
3070 path for non-SLP reduction chains where we cannot use it.
3071 Rework reduc_def/index and vector type deduction. Rework
3072 vector operand gathering during reduction op code-gen.
3073 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
3074 chains dissolve the chain and leave it to non-SLP reduction
3077 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3079 * tree-data-ref.h (dr_alignment): Declare.
3080 * tree-data-ref.c (dr_alignment): New function.
3081 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
3082 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
3084 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
3086 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3088 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
3089 and base_misalignment fields.
3090 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
3091 * tree-data-ref.c: Include builtins.h.
3092 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
3093 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
3094 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
3095 * tree-vect-data-refs.c: Include tree-cfg.h.
3096 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
3097 fields instead of calculating an alignment here.
3098 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
3099 innermost_loop_behavior fields.
3101 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3103 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
3105 (DR_STEP_ALIGNMENT): New macro.
3106 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
3107 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
3108 (create_data_ref): Print it.
3109 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
3110 to tell whether the step preserves vector (mis)alignment.
3111 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
3112 Move the check for an integer step and generalise to all INTEGER_CST.
3113 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
3114 Print the outer step alignment.
3116 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3118 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
3119 with offset_alignment.
3120 (DR_ALIGNED_TO): Delete.
3121 (DR_OFFSET_ALIGNMENT): New macro.
3122 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
3123 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
3124 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
3125 (create_data_ref): Likewise.
3126 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
3127 (vect_analyze_data_refs): Likewise.
3128 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
3129 creating dummy innermost behavior.
3131 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3133 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
3134 with a "innermost_loop_behavior *" and refeence tree.
3135 * tree-data-ref.c (dr_analyze_innermost): Likewise.
3136 (create_data_ref): Update call accordingly.
3137 * tree-predcom.c (find_looparound_phi): Likewise.
3139 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3141 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
3142 fields with dr_wrt_vec_loop.
3143 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
3144 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
3145 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
3146 (vect_dr_behavior): New function.
3147 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
3148 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
3149 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
3150 track whether the step preserves the misalignment.
3151 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
3152 Use vect_dr_behavior.
3153 (vect_setup_realignment): Update call accordingly.
3154 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
3155 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
3156 call to vect_create_addr_base_for_vector_ref.
3157 (vect_create_cond_for_align_checks): Likewise.
3158 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
3159 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
3160 (vect_recog_mask_conversion_pattern): Likewise.
3161 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
3162 (new_stmt_vec_info): Remove redundant zeroing.
3164 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
3166 * common/config/arm/arm-common.c (arm_be8_option): New function.
3167 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
3168 (ISA_ARMv6): Add isa_bit_be8.
3169 * config/arm/arm.h (arm_be8_option): Add prototype.
3170 (BE8_SPEC_FUNCTION): New define.
3171 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
3172 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
3173 (mlittle-endian): Similarly.
3174 (mbe8, mbe32): New options.
3175 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
3176 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
3178 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3180 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
3182 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3184 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
3185 (cleanup_tree_cfg_bb): Use it.
3186 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
3188 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
3190 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3193 * loop-doloop.c (add_test): Update profile.
3195 2017-07-03 Martin Liska <mliska@suse.cz>
3198 * sanopt.c (rewrite_usage_of_param): New function.
3199 (sanitize_rewrite_addressable_params): Likewise.
3200 (pass_sanopt::execute): Call rewrite_usage_of_param.
3202 2017-07-03 Richard Biener <rguenther@suse.de>
3204 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
3205 back to using VIEW_CONVERT_EXPR.
3207 2017-07-03 Martin Liska <mliska@suse.cz>
3210 * doc/extend.texi: Document when a resolver function is
3211 generated for target_clones.
3213 2017-07-03 Martin Liska <mliska@suse.cz>
3215 * asan.c (asan_emit_stack_protection): Unpoison just red zones
3216 and shadow memory of auto variables which are subject of
3217 use-after-scope sanitization.
3218 (asan_expand_mark_ifn): Add do set only when is_poison.
3220 2016-07-03 Richard Biener <rguenther@suse.de>
3222 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
3224 (vect_force_simple_reduction): Record reduction def -> phi mapping.
3225 (vectorizable_reduction): Perform reduction PHI creation when
3226 visiting a reduction PHI and adjust and simplify code generation
3227 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
3228 (vect_transform_loop): Visit reduction PHIs.
3229 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
3230 defs into the SLP tree.
3231 (vect_build_slp_tree): Reduction defs terminate the recursion.
3232 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
3234 (vect_get_vec_defs_for_stmt_copy): Export.
3235 (vect_get_vec_defs): Likewise.
3236 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
3238 (vect_get_vec_defs_for_stmt_copy): Declare.
3239 (vect_get_vec_defs): Likewise.
3241 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3243 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
3244 parameter with a "loop" parameter and use it instead of the
3245 loop containing DR_STMT. Don't check simple_iv when doing
3246 BB analysis. Describe the two analysis modes in the comment.
3248 2017-07-03 Tom de Vries <tom@codesourcery.com>
3250 PR tree-optimization/69468
3251 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
3252 (find_same_succ_bb): Handle ignore_edge_flags.
3254 2017-07-03 Tom de Vries <tom@codesourcery.com>
3256 PR tree-optimization/81192
3257 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
3259 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
3261 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
3263 2017-07-03 Tom de Vries <tom@codesourcery.com>
3265 PR tree-optimization/81192
3266 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
3267 BB_SAME_SUCC (bb) == NULL.
3269 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3271 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
3274 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3276 * dumpfile.c: Include profile-count.h
3277 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
3279 (insert_cond_bb): Update profile.
3280 * tree-cfg.h (insert_cond_bb): Update prototype.
3281 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
3282 * tree-dump.c: Do not include tree-cfg.
3284 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3286 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
3288 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3290 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
3293 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3295 * tree-complex.c (expand_complex_div_wide): update profile.
3297 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3298 Alan Hayward <alan.hayward@arm.com>
3299 David Sherwood <david.sherwood@arm.com>
3301 * Makefile.in (MACHMODE_H): Remove insn-modes.h
3302 (CORETYPES_H): New define.
3303 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
3304 (insn-modes-inline.h, s-modes-inline-h): New rules.
3305 (generated_files): Add insn-modes-inline.h.
3306 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
3307 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
3308 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
3309 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
3310 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
3311 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
3312 (build/gencodes.o, build/genconditions.o): Likewise.
3313 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
3314 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
3315 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
3316 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
3317 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
3318 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
3319 * coretypes.h: Include everything up to real.h for generators.
3320 Include insn-modes.h first. Include wide-int-print.h after
3321 wide-int.h. Include insn-modes-inline.h and then machmode.h.
3322 * machmode.h: Don't include insn-modes.h here.
3323 * function-tests.c: Remove includes of signop.h, machmode.h,
3324 double-int.h and wide-int.h.
3326 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
3328 * optc-save-gen.awk: Likewise.
3329 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
3330 * godump.c: Remove include of wide-int-print.h.
3331 * pretty-print.h: Likewise.
3332 * wide-int-print.cc: Likewise.
3333 * wide-int.cc: Likewise.
3334 * hash-map-tests.c: Remove include of signop.h.
3335 * hash-set-tests.c: Likewise.
3336 * rtl-tests.c: Likewise.
3337 * mkconfig.sh: Remove include of machmode.h.
3338 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
3340 (emit_insn_modes_inline_h): ...this new function. Emit the code
3341 into an insn-modes-inline.h header file, adding appropriate
3342 include guards and end comments.
3343 (emit_insn_modes_c_header): Remove include of machmode.h.
3344 (emit_min_insn_modes_c_header): Include coretypes.h rather than
3346 (main): Handle -i flag and call emit_insn_modes_inline_h when
3349 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3351 * tree-ssa-strlen.c (strinfo): Rename the length field to
3352 nonzero_chars. Add a full_string_p field.
3353 (compare_nonzero_chars, zero_length_string_p): New functions.
3354 (get_addr_stridx): Add an offset_out parameter.
3355 Use compare_nonzero_chars.
3356 (get_stridx): Update accordingly. Use compare_nonzero_chars.
3357 (new_strinfo): Update after above changes to strinfo.
3358 (set_endptr_and_length): Set full_string_p.
3359 (get_string_length): Update after above changes to strinfo.
3360 (unshare_strinfo): Update call to new_strinfo.
3361 (maybe_invalidate): Likewise.
3362 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
3363 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
3364 as a uhwi instead of an shwi. Update after above changes to
3365 strinfo and new_strinfo.
3366 (zero_length_string): Assert that chainsi contains full strings.
3367 Use zero_length_string_p. Update call to new_strinfo.
3368 (adjust_related_strinfos): Update after above changes to strinfo.
3369 Copy full_string_p from origsi.
3370 (adjust_last_stmt): Use zero_length_string_p.
3371 (handle_builtin_strlen): Update after above changes to strinfo and
3372 new_strinfo. Install the lhs as the string length if the previous
3373 entry didn't describe a full string.
3374 (handle_builtin_strchr): Update after above changes to strinfo
3376 (handle_builtin_strcpy): Likewise.
3377 (handle_builtin_strcat): Likewise.
3378 (handle_builtin_malloc): Likewise.
3379 (handle_pointer_plus): Likewise.
3380 (handle_builtin_memcpy): Likewise. Track nonzero characters
3381 that aren't necessarily followed by a nul terminator.
3382 (handle_char_store): Likewise.
3384 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3386 PR tree-optimization/80769
3387 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
3388 for malloc and calloc. Document the new invariant that all related
3389 strinfos have delayed lengths or none do.
3390 (verify_related_strinfos): Move earlier in file.
3391 (set_endptr_and_length): New function, split out from...
3392 (get_string_length): ...here. Also set the lengths of related
3394 (zero_length_string): Assert that chainsi has known (rather than
3396 (adjust_related_strinfos): Likewise.
3398 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3400 PR tree-optimization/81136
3401 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
3402 assert that two references with the same misalignment have the same
3403 compile-time misalignment if those compile-time misalignments
3406 2017-07-01 Andi Kleen <ak@linux.intel.com>
3408 * print-tree.c (print_node): Print all attributes.
3410 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3412 * cfg.c (scale_bbs_frequencies): New function.
3413 * cfg.h (scale_bbs_frequencies): Declare it.
3414 * cfgloopanal.c (single_likely_exit): Cleanup.
3415 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
3417 (scale_loop_profile): Likewise.
3418 (loop_version): Likewise.
3419 (create_empty_loop_on_edge): Update.
3420 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
3421 scale_loop_frequencies, scale_loop_profile, loopify,
3422 loop_version): Update prototypes.
3423 * modulo-sched.c (sms_schedule): Update.
3424 * predict.c (unlikely_executed_edge_p): Also check probability.
3425 (probably_never_executed_edge_p): Fix typo.
3426 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3427 * tree-parloops.c (gen_parallel_loop): Update.
3428 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
3429 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3430 * tree-ssa-loop-split.c (split_loop): Update.
3431 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3432 * tree-vect-loop-manip.c (vect_do_peeling): Update.
3433 (vect_loop_versioning): Update.
3434 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3436 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3438 * trans-mem.c (split_bb_make_tm_edge): Update profile.
3440 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3442 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
3443 to keep profile consistent.
3445 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3447 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
3448 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
3449 * profile-count.h (max_safe_multiplier): Make unsigned.
3450 (profile_count::guessed_zero): New.
3452 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3454 * bb-reorder.c (fix_up_crossing_landing_pad,
3455 fix_crossing_conditional_branches): Use make_single_succ_edge
3456 to keep profile consistent.
3458 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3460 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
3463 2017-07-01 Jakub Jelinek <jakub@redhat.com>
3466 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
3467 the right scopes, make sure cond_jump isn't preserved between multiple
3468 iterations. Search for fallthru edge whenever there are 3+ edges and
3469 use find_fallthru_edge for it.
3471 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3473 Patch by Alexander Monakov <amonakov@ispras.ru>
3474 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
3475 probabilities consistently.
3477 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3479 * pa.c (pa_expand_compare_and_swap_loop): Update call of
3480 emit_cmp_and_jump_insns.
3482 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3485 * tree-inline.c (expand_call_inline): Combine profile statuses.
3487 2017-06-30 Andrew Pinski <apinski@cavium.com>
3489 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
3490 fold_stmt returned true.
3492 2017-06-30 Nathan Sidwell <nathan@acm.org>
3494 * ggc.h (empty_string): Delete.
3495 * cfgexpand.c (expand_asm_stmt): Use plain "".
3496 * optabs.c (expand_asm_memory_barrier): Likewise.
3497 * stringpool.c (empty_string): Delete.
3498 (digit_vector, digit_string): Delete.
3499 (ggc_alloc_string): Use plain "", don't optimize single digit
3500 strings. Use ggc_alloc_atomic.
3502 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
3504 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
3505 comparison set and one other set, use the cost of the non-comparison
3508 2017-06-30 Nathan Sidwell <nathan@acm.org>
3510 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
3513 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
3515 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
3516 loops. Remove now unneeded calls to gimple_switch_set_label() that
3517 just set removed labels to NULL_TREE.
3519 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
3521 * tree-ssanames.c (set_range_info_raw): Abstract from ...
3522 (set_range_info): ...here. Only call set_range_info_raw if domain
3524 (set_nonzero_bits): Call set_range_info_raw.
3525 * tree-ssanames.h (set_range_info_raw): New.
3527 2017-06-30 Jakub Jelinek <jakub@redhat.com>
3530 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
3531 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
3532 of nonimmediate_operand and <store_mask_constraint> instead of m
3533 for the input operand. For V8FI iterator, always split if input
3534 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
3535 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
3536 <store_mask_predicate> instead of register_operand and
3537 <store_mask_constraint> instead of v for the input operand. Make
3538 sure both operands aren't MEMs for if not <mask_applied>.
3540 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
3542 * lto-wrapper.c (copy_file) Close both file descriptors before
3545 2017-06-30 Martin Liska <mliska@suse.cz>
3548 * multiple_target.c (create_dispatcher_calls): Make ifunc
3549 also for function that don't have calls or are not referenced.
3551 2017-06-30 Richard Biener <rguenther@suse.de>
3553 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
3554 analyze the first scalar stmt. Move vector type computation
3555 for the BB case here from ...
3556 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
3557 live operation processing in the SLP case properly.
3559 2017-06-30 Richard Biener <rguenther@suse.de>
3561 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
3563 2017-06-30 Martin Liska <mliska@suse.cz>
3566 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
3567 before BUILT_IN_UNWIND_RESUME when ASAN is used.
3569 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
3571 * doc/invoke.texi (AArch64): Add missing options and remove redundant
3574 2017-06-30 Richard Biener <rguenther@suse.de>
3576 PR tree-optimization/81249
3577 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
3578 condition reduction result to original scalar type.
3580 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3582 * profile-count.h (enum profile_quality): Fix typos and whitespace
3585 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3587 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
3588 type for branch probabilities.
3590 2017-06-29 Julian Brown <julian@codesourcery.com>
3591 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3593 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
3594 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
3595 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
3596 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
3598 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3600 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
3601 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
3602 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
3603 CC usage from generic code to here.
3604 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
3605 CC usage into the target macros.
3607 2017-06-29 Maya Rashish <coypu@sdf.org>
3609 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
3612 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3614 * arm/arm-builtins.c: Include profile-count.h
3615 * except.c (sjlj_emit_function_enter): Use
3616 profile_probability::unlikely.
3618 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3620 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
3621 and tocrel_offset be pointer args rather than implicitly using
3623 (legitimate_constant_pool_address_p, rs6000_emit_move,
3624 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
3625 tocrel_offset and use in toc_relative_expr_p call.
3626 (print_operand, print_operand_address): Use static tocrel_base_oac
3627 and tocrel_offset_oac.
3628 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
3631 2017-06-29 Maya Rashish <coypu@sdf.org>
3633 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
3635 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
3637 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
3638 objects, take into account only the alignment of 'op0' and 'mode1' if
3641 2017-06-29 Steve Ellcey <sellcey@cavium.com>
3643 * ccmp.c (ccmp_tree_comparison_p): New function.
3644 (ccmp_candidate_p): Update to use above function.
3645 (get_compare_parts): New function.
3646 (expand_ccmp_next): Update to use new functions.
3647 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
3649 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
3650 take mode as argument.
3651 * ccmp.h (expand_ccmp_expr): Add mode as argument.
3652 * expr.c (expand_expr_real_1): Pass mode as argument.
3654 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
3656 * combine.c (combine_instructions): Print insns to dump_file, together
3659 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3661 * asan.c (asan_emit_stack_protection): Update.
3662 (create_cond_insert_point): Update.
3663 * auto-profile.c (afdo_propagate_circuit): Update.
3664 * basic-block.h (struct edge_def): Turn probability to
3665 profile_probability.
3666 (EDGE_FREQUENCY): Update.
3667 * bb-reorder.c (find_traces_1_round): Update.
3668 (better_edge_p): Update.
3669 (sanitize_hot_paths): Update.
3670 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
3671 (make_single_succ_edge): Update.
3672 (check_bb_profile): Update.
3673 (dump_edge_info): Update.
3674 (update_bb_profile_for_threading): Update.
3675 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
3676 probabilitycount to 0.
3677 * cfgbuild.c (compute_outgoing_frequencies): Update.
3678 * cfgcleanup.c (try_forward_edges): Update.
3679 (outgoing_edges_match): Update.
3680 (try_crossjump_to_edge): Update.
3681 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
3682 (expand_gimple_tailcall): Update.
3683 (construct_init_block): Use make_single_succ_edge.
3684 (construct_exit_block): Use make_single_succ_edge.
3685 * cfghooks.c (verify_flow_info): Update.
3686 (redirect_edge_succ_nodup): Update.
3687 (split_edge): Update.
3688 (account_profile_record): Update.
3689 * cfgloopanal.c (single_likely_exit): Update.
3690 * cfgloopmanip.c (scale_loop_profile): Update.
3691 (set_zero_probability): Remove.
3692 (duplicate_loop_to_header_edge): Update.
3693 * cfgloopmanip.h (loop_version): Update prototype.
3694 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3695 (force_nonfallthru_and_redirect): Update.
3696 (update_br_prob_note): Update.
3697 (rtl_verify_edges): Update.
3698 (purge_dead_edges): Update.
3699 (rtl_lv_add_condition_to_bb): Update.
3700 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
3701 * cgraphunit.c (init_lowered_empty_function): Update.
3702 (cgraph_node::expand_thunk): Update.
3703 * cilk-common.c: Include profile-count.h
3704 * dojump.c (inv): Remove.
3705 (jumpifnot): Update.
3706 (jumpifnot_1): Update.
3707 (do_jump_1): Update.
3709 (do_jump_by_parts_greater_rtx): Update.
3710 (do_compare_rtx_and_jump): Update.
3711 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
3712 do_jump_1. do_compare_rtx_and_jump): Update prototype.
3713 * dwarf2cfi.c: Include profile-count.h
3714 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
3715 (sjlj_emit_dispatch_table): Likewise.
3716 * explow.c: Include profile-count.h
3717 * expmed.c (emit_store_flag_force): Update.
3718 (do_cmp_and_jump): Update.
3719 * expr.c (compare_by_pieces_d::generate): Update.
3720 (compare_by_pieces_d::finish_mode): Update.
3721 (emit_block_move_via_loop): Update.
3722 (store_expr_with_bounds): Update.
3723 (store_constructor): Update.
3724 (expand_expr_real_2): Update.
3725 (expand_expr_real_1): Update.
3726 * expr.h (try_casesi, try_tablejump): Update prototypes.
3727 * gimple-pretty-print.c (dump_probability): Update.
3728 (dump_profile): New.
3729 (dump_gimple_label): Update.
3730 (dump_gimple_bb_header): Update.
3731 * graph.c (draw_cfg_node_succ_edges): Update.
3732 * hsa-gen.c (convert_switch_statements): Update.
3733 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
3734 (find_if_case_1): Update.
3735 (find_if_case_2): Update.
3736 * internal-fn.c (expand_arith_overflow_result_store): Update.
3737 (expand_addsub_overflow): Update.
3738 (expand_neg_overflow): Update.
3739 (expand_mul_overflow): Update.
3740 (expand_vector_ubsan_overflow): Update.
3741 * ipa-cp.c (good_cloning_opportunity_p): Update.
3742 * ipa-split.c (split_function): Use make_single_succ_edge.
3743 * ipa-utils.c (ipa_merge_profiles): Update.
3744 * loop-doloop.c (add_test): Update.
3745 (doloop_modify): Update.
3746 * loop-unroll.c (compare_and_jump_seq): Update.
3747 (unroll_loop_runtime_iterations): Update.
3748 * lra-constraints.c (lra_inheritance): Update.
3749 * lto-streamer-in.c (input_cfg): Update.
3750 * lto-streamer-out.c (output_cfg): Update.
3751 * mcf.c (adjust_cfg_counts): Update.
3752 * modulo-sched.c (sms_schedule): Update.
3753 * omp-expand.c (expand_omp_for_init_counts): Update.
3754 (extract_omp_for_update_vars): Update.
3755 (expand_omp_ordered_sink): Update.
3756 (expand_omp_for_ordered_loops): Update.
3757 (expand_omp_for_generic): Update.
3758 (expand_omp_for_static_nochunk): Update.
3759 (expand_omp_for_static_chunk): Update.
3760 (expand_cilk_for): Update.
3761 (expand_omp_simd): Update.
3762 (expand_omp_taskloop_for_outer): Update.
3763 (expand_omp_taskloop_for_inner): Update.
3764 * omp-simd-clone.c (simd_clone_adjust): Update.
3765 * optabs.c (expand_doubleword_shift): Update.
3766 (expand_abs): Update.
3767 (emit_cmp_and_jump_insn_1): Update.
3768 (expand_compare_and_swap_loop): Update.
3769 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
3770 * predict.c (predictable_edge_p): Update.
3771 (edge_probability_reliable_p): Update.
3772 (set_even_probabilities): Update.
3773 (combine_predictions_for_insn): Update.
3774 (combine_predictions_for_bb): Update.
3775 (propagate_freq): Update.
3776 (estimate_bb_frequencies): Update.
3777 (force_edge_cold): Update.
3778 * profile-count.c (profile_count::dump): Add missing space into dump.
3779 (profile_count::debug): Add newline.
3780 (profile_count::differs_from_p): Explicitly convert to unsigned.
3781 (profile_count::stream_in): Update.
3782 (profile_probability::dump): New member function.
3783 (profile_probability::debug): New member function.
3784 (profile_probability::differs_from_p): New member function.
3785 (profile_probability::differs_lot_from_p): New member function.
3786 (profile_probability::stream_in): New member function.
3787 (profile_probability::stream_out): New member function.
3788 * profile-count.h (profile_count_quality): Rename to ...
3789 (profile_quality): ... this one.
3790 (profile_probability): New.
3791 (profile_count): Update.
3792 * profile.c (compute_branch_probabilities): Update.
3793 * recog.c (peep2_attempt): Update.
3794 * sched-ebb.c (schedule_ebbs): Update.
3795 * sched-rgn.c (find_single_block_region): Update.
3796 (compute_dom_prob_ps): Update.
3797 (schedule_region): Update.
3798 * sel-sched-ir.c (compute_succs_info): Update.
3799 * stmt.c (struct case_node): Update.
3800 (do_jump_if_equal): Update.
3801 (get_outgoing_edge_probs): Update.
3802 (conditional_probability): Update.
3803 (emit_case_dispatch_table): Update.
3804 (expand_case): Update.
3805 (expand_sjlj_dispatch_table): Update.
3806 (emit_case_nodes): Update.
3807 * targhooks.c: Update.
3808 * tracer.c (better_p): Update.
3809 (find_best_successor): Update.
3810 * trans-mem.c (expand_transaction): Update.
3811 * tree-call-cdce.c: Update.
3812 * tree-cfg.c (gimple_split_edge): Upate.
3813 (move_sese_region_to_fn): Upate.
3814 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
3815 * tree-eh.c (lower_resx): Upate.
3816 (cleanup_empty_eh_move_lp): Upate.
3817 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3818 * tree-inline.c (copy_edges_for_bb): Update.
3819 (copy_cfg_body): Update.
3820 * tree-parloops.c (gen_parallel_loop): Update.
3821 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
3822 (gimple_gen_time_profiler): Update.
3823 * tree-ssa-dce.c (remove_dead_stmt): Update.
3824 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3825 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3826 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3827 (unloop_loops): Update.
3828 (try_peel_loop): Update.
3829 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3830 * tree-ssa-loop-split.c (connect_loops): Update.
3831 (split_loop): Update.
3832 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3833 (hoist_guard): Update.
3834 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3835 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3836 (value_replacement): Update.
3837 * tree-ssa-reassoc.c (branch_fixup): Update.
3838 * tree-ssa-tail-merge.c (replace_block_by): Update.
3839 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3840 (create_edge_and_update_destination_phis): Update.
3841 (compute_path_counts): Update.
3842 (recompute_probabilities): Update.
3843 (update_joiner_offpath_counts): Update.
3844 (freqs_to_counts_path): Update.
3845 (duplicate_thread_path): Update.
3846 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3847 (struct switch_conv_info): Update.
3848 (gen_inbound_check): Update.
3849 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3850 (vect_do_peeling): Update.
3851 (vect_loop_versioning): Update.
3852 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3853 (optimize_mask_stores): Update.
3854 * ubsan.c (ubsan_expand_null_ifn): Update.
3855 * value-prof.c (gimple_divmod_fixed_value): Update.
3856 (gimple_divmod_fixed_value_transform): Update.
3857 (gimple_mod_pow2): Update.
3858 (gimple_mod_pow2_value_transform): Update.
3859 (gimple_mod_subtract): Update.
3860 (gimple_mod_subtract_transform): Update.
3861 (gimple_ic): Update.
3862 (gimple_stringop_fixed_value): Update.
3863 (gimple_stringops_transform): Update.
3864 * value-prof.h: Update.
3866 2017-06-29 Carl Love <cel@us.ibm.com>
3868 * config/rs6000/rs6000-c.c: Add support for built-in functions
3869 vector signed int vec_signed (vector float);
3870 vector signed long long vec_signed (vector double);
3871 vector signed int vec_signed2 (vector double, vector double);
3872 vector signed int vec_signede (vector double);
3873 vector signed int vec_signedo (vector double);
3874 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
3875 instruction generator.
3876 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3877 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
3878 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
3880 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
3881 vunsignede_v2df): Add define_expands.
3882 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
3883 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
3884 VEC_UNSIGNEDO): Add definitions.
3885 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3886 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
3887 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
3888 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
3889 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
3890 * config/rs6000/altivec.h (vec_signed, vec_signed2,
3891 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
3892 vec_unsignede, vec_unsignedo): Add builtin defines.
3893 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
3895 * doc/extend.texi: Update the built-in documentation file for the
3896 new built-in functions.
3898 2017-06-29 Richard Biener <rguenther@suse.de>
3900 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
3901 reduction chains to LOOP_VINFO_REDUCTIONS.
3902 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
3903 SLP reductions after processing reduction chains.
3905 2017-06-29 Nathan Sidwell <nathan@acm.org>
3907 * builtins.c (fold_builtin_FUNCTION): Use
3908 lang_hooks.decl_printable_name.
3910 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
3913 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
3914 with only one label.
3915 * stmt.c (expand_case): Assert NCASES is greater than one.
3917 2017-06-29 Richard Biener <rguenther@suse.de>
3919 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
3921 (group_case_labels): Likewise.
3922 (find_taken_edge): Push sanity checking on val to workers...
3923 (find_taken_edge_cond_expr): ... here
3924 (find_taken_edge_switch_expr): ... and here, handle cases
3925 with just a default label.
3926 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
3927 (group_case_labels): Likewise.
3928 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
3929 group_case_labels does anything cleanup the CFG again.
3931 2017-06-29 Bin Cheng <bin.cheng@arm.com>
3933 PR tree-optimization/81196
3934 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
3935 exit condition comparing two IVs.
3937 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
3939 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
3940 profile to the dummy entry at the end of the list of architectures.
3941 * config/arm/arm-cpu-cdata.h: Regenerated.
3943 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3944 Michael Collison <michael.collison@arm.com>
3947 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
3949 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
3950 (*aarch64_reg_<mode>3_minus_mask): New pattern.
3951 (*aarch64_<optab>_reg_di3_mask2): New pattern.
3952 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
3953 of shift when the shift amount is masked with constant equal to
3954 the size of the mode.
3955 * config/aarch64/predicates.md (subreg_lowpart_operator): New
3958 2017-06-29 Martin Liska <mliska@suse.cz>
3960 * config/i386/i386.opt: Change range from [1,5] to [0,5].
3962 2017-06-29 Yury Gribov <tetra2005@gmail.com>
3965 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
3968 (edge_growth_cache_entry::edge_growth_cache_entry): New
3970 (reset_edge_growth_cache): Update to use constructor.
3972 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3974 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
3975 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
3976 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
3978 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
3980 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
3981 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
3983 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
3985 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
3986 (*-linux-uclibc*): Add t-uclibc tmake_file.
3987 * config/t-musl: New.
3988 * config/t-uclibc: New.
3990 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
3992 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
3994 (gen_comm_data): Emit architectural setting of arch_prof.
3995 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
3997 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
3998 (armv8-m.base, armv8-m.main): Likewise.
3999 * arm-protos.h (arm_build_target): Add profile field.
4000 (arch_option): Likewise.
4001 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
4003 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
4004 arm_active_target.profile.
4006 2017-06-28 Richard Biener <rguenther@suse.de>
4009 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
4010 TYPE_OVERFLOW_WRAPS.
4011 * match.pd (negate_expr_p): Likewise.
4012 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
4013 fold_build2, not fold_binary.
4015 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
4017 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
4018 Convert memory address to Pmode.
4019 (aarch64_print_operand): Assert MEM operands are always Pmode.
4021 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
4024 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
4025 Remove redundant if.
4026 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
4027 * config/arm/aarch-common-protos.h
4028 (aarch_forward_to_shift_is_not_shifted_re): Remove.
4029 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
4031 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
4034 * multiple_target.c (create_dispatcher_calls): Set the default
4035 clone to be static, not public.
4037 2017-06-28 Richard Biener <rguenther@suse.de>
4039 * tree-vect-loop.c (vectorizable_reduction): Move special
4040 cond reduction IV var creation ...
4041 (vect_create_epilog_for_reduction): ... here. Remove induction_index
4042 parameter. Use STMT_VINFO_VECTYPE.
4043 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
4046 2017-06-28 Martin Liska <mliska@suse.cz>
4049 * ipa-visibility.c (non_local_p): Handle visibility.
4051 2017-06-28 Martin Liska <mliska@suse.cz>
4054 * common.opt: Add IntegerRange to various options.
4055 * opt-functions.awk (integer_range_info): New function.
4056 * optc-gen.awk: Add integer_range_info to cl_options struct.
4057 * opts-common.c (decode_cmdline_option): Handle
4058 CL_ERR_INT_RANGE_ARG.
4059 (cmdline_handle_error): Likewise.
4060 * opts.c (print_filtered_help): Show valid interval in
4061 when --help is provided.
4062 * opts.h (struct cl_option): Add range_min and range_max fields.
4063 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
4065 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
4067 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
4068 (x * C EQ/NE y * C): New transformation.
4070 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
4072 * genmultilib (combination_space): Accept '+' in option names.
4074 2017-06-28 Martin Liska <mliska@suse.cz>
4077 * asan.c (instrument_derefs): Bail out inner references
4078 that are hard register variables.
4080 2017-06-28 Jakub Jelinek <jakub@redhat.com>
4083 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
4084 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
4086 2017-06-28 Richard Biener <rguenther@suse.de>
4088 * tree-vectorizer.h (vect_get_vec_defs): Remove.
4089 (vect_get_slp_defs): Adjust.
4090 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
4092 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
4094 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4095 get_initial_defs_for_reduction instead of vect_get_vec_defs.
4096 (vectorizable_reduction): Adjust.
4097 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
4099 (vect_get_slp_defs): Likewise.
4100 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
4101 (vectorizable_bswap): Adjust.
4102 (vectorizable_call): Likewise.
4103 (vectorizable_conversion): Likewise.
4104 (vectorizable_assignment): Likewise.
4105 (vectorizable_shift): Likewise.
4106 (vectorizable_operation): Likewise.
4107 (vectorizable_store): Likewise.
4108 (vectorizable_condition): Likewise.
4109 (vectorizable_comparison): Likewise.
4111 2017-06-28 Michael Collison <michael.collison@arm.com>
4114 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
4116 (arm_gen_movmemqi): Removed unused variable 'i'.
4117 Convert 'for' loop into 'while' loop.
4118 (arm_expand_prologue): Remove last unnecessary set of insn.
4119 (thumb_pop): Remove unused variable 'pushed_words'.
4120 (thumb_exit): Remove last unnecessary set of regs_to_pop.
4122 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4124 * config/s390/predicates.md: Use s390_rel_address_ok_p.
4125 * config/s390/s390-protos.h: Add prototype of
4126 s390_rel_address_ok_p.
4127 * config/s390/s390.c (s390_got_symbol): New function.
4128 (s390_rel_address_ok_p): New function.
4129 (legitimize_pic_address): Use s390_rel_address_ok_p.
4130 (s390_load_got): Use s390_got_symbol.
4131 (s390_option_override): Issue error if
4132 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
4133 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
4135 * config/s390/s390.opt: New option mpic-data-is-text-relative.
4137 2017-06-27 Andrew Pinski <apinski@cavium.com>
4139 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
4140 (X * copysign (1.0, X)): New pattern.
4141 (X * copysign (1.0, -X)): New pattern.
4142 (copysign (-1.0, CST)): New pattern.
4144 2017-06-27 Joseph Myers <joseph@codesourcery.com>
4146 * genmultilib (combination_space): Remove variable.
4147 Validate reuse rules against regular expression for any sequence
4148 of multilib options in any order.
4150 2017-06-27 Michael Collison <michael.collison@arm.com>
4152 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
4153 call aarch64_split_simd_combine.
4154 * (aarch64_combine_internal<mode>): Delete pattern.
4155 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
4156 Allow register and subreg operands.
4158 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4160 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
4161 specific need, just fallback on defaults.
4162 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
4164 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4165 Olivier Hainque <hainque@adacore.com>
4167 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
4169 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
4170 targets. Pick a default if no particular attempt applied.
4171 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
4174 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4176 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
4177 (x86_64-wrs-vxworks7): Likewise.
4179 2017-06-27 Marek Polacek <polacek@redhat.com>
4182 * ubsan.c (instrument_null): Check get_base_address's result for null.
4184 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
4186 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
4188 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
4190 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
4191 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
4192 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
4193 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
4195 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
4196 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
4197 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
4198 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
4199 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
4200 BUILT_IN_FEUPDATEENV): New builtins.
4201 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
4202 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
4203 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
4204 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
4206 (builtin_structptr_types): Adjust size.
4207 * tree.c (builtin_structptr_types): Add four entries.
4209 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4210 Olivier Hainque <hainque@adacore.com>
4212 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
4213 (TLS_SYM): New local macro, forcing reference to __tls__ on
4214 link command lines for VxWorks 7 RTPs, triggering initialization
4216 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
4217 OS features TLS support, true for RTPs on VxWorks 7.
4218 * config/vxworks.c (vxworks_override_options): Setup emutls
4221 2017-06-27 Jakub Jelinek <jakub@redhat.com>
4223 * predict.c (test_prediction_value_range): Use -1U instead of -1
4224 to avoid narrowing conversion warning.
4225 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
4226 to avoid narrowing conversion warning.
4227 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
4229 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
4231 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4233 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
4234 64bit configurations.
4235 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
4236 (SIZE_TYPE): Likewise.
4237 * config/vxworks.c (vxworks_emutls_var_fields): Use
4238 long_unsigned_type_node instead of unsigned_type_node as the offset
4239 field type, which is "pointer" mode in emutls.c.
4241 2017-06-27 Jakub Jelinek <jakub@redhat.com>
4244 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
4247 * gimple-fold.c (replace_call_with_call_and_fold): Handle
4248 gimple_vuse copying separately from gimple_vdef copying.
4250 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4252 * value-prof.c (free_hist): Remove call to memset and the enclosing if
4255 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
4256 Olivier Hainque <hainque@adacore.com>
4258 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
4259 for all vxworks7 targets.
4260 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
4261 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
4262 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
4263 variations for VX6/VX7 and 32/64bits later on in ...
4264 (VXWORKS_LIB_SPEC): Leverage new macros.
4265 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
4266 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
4268 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
4270 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
4271 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
4273 2017-06-26 Carl Love <cel@us.ibm.com>
4275 * config/rs6000/rs6000-c.c: Add support for built-in functions
4276 vector bool char vec_reve (vector bool char);
4277 vector signed char vec_reve (vector signed char);
4278 vector unsigned char vec_reve (vector unsigned char);
4279 vector bool int vec_reve (vector bool int);
4280 vector signed int vec_reve (vector signed int);
4281 vector unsigned int vec_reve (vector unsigned int);
4282 vector bool long long vec_reve (vector bool long long);
4283 vector signed long long vec_reve (vector signed long long);
4284 vector unsigned long long vec_reve (vector unsigned long long);
4285 vector bool short vec_reve (vector bool short);
4286 vector signed short vec_reve (vector signed short);
4287 vector double vec_reve (vector double);
4288 vector float vec_reve (vector float);
4289 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
4290 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
4291 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
4292 (altivec_vreve): New pattern.
4293 * config/rs6000/altivec.h (vec_reve): New define.
4294 * doc/extend.texi (vec_rev): Update the built-in documentation file
4295 for the new built-in functions.
4297 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4299 PR tree-optimization/71815
4300 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
4302 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
4304 (slsr_process_phi): Likewise.
4305 (replace_uncond_cands_and_profitable_phis): Don't replace a
4306 multiply candidate with a stride of 1 (copy or cast).
4307 (phi_incr_cost): Call uses_consumed_by_stmt rather than
4309 (lowest_cost_path): Likewise.
4310 (total_savings): Likewise.
4312 2017-06-26 Richard Biener <rguenther@suse.de>
4315 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
4316 Use def_builtin_pure for all gather builtins.
4318 2017-06-26 Richard Biener <rguenther@suse.de>
4320 PR tree-optimization/81203
4321 * tree-tailcall.c (find_tail_calls): Do not move stmts into
4324 2017-06-26 Marek Polacek <polacek@redhat.com>
4327 * doc/invoke.texi: Document -Wmultistatement-macros.
4329 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
4331 * doc/sourcebuild.texi (ARM-specific attributes): Document new
4332 arm_neon_ok_no_float_abi effective target.
4334 2017-06-26 Richard Biener <rguenther@suse.de>
4336 PR tree-optimization/80928
4337 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
4338 (copy_bbs): Set BB_DUPLICATED flag early.
4339 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
4341 (execute_on_shrinking_pred): Likewise.
4342 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
4343 BB_DUPLICATED blocks.
4344 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
4345 iterate over all PHIs considering removal of *gsi.
4347 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
4349 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
4352 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4354 * config/rs6000/rs6000-string.c: (expand_block_clear,
4355 do_load_for_compare, select_block_compare_mode,
4356 compute_current_alignment, expand_block_compare,
4357 expand_strncmp_align_check, expand_strn_compare,
4358 expand_block_move, rs6000_output_load_multiple)
4359 Move functions related to string/block move/compare
4361 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
4362 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
4363 for this function which is now used in two files.
4364 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
4365 * config.gcc: Add rs6000-string.o to extra_objs for
4366 targets powerpc*-*-* and rs6000*-*-*.
4368 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4371 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
4372 32-bit, since indexed is not valid for DImode.
4373 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
4374 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
4375 (define_peephole2 for Altivec d-form load): Add 32-bit support.
4376 (define_peephole2 for Altivec d-form store): Likewise.
4379 * multiple_target.c (create_dispatcher_calls): Only create the
4380 dispatcher call if the function is the default clone of a
4383 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
4386 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
4387 a call, force the call to not be a tail call.
4389 2017-06-23 Jeff Law <law@redhat.com>
4391 * doc/contrib.texi: Add entry for Steven Pemberton's work on
4394 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
4396 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
4397 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
4398 handling for early expansion of vector shifts (sl,sr,sra,rl).
4399 (builtin_function_type): Add vector shift right instructions
4400 to the unsigned argument list.
4402 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
4404 rtl-optimizatoin/79286
4405 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
4406 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
4407 trap. PIC register plus a const unspec without offset can never trap.
4409 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
4411 * tree.h (builtin_structptr_type): New type.
4412 (builtin_structptr_types): Declare new array.
4413 * tree.c (builtin_structptr_types): New array.
4414 (free_lang_data, build_common_tree_nodes): Use it.
4416 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
4419 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
4422 2017-06-22 Matt Turner <mattst88@gmail.com>
4424 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
4425 Lake models to skylake case. Assume skylake for unknown
4426 models with clflushopt.
4428 2017-06-22 Jeff Law <law@redhat.com>
4430 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
4431 frame sizes that do not satisfy aarch64_uimm12_shift.
4433 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
4435 * profile-count.h (apply_probability,
4436 apply_scale, probability_in): Fix checks for zero.
4438 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4440 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
4441 * doc/cppdiropts.texi (-I @var{dir}): Document it.
4443 2016-06-22 Richard Biener <rguenther@suse.de>
4445 * tree-vect-loop.c (vect_model_reduction_cost): Handle
4446 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
4447 REDUC_MAX_EXPR support.
4448 (vectorizable_reduction): Likewise.
4449 (vect_create_epilog_for_reduction): Likewise.
4451 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
4453 * match.pd (A / (1 << B) -> A >> B): New.
4454 * generic-match-head.c: Include optabs-tree.h.
4455 * gimple-match-head.c: Likewise.
4456 * optabs-tree.h (target_supports_op_p): New.
4457 * optabs-tree.c (target_supports_op_p): New.
4459 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4461 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
4462 $gcc_cv_ld --help output.
4463 (gcc_cv_ld_demangle): Likewise.
4464 (gcc_cv_ld_eh_frame_hdr): Likewise.
4465 (gcc_cv_ld_pie): Likewise.
4466 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
4467 (gcc_cv_ld_buildid): Likewise.
4468 (gcc_cv_ld_sysroot): Likewise.
4469 (ld_bndplt_support): Likewise.
4470 (ld_pushpopstate_support): Likewise.
4471 * configure: Regenerate.
4472 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
4474 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4477 * config/i386/sse.md (round<mode>2): Renumber match_dup and
4478 operands indexes to avoid gap between operands and match_dups.
4480 2017-06-21 Andrew Pinski <apinski@cavium.com>
4482 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
4483 Increment Arith_shift and Arith_shift_reg by 1.
4484 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
4486 * config/aarch64/aarch64.c (thunderx_tunings): Enable
4487 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
4488 (aarch64_strip_extend): Add new argument and test for it.
4489 (aarch64_cheap_mult_shift_p): New function.
4490 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
4491 add a cost if it is true.
4492 Update calls to aarch64_strip_extend.
4493 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
4495 2017-06-21 Andrew Pinski <apinski@cavium.com>
4497 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
4499 (thunderxt88): Likewise.
4500 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
4501 (thunderx_prefetch_tune): New variable.
4502 (thunderx2t99_prefetch_tune): Update for the correct values.
4503 (thunderxt88_tunings): New variable.
4504 (thunderx_tunings): Use thunderx_prefetch_tune instead of
4505 generic_prefetch_tune.
4506 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
4508 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4510 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
4511 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
4512 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
4513 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
4514 (aarch64_atomic_cas<mode>, GPI): Likewise.
4516 2017-06-21 Martin Liska <mliska@suse.cz>
4518 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
4519 statements on cold and hot labels.
4520 * predict.c (tree_estimate_probability_bb): Remove the
4521 prediction from this place.
4523 2017-06-21 Martin Liska <mliska@suse.cz>
4525 PR tree-optimization/79489
4526 * gimplify.c (maybe_add_early_return_predict_stmt): New
4528 (gimplify_return_expr): Call the function.
4529 * predict.c (tree_estimate_probability_bb): Remove handling
4531 * predict.def: Update comment about early return predictor.
4532 * gimple-predict.h (is_gimple_predict): New function.
4533 * predict.def: Change default value of early return to 66.
4534 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
4536 * passes.def: Put pass_strip_predict_hints to the beginning of
4539 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
4541 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
4542 FUNCTION_DECL declarations.
4543 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
4545 (dwaf2out_decl): Likewise.
4546 * godump.c (go_early_global_decl): Skip call to the real debug hook
4547 for FUNCTION_DECL declarations.
4548 * passes.c (rest_of_decl_compilation): Skip call to the
4549 early_global_decl debug hook for FUNCTION_DECL declarations, unless
4550 -fdump-go-spec is passed.
4552 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4554 * config/i386/i386.c (struct builtin_isa): New field pure_p.
4555 Reorder for compactness.
4556 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
4557 (def_builtin_pure, def_builtin_pure2): New functions.
4558 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
4560 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4562 * match.pd (nop_convert): New predicate.
4563 ((A +- CST1) +- CST2): Allow some NOP conversions.
4565 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4568 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
4569 with ctors/dtors if GOVD_SHARED is set.
4571 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4573 * config/aarch64/aarch64.md (movti_aarch64):
4574 Emit mov rather than orr.
4575 (movtf_aarch64): Likewise.
4576 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
4577 Emit mov rather than orr.
4579 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4581 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
4582 Swap alternatives, make integer dup more expensive.
4584 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4586 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
4587 Return true for non-tls symbols.
4589 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
4591 * config/aarch64/aarch64-cores.def (cortex-a55): New.
4592 (cortex-a75): Likewise.
4593 (cortex-a75.cortex-a55): Likewise.
4594 * config/aarch64/aarch64-tune.md: Regenerate.
4595 * doc/invoke.texi (-mtune): Document new values for -mtune.
4597 2017-06-21 Tom de Vries <tom@codesourcery.com>
4599 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
4601 (Effective-Target Keywords, Other attributes): Suggest using
4602 dg-add-options stack_size feature to get stack limit in stack_size
4603 effective target documentation.
4605 2017-06-21 Julian Brown <julian@codesourcery.com>
4606 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4608 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
4609 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
4610 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
4612 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
4613 attribute type list for neon_multiply.
4614 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
4615 attribute type list for neon_multiply.
4616 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
4617 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
4618 attribute type list for neon_multiply.
4619 * config/arm/types.md (crypto_pmull): Add.
4620 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
4621 attribute type list.
4623 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
4625 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
4628 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4630 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
4631 to make sure not to dereference a NULL cost_classes_ptr pointer.
4633 2017-06-20 Carl Love <cel@us.ibm.com>
4635 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4636 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4637 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4638 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4639 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4640 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4641 VMULOSW): New enum "unspec" values.
4642 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
4643 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
4644 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4645 altivec_vmulosw): New patterns.
4646 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4647 VMULOSW): Add definitions.
4649 2017-06-20 Julia Koval <julia.koval@intel.com>
4651 * config/i386/i386.c: Fix rounding expand for new pattern.
4652 * config/i386/subst.md: Fix pattern (parallel -> unspec).
4654 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4656 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
4657 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
4659 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4661 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
4664 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4666 * config/aarch64/aarch64-cores.def: Rearrange to sort by
4667 architecture, then by implementer ID.
4668 * config/aarch64/aarch64-tune.md: Regenerate.
4670 2017-06-20 Richard Biener <rguenther@suse.de>
4673 * fold-const.c (split_tree): Fold to type before negating.
4675 2017-06-20 David Malcolm <dmalcolm@redhat.com>
4677 * diagnostic-show-locus.c
4678 (selftest::test_fixit_deletion_affecting_newline): New function.
4679 (selftest::diagnostic_show_locus_c_tests): Call it.
4681 2017-06-20 Andreas Schwab <schwab@suse.de>
4684 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
4687 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
4689 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
4690 __ARM_FEATURE_COPROC according to support.
4692 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4694 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
4695 Rewritten to avoid overflow for > 32-bit pointers.
4698 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
4699 by removing enum keyword.
4700 (ubsan_type_descriptor): Likewise. Formatting fix.
4703 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
4704 splitter): Require TARGET_SSE2 in the condition.
4706 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
4709 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
4710 for doing vector set of SFmode on ISA 3.0.
4711 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
4712 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
4714 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
4715 SFmode value into a V4SF variable that was extracted from another
4716 V4SF variable without converting the element to double precision
4717 and back to single precision vector format.
4718 (vsx_insert_extract_v4sf_p9_2): Likewise.
4720 2017-06-19 Jakub Jelinek <jakub@redhat.com>
4722 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
4723 in UWHI to avoid undefined overflow.
4726 * ubsan.h (enum ubsan_encode_value_phase): New.
4727 (ubsan_encode_value): Change second argument to
4728 enum ubsan_encode_value_phase with default value of
4729 UBSAN_ENCODE_VALUE_GENERIC.
4730 * ubsan.c (ubsan_encode_value): Change second argument to
4731 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
4732 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
4733 create_tmp_var_raw instead of create_tmp_var and use a
4735 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
4736 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
4737 ubsan_encode_value callers.
4740 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
4741 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
4742 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
4744 2017-06-19 Richard Biener <rguenther@suse.de>
4747 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
4748 estimates if we changed anything.
4750 2017-06-19 Richard Biener <rguenther@suse.de>
4752 PR tree-optimization/80887
4753 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
4754 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
4755 simplified lookups, then reset mprts_hook.
4756 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
4758 (try_to_simplify): Likewise.
4760 2017-06-19 Martin Liska <mliska@suse.cz>
4763 * gimplify.c (gimplify_switch_expr):
4764 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
4766 2017-06-19 Martin Liska <mliska@suse.cz>
4768 * doc/install.texi: Document that PGO runs in 4 stages.
4770 2017-06-19 Martin Liska <mliska@suse.cz>
4773 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
4774 to dispatcher function name.
4775 * multiple_target.c (replace_function_decl): New function.
4776 (create_dispatcher_calls): Redirect both edges and references.
4778 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
4780 * profile-count.c (profile_count::dump): Dump quality.
4781 (profile_count::differs_from_p): Update for unsigned val.
4782 * profile-count.h (profile_count_quality): New enum.
4783 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
4785 2017-06-19 Richard Biener <rguenther@suse.de>
4787 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
4788 struct function as arg.
4789 (estimate_numbers_of_iterations): Export overload with loop arg.
4790 (free_numbers_of_iterations_estimates_loop): Use an overload of
4791 free_numbers_of_iterations_estimates instead.
4792 * tree-cfg.c (remove_bb): Adjust.
4793 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
4794 * tree-parloops.c (gen_parallel_loop): Likewise.
4795 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
4797 (tree_unroll_loops_completely): Likewise.
4798 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
4799 Use an overload instead and export.
4800 (estimated_loop_iterations): Adjust.
4801 (max_loop_iterations): Likewise.
4802 (likely_max_loop_iterations): Likewise.
4803 (estimate_numbers_of_iterations): Take struct function as arg
4805 (loop_exits_before_overflow): Adjust.
4806 (free_numbers_of_iterations_estimates_loop): Use an overload.
4807 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
4808 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
4810 2017-06-19 Richard Biener <rguenther@suse.de>
4813 * ipa-prop.c (find_constructor_constant_at_offset): Handle
4814 RANGE_EXPR conservatively.
4816 2017-06-16 Carl Love <cel@us.ibm.com>
4818 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4819 definitions for vec_float, vec_float2, vec_floato,
4820 vec_floate built-ins.
4821 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
4822 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
4824 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
4825 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
4826 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
4827 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
4828 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
4829 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
4830 vec_floato): Add builtin defines.
4831 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
4832 Update the built-in documentation file for the new built-in
4835 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4837 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
4838 (mthumb): Mark as the negative of -marm.
4840 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4842 * doc/invoke.texi (ARM Options, -mcpu): Document supported
4844 (ARM Options, -mtune): Document that this accepts the same
4845 extension options as -mcpu.
4846 (ARM Options, -mfpu): Document addition of -mfpu=auto.
4848 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4850 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
4851 permitted extensions.
4853 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4855 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
4856 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
4857 (armv8-m.main): Add option +nodsp.
4858 * config/arm/arm-cpu-cdata.h: Regenerated.
4860 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4862 * config/arm/t-fuchsia: New file.
4863 * config.gcc (arm*-*-fuchsia*): Use it.
4865 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4867 * config/arm/t-symbian: Rewrite for new option infrastructure.
4869 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4871 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
4872 (MULTILIB_REQUIRED): Likewise.
4874 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4876 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
4877 (MULTILIB_RESUE): Likewise.
4878 (MULTILIB_MATCHES): Likewise.
4879 (MULTLIB_REQUIRED): Likewise.
4881 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4883 * config/arm/t-rtems: Rewrite for new option framework.
4885 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4887 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
4888 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
4889 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
4890 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
4891 * config/arm/t-multilib: ... here.
4892 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
4893 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
4894 armv7-a and armv8*-a when A-profile libraries have not been built.
4895 * config/arm/t-rmprofile: Rewrite.
4897 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4899 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
4900 with a backslash. Remove the backslash after substituting unescaped
4902 * doc/fragments.texi (MULTILIB_REUSE): Document it.
4904 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4906 * config.gcc: (arm*-*-*): When building a-profile libraries, force
4907 the driver to pass through the default setting of -mfloat-abi.
4908 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
4910 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
4911 (all_feat_combs): New rule.
4912 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
4914 * config/arm/t-aprofile: Rewrite.
4916 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4918 * config/arm/arm.h (FPUTYPE_AUTO): Define.
4919 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
4920 fpu is not specified by the user/command-line.
4921 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
4922 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
4923 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
4924 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
4925 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
4926 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
4928 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4930 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
4931 * config/arm/t-arm-elf: Rewritten.
4933 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4935 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
4936 have some floating-point instructions.
4937 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
4938 (TARGET_MAYBE_HARD_FLOAT): New macro.
4939 * config/arm/arm-builtins.c (arm_init_builtins): Use
4940 TARGET_MAYBE_HARD_FLOAT.
4941 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
4943 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4945 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
4946 (configargs.h): Include it.
4947 (arm_print_hint_for_fpu_option): New function.
4948 (arm_parse_fpu_option): New function.
4949 (candidate_extension): New class.
4950 (arm_canon_for_multilib): New function.
4951 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
4952 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
4953 (ARCH_CANONICAL_SPECS): New macro.
4954 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
4956 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4958 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
4959 are set after handling multilib fragments. Set target_cpu_default2
4962 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4964 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
4966 (arm*-*-*): Set target_cpu_default2 to a quoted string.
4967 * config/arm/parsecpu.awk (check_cpu): Validate any extension
4969 (check_arch): Likewise.
4970 * config/arm/arm.c (arm_configure_build_target): Handle
4971 TARGET_CPU_DEFAULT being a string constant. Scan any feature
4972 options in the default.
4974 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4976 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
4977 when an option is an alias of another.
4978 * config/arm/parsecpu.awk (optalias): New parser token.
4979 (gen_comm_data): Mark non-alias options as such. Emit entries
4980 for extension aliases.
4981 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
4982 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
4983 (armv6kz, armv6zk, armv6t2): Likewise.
4984 (armv7): Make vfpv3-d16 an alias.
4985 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
4987 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
4988 Sort in canonical order.
4989 (armv8-a): Sort in canonical order.
4990 (armv8.1-a, armv8.2-a): Likewise.
4991 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
4993 (cortex-a9): Sort in canonical order.
4994 * config/arm/arm.c (selftests.h): Include it.
4995 (arm_test_cpu_arch_data): New function.
4996 (arm_run_self_tests): New function.
4997 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
4998 (targetm): Move declaration to the end of the file.
4999 * arm-cpu-cdata.h: Regenerated.
5001 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5003 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
5004 call to target_mode_check describing the type of option passed.
5005 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
5006 (arm_target_thumb_only): Use arm_parse_arch_option_name or
5007 arm_parse_cpu_option_name to match parameters against list of
5009 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
5010 arm_arch_core_flags data structure.
5011 * config/arm/arm-cpu_cdata.h: Regenerated.
5013 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5015 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
5017 (arm_print_hint_for_cpu_option): Likewise.
5018 (arm_print_hint_for_arch_option): Likewise.
5019 (arm_parse_cpu_option_name): Likewise.
5020 (arm_parse_arch_option_name): Likewise.
5021 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
5022 of entries in the all_fpus list.
5023 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
5024 (arm_parse_cpu_option_name): Declare.
5025 (arm_parse_arch_option_name): Declare.
5026 (arm_parse_option_features): Declare.
5027 (arm_intialize_isa): Declare.
5028 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
5030 (gen_comm_data): ... here. Make definitions non-static.
5031 * config/arm/arm-cpu-data.h: Regenerated.
5032 * config/arm/arm-cpu-cdata.h: Regenerated.
5034 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5036 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
5037 (cpu_arch_extension): New structure.
5038 (cpu_arch_option, arch_option, cpu_option): New structures.
5039 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
5041 (gen_data): Generate new format data tables.
5042 * config/arm/arm.c (cpu_tune): New structure.
5043 (cpu_option, processors): Delete.
5044 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
5045 (arm_print_hint_for_cpu_option): ... this and ...
5046 (arm_print_hint_for_arch_option): ... this.
5047 (arm_parse_arch_cpu_name): Delete. Replace with ...
5048 (arm_parse_cpu_option_name): ... this and ...
5049 (arm_parse_arch_option_name): ... this.
5050 (arm_unrecognized_feature): Change type of target parameter to
5052 (arm_parse_arch_cpu_features): Delete. Replace with ...
5053 (arm_parse_option_features): ... this.
5054 (arm_configure_build_target): Rework to use new configuration data
5056 (arm_print_tune_info): Rework for new configuration data tables.
5057 * config/arm/arm-cpu-data.h: Regenerated.
5058 * config/arm/arm-cpu.h: Regenerated.
5060 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5062 * Makefile.in (OBJS): Move sbitmap.o from here ...
5063 (OBJS-libcommon): ... to here.
5065 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5067 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
5068 (ISA_ALL_CRYPTO): New macro.
5069 (ISA_ALL_SIMD): New macro
5070 (ISA_ALL_FP): New macro.
5071 * config/arm/arm.c (fpu_bitlist): Update initializer.
5072 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
5074 (arm9e): Add fpu. Add option for nofp
5075 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
5076 (arm926ej-s, arm1026ej-s): Likewise.
5077 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
5078 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
5079 neon-fp16, neon-vfpv4, nofp and nosimd.
5080 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
5081 (cortex-a8): Add fpu. Add option for nofp.
5082 (cortex-a9): Add fpu. Add options for nosimd and nofp.
5083 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
5084 (cortex-r4f): Add fpu.
5085 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
5086 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
5088 (cortex-r8): Likewise.
5089 (cortex-m4): Add fpu. Add option for nofp.
5090 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
5091 (cortex-a17.cortex-a7): Likewise.
5092 (cortex-a32): Add fpu. Add options for crypto and nofp.
5093 (cortex-a35, cortex-a53): Likewise.
5094 (cortex-a57): Add fpu. Add option for crypto.
5095 (cortex-a72, cortex-a73): Likewise.
5096 (exynos-m1): Likewise.
5097 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
5098 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
5099 (cortex-m33): Add fpu. Add option for nofp.
5100 * config/arm/arm-cpu-cdata.h: Regenerated
5101 * config/arm/arm-cpu-data.h: Regenerated.
5103 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5105 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
5106 (armv5te, armv5tej): Likewise.
5107 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
5108 (armv7): Add options fp and vfpv3-d16.
5109 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
5110 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
5112 (armv7ve): Likewise.
5113 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
5114 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
5115 (armv8-a): Add nocrypto option.
5116 (armv8.1-a, armv8.2-a): Likewise.
5117 (armv8-m.main): add options fp, fp.dp and nofp.
5119 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5121 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
5123 (armv8-a+crc): Delete.
5124 (armv8.1-a): Add options simd, crypto and nofp.
5125 (armv8.2-a): Add options fp16, simd, crypto and nofp.
5126 (armv8.2-a+fp16): Delete.
5127 (armv8-m.main): Add option dsp.
5128 (armv8-m.main+dsp): Delete.
5129 (cortex-a8): Add fpu. Add nofp option.
5130 (cortex-a9): Add fpu. Add nofp and nosimd options.
5131 * config/arm/parsecpu.awk (gen_data): Generate option tables and
5132 link to main cpu and architecture data structures.
5133 (gen_comm_data): Only put isa attributes from the main architecture
5135 (option): New statement for architecture and CPU entries.
5136 * arm.c (struct cpu_option): New structure.
5137 (struct processors): Add entry for options.
5138 (arm_unrecognized_feature): New function.
5139 (arm_parse_arch_cpu_name): Ignore any characters after the first
5141 (arm_parse_arch_cpu_feature): New function.
5142 (arm_configure_build_target): Separate out any CPU and architecture
5143 features and parse separately. Don't error out if -mfpu=auto is
5144 used with only an architecture string.
5145 (arm_print_asm_arch_directives): New function.
5146 (arm_file_start): Call it.
5147 * config/arm/arm-cpu-cdata.h: Regenerated.
5148 * config/arm/arm-cpu-data.h: Likewise.
5149 * config/arm/arm-tables.opt: Likewise.
5151 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5153 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
5154 assembler when it is not -mfpu=auto.
5156 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5158 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
5159 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
5160 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
5161 (ASM_CPU_SPEC): Rewrite.
5162 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
5163 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
5164 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
5165 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
5166 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
5167 copied string is NUL-terminated. Also strip any characters prefixed
5169 (arm_rewrite_selected_arch): New function.
5170 (arm_rewrite_march): New function.
5172 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5174 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
5175 (x_arm_cpu_string, x_arm_tune_string): Likewise.
5176 (march, mcpu, mtune): Convert to string-based options.
5177 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
5178 (arm_parse_arch_cpu_name): New function.
5179 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
5180 identify selected architecture or CPU.
5181 (arm_option_save): New function.
5182 (TARGET_OPTION_SAVE): Redefine.
5183 (arm_option_restore): Restore string options.
5184 (arm_option_print): Print string options.
5186 2017-06-16 Martin Sebor <msebor@redhat.com>
5188 PR tree-optimization/80933
5189 PR tree-optimization/80934
5190 * builtins.c (fold_builtin_3): Do not handle bcmp here.
5191 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
5192 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
5193 (gimple_fold_builtin): Call them.
5195 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5197 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
5198 as unlikely; update profile.
5200 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5202 * predict.c (force_edge_cold): Handle declaring edges impossible
5205 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5207 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
5209 (try_unroll_loop_completely): Fix reporting.
5211 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5213 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
5215 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
5218 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
5219 if given a non-constant argument for an intrinsic which requires a
5222 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5224 * profile.c (compare_freqs): New function.
5225 (branch_prob): Sort edge list.
5226 (find_spanning_tree): Assume that the list is priority sorted.
5228 2017-06-16 Richard Biener <rguenther@suse.de>
5230 PR tree-optimization/81090
5231 * passes.def (pass_record_bounds): Remove.
5232 * tree-pass.h (make_pass_record_bounds): Likewise.
5233 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
5234 make_pass_record_bounds): Likewise.
5235 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
5236 not free niter estimates at the beginning but at the end.
5237 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
5239 2017-06-16 Richard Biener <rguenther@suse.de>
5241 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
5242 initializer to workaround ICE in host GCC 4.8.
5244 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5246 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
5248 (clone_inlined_nodes): Update.
5250 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5252 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
5253 prefetch settings, and enable prefetching by default at -O3.
5255 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5257 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5258 Set flag_prefetch_loop_arrays according to tuning data.
5260 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5262 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
5264 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
5265 [Unrelated to main purpose of the patch] Place the pointer field last
5266 to enable type checking errors when tune structure are wrongly merged.
5267 * config/aarch64/aarch64.c (generic_prefetch_tune,)
5268 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
5269 (thunderx2t99_prefetch_tune): New tune constants.
5270 (tune_params *_tunings): Update all tunings (no functional change).
5271 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
5272 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
5273 from tunings structures.
5275 2017-06-16 Jakub Jelinek <jakub@redhat.com>
5278 * ubsan.c (instrument_null): Add T argument, use it instead
5279 of computing it based on IS_LHS.
5280 (instrument_object_size): Likewise.
5281 (pass_ubsan::execute): Adjust instrument_null and
5282 instrument_object_size callers to pass gimple_get_lhs or
5283 gimple_assign_rhs1 result to it. Use instrument_null instead of
5284 calling get_base_address and instrument_mem_ref. Handle
5285 aggregate call arguments for object-size sanitization.
5287 2017-06-16 Yury Gribov <tetra2005@gmail.com>
5289 PR tree-optimization/81089
5290 * tree-vrp.c (is_masked_range_test): Validate operands of
5293 2017-06-15 Martin Sebor <msebor@redhat.com>
5296 * dumpfile.c (dump_register): Avoid calling memset to initialize
5297 a class with a default ctor.
5298 * gcc.c (struct compiler): Remove const qualification.
5299 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
5300 * hash-table.h: Ditto.
5301 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
5303 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
5304 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
5306 * params.h (struct param_info): Make struct members non-const.
5307 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
5308 with default initialization.
5309 * vec.h (vec_copy_construct, vec_default_construct): New helper
5311 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
5312 with vec_copy_construct.
5313 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
5314 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
5315 * doc/invoke.texi (-Wclass-memaccess): Document.
5317 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5319 * emit-rtl.h (is_leaf): Update comment about local
5322 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5325 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
5326 for a variable to have a section before checking if the section has a
5328 Set section to.persistent if persistent attribute is set.
5329 Warn if .persistent attribute is used on an automatic variable.
5331 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
5333 PR rtl-optimization/80474
5334 * reorg.c (update_block): Do not ignore instructions in a delay slot.
5336 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
5338 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
5341 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
5343 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
5344 (casesi): Emit bounds checking as RTL.
5345 (casesi_internal_mips16_<mode>): Remove bounds checking.
5347 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
5349 * config/xtensa/xtensa.c (xtensa_option_override): Append
5350 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
5351 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
5352 xtensa_doloop_hooks): Define unconditionally.
5353 (xtensa_reorg_loops): Only call reorg_loops in the presence of
5355 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
5356 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
5357 for it in xtensa_option_override.
5358 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
5359 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
5361 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
5363 * doc/cppopts.texi: Document '-' special value to -MF.
5365 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
5367 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
5368 (cortex_a53_fconst): Likewise.
5369 (cortex_a53_fpmul): Likewise.
5370 (cortex_a53_f_load_64): Likewise.
5371 (cortex_a53_f_load_many): Likewise.
5372 (cortex_a53_advsimd_alu): Likewise.
5373 (cortex_a53_advsimd_alu_q): Likewise.
5374 (cortex_a53_advsimd_mul): Likewise.
5375 (cortex_a53_advsimd_mul_q): Likewise.
5376 (fpmac bypass): Add new bypass for fpmac-fpmac case.
5377 Add missing fmul, r2f_cvt and fconst cases.
5379 2017-06-14 Richard Biener <rguenther@suse.de>
5382 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
5384 (fold_binary_loc): When associating do not treat pre-existing
5385 TREE_OVERFLOW on literal constants as a reason to allow
5386 TREE_OVERFLOW on associated literal constants.
5388 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5390 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
5391 (MASK_FEATURES): New macro.
5392 * config/sparc/sparc.c (sparc_option_override): Remove the special
5393 handling of -mfpu and generalize it to all MASK_FEATURES switches.
5395 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5397 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
5398 a division of 0 if non-call exceptions are enabled.
5400 2017-06-14 Andrew Pinski <apinski@cavium.com>
5401 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5404 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
5405 Improve vector initialization code gen for only variable case.
5407 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5409 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
5411 2017-06-14 Richard Biener <rguenther@suse.de>
5413 PR tree-optimization/81083
5414 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
5417 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5419 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
5420 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
5421 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
5422 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
5423 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
5424 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
5426 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5428 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
5429 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
5431 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5433 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
5435 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5437 * config/rs6000/t-rtems: Don't handle SPE.
5439 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5441 * config/rs6000/t-linux: Don't handle SPE.
5443 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5445 * config/rs6000/eabispe.h: Delete file.
5447 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5449 * config/rs6000/t-spe: Delete file.
5451 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5453 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
5454 (rs6000_legitimate_offset_address_p): Return false for anything in
5455 V2SImode or V2SFmode.
5457 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5459 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
5460 except V2SF and V2SI. Rearrange the vector modes, and add comments.
5461 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
5463 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
5464 (rs6000_legitimate_offset_address_p): Ditto.
5465 (rs6000_emit_move): Ditto.
5466 (rs6000_init_builtins): Remove V4HI_type_node.
5468 2017-06-13 Martin Liska <mliska@suse.cz>
5471 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
5472 (gate_asan): Likewise.
5473 * asan.h (asan_no_sanitize_address_p): Remove the function.
5474 (sanitize_flags_p): New function.
5475 * builtins.def: Fix coding style.
5476 * common.opt: Use renamed enum value.
5477 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
5478 * doc/extend.texi: Document no_sanitize attribute.
5479 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
5480 to SANITIZE_UNDEFINED_NONDEFAULT.
5481 * gcc.c (sanitize_spec_function): Use the renamed enum value.
5482 * gimple-fold.c (optimize_atomic_compare_exchange_p):
5483 Use sanitize_flags_p.
5484 * gimplify.c (gimplify_function_tree): Likewise.
5485 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
5486 * opts.c (parse_no_sanitize_attribute): New function.
5487 (common_handle_option): Use renamed enum value.
5488 * opts.h (parse_no_sanitize_attribute): Declare.
5489 * tree.c (sanitize_flags_p): New function.
5490 * tree.h: Declared here.
5491 * tsan.c: Use sanitize_flags_p.
5492 * ubsan.c (ubsan_expand_null_ifn): Likewise.
5493 (instrument_mem_ref): Likewise.
5494 (instrument_bool_enum_load): Likewise.
5495 (do_ubsan_in_current_function): Remove the function.
5496 (pass_ubsan::execute): Use sanitize_flags_p.
5497 * ubsan.h: Remove do_ubsan_in_current_function
5498 * tree-cfg.c (print_no_sanitize_attr_value): New function.
5499 (dump_function_to_file): Use it here.
5501 2017-06-13 Martin Jambor <mjambor@suse.cz>
5503 PR tree-optimization/80803
5504 PR tree-optimization/81063
5505 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
5506 (propagate_subaccesses_across_link): Enqueue subtree whenever
5507 necessary instead of relying on the caller.
5509 2017-06-13 Martin Jambor <mjambor@suse.cz>
5511 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
5512 that have a first_link.
5513 (sort_and_splice_var_accesses): Do not check first_link before
5515 (subtree_mark_written_and_enqueue): Likewise.
5516 (propagate_all_subaccesses): Likewise and do not stop at first
5517 parent with a first_link.
5519 2017-06-13 Martin Jambor <mjambor@suse.cz>
5521 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
5524 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5526 * match.pd: New pattern.
5528 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5530 * tree-vrp.c (is_masked_range_test): New function.
5531 (register_edge_assert_for): Determine ranges for
5534 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5536 PR tree-optimization/67328
5537 * fold-const.c (maskable_range_p): New function.
5538 (build_range_check): Generate bittests if possible.
5540 2017-06-13 Martin Liska <mliska@suse.cz>
5542 * gimple-pretty-print.c (dump_probability): Add new argument.
5543 (dump_edge_probability): Dump both probability and count.
5544 (dump_gimple_label): Likewise.
5545 (dump_gimple_bb_header): Likewise.
5547 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
5550 * config/avr/avr-devices.c: Fix indentation.
5551 * config/avr/gen-avr-mmcu-specs.c: Dito.
5553 2017-06-13 Richard Biener <rguenther@suse.de>
5555 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
5556 instead get vector type from stmt_info.
5557 (vectorizable_reduction): Adjust. Remove dead code.
5559 2017-06-13 Richard Biener <rguenther@suse.de>
5562 * fold-const.c (extract_muldiv_1): Remove bogus distribution
5563 case of C * (x * C2 + C3).
5564 (fold_addr_of_array_ref_difference): Properly fold index difference.
5566 2017-06-12 David S. Miller <davem@davemloft.net>
5569 * config/sparc/sparc.md (return expander): Emit frame blockage if
5570 function uses alloca.
5572 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
5574 * combine.c (make_field_assignment): Check len rather than the mode
5575 precision when calling force_to_mode.
5577 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
5579 Support multilibs and devices that see flash in RAM address range.
5582 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
5583 (avr_mcu_t) <flash_pm_offset>: New field.
5584 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
5585 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
5586 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
5587 (AVR_TINY_PM_OFFSET): Remove macro.
5588 * config/avr/avr.opt (-mshort-calls): New option.
5589 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
5590 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
5591 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
5592 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
5593 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
5594 instead of avr_arch->have_jmp_call.
5595 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
5596 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
5597 avr_arch->flash_pm_offset to define.
5598 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
5599 new field flash_pm_offset. Add entry for avrxmega3.
5600 (avr_texinfo): Add entry for avrxmega3.
5601 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
5602 attiny212, attiny214,
5603 attiny412, attiny414, attiny416, attiny417,
5604 attiny814, attiny816, attiny817,
5605 attiny1614, attiny1616, attiny1617,
5606 attiny3214, attiny3216, attiny3217.
5607 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
5608 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
5609 (avr_print_operand_address) [AVR_TINY]: Same.
5610 (avr_asm_init_sections) <readonly_data_section>: Only patch
5611 callback if avr_arch->flash_pm_offset = 0.
5612 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
5613 for rodata if avr_arch->flash_pm_offset != 0.
5614 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
5615 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
5616 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
5617 (m_options): Append opt_rcall.
5618 (m_dirnames): Append dir_rcall.
5619 * config/avr/t-multilib: Regenerate.
5621 * configure.ac [target=avr]: Check whether avrxmega3 default
5622 linker description file works as needed.
5623 * configure: Regenerate.
5624 * doc/avr-mmcu.texi: Regenerate.
5625 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
5626 <__AVR_ARCH__>: Document avrxmega3 and 103.
5627 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
5628 <__AVR_SHORT_CALLS__>: Document it.
5629 <__AVR_PM_BASE_ADDRESS__>: Document it.
5630 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
5631 (AVR Variable Attributes) <progmem>: Document this is
5632 not needed for avrxmega3.
5633 (AVR Named Address Spaces) <__flash>: Dito.
5635 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
5637 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
5639 2017-06-12 Doug Rupp <rupp@adacore.com>
5641 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
5642 Append vxworks-stdint.h to the tm_file list.
5643 * config/vxworks-stdint.h: New file.
5645 2017-06-12 Martin Liska <mliska@suse.cz>
5647 PR tree-optimization/81041
5648 * tree-profile.c (gimple_gen_ic_func_profiler):
5649 Create an extra BB in profile-generate
5650 (gimple_gen_time_profiler): Likewise.
5652 2017-06-12 Jakub Jelinek <jakub@redhat.com>
5654 PR tree-optimization/81003
5655 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
5656 (update_range_test): Use it instead of force_gimple_operand_gsi.
5658 2017-06-12 Richard Biener <rguenther@suse.de>
5660 PR tree-optimization/81053
5661 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
5662 with backedge value not defined in loop. Simplify def stmt
5665 2017-06-11 Tom de Vries <tom@codesourcery.com>
5668 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
5670 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
5671 nvptx_cannot_force_const_mem.
5673 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5675 * opts.c (finish_options): Move test for flag_split_stack after
5676 it has been initialized.
5678 2017-06-11 Jason Merrill <jason@redhat.com>
5680 * tree.h (id_equal): New.
5681 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
5682 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
5683 instead of strcmp of IDENTIFIER_POINTER.
5685 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5687 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
5688 (mark_all_inlined_calls_cdtor): Fix formating.
5689 (inline_transform): Rescale profile before inlining.
5691 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5693 * cgraph.h (cgraph_edge::clone): Update prototype.
5694 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
5695 (cgraph_node::create_clone): Update.
5696 (cgraph_node::create_version_clone): Update.
5697 * tree-inline.c (copy_bb): Update.
5698 (expand_call_inline): Update.
5700 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5702 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
5703 factored out from ...
5704 (rs6000_emit_prologue): ... here.
5706 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5708 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
5709 factored out from ...
5710 (rs6000_emit_prologue): ... here.
5712 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5714 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
5716 (handle_missing_profiles): Fix computation of tp_first_run.
5717 (counts_to_freqs): Do not touch freqs when count is 0.
5719 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5721 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
5724 2017-06-10 Tom de Vries <tom@codesourcery.com>
5726 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
5727 attributes): Document signal effective target.
5729 2017-06-10 Tom de Vries <tom@codesourcery.com>
5731 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
5732 Document effective target stack_size.
5734 2017-06-09 David Malcolm <dmalcolm@redhat.com>
5736 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
5737 to the edit_context if they can be auto-applied.
5739 2017-06-9 Ian Lance Taylor <iant@golang.org>
5741 * opts.c (finish_options): If -fsplit-stack, disable implicit
5742 -forder-blocks-and-partition.
5743 * doc/invoke.texi (Optimize Options): Document that when using
5744 -fsplit-stack -forder-blocks-and-partition is not implicitly
5747 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5749 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
5750 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
5751 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
5752 * builtins.def (abort, trap, unreachable): Declare cold.
5753 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
5754 * tree-core.h (ECF_COLD): New.
5755 * tree.c (set_call_expr_flags): Handle ECF_COLD.
5756 (build_common_builtin_nodes): Mark unreachable and abort as cold.
5758 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5760 * predict.c (unlikely_executed_stmt_p): Cleanup.
5762 2017-06-09 Richard Biener <rguenther@suse.de>
5764 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
5765 model if the ref is always written to.
5767 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5769 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
5771 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5773 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
5776 2017-06-09 Tom de Vries <tom@codesourcery.com>
5779 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
5780 "target cannot support label values" when encountering LABEL_REF.
5782 2017-06-09 Martin Liska <mliska@suse.cz>
5784 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
5785 (gimple_gen_ic_func_profiler): Emit direct comparison
5786 of __gcov_indirect_call_callee with NULL.
5787 (gimple_gen_time_profiler): Change probability from
5788 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
5790 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5792 * profile.c (edge_gcov_counts): Turn to pointer.
5793 (compute_branch_probabilities, compute_branch_probabilities): Update.
5794 (branch_prob): Do not clear edge_gcov_count.
5795 * profile.h (edge_gcov_counts): Turn to pointer.
5796 (edge_gcov_count): Update.
5798 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5800 * gimple.h (gimple_check_failed): Mark cold.
5802 2017-06-09 Richard Biener <rguenther@suse.de>
5804 PR tree-optimization/66623
5805 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
5806 refactor check_reduction into two parts, properly computing
5807 whether we have to check reduction validity for outer loop
5810 2017-06-09 Richard Biener <rguenther@suse.de>
5812 PR tree-optimization/79483
5813 * graphite-scop-detection.c (order): New global.
5814 (get_order): Compute bb to order mapping that satisfies code
5815 generation constraints.
5816 (cmp_pbbs): New helper.
5817 (build_scops): Start domwalk at entry block, sort generated
5820 2017-06-09 Richard Biener <rguenther@suse.de>
5823 * ipa-polymorphic-call.c
5824 (ipa_polymorphic_call_context::restrict_to_inner_class):
5825 Skip FIELD_DECLs with error_mark_node type.
5826 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
5829 2017-06-09 Martin Liska <mliska@suse.cz>
5831 * predict.c (struct branch_predictor): New struct.
5832 (test_prediction_value_range): New test.
5833 (predict_c_tests): New function.
5834 * selftest-run-tests.c (selftest::run_tests): Run the function.
5835 * selftest.h: Declare new tests.
5837 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
5840 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
5841 gen_add3_insn did not fail.
5842 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
5843 r0, construct that number in a temporary reg and add that reg to r0.
5844 If asked to put the result in r0 as well, fail.
5846 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
5848 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5849 for early expansion of vec_eqv.
5851 2017-06-08 Jakub Jelinek <jakub@redhat.com>
5854 * ubsan.c (instrument_null): Avoid pointless code temporary.
5855 (pass_ubsan::execute): Instrument aggregate arguments of calls.
5857 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
5861 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
5864 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
5867 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5869 * predict.c (unlikely_executed_edge_p): Move ahead.
5870 (probably_never_executed_edge_p): Use it.
5872 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5875 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
5876 gimple_call_builtin_p call.
5878 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5880 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
5881 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
5882 rtl_check_failed_type2, rtl_check_failed_code1,
5883 rtl_check_failed_code2, rtl_check_failed_code_mode,
5884 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
5885 rtvec_check_failed_bounds, rtl_check_failed_flag,
5886 _fatal_insn_not_found, _fatal_insn): Likewise.
5887 * tree.h (tree_contains_struct_check_failed,
5888 tree_check_failed, tree_not_check_failed,
5889 tree_class_check_failed, tree_range_check_failed,
5890 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
5891 tree_vec_elt_check_failed, phi_node_elt_check_failed,
5892 tree_operand_check_failed, omp_clause_check_failed,
5893 omp_clause_operand_check_failed, omp_clause_range_check_failed):
5896 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5898 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
5899 flag_branch_probabilities.
5900 * ipa-inline.c (edge_badness): Likewise.
5901 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
5902 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
5903 * predict.c (maybe_hot_frequency_p): Likewise.
5904 (probably_never_executed): Likewise.
5905 * sched-ebb.c (schedule_ebbs): Likewise.
5906 * sched-rgn.c (find_single_block_region): Likewise.
5907 * tracer.c (tail_duplicate): Likewise.
5909 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5911 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
5912 longer requires x_flag_profile_use.
5914 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5916 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
5917 instead of flag_reorder_blocks_and_partition.
5918 * dbxout.c (dbxout_function_end): Likewise.
5919 * dwarf2out.c (gen_subprogram_die): Likewise.
5920 * haifa-sched.c (sched_create_recovery_edges): Likewise.
5921 * hw-doloop.c (reorg_loops): Likewise.
5922 * varasm.c (assemble_start_function,
5923 assemble_end_function): Likewise.
5924 (decide_function_section): Do not check for
5925 flag_reorder_blocks_and_partition.
5927 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5929 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
5931 (chkp_get_hard_register_fake_addr_expr): Ditto.
5932 (chkp_build_addr_expr): Add check for hard reg case.
5933 (chkp_parse_array_and_component_ref): Ditto.
5934 (chkp_find_bounds_1): Ditto.
5935 (chkp_process_stmt): Don't generate bounds store for
5938 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5940 * predict.c (maybe_hot_bb_p): Do not check profile status.
5941 (maybe_hot_edge_p): Likewise.
5942 (probably_never_executed): Check for zero counts even if profile
5944 (unlikely_executed_edge_p): New function.
5945 (unlikely_executed_stmt_p): New function.
5946 (unlikely_executed_bb_p): New function.
5947 (set_even_probabilities): Use unlikely predicates.
5948 (combine_predictions_for_bb): Likewise.
5949 (predict_paths_for_bb): Likewise.
5950 (predict_paths_leading_to_edge): Likewise.
5951 (determine_unlikely_bbs): New function.
5952 (estimate_bb_frequencies): Use it.
5953 (compute_function_frequency): Use zero counts even if profile is
5955 * profile-count.h: Fix typo.
5957 2017-08-08 Julia Koval <julia.koval@intel.com>
5959 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
5960 _mm512_mask_cvtsepi16_storeu_epi8,
5961 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
5962 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
5963 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
5964 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
5965 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
5966 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
5967 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
5968 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
5969 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
5970 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
5971 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
5972 __builtin_ia32_pmovuswb256mem_mask,
5973 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
5974 __builtin_ia32_pmovwb512mem_mask): New builtins.
5976 2017-08-08 Julia Koval <julia.koval@intel.com>
5978 PR target/73350,80862
5979 * config/i386/subst.md (round): Fix round pattern.
5980 * config/i386/i386.c (ix86_erase_embedded_rounding):
5981 Fix erasing rounding for the fixed pattern.
5983 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5985 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
5987 2017-06-08 Martin Liska <mliska@suse.cz>
5989 PR gcov-profile/80911
5990 * gcov.c (block_info::block_info): New constructor.
5992 2017-06-07 Carl Love <cel@us.ibm.com>
5994 * config/rs6000/rs6000-c: The return type of the following
5995 built-in functions was implemented as int not long long. Fix sign
5996 of return value for the unsigned version of vec_mulo and vec_mule.
5997 vector unsigned long long vec_bperm (vector unsigned long long,
5998 vector unsigned char)
5999 vector signed long long vec_mule (vector signed int,
6001 vector unsigned long long vec_mule (vector unsigned int,
6002 vector unsigned int)
6003 vector signed long long vec_mulo (vector signed int,
6005 vector unsigned long long vec_mulo (vector unsigned int,
6006 vector unsigned int)
6007 * doc/extend.texi: Fix the documentation for the built-in
6010 2017-06-07 Carl Love <cel@us.ibm.com>
6013 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
6016 2017-06-07 Carl Love <cel@us.ibm.com>
6018 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
6019 support, Generate doublehv for signed int/float for BE case only.
6021 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
6023 * doc/invoke.texi (mcx16): Rewrite.
6025 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6027 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
6028 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
6029 *mov<mode>_softfloat, and an anonymous splitter): Use
6030 nonimmediate_operand instead of rs6000_nonimmediate_operand.
6032 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6034 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
6036 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
6037 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
6038 (rs6000_debug_reg_global): Adjust.
6039 (rs6000_init_hard_regno_mode_ok): Adjust.
6040 (rs6000_dbx_register_number): Adjust.
6041 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
6042 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
6043 Remove SPE_ACC and SPEFSCR.
6044 (REG_ALLOC_ORDER): Ditto.
6045 (FRAME_POINTER_REGNUM): Change to 111.
6046 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
6047 (REG_CLASS_NAMES): Ditto.
6048 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
6049 (REGISTER_NAMES): Ditto.
6050 (ADDITIONAL_REG_NAMES): Ditto.
6051 (rs6000_reg_names): Ditto.
6052 * config/rs6000/rs6000.md: Renumber some register number
6055 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6057 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
6059 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
6060 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
6062 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
6063 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
6064 Delete the SPE high registers.
6065 (REG_ALLOC_ORDER): Ditto.
6066 (enum reg_class): Remove SPE_HIGH_REGS.
6067 (REG_CLASS_NAMES): Ditto.
6068 (REG_CLASS_CONTENTS): Delete the SPE high registers.
6069 (REGISTER_NAMES): Ditto.
6070 (rs6000_reg_names): Ditto.
6071 * doc/tm.texi.in: Remove SPE as example.
6072 * doc/tm.texi: Regenerate.
6074 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6076 * config/rs6000/8540.md (ppc8540_brinc): Delete.
6077 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
6078 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
6079 * config/rs6000/rs6000.md (type): Remove "brinc".
6081 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6083 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
6084 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
6085 * config/rs6000/linuxspe.h: Delete file.
6086 * config/rs6000/rs6000.md: Don't include spe.md.
6087 * config/rs6000/spe.h: Delete file.
6088 * config/rs6000/spe.md: Delete file.
6089 * config/rs6000/t-rs6000: Remove spe.md.
6091 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6093 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
6094 (reg_or_none500mem_operand): Delete.
6095 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
6096 instead of reg_or_none500mem_operand.
6098 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6100 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
6101 handling of SPE flags.
6102 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
6104 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6106 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
6108 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
6109 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
6110 paired_divv2sf3): Similar.
6111 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
6112 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
6113 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
6115 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
6116 Rename the paired_* instruction patterns.
6117 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
6119 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
6120 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
6121 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
6122 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
6123 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
6125 (struct machine_function): Delete field spe_insn_chain_scanned_p.
6126 (spe_func_has_64bit_regs_p): Delete.
6127 (spe_expand_predicate_builtin): Delete.
6128 (spe_expand_evsel_builtin): Delete.
6129 (TARGET_DWARF_REGISTER_SPAN): Do not define.
6130 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
6131 (invalid_e500_subreg): Delete.
6132 (rs6000_legitimize_address): Always force_reg op2 as well, for
6133 paired single memory accesses.
6134 (rs6000_member_type_forces_blk): Delete.
6135 (rs6000_spe_function_arg): Delete.
6136 (rs6000_expand_unop_builtin): Delete SPE handling.
6137 (rs6000_expand_binop_builtin): Ditto.
6138 (spe_expand_stv_builtin): Delete.
6139 (bdesc_2arg_spe): Delete.
6140 (spe_expand_builtin): Delete.
6141 (spe_expand_predicate_builtin): Delete.
6142 (spe_expand_evsel_builtin): Delete.
6143 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
6144 (spe_init_builtins): Delete.
6145 (spe_func_has_64bit_regs_p): Delete.
6146 (savres_routine_name): Delete "info" parameter. Adjust callers.
6147 (rs6000_emit_stack_reset): Ditto.
6148 (rs6000_dwarf_register_span): Delete.
6149 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
6150 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
6151 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
6153 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
6155 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
6156 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
6157 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
6158 mulv2sf3, divv2sf3): Delete expanders.
6160 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6162 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
6164 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6166 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
6167 * config/rs6000/rs6000.c: Ditto.
6169 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6171 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
6172 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
6173 comparison_operator.
6175 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6177 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
6178 * config/rs6000/rs6000.opt: Ditto.
6179 * config/rs6000/t-rtems: Ditto.
6181 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6183 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
6184 TARGET_E500_SINGLE by 0, simplify.
6185 * config/rs6000/rs6000.c: Ditto.
6186 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
6187 (spe_build_register_parallel): Delete.
6188 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
6189 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
6190 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
6191 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
6192 (E500_CONVERT): Delete.
6193 * config/rs6000/spe.md: Remove many patterns and all define_constants.
6195 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6197 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
6198 * config/rs6000/dfp.md: Ditto.
6199 (negdd2, *negdd2_fpr): Merge.
6200 (absdd2, *absdd2_fpr): Merge.
6201 (negtd2, *negtd2_fpr): Merge.
6202 (abstd2, *abstd2_fpr): Merge.
6203 * config/rs6000/e500.h: Delete file.
6204 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
6205 TARGET_FPRS by 1 and simplify.
6206 * config/rs6000/rs6000-c.c: Ditto.
6207 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
6209 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
6211 * config/rs6000/rs6000.md: Ditto.
6212 (floatdidf2, *floatdidf2_fpr): Merge.
6213 (move_from_CR_gt_bit): Delete.
6214 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
6215 (E500_CR_IOR_COMPARE): Delete.
6216 (All patterns that require !TARGET_FPRS): Delete.
6217 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
6219 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6221 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
6223 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6225 * graphds.c (add_edge): Intitialize edge's attached data.
6226 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
6227 pointer parameter. Call pointed function on each edge during
6228 graph traversing. Skip traversing the edge when the function
6230 (graphds_dfs, graphds_scc): Ditto.
6231 (for_each_edge): New parameter. Pass the new parameter to callback
6233 * graphds.h (skip_edge_callback): New function pointer type.
6234 (graphds_dfs, graphds_scc): New function pointer parameter.
6235 (graphds_edge_callback, for_each_edge): New parameter.
6237 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6239 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
6240 out code checking if runtime alias check is possible to below ...
6241 Call the new function.
6242 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
6243 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
6245 2017-06-07 Marek Polacek <polacek@redhat.com>
6248 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
6249 TYPE_OVERFLOW_WRAPS check.
6251 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6253 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
6254 if versioning is required.
6255 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
6256 peeling with the check for versioning.
6258 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6260 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
6261 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
6262 Set true to new parameter if new ssa variable is defined.
6263 (vect_gen_vector_loop_niters): Refactor. Set range information
6264 for the new vector loop bound variable.
6265 (vect_do_peeling): Ditto.
6267 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6269 * tree-affine.c (ssa.h): Include header file.
6270 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
6271 has wrapping overflow behavior.
6273 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6275 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
6277 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6279 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
6280 (tree_to_aff_combination): ... here.
6282 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6284 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
6285 reg_pressure model function.
6286 (ivopts_global_cost_for_size): Delete.
6287 (determine_set_costs, iv_ca_recount_cost): Call new model function
6288 ivopts_estimate_reg_pressure.
6290 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6292 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
6293 expensive than udiv. Remove floating point cases from mod.
6295 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6297 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
6300 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6302 * config/aarch64/aarch64.md
6303 (copysignsf3): Fix mask generation.
6305 2017-06-07 Jakub Jelinek <jakub@redhat.com>
6307 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
6309 (class dump_manager): Add register_dumps method.
6310 * dumpfile.c: Include langhooks.h.
6311 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
6312 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
6313 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
6314 (dump_manager::dump_register): Start with 512 entries instead of 32.
6315 (dump_manager::register_dumps): New method.
6316 * toplev.c (general_init): Instead of invoking register_dumps
6317 langhook, invoke register_dumps method on the dump manager.
6318 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
6321 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
6323 * doc/md.texi: Clarify the restrictions on a define_insn condition.
6324 Say that # requires an associated define_split to exist, and that
6325 the define_split must be suitable for use after register allocation.
6327 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6329 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
6330 (compute_outgoing_frequencies): Also initialize zero counts.
6331 (find_many_sub_basic_blocks): Do not produce uninitialized profile
6332 around loops; preserve more of profile when nothing changes.
6334 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
6336 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
6338 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
6339 * config/arm/arm-cpu-cdata.h: Regenerate.
6340 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
6341 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
6342 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
6343 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
6344 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
6346 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
6348 * config/arm/t-rmprofile: Likewise.
6349 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
6351 2017-06-06 David S. Miller <davem@davemloft.net>
6354 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
6355 blockage if function uses alloca.
6357 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
6359 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
6360 New "uid" fields to hold pretty-print IDs of group and ref.
6361 Memory references are now identified as <group_id>:<ref_id>
6362 instead of using [random] addresses.
6363 (dump_mem_details): Simplify, no functional change.
6364 (dump_mem_ref): Simplify and make output more concise.
6365 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
6366 (find_or_create_group): Initialize group uid.
6367 (record_ref): Initialize ref uid. Improve debug output.
6368 (prune_group_by_reuse, should_issue_prefetch_p,)
6369 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
6370 (mark_nontemporal_store, determine_loop_nest_reuse):
6371 Improve debug output.
6373 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
6375 * dbgcnt.def (prefetch): New debug counter.
6376 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
6377 (schedule_prefetches): Stop issueing prefetches if debug counter
6380 2017-06-06 Tom de Vries <tom@codesourcery.com>
6382 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
6383 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
6385 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6387 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
6388 Use aarch64_reg_or_zero predicate for operand 4.
6389 (aarch64_compare_and_swap<mode> define_insn_and_split):
6390 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
6391 (aarch64_store_exclusive<mode>): Likewise for operand 2.
6393 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
6395 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
6396 (arm_compute_save_core_reg_mask): This.
6397 (thumb1_compute_save_reg_mask): Rename into ...
6398 (thumb1_compute_save_core_reg_mask): This.
6399 (arm_compute_save_reg0_reg12_mask): Adapt comment.
6400 (arm_compute_frame_layout): Likewise.
6402 2017-06-06 Richard Biener <rguenther@suse.de>
6404 PR tree-optimization/80974
6405 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
6406 keep or clear leaders SSA info.
6408 2017-06-06 Tom de Vries <tom@codesourcery.com>
6410 * config/nvptx/nvptx.c (split_mode_p): New function.
6411 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
6413 2017-06-06 Tom de Vries <tom@codesourcery.com>
6415 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
6417 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6420 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
6423 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6425 * shrink-wrap.c (handle_simple_exit): Update profile.
6426 (try_shrink_wrapping): Upate profile.
6428 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6430 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
6431 (tree_guess_outgoing_edge_probabilities): New.
6432 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
6433 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
6435 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6437 * ipa-split.c (split_function): Initialize return bb profile.
6439 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6441 * profile.c (compute_branch_probabilities): Also initialize
6444 2017-06-06 Richard Biener <rguenther@suse.de>
6446 PR tree-optimization/80928
6447 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
6448 (vect_analyze_loop_operations): Properly guard analysis for
6450 (vect_transform_loop): Likewise.
6451 (vect_analyze_loop_2): Also reset SLP type on PHIs.
6452 (vect_model_induction_cost): Do not cost for pure SLP.
6453 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
6454 of induction in inner loop vectorization.
6455 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
6456 (vect_get_and_check_slp_defs): Handle vect_induction_def.
6457 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
6459 (vect_analyze_slp_cost_1): Cost induction.
6460 (vect_detect_hybrid_slp_stmts): Handle PHIs.
6461 (vect_get_slp_vect_defs): Likewise.
6462 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
6463 (vect_transform_stmt): Handle SLP reductions.
6464 * tree-vectorizer.h (vectorizable_induction): Adjust.
6466 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6468 * config/rs6000/rs6000.c (make_resolver_func): Update
6469 init_lowered_empty_function call.
6471 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
6473 * doc/invoke.texi: Document the -fprofile-abs-path option.
6474 * common.opt (fprofile-abs-path): New option.
6475 * gcov-io.h (gcov_write_filename): Declare.
6476 * gcov-io.c (gcov_write_filename): New function.
6477 * coverage.c (coverage_begin_function): Use gcov_write_filename.
6478 * profile.c (output_location): Likewise.
6480 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6482 * shring-wrap.c: Revert accidental commit.
6484 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
6486 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
6488 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6490 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
6492 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
6493 profile in callgraph edge.
6494 * profile-count.h (apply_probability): If THIS is 0, then result is 0
6495 (apply_scale): Likewise.
6496 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
6497 Also scale profile when inlining function with zero profile.
6498 (initialize_cfun): Update exit block profile even when it is zero.
6499 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
6500 when profile is read.
6502 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6504 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
6505 (CLONE_*): New constants to define the processors we can generate
6506 code for with the target_clone attribute.
6507 (rs6000_clone_map): New array to identify which clone processors
6508 the current program is running on.
6509 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
6510 target_clone attribute.
6511 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
6512 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
6513 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
6514 (cpu_expand_builtin): Add support for target_clone attribute.
6515 (rs6000_valid_attribute_p): Allow "default" attribute.
6516 (get_decl_name): New debug function to simplify printing the
6517 current function name in debugging statements.
6518 (rs6000_clone_priority): New functions to support the target_clone
6519 attribute, and be able to generate code to switch between ISA 2.05
6520 through ISA 3.0 (power6 through power9).
6521 (rs6000_compare_version_priority): Likewise.
6522 (rs6000_get_function_versions_dispatcher): Likewise.
6523 (make_resolver_func): Likewise.
6524 (add_condition_to_bb): Likewise.
6525 (dispatch_function_versions): Likewise.
6526 (rs6000_generate_version_dispatcher_body): Likewise.
6527 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
6528 (fusion_gpr_load_p): Fix a spacing issue.
6529 * doc/extend.texi (Common Function Attributes): Document that the
6530 PowerPC supports the target_clone attribute.
6532 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
6534 * config/arm/arm.h: explain F symbol found in description of ARM
6535 register allocation in its legend.
6537 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6539 * config/mips/frame-header-opt.c: Include profile-count.h.
6540 * config/riscv/riscv.c: Include profile-count.h
6542 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6544 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
6546 (sm_set_flag_if_changed): Add bbs field.
6547 (execute_sm_if_changed_flag_set): Pass BBS.
6548 (execute_sm): Update.
6550 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6552 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6555 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6557 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
6558 (peephole2): New peephole2 to emit the above.
6559 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
6561 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6563 * config/aarch64/aarch64.c (define_peephole2 above
6564 *sub_<shift>_<mode>): New peephole.
6566 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
6568 * config/i386/i386.c (make_resolver_func): Update.
6569 * Makefile.in: Add profile-count.h and profile-count.o
6570 * auto-profile.c (afdo_indirect_call): Update to new API.
6571 (afdo_set_bb_count): Update.
6572 (afdo_propagate_edge): Update.
6573 (afdo_propagate_circuit): Update.
6574 (afdo_calculate_branch_prob): Update.
6575 (afdo_annotate_cfg): Update.
6576 * basic-block.h: Include profile-count.h
6577 (struct edge_def): Turn count to profile_count.
6578 (struct basic_block_def): Likewie.
6579 (REG_BR_PROB_BASE): Move to profile-count.h
6580 (RDIV): Move to profile-count.h
6581 * bb-reorder.c (max_entry_count): Turn to profile_count.
6582 (find_traces): Update.
6583 (rotate_loop):Update.
6584 (connect_traces):Update.
6585 (sanitize_hot_paths):Update.
6586 * bt-load.c (migrate_btr_defs): Update.
6587 * cfg.c (RDIV): Remove.
6588 (init_flow): Use alloc_block.
6589 (alloc_block): Uninitialize count.
6590 (unchecked_make_edge): Uninitialize count.
6591 (check_bb_profile): Update.
6592 (dump_edge_info): Update.
6593 (dump_bb_info): Update.
6594 (update_bb_profile_for_threading): Update.
6595 (scale_bbs_frequencies_int): Update.
6596 (scale_bbs_frequencies_gcov_type): Update.
6597 (scale_bbs_frequencies_profile_count): New.
6598 * cfg.h (update_bb_profile_for_threading): Update.
6599 (scale_bbs_frequencies_profile_count): Declare.
6600 * cfgbuild.c (compute_outgoing_frequencies): Update.
6601 (find_many_sub_basic_blocks): Update.
6602 * cfgcleanup.c (try_forward_edges): Update.
6603 (try_crossjump_to_edge): Update.
6604 * cfgexpand.c (expand_gimple_tailcall): Update.
6605 (construct_exit_block): Update.
6606 * cfghooks.c (verify_flow_info): Update.
6607 (dump_bb_for_graph): Update.
6608 (split_edge): Update.
6609 (make_forwarder_block): Update.
6610 (duplicate_block): Update.
6611 (account_profile_record): Update.
6612 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
6613 (get_estimated_loop_iterations): Update.
6614 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
6615 (single_likely_exit): Update.
6616 * cfgloopmanip.c (scale_loop_profile): Update.
6618 (set_zero_probability): Update.
6619 (lv_adjust_loop_entry_edge): Update.
6620 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
6621 (purge_dead_edges): Update.
6622 (rtl_account_profile_record): Update.
6623 * cgraph.c (cgraph_node::create): Uninitialize count.
6624 (symbol_table::create_edge): Uninitialize count.
6625 (cgraph_update_edges_for_call_stmt_node): Update.
6626 (cgraph_edge::dump_edge_flags): Update.
6627 (cgraph_node::dump): Update.
6628 (cgraph_edge::maybe_hot_p): Update.
6629 * cgraph.h: Include profile-count.h
6630 (create_clone), create_edge, create_indirect_edge): Update.
6631 (cgraph_node): Turn count to profile_count.
6632 (cgraph_edge0: Likewise.
6633 (make_speculative, clone): Update.
6634 (create_edge): Update.
6635 (init_lowered_empty_function): Update.
6636 * cgraphclones.c (cgraph_edge::clone): Update.
6637 (duplicate_thunk_for_node): Update.
6638 (cgraph_node::create_clone): Update.
6639 * cgraphunit.c (cgraph_node::analyze): Update.
6640 (cgraph_node::expand_thunk): Update.
6641 * final.c (dump_basic_block_info): Update.
6642 * gimple-streamer-in.c (input_bb): Update.
6643 * gimple-streamer-out.c (output_bb): Update.
6644 * graphite.c (print_global_statistics): Update.
6645 (print_graphite_scop_statistics): Update.
6646 * hsa-brig.c: Include basic-block.h.
6647 * hsa-dump.c: Include basic-block.h.
6648 * hsa-gen.c (T sum_slice): Update.
6649 (convert_switch_statements):Update.
6650 * hsa-regalloc.c: Include basic-block.h.
6651 * ipa-chkp.c (chkp_produce_thunks): Update.
6652 * ipa-cp.c (struct caller_statistics): Update.
6653 (init_caller_stats): Update.
6654 (gather_caller_stats): Update.
6655 (ipcp_cloning_candidate_p): Update.
6656 (good_cloning_opportunity_p): Update.
6657 (get_info_about_necessary_edges): Update.
6658 (dump_profile_updates): Update.
6659 (update_profiling_info): Update.
6660 (update_specialized_profile): Update.
6661 (perhaps_add_new_callers): Update.
6662 (decide_about_value): Update.
6663 (ipa_cp_c_finalize): Update.
6664 * ipa-devirt.c (struct odr_type_warn_count): Update.
6665 (struct decl_warn_count): Update.
6666 (struct final_warning_record): Update.
6667 (possible_polymorphic_call_targets): Update.
6668 (ipa_devirt): Update.
6669 * ipa-fnsummary.c (redirect_to_unreachable): Update.
6670 * ipa-icf.c (sem_function::merge): Update.
6671 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
6672 * ipa-inline.c (compute_uninlined_call_time): Update.
6673 (compute_inlined_call_time): Update.
6674 (want_inline_small_function_p): Update.
6675 (want_inline_self_recursive_call_p): Update.
6676 (edge_badness): Update.
6677 (lookup_recursive_calls): Update.
6678 (recursive_inlining): Update.
6679 (inline_small_functions): Update.
6680 (dump_overall_stats): Update.
6681 (dump_inline_stats): Update.
6682 * ipa-profile.c (ipa_profile_generate_summary): Update.
6683 (ipa_propagate_frequency): Update.
6684 (ipa_profile): Update.
6685 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
6686 * ipa-utils.c (ipa_merge_profiles): Update.
6687 * loop-doloop.c (doloop_modify): Update.
6688 * loop-unroll.c (report_unroll): Update.
6689 (unroll_loop_runtime_iterations): Update.
6690 * lto-cgraph.c (lto_output_edge): Update.
6691 (lto_output_node): Update.
6692 (input_node): Update.
6693 (input_edge): Update.
6694 (merge_profile_summaries): Update.
6695 * lto-streamer-in.c (input_cfg): Update.
6696 * lto-streamer-out.c (output_cfg): Update.
6697 * mcf.c (create_fixup_graph): Update.
6698 (adjust_cfg_counts): Update.
6699 (sum_edge_counts): Update.
6700 * modulo-sched.c (sms_schedule): Update.
6701 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
6702 * predict.c (maybe_hot_count_p): Update.
6703 (probably_never_executed): Update.
6704 (dump_prediction): Update.
6705 (combine_predictions_for_bb): Update.
6706 (propagate_freq): Update.
6707 (handle_missing_profiles): Update.
6708 (counts_to_freqs): Update.
6709 (rebuild_frequencies): Update.
6710 (force_edge_cold): Update.
6711 * predict.h: Include profile-count.h
6712 (maybe_hot_count_p, counts_to_freqs): UPdate.
6713 * print-rtl-function.c: Do not include cfg.h
6714 * print-rtl.c: Include basic-block.h
6715 * profile-count.c: New file.
6716 * profile-count.h: New file.
6717 * profile.c (is_edge_inconsistent): Update.
6718 (correct_negative_edge_counts): Update.
6719 (is_inconsistent): Update.
6720 (set_bb_counts): Update.
6721 (read_profile_edge_counts): Update.
6722 (compute_frequency_overlap): Update.
6723 (compute_branch_probabilities): Update; Initialize and deinitialize
6725 (branch_prob): Update.
6726 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
6727 (edge_gcov_count): New.
6728 (bb_gcov_count): New.
6729 * shrink-wrap.c (try_shrink_wrapping): Update.
6730 * tracer.c (better_p): Update.
6731 * trans-mem.c (expand_transaction): Update.
6732 (ipa_tm_insert_irr_call): Update.
6733 (ipa_tm_insert_gettmclone_call): Update.
6734 * tree-call-cdce.c: Update.
6735 * tree-cfg.c (gimple_duplicate_sese_region): Update.
6736 (gimple_duplicate_sese_tail): Update.
6737 (gimple_account_profile_record): Update.
6738 (execute_fixup_cfg): Update.
6739 * tree-inline.c (copy_bb): Update.
6740 (copy_edges_for_bb): Update.
6741 (initialize_cfun): Update.
6742 (freqs_to_counts): Update.
6743 (copy_cfg_body): Update.
6744 (expand_call_inline): Update.
6745 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
6746 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
6747 (try_unroll_loop_completely): Update.
6748 (try_peel_loop): Update.
6749 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6750 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
6751 * tree-ssa-loop-split.c (connect_loops): Update.
6752 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
6753 * tree-ssa-reassoc.c (branch_fixup): Update.
6754 * tree-ssa-tail-merge.c (replace_block_by): Update.
6755 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
6756 (compute_path_counts): Update.
6757 (update_profile): Update.
6758 (recompute_probabilities): Update.
6759 (update_joiner_offpath_counts): Update.
6760 (estimated_freqs_path): Update.
6761 (freqs_to_counts_path): Update.
6762 (clear_counts_path): Update.
6763 (ssa_fix_duplicate_block_edges): Update.
6764 (duplicate_thread_path): Update.
6765 * tree-switch-conversion.c (case_bit_test_cmp): Update.
6766 (struct switch_conv_info): Update.
6767 * tree-tailcall.c (decrease_profile): Update.
6768 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
6769 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6770 * value-prof.c (check_counter): Update.
6771 (gimple_divmod_fixed_value): Update.
6772 (gimple_mod_pow2): Update.
6773 (gimple_mod_subtract): Update.
6774 (gimple_ic_transform): Update.
6775 (gimple_stringop_fixed_value): Update.
6776 * value-prof.h (gimple_ic): Update.
6778 2017-06-02 Carl Love <cel@us.ibm.com>
6780 * config/rs6000/rs6000-c: Add support for built-in functions
6781 vector double vec_doublee (vector signed int);
6782 vector double vec_doublee (vector unsigned int);
6783 vector double vec_doublee (vector float);
6784 vector double vec_doubleh (vector signed int);
6785 vector double vec_doubleh (vector unsigned int);
6786 vector double vec_doubleh (vector float);
6787 vector double vec_doublel (vector signed int);
6788 vector double vec_doublel (vector unsigned int);
6789 vector double vec_doublel (vector float);
6790 vector double vec_doubleo (vector signed int);
6791 vector double vec_doubleo (vector unsigned int);
6792 vector double vec_doubleo (vector float);.
6793 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
6794 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
6796 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
6797 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
6798 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
6800 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
6801 vec_doublel, vec_doubleh.
6802 * doc/extend.texi: Update the built-in documentation file for the
6803 new built-in functions.
6805 2017-06-02 David Malcolm <dmalcolm@redhat.com>
6808 * ipa-inline-analysis.c (free_growth_caches): Set
6809 edge_removal_hook_holder to NULL after removing it.
6811 2017-06-02 Sudakshina Das <sudi.das@arm.com>
6813 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
6814 comparision with zero.
6816 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6817 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6818 for early expansion of vec_min and vec_max builtins.
6819 (builtin_function_type): Add min/max unsigned variants to those
6820 identified as having unsigned arguments.
6822 2017-06-02 Olivier Hainque <hainque@adacore.com>
6824 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
6826 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6828 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
6829 Use VALL_F16 iterator rather than VALL.
6831 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6833 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
6834 Emit CBNZ inside loop when doing a strong exchange and comparing
6835 against zero. Generate the CC flags after the loop.
6837 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
6839 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
6840 (dl_section_ref): New.
6841 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
6842 On AIX, append an expression to subtract the size of the
6843 section length to dl_section_ref.
6845 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6847 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6848 for early expansion of vector absolute builtins.
6850 2017-06-02 Richard Biener <rguenther@suse.de>
6852 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
6853 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
6855 2017-06-02 Richard Biener <rguenther@suse.de>
6857 PR tree-optimization/80948
6858 * tree-tailcall.c (find_tail_calls): Track stmts to move in
6861 2017-06-02 Richard Biener <rguenther@suse.de>
6863 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
6865 * tree-vect-stmts.c (process_use): Do not mark backedge defs
6866 for inductions as relevant.
6868 2017-06-02 Richard Biener <rguenther@suse.de>
6870 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
6871 (vectorizable_induction): ... this. Remove dead code.
6873 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
6875 * builtins. (expand_builtin_alloca): Remove second parameter and
6876 infer its value from the first parameter instead.
6877 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
6879 2017-06-02 Jakub Jelinek <jakub@redhat.com>
6881 PR rtl-optimization/80903
6882 * loop-doloop.c (add_test): Unshare sequence.
6884 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
6886 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
6888 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
6890 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
6892 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
6893 xlogue_layout::get_instance, logue_layout::xlogue_layout,
6894 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
6895 (xlogue_layout::get_stub_rtx): Make static.
6896 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
6897 (xlogue_layout::compute_stub_managed_regs): Rename to...
6898 (xlogue_layout::count_stub_managed_regs): ...this.
6899 (xlogue_layout::is_stub_managed_reg): New function.
6900 (xlogue_layout::m_stub_names): Rename to...
6901 (xlogue_layout::s_stub_names): ...this, make static.
6902 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
6903 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
6904 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
6905 xlogue_layout::s_stub_names): Instantiate statics.
6906 (stub_managed_regs): Remove.
6907 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
6908 (disable_call_ms2sysv_xlogues): Rename to...
6909 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
6910 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
6912 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
6913 change after reload_completed.
6914 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
6916 (ix86_expand_prologue): Likewise.
6917 (ix86_expand_epilogue): Likewise.
6918 (ix86_expand_split_stack_prologue): Likewise.
6919 (ix86_compute_frame_layout): Remove frame parameter ...
6920 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
6921 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
6923 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
6924 (ix86_frame): Move from here ...
6925 * config/i386/i386.h (ix86_frame): ... to here.
6926 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
6927 complete ix86_frame data structure instead. Remove some_ld_name.
6929 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
6931 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
6932 symbols that hold a DECL_VALUE_EXPR.
6934 2017-06-01 Martin Jambor <mjambor@suse.cz>
6936 PR tree-optimization/80898
6937 * tree-sra.c (process_subtree_disqualification): Removed.
6938 (disqualify_candidate): Do not acll
6939 process_subtree_disqualification.
6940 (subtree_mark_written_and_enqueue): New function.
6941 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
6942 RHS has been disqualified and re-queue LHS if necessary. Apart
6943 from that, ignore disqualified RHS.
6945 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6947 * config/s390/s390.c (s390_emit_epilogue): Disable early return
6948 address fetch for z10 or later.
6950 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6952 * config/arc/arc.md (tst_movb): Add guard when splitting.
6954 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6956 * config/arc/arc.c (arc_can_eliminate): Test against
6957 arc_frame_pointer_needed.
6959 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6961 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
6962 to prevent store reordering.
6963 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
6964 (type): Add block type.
6965 (stack_tie): Define special instruction to be used in
6968 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6970 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
6971 constraint. It is not valid for the pattern.
6972 (noncommutative_binary_comparison): Likewise.
6974 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6976 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
6979 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6981 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
6982 be used by the reg-alloc.
6984 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6986 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
6987 reg-alloc when having mul64 or mul32x16 instructions.
6988 (mulsidi3): Likewise.
6989 (umulsidi3): Likewise.
6990 (mulsi32x16): New pattern.
6991 (mulsi64): Likewise.
6992 (mulsidi64): Likewise.
6993 (umulsidi64): Likewise.
6994 (MUL32x16_REG): Define.
6995 (mul64_600): Use MUL32x16_REG.
6996 (mac64_600): Likewise.
6997 (umul64_600): Likewise.
6998 (umac64_600): Likewise.
7000 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7002 * config/arc/arc.md (mulsi3_700): Make it commutative.
7004 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
7006 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
7008 (*sign_extendsidi2_insn): Likewise for movstosw.
7010 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
7012 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
7013 the type of the input discriminant value. Convert the
7014 discriminant value of signedness vary.
7016 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
7018 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
7019 Add to -Wall section.
7021 2017-06-01 Richard Biener <rguenther@suse.de>
7024 * fold-const.c (fold_plusminus_mult_expr): If the factored
7025 factor may be zero use a wrapping type for the inner operation.
7026 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
7027 and handle moved defs.
7028 (process_assignment): Properly guard the unary op case. Return a
7029 tri-state indicating that moving the stmt before the call may allow
7030 to continue. Pass through to_move.
7031 (find_tail_calls): Handle moving unrelated defs before
7034 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
7037 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
7038 splitter result in the canonical way.
7040 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
7042 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
7043 also for 32bit target. Update insn attributes.
7044 (zero-extendsidi2 splitter): Allow all registers for operand 1.
7046 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
7048 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
7049 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
7050 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
7051 (_mm_maskz_min_ss): New intrinsics.
7053 2017-05-31 Martin Liska <mliska@suse.cz>
7055 * tree-vect-loop.c (vect_create_epilog_for_reduction):
7056 Change comment style to one we normally use.
7057 (vectorizable_reduction): Likewise.
7058 (vectorizable_induction): Likewise.
7059 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7060 (vectorizable_call): Likewise.
7061 (vectorizable_simd_clone_call): Likewise.
7062 (vectorizable_conversion): Likewise.
7063 (vectorizable_assignment): Likewise.
7064 (vectorizable_shift): Likewise.
7065 (vectorizable_operation): Likewise.
7066 (vectorizable_store): Likewise.
7067 (vectorizable_load): Likewise.
7068 * tree-vectorizer.h: Likewise.
7070 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
7072 * passes.c (emergency_dump_function): New.
7073 * tree-pass.h (emergency_dump_function): Declare.
7074 * plugin.c (plugins_internal_error_function): Remove.
7075 * plugin.h (plugins_internal_error_function): Remove declaration.
7076 * toplev.c (internal_error_function): New static function. Use it...
7077 (general_init): ...here.
7079 2017-05-31 Graham Markall <graham.markall@embecosm.com>
7081 * config/arc/arc.c (arc_print_operand): Handle constant operands.
7082 (arc_rtx_costs): Add costs for new patterns.
7083 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
7084 * config/arc/predicates.md: Add _1_2_3_operand predicate.
7086 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
7088 * tree-ssa-strlen.c (get_next_strinfo): New function.
7089 (get_stridx_plus_constant): Use it.
7090 (zero_length_string): Likewise.
7091 (adjust_related_strinfos): Likewise.
7092 (adjust_last_stmt): Likewise.
7094 2017-05-31 Richard Biener <rguenther@suse.de>
7097 * config/i386/i386.c (ix86_expand_builtin): Remove assert
7098 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
7100 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
7102 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
7103 loop_vinfo argument and use of dependence distance vectors.
7104 Check instead whether the two references differ only in their
7105 initial value and assume that they have the same alignment if the
7106 difference is a multiple of the vector alignment.
7107 (vect_analyze_data_refs_alignment): Update call accordingly.
7109 2017-05-31 Martin Liska <mliska@suse.cz>
7112 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
7114 2017-05-31 Bin Cheng <bin.cheng@arm.com>
7116 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
7117 (create_intersect_range_checks): Move from ...
7118 * tree-data-ref.c (create_intersect_range_checks_index)
7119 (create_intersect_range_checks): ... to here.
7120 (create_runtime_alias_checks): New function factored from ...
7121 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
7122 here. Call above function.
7123 * tree-data-ref.h (create_runtime_alias_checks): New function.
7125 2017-05-31 Bin Cheng <bin.cheng@arm.com>
7127 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
7128 segment length for dr_b and compute it in wide_int.
7130 2017-05-31 Richard Biener <rguenther@suse.de>
7132 PR tree-optimization/80906
7133 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
7134 and pass through iv_map.
7135 (copy_bb_and_scalar_dependences): Adjust.
7136 (translate_pending_phi_nodes): Likewise.
7137 (copy_loop_close_phi_args): Handle code-generating IVs instead
7140 2017-05-30 David Malcolm <dmalcolm@redhat.com>
7142 * diagnostic-color.c (color_dict): Add "type-diff".
7143 (parse_gcc_colors): Update comment.
7144 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
7145 -fdiagnostics-show-template-tree and -fno-elide-type.
7146 (GCC_COLORS): Add type-diff to example.
7148 (-fdiagnostics-show-template-tree): New.
7149 (-fno-elide-type): New.
7150 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
7151 the pp_format_decoder callback. Call any m_format_postprocessor's
7153 (pretty_printer::pretty_printer): Initialize
7154 m_format_postprocessor.
7155 (pretty_printer::~pretty_printer): Delete any
7156 m_format_postprocessor.
7157 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
7158 (class format_postprocessor): New class.
7159 (struct pretty_printer::format_decoder): Document the new parameters.
7160 (struct pretty_printer::m_format_postprocessor): New field.
7161 * tree-diagnostic.c (default_tree_printer): Update for new
7162 bool and const char ** params.
7163 * tree-diagnostic.h (default_tree_printer): Likewise.
7165 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
7167 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
7168 (lwa_operand): Delete rs6000_gen_cell_microcode test.
7169 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
7170 rs6000_gen_cell_microcode code.
7171 (rs6000_final_prescan_insn): Delete.
7172 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
7173 "warn-cell-microcode" entries.
7174 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
7175 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
7176 throughout. Change cc_reg_not_micro_cr0_operand to
7177 cc_reg_not_cr0_operand throughout.
7178 (*extendhi<mode>2_noload): Delete.
7179 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
7180 (mwarn-cell-microcode): Delete.
7181 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
7182 -mgen-cell-microcode and -mwarn-cell-microcode.
7184 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
7187 * config/i386/constraints.md (Yd): New constraint.
7189 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
7190 and (?Yd, r) alternatives. Update insn attributes.
7191 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
7192 and (?*Yd, r) alternatives. Update insn attributes.
7193 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
7195 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
7197 * gimplify.c (gimplify_modify_expr): Don't create a
7198 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
7201 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
7203 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
7205 2017-05-30 Richard Biener <rguenther@suse.de>
7207 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
7208 and reduc_def fields.
7209 (STMT_VINFO_REDUC_TYPE): New define.
7210 (STMT_VINFO_REDUC_DEF): Likewise.
7211 (vect_force_simple_reduction): Adjust prototype.
7212 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
7213 (vect_is_simple_reduction): Remove check_reduction argument.
7214 (vect_force_simple_reduction): Adjust and set
7215 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
7216 (vectorizable_reduction): Do not re-do reduction analysis
7217 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
7218 * tree-parloops.c (gather_scalar_reductions): Adjust.
7220 2017-05-30 Richard Biener <rguenther@suse.de>
7223 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
7226 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
7228 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
7229 Introduce unknown_misalignment parameter and remove vf.
7230 (vect_peeling_hash_get_lowest_cost):
7231 Pass unknown_misalignment parameter.
7232 (vect_enhance_data_refs_alignment):
7233 Fix unsupportable data ref treatment.
7235 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7237 * tree-vect-data-refs.c (vect_get_data_access_cost):
7238 Workaround for SLP handling.
7239 (vect_enhance_data_refs_alignment):
7240 Compute costs for doing no peeling at all, compare to the best
7241 peeling costs so far and avoid peeling if cheaper.
7243 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7245 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
7246 Return peeling info and set costs to zero for unlimited cost
7248 (vect_enhance_data_refs_alignment): Also inspect all datarefs
7249 with unknown misalignment. Compute and costs for unknown
7250 misalignment, compare them to the costs for known misalignment
7251 and choose the cheapest for peeling.
7253 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7255 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
7256 (vect_get_peeling_costs_all_drs): Create function.
7257 (vect_peeling_hash_get_lowest_cost):
7258 Use vect_get_peeling_costs_all_drs.
7259 (vect_peeling_supportable): Create function.
7260 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
7262 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7264 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
7265 DR_HAS_NEGATIVE_STEP.
7266 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
7267 (vect_enhance_data_refs_alignment): Use.
7268 (vect_duplicate_ssa_name_ptr_info): Use.
7269 * tree-vectorizer.h (dr_misalignment): Use.
7270 (known_alignment_for_access_p): Use.
7272 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7275 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
7277 (has_section_name): New function.
7279 2017-05-30 Martin Liska <mliska@suse.cz>
7282 * auto-profile.c (get_function_decl_from_block): Fix
7285 2017-05-30 Richard Biener <rguenther@suse.de>
7288 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
7290 2017-05-30 Martin Liska <mliska@suse.cz>
7292 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
7293 * dumpfile.h (struct dump_file_info): Remove ctors.
7295 2017-05-30 Martin Liska <mliska@suse.cz>
7297 * predict.def: Fix GNU coding style.
7299 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
7301 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
7302 Mark 'to' argument with ATTRIBUTE_UNUSED.
7304 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
7306 * config/xtensa/xtensa.c (xtensa_emit_call): Use
7307 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
7308 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
7311 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
7313 * doc/install.texi (Options specification): Restore entry of
7314 --enable-sjlj-exceptions.
7316 2017-05-27 Michael Eager <eager@eagercon.com>
7319 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
7321 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
7323 * config/microblaze/microblaze.h
7324 (FIXED_REGISTERS): Update in macro.
7325 (CALL_USED_REGISTERS): Update in macro.
7327 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
7329 * doc/install.texi: Add links to macOS binary distributions.
7331 2017-05-27 Jakub Jelinek <jakub@redhat.com>
7335 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7337 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
7339 2017-05-26 Martin Liska <mliska@suse.cz>
7341 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
7343 2017-05-26 Martin Liska <mliska@suse.cz>
7345 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
7346 always leading ';; '.
7347 (dump_bb_info): Likewise.
7348 (brief_dump_cfg): Likewise.
7349 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
7350 * dumpfile.c: Remove usage of TDF_VERBOSE.
7351 * dumpfile.h (enum dump_kind): Likewise.
7352 (dump_gimple_bb_header): Do not use TDF_COMMENT.
7353 * print-tree.c (debug_verbose): Remove.
7354 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
7355 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
7356 * tree-diagnostic.c (default_tree_printer): Replace
7357 TDF_DIAGNOSTIC with TDF_SLIM.
7359 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7361 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
7362 in parameter loop, rather than loop_vinfo.
7363 (create_intersect_range_checks): Ditto.
7364 (vect_create_cond_for_alias_checks): Update call to above functions.
7366 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7368 PR tree-optimization/80815
7369 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
7370 for merging runtime alias checks. Handle negative DR_STEPs.
7372 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7374 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
7376 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
7377 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
7378 out code pruning runtime alias checks.
7379 * tree-data-ref.c (prune_runtime_alias_test_list): New function
7380 factored out from above.
7381 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
7383 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
7385 (prune_runtime_alias_test_list): New decalaration.
7387 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7389 * tree-vect-data-refs.c (compare_tree): Rename and move ...
7390 * tree-data-ref.c (data_ref_compare_tree): ... to here.
7391 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
7392 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
7393 (operator==, comp_dr_with_seg_len_pair): Ditto.
7394 (vect_prune_runtime_alias_test_list): Ditto.
7396 2017-05-26 Martin Liska <mliska@suse.cz>
7399 * params.def: Bound partial-inlining-entry-probability param.
7401 2017-05-26 Marek Polacek <polacek@redhat.com>
7404 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
7407 2017-05-26 Richard Biener <rguenther@suse.de>
7409 PR tree-optimization/80842
7410 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
7413 2017-05-26 Richard Biener <rguenther@suse.de>
7415 PR tree-optimization/80844
7416 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
7418 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
7420 * doc/md.texi (Machine Constraints): Update x86 family
7421 machine constraints section to match 'config/i386/constraints.md'.
7423 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
7425 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
7427 2017-05-25 Nathan Sidwell <nathan@acm.org>
7429 * doc/invoke.texi (--enable-languages): Update documentation.
7431 2017-05-25 Martin Liska <mliska@suse.cz>
7433 * dumpfile.c: Add TDF_FOLDING.
7434 * dumpfile.h (enum dump_kind): Likewise.
7435 * genmatch.c (dt_simplify::gen_1): Use it.
7437 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7439 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
7441 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7443 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
7444 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
7446 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7448 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
7449 * match.pd (X == C): Rewrite it here.
7450 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
7451 with_certain_nonzero_bits2): New predicates.
7452 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
7454 2017-05-24 Nathan Sidwell <nathan@acm.org>
7456 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
7459 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
7462 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
7464 * config/powerpcspe: New port. Files are copied from the rs6000
7465 port, with "rs6000" in filenames replaced by "powerpcspe".
7467 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
7469 PR rtl-optimization/80754
7470 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
7472 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
7474 * config/sparc/sparc.md (length): Return the correct value for -mflat
7475 sibcalls to match output_sibcall.
7477 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
7481 * config/rs6000/rs6000.c (struct machine_function): Add new field
7483 (rs6000_get_separate_components): Init that field, use it.
7484 (rs6000_components_for_bb): Use the field.
7486 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
7488 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
7490 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
7493 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
7495 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7498 * config/s390/s390.c (s390_check_qrst_address): Check incoming
7499 address against address_operand predicate.
7500 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
7502 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
7504 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
7505 parameters passed indirectly.
7507 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
7509 * config/i386/i386.md (*movdi_internal): Remove SSE4
7510 alternative 18 (?r, *v). Update insn attributes.
7511 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
7512 Update insn attributes.
7513 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
7514 Update insn attributes.
7515 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
7516 alternative 1 (r, v). Remove isa attribute.
7517 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
7518 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
7519 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
7521 2017-05-23 Tom de Vries <tom@codesourcery.com>
7523 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
7526 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7528 * cgraphunit.c (symbol_table::process_new_functions): Update.
7529 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
7530 (inline_generate_summary): Rename to ...
7531 (ipa_fn_summary_generate): ... this one.
7532 (inline_read_summary): Rename to ...
7533 (ipa_fn_summary_read): ... this one.
7534 (inline_write_summary): Rename to ...
7535 (ipa_fn_summary_write): ... this one.
7536 (inline_free_summary): Rename to ...
7537 (ipa_free_fn_summary): ... this one.
7538 (pass_data_local_fn_summary, pass_local_fn_summary,
7539 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
7540 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
7541 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
7542 make_pass_ipa_fn_summary): New.
7543 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
7544 inline_write_summary, inline_free_summary): Remove.
7545 (ipa_free_fn_summary) : New.
7546 * ipa-inline.c (ipa_inline): Update.
7547 (pass_ipa_inline): Do not generate summaries.
7548 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
7550 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
7551 and add pass_ipa_fn_summary.
7552 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
7554 (make_pass_inline_parameters): Remove.
7556 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
7558 * omp-low.c (struct omp_context): Remove "default_kind" member.
7561 * omp-offload.c (execute_oacc_device_lower): Remove the
7562 parallelism dimensions function attributes for unparallelized
7563 OpenACC kernels constructs.
7565 2017-05-23 Martin Liska <mliska@suse.cz>
7567 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
7569 (cgraph_edge::make_speculative): Likewise.
7570 (cgraph_edge::resolve_speculation): Likewise.
7571 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
7572 (cgraph_node::dump): Likewise.
7573 * cgraph.h: Likewise.
7574 * cgraphunit.c (analyze_functions): Likewise.
7575 (symbol_table::compile): Likewise.
7576 * ipa-cp.c (print_all_lattices): Likewise.
7577 (determine_versionability): Likewise.
7578 (initialize_node_lattices): Likewise.
7579 (ipcp_verify_propagated_values): Likewise.
7580 (estimate_local_effects): Likewise.
7581 (update_profiling_info): Likewise.
7582 (create_specialized_node): Likewise.
7583 (perhaps_add_new_callers): Likewise.
7584 (decide_about_value): Likewise.
7585 (decide_whether_version_node): Likewise.
7586 (identify_dead_nodes): Likewise.
7587 (ipcp_store_bits_results): Likewise.
7588 * ipa-devirt.c (dump_targets): Likewise.
7589 (ipa_devirt): Likewise.
7590 * ipa-icf.c (sem_item::dump): Likewise.
7591 (sem_function::equals): Likewise.
7592 (sem_variable::equals): Likewise.
7593 (sem_item_optimizer::read_section): Likewise.
7594 (sem_item_optimizer::execute): Likewise.
7595 (congruence_class::dump): Likewise.
7596 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
7597 (dump_inline_summary): Likewise.
7598 (estimate_node_size_and_time): Likewise.
7599 (inline_analyze_function): Likewise.
7600 * ipa-inline-transform.c (inline_call): Likewise.
7601 * ipa-inline.c (report_inline_failed_reason): Likewise.
7602 (want_early_inline_function_p): Likewise.
7603 (edge_badness): Likewise.
7604 (update_edge_key): Likewise.
7605 (inline_small_functions): Likewise.
7606 * ipa-profile.c (ipa_profile): Likewise.
7607 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
7608 (ipa_make_edge_direct_to_target): Likewise.
7609 (remove_described_reference): Likewise.
7610 (ipa_impossible_devirt_target): Likewise.
7611 (propagate_controlled_uses): Likewise.
7612 (ipa_print_node_params): Likewise.
7613 (ipcp_transform_function): Likewise.
7614 * ipa-pure-const.c (pure_const_read_summary): Likewise.
7615 (propagate_pure_const): Likewise.
7616 * ipa-reference.c (generate_summary): Likewise.
7617 (read_write_all_from_decl): Likewise.
7618 (propagate): Likewise.
7619 (ipa_reference_read_optimization_summary): Likewise.
7620 * ipa-utils.c (ipa_merge_profiles): Likewise.
7621 * ipa.c (walk_polymorphic_call_targets): Likewise.
7622 (symbol_table::remove_unreachable_nodes): Likewise.
7623 (ipa_single_use): Likewise.
7624 * passes.c (execute_todo): Likewise.
7625 * predict.c (drop_profile): Likewise.
7626 * symtab.c (symtab_node::get_dump_name): New function.
7627 (symtab_node::dump_name): Likewise.
7628 (symtab_node::dump_asm_name): Likewise.
7629 (symtab_node::dump_references): Likewise.
7630 (symtab_node::dump_referring): Likewise.
7631 (symtab_node::dump_base): Likewise.
7632 (symtab_node::debug_symtab): Likewise.
7633 * tree-sra.c (convert_callers_for_node): Likewise.
7634 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
7635 * value-prof.c (init_node_map): Likewise.
7637 2017-05-23 Martin Liska <mliska@suse.cz>
7639 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
7640 and symtab_node::debug_symtab to symbol_table::debug.
7641 * cgraphunit.c (analyze_functions): Use the renamed function.
7642 (symbol_table::compile): Likewise.
7643 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
7644 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
7645 * passes.c (execute_todo): Likewise.
7646 * symtab.c (symbol_table::dump): New function.
7647 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
7649 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7651 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
7652 that nonconst implies exec.
7654 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7656 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7657 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7658 (inline_edge_summary_vec): Turn into ...
7659 (ipa_call_summaries): ... this one.
7660 (redirect_to_unreachable, edge_set_predicate,
7661 evaluate_properties_for_edge, inline_summary_alloc,
7662 reset_ipa_call_summary, reset_inline_summary,
7663 inline_summary_t::duplicate): Update.
7664 (inline_edge_duplication_hook): Turn to ...
7665 (ipa_call_summary_t::duplicate): ... this one.
7666 (inline_edge_removal_hook): Turn to ...
7667 (ipa_call_summary_t::remove): ... this one.
7668 (dump_inline_edge_summary): Turn to ...
7669 (dump_ipa_call_summary): ... this one.
7670 (estimate_function_body_sizes): Update.
7671 (inline_update_callee_summaries): Update.
7672 (remap_edge_change_prob): Update.
7673 (remap_edge_summaries): Update.
7674 (inline_merge_summary): Update.
7675 (do_estimate_edge_time): Update.
7676 (inline_generate_summary): Update.
7677 (inline_read_section): Update.
7678 (inline_read_summary): Update.
7679 (inline_free_summary): Update.
7680 * ipa-inline.c (can_inline_edge_p): Update.
7681 (compute_inlined_call_time): Update.
7682 (want_inline_small_function_p): Update.
7683 (edge_badness): Update.
7684 (early_inliner): Update.
7685 * ipa-inline.h (inline_edge_summary): Turn to ...
7686 (ipa_call_summary): ... this one.
7687 (ipa_call_summary_t): New class.
7688 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
7689 (ipa_call_summaries): New.
7690 (inline_edge_summary): Remove.
7691 (estimate_edge_growth): Update.
7692 * ipa-profile.c (ipa_propagate_frequency_1): Update.
7693 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7694 * ipa-split.c (execute_split_functions): Update.
7695 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
7697 2017-05-23 Tom de Vries <tom@codesourcery.com>
7699 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7700 attributes): Document rdrand effective target.
7702 2017-05-23 Tom de Vries <tom@codesourcery.com>
7704 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7705 attributes): Sort alphabetically.
7707 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
7709 * config/avr/genmultilib.awk: Use gsub instead of gensub.
7711 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
7714 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
7715 V2DF/V2DI splat into two separate patterns, one that handles
7716 registers, and the other that only handles memory. Drop support
7717 for splatting from a GPR on ISA 2.07 and then splitting the
7718 splat into direct move and splat.
7719 (vsx_splat_<mode>_reg): Likewise.
7720 (vsx_splat_<mode>_mem): Likewise.
7722 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
7724 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
7726 2017-05-22 Jakub Jelinek <jakub@redhat.com>
7729 * omp-low.c (finish_taskreg_remap): New function.
7730 (finish_taskreg_scan): If unit size of ctx->record_type
7731 is non-constant, unshare the size expression and replace
7732 decls in it with possible outer var refs.
7735 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
7736 GOVD_SHARED rather than GOVD_PRIVATE with it.
7737 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
7738 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
7741 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
7742 as last argument to build_outer_var_ref for pointer bases of array
7745 2017-05-19 Martin Sebor <msebor@redhat.com>
7747 * print-tree.c (print_node): Print DECL_READ_P flag.
7749 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7751 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
7752 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
7753 * cgraph.c: Likewise.
7754 * cgraphunit.c: Likewise.
7755 * gengtype.c: Likewise.
7756 * ipa-cp.c: Likewise.
7757 * ipa-devirt.c: Likewise.
7758 * ipa-icf.c: Likewise.
7759 * ipa-predicate.c: Likewise.
7760 * ipa-profile.c: Likewise.
7761 * ipa-prop.c: Likewise.
7762 * ipa-split.c: Likewise.
7764 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
7765 edge_predicate_pool, dump_inline_hints,
7766 inline_summary::account_size_time, redirect_to_unreachable,
7767 edge_set_predicate, set_hint_predicate,
7768 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
7769 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
7770 inline_summary_t::remove, remap_hint_predicate_after_duplication,
7771 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
7772 ipa_call_summary_t::remove, initialize_growth_caches,
7773 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
7774 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
7775 mark_modified, unmodified_parm_1, unmodified_parm,
7776 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
7777 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
7778 compute_bb_predicates, will_be_nonconstant_expr_predicate,
7779 will_be_nonconstant_predicate, record_modified_bb_info,
7780 get_minimal_bb, record_modified, param_change_prob,
7781 phi_result_unknown_predicate, predicate_for_phi_result,
7782 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
7783 estimate_function_body_sizes, compute_inline_parameters,
7784 compute_inline_parameters_for_curren, pass_data_inline_parameters,
7785 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
7786 inline_update_callee_summaries, remap_edge_change_prob,
7787 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
7788 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
7789 inline_analyze_function, inline_summary_t::insert,
7790 inline_generate_summary, read_ipa_call_summary, inline_read_section,
7791 inline_read_summary, write_ipa_call_summary, inline_write_summary,
7792 inline_free_summary): Move to ipa-fnsummary.h
7793 (predicate_t): Remove.
7794 * ipa-fnsummary.c: New file.
7795 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
7796 (enum inline_hints_vals, inline_hints, agg_position_info,
7797 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
7798 inline_summaries, ipa_call_summary, ipa_call_summary_t,
7799 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
7800 dump_inline_summary, dump_inline_hints, inline_generate_summary,
7801 inline_read_summary, inline_write_summary, inline_free_summary,
7802 inline_analyze_function, initialize_inline_failed,
7803 inline_merge_summary, inline_update_overall_summary,
7804 compute_inline_parameters): Move to ipa-fnsummary.h
7805 * ipa-fnsummary.h: New file.
7806 * ipa-inline-transform.h: Include ipa-inline.h.
7807 * ipa-inline.c: LIkewise.
7809 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7811 * ipa-inline.c (edge_badness): Use inlined_time instead of
7812 inline_summaries->get.
7814 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7816 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
7818 2017-05-22 Nathan Sidwell <nathan@acm.org>
7820 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
7821 (fdump-lang): Document 'raw' option.
7822 * dumpfile.h (TDI_tu): Delete.
7823 * dumpfile.c (dump_files): Remove translation-unit.
7824 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
7826 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
7828 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
7829 command option from $(AWK) call.
7830 * config/avr/genmultilib.awk: Simplify and rewrite so that it
7831 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
7832 [FORMAT]: Remove handling of variable.
7833 * config/avr/t-multilib: Regenerate.
7835 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7837 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
7839 (dump_inline_summary): Do not print self_time.
7840 (estimate_function_body_sizes): Do not set self_time.
7841 (compute_inline_parameters): Likewise.
7842 (inline_read_section, inline_write_summary): Do not stream self_time.
7843 * ipa-inline.h (inline_summary): Drop self_time.
7845 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7847 * ipa-inline-analysis.c (account_size_time): Rename to ...
7848 (inline_summary::account_size_time): ... this one.
7849 (reset_ipa_call_summary): Turn to ...
7850 (ipa_call_summary::reset): ... this one.
7851 (reset_inline_summary): Turn to ...
7852 (inline_summary::reset): ... this one.
7853 (inline_summary_t::remove): Update.
7854 (inline_summary_t::duplicate): Update.
7855 (ipa_call_summary_t::remove): Update.
7856 (dump_inline_summary): Update.
7857 (estimate_function_body_sizes): Update.
7858 (compute_inline_parameters): Update.
7859 (estimate_node_size_and_time): Update.
7860 (inline_merge_summary): Update.
7861 (inline_update_overall_summary): Update.
7862 (inline_read_section): Update.
7863 (inline_write_summary): Update.
7864 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
7865 add account_size_time and reset member functions.
7866 (ipa_call_summary): Add reset function.
7867 * ipa-predicate.h (predicate::operator &): Constify.
7869 2017-05-22 Richard Biener <rguenther@suse.de>
7871 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
7873 2017-05-19 Jason Merrill <jason@redhat.com>
7875 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
7877 2017-05-19 Marek Polacek <polacek@redhat.com>
7880 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
7881 TYPE_OVERFLOW_WRAPS checks.
7883 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
7885 * tree-core.h (enum omp_clause_default_kind): Add
7886 "OMP_CLAUSE_DEFAULT_PRESENT".
7887 * tree-pretty-print.c (dump_omp_clause): Handle it.
7888 * gimplify.c (enum gimplify_omp_var_data): Add
7889 "GOVD_MAP_FORCE_PRESENT".
7890 (gimplify_adjust_omp_clauses_1): Map it to
7891 "GOMP_MAP_FORCE_PRESENT".
7892 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
7894 * gimplify.c (oacc_default_clause): Clarify.
7896 2017-05-19 Nathan Sidwell <nathan@acm.org>
7898 LANG_HOOK_REGISTER_DUMPS
7899 * toplev.c (general_init): Call register dump lang hook.
7900 * doc/invoke.texi: Document -fdump-lang option family.
7901 * dumpfile.c (dump_files): Remove class dump here.
7902 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
7903 * dumpfile.h (tree_dump_index): Remove TDI_class.
7904 * langhooks-def.h (lhd_register_dumps): Declare.
7905 (LANG_HOOKS_REGISTER_DUMPS): Define.
7906 (LANG_HOOKS_INITIALIZER): Add it.
7907 * langhooks.c (lhd_register_dumps): Define.
7908 * langhooks.h (struct lang_hooks): Add register_dumps.
7910 2017-05-19 Nathan Sidwell <nathan@acm.org>
7912 * context.h (context::set_passes): New.
7913 * context.c (context::context): Do not create pass manager.
7914 * toplev.c (general_init): Create pass manager here.
7916 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
7918 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
7919 use this splitter if two add or or instructions would also work for
7920 the constant we want to generate.
7922 2017-05-19 Richard Biener <rguenther@suse.de>
7925 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
7926 predicate evaluation.
7928 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
7930 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
7932 * ipa-inline.c (want_inline_small_function_p): Do not cast to
7935 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
7937 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7938 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7939 (inline_edge_summary_vec): Turn into ...
7940 (ipa_call_summaries): ... this one.
7941 (redirect_to_unreachable, edge_set_predicate,
7942 evaluate_properties_for_edge, inline_summary_alloc,
7943 reset_ipa_call_summary, reset_inline_summary,
7944 inline_summary_t::duplicate): Update.
7945 (inline_edge_duplication_hook): Turn to ...
7946 (ipa_call_summary_t::duplicate): ... this one.
7947 (inline_edge_removal_hook): Turn to ...
7948 (ipa_call_summary_t::remove): ... this one.
7949 (dump_inline_edge_summary): Turn to ...
7950 (dump_ipa_call_summary): ... this one.
7951 (estimate_function_body_sizes): Update.
7952 (inline_update_callee_summaries): Update.
7953 (remap_edge_change_prob): Update.
7954 (remap_edge_summaries): Update.
7955 (inline_merge_summary): Update.
7956 (do_estimate_edge_time): Update.
7957 (inline_generate_summary): Update.
7958 (inline_read_section): Update.
7959 (inline_read_summary): Update.
7960 (inline_free_summary): Update.
7961 * ipa-inline.c (can_inline_edge_p): Update.
7962 (compute_inlined_call_time): Update.
7963 (want_inline_small_function_p): Update.
7964 (edge_badness): Update.
7965 (early_inliner): Update.
7966 * ipa-inline.h (inline_edge_summary): Turn to ...
7967 (ipa_call_summary): ... this one.
7968 (ipa_call_summary_t): New class.
7969 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
7970 (ipa_call_summaries): New.
7971 (inline_edge_summary): Remove.
7972 (estimate_edge_growth): Update.
7973 * ipa-profile.c (ipa_propagate_frequency_1): Update.
7974 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7975 * ipa-split.c (execute_split_functions): Update.
7976 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
7978 2017-05-19 Richard Biener <rguenther@suse.de>
7981 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
7983 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
7985 * config/rs6000/rs6000.c (struct machine_function): Add field
7986 fpr_is_wrapped_separately.
7987 (rs6000_get_separate_components): Use 64 components. Handle the
7989 (rs6000_components_for_bb): Handle the FPR components.
7990 (rs6000_emit_prologue_components): Handle the FPR components.
7991 (rs6000_emit_epilogue_components): Handle the FPR components.
7992 (rs6000_set_handled_components): Handle the FPR components.
7993 (rs6000_emit_prologue): Don't output prologue code for those FPRs
7994 that are already separately shrink-wrapped.
7995 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
7996 that are already separately shrink-wrapped.
7998 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
8001 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
8004 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
8005 (define_peephole2 for Altivec d-form load): Add peepholes to catch
8006 cases where the register allocator uses a move and an offsettable
8007 memory operation to/from a FPR register on ISA 2.06/2.07.
8008 (define_peephole2 for Altivec d-form store): Likewise.
8010 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
8013 * config/i386/mmx.md (*mov<mode>_internal): Enable
8014 alternatives 11, 12, 13 and 14 also for 32bit targets.
8015 Remove alternatives 15, 16, 17 and 18.
8016 * config/i386/sse.md (vec_concatv2di): Change
8017 alternative (!x, *y) to (x, ?!*Yn).
8019 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
8021 * dumpfile.h (enum dump_kind): Remove stray comma.
8023 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
8025 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
8026 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
8027 predicate::num_conditions
8028 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
8029 (CHANGED): turn into predicate::changed.
8030 (agg_position_info): Move to ipa-predicate.h
8031 (add_condition, predicate::add_clause, predicate::operator &=,
8032 predicate::or_with, predicate::evaluate, predicate::probability,
8033 dump_condition, dump_clause, predicate::dump,
8034 predicate::remap_after_duplication, predicate::remap_after_inlining,
8035 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
8036 (evaluate_conditions_for_known_args): Update.
8037 (set_cond_stmt_execution_predicate): Update.
8038 * ipa-inline.h: Include ipa-predicate.h
8039 (condition, inline_param_summary, conditions, agg_position_info,
8040 predicate): Move to ipa-predicate.h
8041 * ipa-predicate.c: New file.
8042 * ipa-predicate.h: New file.
8044 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
8046 * final.c (leaf_function_p): Check we are not in a sequence.
8048 2017-05-18 Martin Liska <mliska@suse.cz>
8050 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
8051 * dumpfile.c (dump_register): Use new enum dump_kind.
8052 (get_dump_file_name): Likewise.
8053 (dump_enable_all): Likewise.
8054 (dump_switch_p_1): Likewise.
8055 (enable_rtl_dump_file): Remove usage of TDF_RTL.
8056 * dumpfile.h (enum dump_kind): New enum type.
8057 (struct dump_file_info): Create constructor and
8058 format fields and comments.
8059 * passes.c (pass_manager::register_one_dump_file):
8061 * statistics.c (statistics_early_init): Likewise.
8062 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
8063 TDF_TREE with TDF_SLIM.
8064 (gather_memory_references_ref): Likewise.
8066 2017-05-18 Martin Liska <mliska@suse.cz>
8068 * vec.h (struct vnull): Use it.
8070 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
8072 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
8073 (true_predicate, false_predicate, true_predicate_p,
8074 false_predicate_p): Remove.
8075 (single_cond_predicate, not_inlined_predicate): Turn to member function
8077 (add_condition): Update.
8078 (add_clause): Turn to...
8079 (predicate::add_clause): ... this one; update; allow passing NULL
8081 (and_predicates): Turn to ...
8082 (predicate::operator &=): ... this one.
8083 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
8084 (or_predicates): Turn to ...
8085 (predicate::or_with): ... this one.
8086 (evaluate_predicate): Turn to ...
8087 (predicate::evaluate): ... this one.
8088 (predicate_probability): Turn to ...
8089 (predicate::probability): ... this one.
8090 (dump_condition): Update.
8091 (dump_predicate): Turn to ...
8092 (predicate::dump): ... this one.
8093 (account_size_time): Update.
8094 (edge_set_predicate): Update.
8095 (set_hint_predicate): UPdate.
8096 (evaluate_conditions_for_known_args): Update.
8097 (evaluate_properties_for_edge): Update.
8098 (remap_predicate_after_duplication): Turn to...
8099 (predicate::remap_after_duplication): ... this one.
8100 (remap_hint_predicate_after_duplication): Update.
8101 (inline_summary_t::duplicate): UPdate.
8102 (dump_inline_edge_summary): Update.
8103 (dump_inline_summary): Update.
8104 (set_cond_stmt_execution_predicate): Update.
8105 (set_switch_stmt_execution_predicate): Update.
8106 (compute_bb_predicates): Update.
8107 (will_be_nonconstant_expr_predicate): Update.
8108 (will_be_nonconstant_predicate): Update.
8109 (phi_result_unknown_predicate): Update.
8110 (predicate_for_phi_result): Update.
8111 (array_index_predicate): Update.
8112 (estimate_function_body_sizes): Update.
8113 (estimate_node_size_and_time): Update.
8114 (estimate_ipcp_clone_size_and_time): Update.
8115 (remap_predicate): Rename to ...
8116 (predicate::remap_after_inlining): ... this one.
8117 (remap_hint_predicate): Update.
8118 (inline_merge_summary): Update.
8119 (inline_update_overall_summary): Update.
8120 (estimate_size_after_inlining): Update.
8121 (read_predicate): Rename to ...
8122 (predicate::stream_in): ... this one.
8123 (read_inline_edge_summary): Update.
8124 (write_predicate): Rename to ...
8125 (predicate::stream_out): ... this one.
8126 (write_inline_edge_summary): Update.
8127 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
8128 (clause_t): Turn to uint32_t
8129 (predicate): Turn to class; implement constructor and operators
8131 (size_time_entry): Update.
8132 (inline_summary): Update.
8133 (inline_edge_summary): Update.
8135 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
8137 * fold-const.c (fold_binary_loc): Move transformation...
8138 * match.pd (C - X CMP X): ... here.
8140 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
8142 * config/sparc/sparc.c (sparc_option_override): Set function
8143 alignment for -mcpu=niagara7 to 64 to match the I$ line.
8144 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
8146 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
8148 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
8150 2017-05-18 Marek Polacek <polacek@redhat.com>
8153 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
8154 (pass_ubsan::execute): Call gimple_assign_single_p instead of
8155 gimple_assign_load_p.
8157 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
8160 * real.c (do_compare): Give decimal_do_compare preference over
8161 comparing just the signs.
8163 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
8165 * doc/md.texi (Canonicalization of Instructions): Describe the
8166 canonical form of instructions that inherently set a condition
8169 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
8172 * tree-cfg.c: Move deletion of unreachable case statements to after
8173 the merging of consecutive case labels.
8175 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
8177 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
8178 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
8179 restoring of callee-saved registers.
8181 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
8183 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
8184 * config/visium/visium.c (single_set_and_flags): Likewise.
8185 * config/visium/visium.md (Substitutions): Likewise.
8187 2017-05-17 Martin Liska <mliska@suse.cz>
8189 * cfg.c: Introduce dump_flags_t type and
8190 use it instead of int type.
8192 * cfghooks.c: Likewise.
8193 * cfghooks.h (struct cfg_hooks): Likewise.
8194 * cfgrtl.c: Likewise.
8195 * cfgrtl.h: Likewise.
8196 * cgraph.c (cgraph_node::get_body): Likewise.
8197 * coretypes.h: Likewise.
8198 * domwalk.c: Likewise.
8199 * domwalk.h: Likewise.
8200 * dumpfile.c (struct dump_option_value_info): Likewise.
8201 (dump_enable_all): Likewise.
8202 (dump_switch_p_1): Likewise.
8203 (opt_info_switch_p): Likewise.
8204 * dumpfile.h (enum tree_dump_index): Likewise.
8205 (struct dump_file_info): Likewise.
8206 * genemit.c: Likewise.
8207 * generic-match-head.c: Likewise.
8208 * gengtype.c (open_base_files): Likewise.
8209 * gimple-pretty-print.c: Likewise.
8210 * gimple-pretty-print.h: Likewise.
8211 * graph.c (print_graph_cfg): Likewise.
8212 * graphite-scop-detection.c (dot_all_sese): Likewise.
8213 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
8214 * loop-unroll.c (report_unroll): Likewise.
8215 * passes.c (pass_manager::register_one_dump_file): Likewise.
8216 * print-tree.c: Likewise.
8217 * statistics.c: Likewise.
8218 * tree-cfg.c: Likewise.
8219 * tree-cfg.h: Likewise.
8220 * tree-dfa.c: Likewise.
8221 * tree-dfa.h: Likewise.
8222 * tree-dump.c (dump_function): Likewise.
8223 * tree-dump.h (struct dump_info): Likewise.
8224 * tree-pretty-print.c: Likewise.
8225 * tree-pretty-print.h: Likewise.
8226 * tree-ssa-live.c: Likewise.
8227 * tree-ssa-live.h: Likewise.
8228 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
8229 * tree-vect-loop.c: Likewise.
8230 * tree-vect-slp.c: Likewise.
8232 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
8233 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8235 PR tree-optimization/80457
8236 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
8237 of all arguments to a statement as scalar_to_vec operations.
8238 (vectorizable_call): Adjust call to vect_model_simple_cost for
8240 (vectorizable_conversion): Likewise.
8241 (vectorizable_assignment): Likewise.
8242 (vectorizable_shift): Likewise.
8243 (vectorizable_operation): Likewise.
8244 (vectorizable_comparison): Likewise.
8245 (vect_is_simple_cond): Record the def types for operands.
8246 (vectorizable_condition): Likewise, call vect_model_simple_cost.
8247 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
8248 for statement argument count.
8250 2017-05-16 Carl Love <cel@us.ibm.com>
8252 * config/rs6000/rs6000-c: Add support for built-in functions
8253 vector unsigned long long vec_bperm (vector unsigned long long,
8254 vector unsigned char)
8255 vector signed long long vec_mule (vector signed int,
8257 vector unsigned long long vec_mule (vector unsigned int,
8258 vector unsigned int)
8259 vector signed long long vec_mulo (vector signed int,
8261 vector unsigned long long vec_mulo (vector unsigned int,
8262 vector unsigned int)
8263 vector signed char vec_sldw (vector signed char,
8266 vector unsigned char vec_sldw (vector unsigned char,
8267 vector unsigned char,
8269 vector signed short vec_sldw (vector signed short,
8270 vector signed short,
8272 vector unsigned short vec_sldw (vector unsigned short,
8273 vector unsigned short,
8275 vector signed int vec_sldw (vector signed int,
8278 vector unsigned int vec_sldw (vector unsigned int,
8279 vector unsigned int,
8281 vector signed long long vec_sldw (vector signed long long,
8282 vector signed long long,
8284 vector unsigned long long vec_sldw (vector unsigned long long,
8285 vector unsigned long long,
8287 * config/rs6000/rs6000-c: Add support for built-in functions
8288 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
8289 * config/rs6000/altivec.h: Add defintion for vec_sldw.
8290 * doc/extend.texi: Update the built-in documentation for the
8291 new built-in functions.
8293 2017-05-16 Marek Polacek <polacek@redhat.com>
8297 * tree.c (save_expr): Don't fold the expression.
8299 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
8301 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
8302 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
8303 and (?*y,m). Update insn attributes.
8305 2017-05-16 Martin Liska <mliska@suse.cz>
8307 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
8308 flags argument of print_gimple_stmt, print_gimple_expr,
8309 print_generic_stmt and print_generic_expr.
8310 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
8311 * coretypes.h: Likewise.
8312 * except.c (dump_eh_tree): Likewise.
8313 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8314 * gimple-pretty-print.h: Likewise.
8315 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
8316 (backprop::push_to_worklist): Likewise.
8317 (backprop::pop_from_worklist): Likewise.
8318 (backprop::process_use): Likewise.
8319 (backprop::intersect_uses): Likewise.
8320 (note_replacement): Likewise.
8321 * gimple-ssa-store-merging.c
8322 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
8323 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
8324 (pass_store_merging::execute): Likewise.
8325 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
8326 (ssa_base_cand_dump_callback): Likewise.
8327 (dump_incr_vec): Likewise.
8328 (replace_refs): Likewise.
8329 (replace_mult_candidate): Likewise.
8330 (create_add_on_incoming_edge): Likewise.
8331 (create_phi_basis): Likewise.
8332 (insert_initializers): Likewise.
8333 (all_phi_incrs_profitable): Likewise.
8334 (introduce_cast_before_cand): Likewise.
8335 (replace_one_candidate): Likewise.
8336 * gimplify.c (gimplify_expr): Likewise.
8337 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
8338 (set_rename): Likewise.
8339 (rename_uses): Likewise.
8340 (copy_loop_phi_nodes): Likewise.
8341 (add_close_phis_to_merge_points): Likewise.
8342 (copy_loop_close_phi_args): Likewise.
8343 (copy_cond_phi_args): Likewise.
8344 (graphite_copy_stmts_from_block): Likewise.
8345 (translate_pending_phi_nodes): Likewise.
8346 * graphite-poly.c (print_pdr): Likewise.
8347 (dump_gbb_cases): Likewise.
8348 (dump_gbb_conditions): Likewise.
8349 (print_scop_params): Likewise.
8350 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
8351 (build_cross_bb_scalars_use): Likewise.
8352 (gather_bbs::before_dom_children): Likewise.
8353 * hsa-dump.c (dump_hsa_immed): Likewise.
8354 * ipa-cp.c (print_ipcp_constant_value): Likewise.
8355 (get_replacement_map): Likewise.
8356 * ipa-inline-analysis.c (dump_condition): Likewise.
8357 (estimate_function_body_sizes): Likewise.
8358 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
8359 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
8360 * ipa-prop.c (ipa_dump_param): Likewise.
8361 (ipa_print_node_jump_functions_for_edge): Likewise.
8362 (ipa_modify_call_arguments): Likewise.
8363 (ipa_modify_expr): Likewise.
8364 (ipa_dump_param_adjustments): Likewise.
8365 (ipa_dump_agg_replacement_values): Likewise.
8366 (ipcp_modif_dom_walker::before_dom_children): Likewise.
8367 * ipa-pure-const.c (check_stmt): Likewise.
8368 (pass_nothrow::execute): Likewise.
8369 * ipa-split.c (execute_split_functions): Likewise.
8370 * omp-offload.c (dump_oacc_loop_part): Likewise.
8371 (dump_oacc_loop): Likewise.
8372 * trans-mem.c (tm_log_emit): Likewise.
8373 (tm_memopt_accumulate_memops): Likewise.
8374 (dump_tm_memopt_set): Likewise.
8375 (dump_tm_memopt_transform): Likewise.
8376 * tree-cfg.c (gimple_verify_flow_info): Likewise.
8377 (print_loop): Likewise.
8378 * tree-chkp-opt.c (chkp_print_addr): Likewise.
8379 (chkp_gather_checks_info): Likewise.
8380 (chkp_get_check_result): Likewise.
8381 (chkp_remove_check_if_pass): Likewise.
8382 (chkp_use_outer_bounds_if_possible): Likewise.
8383 (chkp_reduce_bounds_lifetime): Likewise.
8384 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
8385 (chkp_mark_completed_bounds): Likewise.
8386 (chkp_register_incomplete_bounds): Likewise.
8387 (chkp_mark_invalid_bounds): Likewise.
8388 (chkp_maybe_copy_and_register_bounds): Likewise.
8389 (chkp_build_returned_bound): Likewise.
8390 (chkp_get_bound_for_parm): Likewise.
8391 (chkp_build_bndldx): Likewise.
8392 (chkp_get_bounds_by_definition): Likewise.
8393 (chkp_generate_extern_var_bounds): Likewise.
8394 (chkp_get_bounds_for_decl_addr): Likewise.
8395 * tree-chrec.c (chrec_apply): Likewise.
8396 * tree-data-ref.c (dump_data_reference): Likewise.
8397 (dump_subscript): Likewise.
8398 (dump_data_dependence_relation): Likewise.
8399 (analyze_overlapping_iterations): Likewise.
8400 * tree-inline.c (expand_call_inline): Likewise.
8401 (tree_function_versioning): Likewise.
8402 * tree-into-ssa.c (dump_defs_stack): Likewise.
8403 (dump_currdefs): Likewise.
8404 (dump_names_replaced_by): Likewise.
8405 (dump_update_ssa): Likewise.
8406 (update_ssa): Likewise.
8407 * tree-object-size.c (pass_object_sizes::execute): Likewise.
8408 * tree-parloops.c (build_new_reduction): Likewise.
8409 (try_create_reduction_list): Likewise.
8410 (ref_conflicts_with_region): Likewise.
8411 (oacc_entry_exit_ok_1): Likewise.
8412 (oacc_entry_exit_single_gang): Likewise.
8413 * tree-pretty-print.h: Likewise.
8414 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
8415 (get_scalar_evolution): Likewise.
8416 (add_to_evolution): Likewise.
8417 (get_loop_exit_condition): Likewise.
8418 (analyze_evolution_in_loop): Likewise.
8419 (analyze_initial_condition): Likewise.
8420 (analyze_scalar_evolution): Likewise.
8421 (instantiate_scev): Likewise.
8422 (number_of_latch_executions): Likewise.
8423 (gather_chrec_stats): Likewise.
8424 (final_value_replacement_loop): Likewise.
8425 (scev_const_prop): Likewise.
8426 * tree-sra.c (dump_access): Likewise.
8427 (disqualify_candidate): Likewise.
8428 (create_access): Likewise.
8430 (maybe_add_sra_candidate): Likewise.
8431 (create_access_replacement): Likewise.
8432 (analyze_access_subtree): Likewise.
8433 (analyze_all_variable_accesses): Likewise.
8434 (sra_modify_assign): Likewise.
8435 (initialize_constant_pool_replacements): Likewise.
8436 (find_param_candidates): Likewise.
8437 (decide_one_param_reduction): Likewise.
8438 (replace_removed_params_ssa_names): Likewise.
8439 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
8440 * tree-ssa-copy.c (dump_copy_of): Likewise.
8441 (copy_prop_visit_cond_stmt): Likewise.
8442 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
8443 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
8444 (record_equivalences_from_stmt): Likewise.
8445 * tree-ssa-dse.c (compute_trims): Likewise.
8446 (delete_dead_call): Likewise.
8447 (delete_dead_assignment): Likewise.
8448 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
8449 (forward_propagate_into_cond): Likewise.
8450 (pass_forwprop::execute): Likewise.
8451 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8452 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
8454 (move_computations_worker): Likewise.
8455 (execute_sm): Likewise.
8456 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
8457 (remove_exits_and_undefined_stmts): Likewise.
8458 (remove_redundant_iv_tests): Likewise.
8459 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
8460 (adjust_iv_update_pos): Likewise.
8461 * tree-ssa-math-opts.c (bswap_replace): Likewise.
8462 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
8463 (value_replacement): Likewise.
8464 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
8465 * tree-ssa-pre.c (print_pre_expr): Likewise.
8466 (get_representative_for): Likewise.
8467 (create_expression_by_pieces): Likewise.
8468 (insert_into_preds_of_block): Likewise.
8469 (eliminate_insert): Likewise.
8470 (eliminate_dom_walker::before_dom_children): Likewise.
8471 (eliminate): Likewise.
8472 (remove_dead_inserted_code): Likewise.
8473 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
8474 * tree-ssa-reassoc.c (get_rank): Likewise.
8475 (eliminate_duplicate_pair): Likewise.
8476 (eliminate_plus_minus_pair): Likewise.
8477 (eliminate_not_pairs): Likewise.
8478 (undistribute_ops_list): Likewise.
8479 (eliminate_redundant_comparison): Likewise.
8480 (update_range_test): Likewise.
8481 (optimize_range_tests_var_bound): Likewise.
8482 (optimize_vec_cond_expr): Likewise.
8483 (rewrite_expr_tree): Likewise.
8484 (rewrite_expr_tree_parallel): Likewise.
8485 (linearize_expr): Likewise.
8486 (break_up_subtract): Likewise.
8487 (linearize_expr_tree): Likewise.
8488 (attempt_builtin_powi): Likewise.
8489 (attempt_builtin_copysign): Likewise.
8490 (transform_stmt_to_copy): Likewise.
8491 (transform_stmt_to_multiply): Likewise.
8492 (dump_ops_vector): Likewise.
8493 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
8494 (print_scc): Likewise.
8495 (set_ssa_val_to): Likewise.
8496 (visit_reference_op_store): Likewise.
8497 (visit_use): Likewise.
8498 (sccvn_dom_walker::before_dom_children): Likewise.
8499 (run_scc_vn): Likewise.
8500 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
8502 (expr_hash_elt::print): Likewise.
8503 (const_and_copies::pop_to_marker): Likewise.
8504 (const_and_copies::record_const_or_copy_raw): Likewise.
8505 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
8506 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
8507 (dump_predicates): Likewise.
8508 (find_uninit_use): Likewise.
8509 (warn_uninitialized_phi): Likewise.
8510 (pass_late_warn_uninitialized::execute): Likewise.
8511 * tree-ssa.c (verify_vssa): Likewise.
8512 (verify_ssa): Likewise.
8513 (maybe_optimize_var): Likewise.
8514 * tree-vrp.c (dump_value_range): Likewise.
8515 (dump_all_value_ranges): Likewise.
8516 (dump_asserts_for): Likewise.
8517 (register_edge_assert_for_2): Likewise.
8518 (vrp_visit_cond_stmt): Likewise.
8519 (vrp_visit_switch_stmt): Likewise.
8520 (vrp_visit_stmt): Likewise.
8521 (vrp_visit_phi_node): Likewise.
8522 (simplify_cond_using_ranges_1): Likewise.
8523 (fold_predicate_in): Likewise.
8524 (evrp_dom_walker::before_dom_children): Likewise.
8525 (evrp_dom_walker::push_value_range): Likewise.
8526 (evrp_dom_walker::pop_value_range): Likewise.
8527 (execute_early_vrp): Likewise.
8529 2017-05-16 Richard Biener <rguenther@suse.de>
8531 * dwarf2out.c (loc_list_from_tree_1): Do not create
8532 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
8534 2017-05-16 Richard Biener <rguenther@suse.de>
8536 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
8538 (note_variable_value_in_expr): If we resolved the decl ref
8539 do not push to the stack.
8541 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
8543 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
8544 operations in fast-math mode.
8545 (vaddq_f16): Likewise.
8546 (vmul_f16): Likewise.
8547 (vmulq_f16): Likewise.
8548 (vsub_f16): Likewise.
8549 (vsubq_f16): Likewise.
8550 * config/arm/neon.md (add<mode>3): New.
8552 (fma:<VH:mode>3): New. Also remove outdated comment.
8555 2017-05-16 Martin Liska <mliska@suse.cz>
8559 * ipa-devirt.c (warn_types_mismatch): Fix typo.
8560 (odr_types_equivalent_p): Likewise.
8562 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
8564 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
8566 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
8569 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
8570 non-interunit SSE move alternatives with '?'.
8571 (zero-extendsidi peephole2): New peephole to skip intermediate
8572 general register in SSE zero-extend sequence.
8574 2017-05-15 Jeff Law <law@redhat.com>
8576 * reorg.c (relax_delay_slots): Create a new variable to hold
8577 the temporary target rather than clobbering TARGET_LABEL.
8579 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
8580 missing argument to extract_bit_field call.
8581 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
8583 2017-05-15 Martin Liska <mliska@suse.cz>
8586 * gcc.c (process_command): Do not allow empty argument of -o option.
8588 2017-05-15 Renlin Li <renlin.li@arm.com>
8590 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
8591 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
8592 * config/aarch64/constraints.md (Usf): Add long call check.
8593 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
8594 (call_value): Likewise.
8595 (sibcall): Likewise.
8596 (sibcall_value): Likewise.
8598 (call_value_insn): New.
8599 (sibcall_insn): Update rtx pattern.
8600 (sibcall_value_insn): Likewise.
8601 (call_internal): Remove.
8602 (call_value_internal): Likewise.
8603 (sibcall_internal): Likewise.
8604 (sibcall_value_internal): Likewise.
8605 (call_reg): Likewise.
8606 (call_symbol): Likewise.
8607 (call_value_reg): Likewise.
8608 (call_value_symbol): Likewise.
8610 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
8613 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
8615 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
8617 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
8618 compatible with CCGOCmode and with CCZmode.
8620 2017-05-14 Martin Sebor <msebor@redhat.com>
8623 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
8624 (gimple_fold_builtin_snprintf): Same.
8625 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
8626 (gimple_fold_builtin_snprintf): Same.
8627 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
8629 (is_call_safe): New function.
8630 (try_substitute_return_value): Call it.
8631 (try_simplify_call): New function.
8632 (pass_sprintf_length::handle_gimple_call): Call it.
8634 2017-05-14 Martin Sebor <msebor@redhat.com>
8637 * builtins.c (expand_builtin_stpncpy): Simplify.
8639 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
8641 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
8642 * config/i386/i386.h
8643 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
8644 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
8645 (struct machine_function): Add new members call_ms2sysv,
8646 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
8647 (struct machine_frame_state): New fields sp_realigned and
8648 sp_realigned_offset.
8649 * config/i386/i386.c
8650 (enum xlogue_stub): New enum.
8651 (enum xlogue_stub_sets): New enum.
8652 (class xlogue_layout): New class.
8653 (struct ix86_frame): New fields stack_realign_allocate_offset,
8654 stack_realign_offset and outlined_save_offset. Modify comments to
8655 detail stack layout when using out-of-line stubs.
8656 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
8657 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
8658 -mcall-ms2sysv-xlogues.
8659 (stub_managed_regs): New static variable.
8660 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
8661 registers managed by out-of-line stub.
8662 (disable_call_ms2sysv_xlogues): New function.
8663 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
8664 m->call_ms2sysv when appropriate and compute frame layout for
8666 (sp_valid_at, fp_valid_at): New inline functions.
8667 (choose_basereg): New function.
8668 (choose_baseaddr): Add align parameter, use choose_basereg and modify
8670 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
8671 Use align parameter of choose_baseaddr to generated aligned SSE movs
8673 (pro_epilogue_adjust_stack): Modify to track
8674 machine_frame_state::sp_realigned.
8675 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
8676 (ix86_nsaved_sseregs): Likewise.
8677 (ix86_emit_save_regs): Likewise.
8678 (ix86_emit_save_regs_using_mov): Likewise.
8679 (ix86_emit_save_sse_regs_using_mov): Likewise.
8680 (get_scratch_register_on_entry): Likewise.
8681 (gen_frame_set): New function.
8682 (gen_frame_load): Likewise.
8683 (gen_frame_store): Likewise.
8684 (emit_outlined_ms2sysv_save): Likewise.
8685 (emit_outlined_ms2sysv_restore): Likewise.
8686 (ix86_expand_prologue): Modify stack re-alignment code and call
8687 emit_outlined_ms2sysv_save when appropriate.
8688 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
8689 parameter rtx_insn *insn, which allows the function to be used to only
8691 (ix86_expand_epilogue): Modify validity checks of frame and stack
8692 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
8693 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
8694 * config/i386/predicates.md
8695 (save_multiple): New predicate.
8696 (restore_multiple): Likewise.
8697 * config/i386/sse.md
8698 (save_multiple<mode>): New pattern.
8699 (save_multiple_realign<mode>): Likewise.
8700 (restore_multiple<mode>): Likewise.
8701 (restore_multiple_and_return<mode>): Likewise.
8702 (restore_multiple_leave_return<mode>): Likewise.
8703 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
8705 2017-05-14 Julia Koval <julia.koval@intel.com>
8707 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
8708 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
8709 (__builtin_ia32_xsetbv): New builtins.
8710 * config/i386/i386.c (ix86_expand_special_args_builtin):
8712 (ix86_expand_builtin): Special expand for new intrinsics.
8713 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
8714 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
8715 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
8717 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8719 * cfganal.c (inverted_post_order_compute): Change argument type
8721 * cfganal.h (inverted_post_order_compute): Adjust prototype.
8722 * df-core.c (rest_of_handle_df_initialize): Adjust.
8723 (rest_of_handle_df_finish): Likewise.
8724 (df_analyze_1): Likewise.
8725 (df_analyze): Likewise.
8726 (loop_inverted_post_order_compute): Change argument to be a vec *.
8727 (df_analyze_loop): Adjust.
8728 (df_get_n_blocks): Likewise.
8729 (df_get_postorder): Likewise.
8730 * df.h (struct df_d): Change field to be a vec.
8731 * lcm.c (compute_laterin): Adjust.
8732 (compute_available): Likewise.
8733 * lra-lives.c (lra_create_live_ranges_1): Likewise.
8734 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
8735 * tree-ssa-pre.c (compute_antic): Likewise.
8737 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8739 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
8740 (depth_first_search::depth_first_search): Change structure init
8741 function to this constructor.
8742 (depth_first_search::add_bb): Rename function to this member.
8743 (depth_first_search::execute): Likewise.
8744 (flow_dfs_compute_reverse_finish): Adjust.
8746 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8748 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
8749 (longest_simple_path): Likewise.
8750 * shrink-wrap.c (spread_components): Likewise.
8751 (disqualify_problematic_components): Likewise.
8752 (emit_common_heads_for_components): Likewise.
8753 (emit_common_tails_for_components): Likewise.
8754 (insert_prologue_epilogue_for_components): Likewise.
8756 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8758 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
8761 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8763 * df-core.c (df_set_blocks): Start using auto_bitmap.
8764 (df_compact_blocks): Likewise.
8765 * df-problems.c (df_rd_confluence_n): Likewise.
8766 * df-scan.c (df_insn_rescan_all): Likewise.
8767 (df_process_deferred_rescans): Likewise.
8768 (df_update_entry_block_defs): Likewise.
8769 (df_update_exit_block_uses): Likewise.
8770 (df_entry_block_bitmap_verify): Likewise.
8771 (df_exit_block_bitmap_verify): Likewise.
8772 (df_scan_verify): Likewise.
8773 * lra-constraints.c (lra_constraints): Likewise.
8774 (undo_optional_reloads): Likewise.
8775 (lra_undo_inheritance): Likewise.
8776 * lra-remat.c (calculate_gen_cands): Likewise.
8777 (do_remat): Likewise.
8778 * lra-spills.c (assign_spill_hard_regs): Likewise.
8779 (spill_pseudos): Likewise.
8780 * tree-ssa-pre.c (bitmap_set_and): Likewise.
8781 (bitmap_set_subtract_values): Likewise.
8783 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8785 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
8786 management with auto_bitmap.
8787 (fix_inter_tick): Likewise.
8788 (fix_recovery_deps): Likewise.
8789 * ira.c (add_store_equivs): Likewise.
8790 (find_moveable_pseudos): Likewise.
8791 (split_live_ranges_for_shrink_wrap): Likewise.
8792 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
8793 (rtx_reuse_manager::seen_def_p): Likewise.
8794 (rtx_reuse_manager::set_seen_def): Likewise.
8795 * print-rtl.h (class rtx_reuse_manager): Likewise.
8797 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8799 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
8801 (migrate_btr_def): Likewise.
8802 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
8803 * df-core.c (loop_post_order_compute): Likewise.
8804 (loop_inverted_post_order_compute): Likewise.
8805 * hsa-common.h: Likewise.
8806 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
8807 * init-regs.c (initialize_uninitialized_regs): Likewise.
8808 * ipa-inline.c (resolve_noninline_speculation): Likewise.
8809 (inline_small_functions): Likewise.
8810 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
8811 * ira.c (combine_and_move_insns): Likewise.
8812 (build_insn_chain): Likewise.
8813 * loop-invariant.c (find_invariants): Likewise.
8814 * lower-subreg.c (propagate_pseudo_copies): Likewise.
8815 * predict.c (tree_predict_by_opcode): Likewise.
8816 (predict_paths_leading_to): Likewise.
8817 (predict_paths_leading_to_edge): Likewise.
8818 (estimate_loops_at_level): Likewise.
8819 (estimate_loops): Likewise.
8820 * shrink-wrap.c (try_shrink_wrapping): Likewise.
8821 (spread_components): Likewise.
8822 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
8823 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
8824 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
8825 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
8826 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
8827 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
8828 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
8829 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
8830 (mark_threaded_blocks): Likewise.
8831 (thread_through_all_blocks): Likewise.
8832 * tree-ssa.c (verify_ssa): Likewise.
8833 (execute_update_addresses_taken): Likewise.
8834 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
8836 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8838 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
8840 (post_order_compute): Likewise.
8841 (inverted_post_order_compute): Likewise.
8842 (pre_and_rev_post_order_compute_fn): Likewise.
8844 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8846 * genrecog.c (int_set::int_set): Explicitly construct our
8847 auto_vec base class.
8848 * vec.h (auto_vec::auto_vec): New constructor.
8850 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8852 * bitmap.h (class auto_bitmap): New constructor taking
8853 bitmap_obstack * argument.
8855 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8857 * bitmap.h (class auto_bitmap): Change type of m_bits to
8858 bitmap_head, and adjust ctor / dtor and member operators.
8860 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
8862 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
8863 when returned register mode doesn't match original mode.
8865 2017-05-12 Jeff Law <law@redhat.com>
8866 Jakub Jelinek <jakub@redhat.com>
8868 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
8869 we look for cc setter after the compare-elim changes.
8870 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
8871 within the vector to match what compare-elim now expects.
8872 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
8873 (xorsi3_flags, one_cmplsi2_flags): Likewise.
8875 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
8876 after the compare-elim changes.
8877 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
8878 the vector to match what compare-elim now expects.
8879 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
8880 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
8881 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
8882 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
8883 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
8885 * config/visium/visium.c (single_set_and_flags): Fix where
8886 we look for cc setter after the compare-elim changes.
8887 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
8888 with the vector to match what compare-elim now expects.
8889 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
8890 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
8891 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
8892 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
8893 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
8894 (neg<mode>2_insn_set_overflow): Likewise.
8896 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
8899 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
8900 maybe_expand_insn call, set ops[0].target. If still set after call,
8901 set alt_rtl. Add extra arg to recursive calls.
8902 (extract_bit_field): Add alt_rtl argument. Pass to
8904 * expmed.h (extract_bit_field): Fix prototype.
8905 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
8906 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
8907 to extract_bit_field_calls.
8908 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
8909 Pass alt_rtl to extract_bit_field calls.
8910 * calls.c (store_unaligned_arguments_into_psuedos)
8911 load_register_parameters): Pass extra NULL to extract_bit_field calls.
8912 * optabs.c (maybe_legitimize_operand): Clear op->target when call
8914 * optabs.h (struct expand_operand): Add target bitfield.
8916 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
8918 * compare-elim.c (try_eliminate_compare): Canonicalize
8919 operation with embedded compare to
8920 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
8921 (set (reg) (operation)].
8923 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
8925 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
8928 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
8929 cost of adding a carry flag for ADC instruction.
8930 [case MINUS]: Ignore the cost of subtracting a carry flag
8931 for SBB instruction.
8933 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
8935 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
8937 * config/rs6000/bmiintrin.h: New file.
8938 * config/rs6000/bmi2intrin.h: New file.
8939 * config/rs6000/x86intrin.h: New file.
8941 2017-05-12 Jeff Law <law@redhat.com>
8943 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
8946 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
8949 * tree-cfg.c: Remove cfg edges of unreachable case statements.
8951 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8953 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
8954 early expansion of vector divide builtins.
8955 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
8956 builtins identified as having unsigned arguments.
8958 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8960 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
8961 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
8962 expansion of vector logical operations (and, andc, or, xor,
8965 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8967 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
8968 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
8970 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8972 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
8973 early GIMPLE expansion of vector multiplies.
8975 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
8977 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
8978 TARGET_HAVE_MOVT conditional.
8979 (movt splitter): Likewise.
8981 2017-05-12 Richard Biener <rguenther@suse.de>
8983 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
8984 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8985 Fold all stmts not inplace.
8987 2017-05-12 Richard Biener <rguenther@suse.de>
8989 PR tree-optimization/80713
8990 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
8991 inserted_exprs bit for not removed stmts.
8993 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
8996 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
8997 parallelized" attribute for parallelized OpenACC kernels.
8998 * omp-offload.c (execute_oacc_device_lower): Use it.
9000 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
9001 Set "oacc kernels" attribute.
9002 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
9003 parameter. Adjust all users.
9004 (oacc_fn_attrib_kernels_p): Remove function.
9005 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
9006 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
9007 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
9008 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
9009 assert "oacc kernels" attribute is set.
9011 2017-05-11 Carl Love <cel@us.ibm.com>
9013 * config/rs6000/rs6000-c: Add support for built-in functions
9014 vector unsigned char vec_popcnt (vector signed char)
9015 vector unsigned char vec_popcnt (vector unsigned char)
9016 vector unsigned short vec_popcnt (vector signed short)
9017 vector unsigned short vec_popcnt (vector unsigned short)
9018 vector unsigned int vec_popcnt (vector signed int)
9019 vector unsigned int vec_popcnt (vector unsigned int)
9020 vector unsigned long long vec_popcnt (vector signed long long)
9021 vector unsigned long long vec_popcnt (vector unsigned long long)
9022 vector signed long long vec_slo (vector signed long long,
9024 vector signed long long vec_slo (vector signed long long,
9025 vector unsigned char)
9026 vector unsigned long long vec_slo (vector unsigned long long,
9028 vector unsigned long long vec_slo (vector unsigned long long,
9029 vector unsigned char)
9030 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
9031 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
9032 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
9033 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
9034 * doc/extend.texi: Update the built-in documentation file for the
9035 new built-in functions.
9037 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
9039 * attribs.h (sorted_attr_string): Move machine independent
9040 functions for target clone support from the i386 port to common
9041 code. Rename ix86_function_versions to common_function_versions.
9042 Rename make_name to make_unique_name.
9043 (common_function_versions): Likewise.
9044 (make_unique_name): Likewise.
9045 (make_dispatcher_decl): Likewise.
9046 (is_function_default_version): Likewise.
9047 * attribs.c (attr_strcmp): Likewise.
9048 (sorted_attr_string): Likewise.
9049 (common_function_versions): Likewise.
9050 (make_unique_name): Likewise.
9051 (make_dispatcher_decl): Likewise.
9052 (is_function_default_version): Likewise.
9053 * config/i386/i386.c (attr_strcmp): Likewise.
9054 (sorted_attr_string): Likewise.
9055 (ix86_function_versions): Likewise.
9056 (make_name): Likewise.
9057 (make_dispatcher_decl): Likewise.
9058 (is_function_default_version): Likewise.
9059 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
9061 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9064 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
9065 Account for direct move costs for vec_construct of integer
9068 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
9071 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
9072 (UNSPEC_STX_ATOMIC): Ditto.
9073 (loaddi_via_sse): New insn.
9074 (storedi_via_sse): Ditto.
9075 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
9076 Update corresponding peephole2 patterns.
9077 (atomic_storedi_fpu): Ditto.
9079 2017-05-11 Julia Koval <julia.koval@intel.com>
9081 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
9082 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
9084 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
9085 (__builtin_ia32_rsqrt14ss_mask): New builtins.
9086 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
9088 2017-05-11 Nathan Sidwell <nathan@acm.org>
9090 * graphite-poly.c: Include dumpfile.h.
9092 * dumpfle.h (dump_function): Declare here ...
9093 * tree-dump.h (dump_function): ... not here.
9094 * dumpfile.c: #include tree-cfg.h.
9095 (dump_function): Move here from ...
9096 * tree-dump.c (dump_function): ... here.
9097 * gimplify.c: #include splay-tree.h, not tree-dump.h.
9098 * graphite-poly.c: Don't include tree-dump.h.
9099 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
9100 * print-tree.c: Likewise.
9101 * stor-layout.c: Likewise.
9102 * tree-nested.c: Likewise.
9104 * dumpfile.c (dump_start): Use TDF_FLAGS.
9105 (dump_enable_all): Fix TDF_KIND check thinko.
9107 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
9109 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9110 array entries to represent two legal parameterizations of the
9111 overloaded __builtin_cmpb function, as represented by the
9112 P6_OV_BUILTIN_CMPB constant.
9113 (altivec_resolve_overloaded_builtin): Add special case handling
9114 for the __builtin_cmpb function, as represented by the
9115 P6_OV_BUILTIN_CMPB constant.
9116 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
9117 (BU_P6_64BIT_2): New macro.
9118 (BU_P6_OVERLOAD_2): New macro
9119 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
9120 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
9121 (CMPB): Add overload support to represent both 32-bit and 64-bit
9122 compare-bytes function.
9123 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
9124 support for TARGET_CMPB.
9125 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
9126 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
9127 documentation of the __builtin_cmpb overloaded built-in function.
9129 2017-05-11 Richard Biener <rguenther@suse.de>
9131 PR tree-optimization/80705
9132 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
9133 bases are not vectorizable.
9135 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9137 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
9138 when counting register pressure.
9140 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9142 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
9143 (iv_ca_more_deps): Renamed to ...
9144 (iv_ca_compare_deps): ... this.
9145 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
9147 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9149 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
9151 (determine_group_iv_costs): ... here.
9152 (find_inv_vars_cb): Record inv var if it's not recorded before.
9154 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9156 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
9157 (get_shiftadd_cost): Ditto.
9159 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9161 * tree-ssa-address.c: Include header file.
9162 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
9164 (add_to_parts): Refactor.
9165 (addr_to_parts): New parameter. Update use of move_hint_to_base.
9166 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
9169 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9171 PR tree-optimization/53090
9172 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
9174 (extract_cond_operands): Detect condition with IV on both sides
9175 and return COMP_IV_EXPR_2.
9176 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
9177 (rewrite_use_compare): Simplify by removing call to function
9178 extract_cond_operands.
9180 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9182 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
9183 (extract_cond_operands): Detect condition comparing against non-
9184 invariant bound and return appropriate enum value.
9185 (find_interesting_uses_cond): Update use of extract_cond_operands.
9186 Handle its return value accordingly.
9187 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
9189 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9191 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
9192 nonlinear iv_use computation in loop invariant sensitive way.
9194 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9196 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
9197 (find_iv_candidates): Call relate_compare_use_with_all_cands.
9199 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9201 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
9202 (dump_cand): Support iv_cand.inv_exprs.
9203 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
9205 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
9208 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9210 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
9212 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
9213 as local function. Include necessary header files.
9214 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
9216 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9218 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
9220 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9222 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
9223 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
9224 RSHIFT_EXPR and BIT_NOT_EXPR.
9226 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9228 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
9229 (adjust_setup_cost): New parameter supporting round up adjustment.
9230 (struct address_cost_data): Delete.
9231 (force_expr_to_var_cost): Don't bound cost with spill_cost.
9232 (split_address_cost, ptr_difference_cost): Delete.
9233 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
9234 (struct ainc_cost_data): New struct.
9235 (get_address_cost_ainc): New function.
9236 (get_address_cost, get_computation_cost): Reimplement.
9237 (determine_group_iv_cost_address): Record inv_expr for all uses of
9239 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
9240 (iv_ca_has_deps): Reimplemented to ...
9241 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
9243 (iv_ca_extend): Call iv_ca_more_deps.
9245 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9247 * tree-ssa-address.c (struct mem_address): Move to header file.
9248 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
9249 * tree-ssa-address.h (struct mem_address): Move from C file.
9250 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
9252 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9254 * tree-affine.h (aff_combination_type): New interface.
9255 (aff_combination_zero_p): Remove static.
9256 (aff_combination_const_p): New interface.
9257 (aff_combination_singleton_var_p): New interfaces.
9259 2017-05-11 Richard Biener <rguenther@suse.de>
9261 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9262 Skip unreachable blocks and destinations.
9263 (eliminate): Move stmt removal and fixup ...
9264 (fini_eliminate): ... here. Skip inserted exprs.
9265 (pass_pre::execute): Move fini_pre after fini_eliminate.
9266 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
9267 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
9268 PRE to get rid of dead code that has invalid SSA form and
9269 split critical edges again.
9271 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9273 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
9275 2017-05-11 Richard Biener <rguenther@suse.de>
9277 * passes.c (execute_function_todo): Verify loops if they are
9278 said to be up-to-date.
9279 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
9280 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
9282 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
9285 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
9286 handle calling assemble_external ourself.
9289 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
9290 modes with zero size. Enhance comment.
9292 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9294 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
9295 built-ins for vec_xl and vec_xst with short and char pointer
9298 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
9300 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
9301 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
9302 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
9303 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
9304 (_mm_maskz_min_round_ss): New intrinsics.
9305 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
9306 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
9307 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
9308 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
9309 (__builtin_ia32_minss_mask_round): New builtins.
9310 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9311 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9312 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
9314 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
9315 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
9317 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
9320 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
9322 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
9323 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
9324 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
9325 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
9326 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
9327 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
9328 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
9329 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9330 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
9331 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
9332 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
9333 (__builtin_ia32_mulss_mask_round): New builtins.
9334 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9335 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9336 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
9338 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
9339 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
9341 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
9344 2017-05-10 Julia Koval <julia.koval@intel.com>
9346 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
9347 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
9348 (_mm256_setr_m128i): New intrinsics.
9350 2017-05-10 Julia Koval <julia.koval@intel.com>
9352 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
9353 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
9354 (_mm_maskz_rcp14_ss): New intrinsics.
9355 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
9356 (__builtin_ia32_rcp14ss_mask): New builtins.
9357 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
9359 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
9361 PR tree-optimization/51513
9362 * tree-cfg.c (gimple_seq_unreachable_p): New function.
9363 (assert_unreachable_fallthru_edge_p): Use it.
9364 (group_case_labels_stmt): Likewise.
9365 * tree-cfg.h: Prototype it.
9366 * stmt.c: Include cfghooks.h and tree-cfg.h.
9367 (emit_case_dispatch_table) <gap_label>: New local variable.
9368 Use it to fill dispatch table gaps.
9369 Test for default_label before updating probabilities.
9370 (expand_case) <default_label>: Remove unneeded initialization.
9371 Test for unreachable default case statement and remove its edge.
9372 Set default_label accordingly.
9373 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
9375 2017-05-10 Carl Love <cel@us.ibm.com>
9377 * config/rs6000/rs6000-c: Add support for built-in functions
9378 vector signed char vec_neg (vector signed char)
9379 vector signed short int vec_neg (vector short int)
9380 vector signed int vec_neg (vector signed int)
9381 vector signed long long vec_neg (vector signed long long)
9382 vector float vec_neg (vector float)
9383 vector double vec_neg (vector double)
9384 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
9386 * config/rs6000/altivec.h: Add define for vec_neg
9387 * doc/extend.texi: Update the built-in documentation for the
9388 new built-in functions.
9390 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9392 PR tree-optimization/77644
9393 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
9395 2017-05-10 Nathan Sidwell <nathan@acm.org>
9397 * dumpfile.h (TDI_lang_all): New.
9398 (TDF_KIND): New. Renumber others
9399 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
9401 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
9403 (get_dump_file_name): Adjust suffix generation.
9404 (dump_enable_all): Use TDF_KIND.
9405 * doc/invoke.texi (-fdump-lang-all): Document.
9407 * dumpfile.h: Tabify.
9409 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
9412 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
9413 Move member access before delete.
9415 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
9417 * tree-inline.c (expand_call_inline): Split block at stmt
9420 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9423 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
9424 are now unused after splitting mov{sf,sd}_hardfloat.
9425 (f32_lr2): Likewise.
9427 (f32_lm2): Likewise.
9429 (f32_li2): Likewise.
9432 (f32_sr2): Likewise.
9434 (f32_sm2): Likewise.
9436 (f32_si2): Likewise.
9439 (f32_vsx): Likewise.
9441 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
9442 For movsf, order stores so the VSX stores occur before the GPR
9443 store which encourages the register allocator to use a traditional
9444 FPR instead of a GPR. For movsd, order the stores so that the GPR
9445 store comes before the VSX stores to allow the power6 to work.
9446 This is due to the power6 not having a 32-bit integer store
9447 instruction from a FPR.
9448 (movsf_hardfloat): Likewise.
9449 (movsd_hardfloat): Likewise.
9451 2017-05-09 Martin Sebor <msebor@redhat.com>
9453 PR translation/80280
9454 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
9457 PR translation/80280
9458 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
9459 data member added in r247778.
9460 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
9462 2017-05-09 Nathan Sidwell <nathan@acm.org>
9464 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
9466 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
9469 2017-05-09 Marek Polacek <polacek@redhat.com>
9471 * doc/invoke.texi: Fix typo.
9473 2017-05-09 Richard Biener <rguenther@suse.de>
9475 * tree-vrp.c (vrp_val_is_max): Adjust comment.
9476 (vrp_val_is_min): Likewise.
9477 (set_value_range_to_value): Likewise.
9478 (set_value_range_to_nonnegative): Likewise.
9479 (gimple_assign_nonzero_p): Likewise.
9480 (gimple_stmt_nonzero_p): Likewise.
9481 (vrp_int_const_binop): Likewise. Remove unreachable case.
9482 (adjust_range_with_scev): Adjust comments.
9483 (compare_range_with_value): Likewise.
9484 (extract_range_from_phi_node): Likewise.
9485 (test_for_singularity): Likewise.
9487 2017-05-09 Richard Biener <rguenther@suse.de>
9489 * tree-vrp.c (get_single_symbol): Add assert that we don't
9490 get overflowed constants as invariant part.
9491 (compare_values_warnv): Add comment before the TREE_NO_WARNING
9492 checks. Use wi::cmp instead of recursing for integer constants.
9493 (compare_values): Just ignore whether we assumed undefined
9494 overflow instead of failing the compare.
9495 (extract_range_for_var_from_comparison_expr): Add comment before the
9496 TREE_NO_WARNING sets.
9497 (test_for_singularity): Likewise.
9498 (extract_range_from_comparison): Do not disable optimization
9499 when we assumed undefined overflow.
9500 (extract_range_basic): Remove init of unused var.
9502 2017-05-09 Richard Biener <rguenther@suse.de>
9504 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
9505 (extract_range_from_multiplicative_op_1): Adjust.
9506 (extract_range_from_binary_expr_1): Use int_const_binop.
9508 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
9511 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
9512 rs6000_store_data_bypass_p in seven define_bypass directives and
9513 in several comments.
9514 * config/rs6000/rs6000-protos.h: Add prototype for
9515 rs6000_store_data_bypass_p function.
9516 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
9517 function implements slightly different (rs6000-specific) semantics
9518 than store_data_bypass_p, returning false rather than aborting
9519 with assertion error when arguments do not satisfy the
9520 requirements of store data bypass.
9521 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
9522 rs6000_store_data_bypass_p.
9524 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
9526 * config/xtensa/xtensa-protos.h
9527 (xtensa_initial_elimination_offset): New declaration.
9528 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
9529 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
9530 macro definition, add case for FRAME_POINTER_REGNUM when
9531 FRAME_GROWS_DOWNWARD.
9532 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
9533 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
9534 xtensa_initial_elimination_offset.
9536 2017-05-08 Nathan Sidwell <nathan@acm.org>
9538 * doc/invoke.texi: Alphabetize -fdump options.
9540 2017-05-08 Martin Sebor <msebor@redhat.com>
9542 PR translation/80280
9543 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
9545 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
9547 * target.def (compute_frame_layout): New optional target hook.
9548 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
9549 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
9550 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
9552 * reload1.c (verify_initial_elim_offsets): Likewise.
9553 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
9554 (use_simple_return_p): Call arm_compute_frame_layout if needed.
9555 (arm_get_frame_offsets): Split up into this ...
9556 (arm_compute_frame_layout): ... and this function.
9558 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
9560 * config/aarch64/constraints.md (Usa): New constraint.
9561 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
9563 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
9565 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
9566 with_multilib_list after it has been checked.
9568 2017-05-08 Richard Biener <rguenther@suse.de>
9570 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
9571 (bitmap_set_subtract_values): Likewise.
9573 2017-05-08 Richard Biener <rguenther@suse.de>
9575 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
9576 (gimple_assign_nonzero): ... this and remove strict_overflow_p
9578 (gimple_stmt_nonzero_warnv_p): Rename to ...
9579 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
9581 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
9582 (extract_range_basic): Adjust, do not disable propagation on
9583 strict overflow sensitive simplification.
9584 (vrp_visit_cond_stmt): Likewise.
9586 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
9588 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
9589 body size unconditionally.
9591 2017-05-07 Jeff Law <law@redhat.com>
9594 2017-05-06 Jeff Law <law@redhat.com>
9595 PR tree-optimization/78496
9596 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9599 PR tree-optimization/78496
9600 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9601 (simplify_stmt_using_ranges): Call it.
9602 (vrp_dom_walker::before_dom_children): Extract equivalences
9603 from an ASSERT_EXPR with an equality comparison against a
9606 2017-05-06 Jeff Law <law@redhat.com>
9608 PR tree-optimization/78496
9609 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9612 PR tree-optimization/78496
9613 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9614 (simplify_stmt_using_ranges): Call it.
9615 (vrp_dom_walker::before_dom_children): Extract equivalences
9616 from an ASSERT_EXPR with an equality comparison against a
9619 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9621 * lra-constraints.c (lra_copy_reg_equiv): New function.
9622 (split_reg): Use it to copy equivalence information from the
9623 original register to the spill register.
9625 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9627 PR rtl-optimization/75964
9628 * simplify-rtx.c (simplify_const_relational_operation): Remove
9629 invalid handling of comparisons of integer ABS.
9631 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
9633 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
9635 (init_regs): Remove declaration.
9636 (function_arg_advance_32): Initialize error_p as boolean variable.
9638 2017-05-05 Nathan Sidwell <nathan@acm.org>
9640 * store-motion.c (remove_reachable_equiv_notes): Reformat long
9641 lines. Use for (;;).
9643 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9645 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
9646 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
9647 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
9648 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
9649 VF=2 that require versioning.
9651 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9653 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
9656 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9658 * diagnostic.h (diagnostic_override_option_index): Convert from
9659 macro to inline function.
9661 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9663 * diagnostic.c (last_module_changed_p): New function.
9664 (set_last_module): New function.
9665 (diagnostic_report_current_module): Convert macro usage to
9666 the above functions.
9667 * diagnostic.h (diagnostic_context::last_module): Strengthen
9668 from const line_map * to const line_map_ordinary *.
9669 (diagnostic_last_module_changed): Delete macro.
9670 (diagnostic_set_last_module): Delete macro.
9672 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9674 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
9675 with diagnostic_report_diagnostic.
9676 (diagnostic_n_impl_richloc): Likewise.
9677 * diagnostic.h (report_diagnostic): Delete macro.
9678 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
9679 with diagnostic_report_diagnostic.
9680 * substring-locations.c (format_warning_va): Likewise.
9682 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9684 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
9685 save/restor of format_spec. Move option-printing code to...
9686 (print_option_information): ...this new function, and
9687 reimplement by simply printing to the pretty_printer,
9688 rather than appending to the format string.
9690 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9692 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
9693 handling logic into...
9694 (update_effective_level_from_pragmas): ...this new function.
9696 2017-05-04 Andrew Waterman <andrew@sifive.com>
9698 * config/riscv/riscv.opt (mstrict-align): New option.
9699 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
9700 (SLOW_UNALIGNED_ACCESS): Define.
9701 (riscv_slow_unaligned_access): Declare.
9702 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
9704 (riscv_slow_unaligned_access): New variable.
9705 (rocket_tune_info): Set slow_unaligned_access to true.
9706 (optimize_size_tune_info): Set slow_unaligned_access to false.
9707 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
9708 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
9709 (riscv_option_override): Set riscv_slow_unaligned_access.
9710 * doc/invoke.texi: Add -mstrict-align to RISC-V.
9712 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
9714 * config/riscv/riscv.md: Unify indentation.
9716 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
9721 * config/rs6000/rs6000.md (u code attribute): Add FIX and
9723 (extendsi<mode>2): Add support for doing sign extension via
9724 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
9725 don't have ISA 3.0 instructions.
9726 (extendsi<mode>2 splitter): Likewise.
9727 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
9728 generate the normal insns since SImode can now go in vector
9729 registers. Disallow the special UNSPECs needed for previous
9730 machines to hide SImode being used. Add new insns
9731 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
9732 (fix_trunc<mode>si2_stfiwx): Likewise.
9733 (fix_trunc<mode>si2_internal): Likewise.
9734 (fixuns_trunc<mode>si2): Likewise.
9735 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9736 (fctiw<u>z_<mode>_smallint): Likewise.
9737 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
9738 of floating point to 32-bit integer from doing a direct move to
9739 the GPR registers to do a store.
9740 (fctiwz_<mode>): Break long line.
9742 2017-05-05 Bin Cheng <bin.cheng@arm.com>
9744 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
9745 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
9746 (addr_list, addr_offset_valid_p): New.
9747 (split_address_groups): Check offset validity with above function.
9748 (gt-tree-ssa-loop-ivopts.h): Include header file.
9750 2017-05-05 Nathan Sidwell <nathan@acm.org>
9752 * config.gcc (arm*-*-*): Add missing 'fi'.
9754 2017-05-05 Steve Ellcey <sellcey@cavium.com>
9756 * doc/invoke.texi (-fopt-info): Explicitly say order of options
9757 included in -fopt-info does not matter.
9758 * doc/optinfo.texi (-fopt-info): Fix description of default
9759 behavour. Explicitly say order of options included in -fopt-info
9762 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
9764 * config.gcc: Allow combinations of aprofile and rmprofile values for
9765 --with-multilib-list.
9766 * config/arm/t-multilib: New file.
9767 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
9768 variables. Remove setting of ISA and floating-point ABI in
9769 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
9770 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
9771 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
9772 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
9774 * config/arm/t-rmprofile: Likewise except for the matches changes.
9775 * doc/install.texi (--with-multilib-list): Document the combination of
9776 aprofile and rmprofile values and warn about pitfalls in doing that.
9778 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9780 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
9781 (movdi_aarch64): Likewise.
9783 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9785 PR tree-optimization/80632
9786 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
9788 (build_arrays): Initialize it for virtual phis.
9789 (fix_phi_nodes): Use it for virtual phis.
9791 PR tree-optimization/80558
9792 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
9793 [x, y] op z into [x op, y op z] for op & or | if conditions
9796 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
9797 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9800 * config/arm/arm.md (use_literal_pool): Remove.
9801 (64-bit immediate split): No longer takes cost into consideration
9802 if arm_disable_literal_pool is enabled.
9803 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
9804 used when arm_disable_literal_pool is enabled.
9805 (arm_max_const_double_inline_cost): Remove use of
9806 arm_disable_literal_pool.
9807 (push_minipool_fix): Add assert.
9808 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
9809 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
9810 (no_literal_pool_sf_immediate): New.
9812 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9814 PR tree-optimization/80613
9815 * tree-ssa-dce.c (propagate_necessity): Remove cases for
9816 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
9818 2017-05-05 Richard Biener <rguenther@suse.de>
9820 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
9822 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
9824 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
9825 of this flag from insn conditions due to removal from r247495.
9827 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9829 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
9831 (arm_early_store_addr_dep_ptr): Likewise.
9832 * config/arm/aarch-common-protos.h
9833 (arm_early_load_addr_dep_ptr): Add prototype.
9834 (arm_early_store_addr_dep_ptr): Likewise.
9835 * config/arm/cortex-a53.md: Add new bypasses.
9837 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9839 * tree.c (next_type_uid): Change type to unsigned.
9840 (type_hash_canon): Decrement back next_type_uid if
9841 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
9842 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
9845 2017-05-04 Martin Sebor <msebor@redhat.com>
9847 * builtins.c: Fix a trivial typo in a comment.
9850 * builtins.c (check_sizes): Adjust to handle reading past the end.
9851 Avoid printing excessive upper bound of ranges. Use %E to print
9852 tree nodes instead of converting them to %wu.
9853 (expand_builtin_memchr): New function.
9854 (compute_dest_size): Rename...
9855 (compute_objsize): ...to this.
9856 (expand_builtin_memcpy): Adjust.
9857 (expand_builtin_mempcpy): Adjust.
9858 (expand_builtin_strcat): Adjust.
9859 (expand_builtin_strcpy): Adjust.
9860 (check_strncat_sizes): Adjust.
9861 (expand_builtin_strncat): Adjust.
9862 (expand_builtin_strncpy): Adjust and simplify.
9863 (expand_builtin_memset): Adjust.
9864 (expand_builtin_bzero): Adjust.
9865 (expand_builtin_memcmp): Adjust.
9866 (expand_builtin): Handle memcmp.
9867 (maybe_emit_chk_warning): Check strncat just once.
9869 2017-05-04 Martin Sebor <msebor@redhat.com>
9871 PR preprocessor/79214
9874 * builtins.c (check_sizes): Add inlining context and issue
9875 warnings even when -Wno-system-headers is set.
9876 (check_strncat_sizes): Same.
9877 (expand_builtin_strncat): Same.
9878 (expand_builtin_memmove): New function.
9879 (expand_builtin_stpncpy): Same.
9880 (expand_builtin): Handle memmove and stpncpy.
9882 2017-05-04 Bin Cheng <bin.cheng@arm.com>
9884 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
9885 which is not used any more.
9887 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9889 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
9891 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9893 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
9894 (cortexa53_tunings): Likewise.
9895 (cortexa57_tunings): Likewise.
9896 (cortexa72_tunings): Likewise.
9897 (cortexa73_tunings): Likewise.
9899 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9901 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
9902 Set loop alignment to 8.
9904 2017-05-04 Martin Sebor <msebor@redhat.com>
9906 PR translation/80280
9907 * builtins.c (expand_builtin_object_size): Add missing quoting to
9908 %D and like directives.
9909 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
9910 (hsa_type_for_tree_type): Same.
9911 (verify_function_arguments): Same.
9912 * symtab.c (symbol_table::change_decl_assembler_name): Same.
9913 * varasm.c (get_section): Same.
9916 2017-05-04 Martin Sebor <msebor@redhat.com>
9918 PR translation/80280
9919 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
9921 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9923 * config/aarch64/aarch64.c (generic_addrcost_table):
9924 Change HI/TI mode setting.
9926 2017-05-04 Martin Jambor <mjambor@suse.cz>
9928 PR tree-optimization/80622
9929 * tree-sra.c (comes_initialized_p): New function.
9930 (build_accesses_from_assign): Only set write lazily when
9931 comes_initialized_p is false.
9932 (analyze_access_subtree): Use comes_initialized_p.
9933 (propagate_subaccesses_across_link): Assert !comes_initialized_p
9934 instead of testing for PARM_DECL.
9936 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9938 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
9939 constraint on operand 0 to allow more general addressing modes.
9940 Adjust output template.
9941 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
9943 * config/aarch64/aarch64-protos.h
9944 (aarch64_address_valid_for_prefetch_p): Declare prototype.
9945 * config/aarch64/constraints.md (Dp): New address constraint.
9946 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
9949 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
9951 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
9952 update use of estimate_ipcp_clone_size_and_time.
9953 (estimate_local_effects): Update use of
9954 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
9955 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
9956 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
9957 Return nonspecialized time.
9959 2017-05-04 Richard Biener <rguenther@suse.de>
9961 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
9962 for the last VUSE which def dominates the PHI. Directly call
9964 (get_continuation_for_phi_1): Remove.
9966 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
9968 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
9969 to explain the use of truncating division. Cap the number of
9970 iterations to the maximum given by nb_iterations_upper_bound,
9973 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
9975 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
9976 * configure: Regenerate.
9977 * config.in: Regenerate.
9978 * config/i386/driver-mingw32.c: new file.
9979 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
9980 * config.host: Link driver-mingw32.o on MinGW host.
9981 * doc/install.texi: Document new --enable-mingw-wildcard configure
9984 2017-05-04 Marek Polacek <polacek@redhat.com>
9986 PR tree-optimization/80612
9987 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
9989 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9990 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
9992 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
9993 (movt splitter): Likewise.
9994 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
9995 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
9996 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
9997 block for Thumb-1 with MOVT.
9998 (thumb2_legitimate_address_p): Move code block ...
9999 (can_avoid_literal_pool_for_label_p): ... into this new function.
10000 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
10002 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
10003 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
10004 "M-profile targets with the MOVT instruction".
10006 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10008 * config/arm/arm-builtins.c (arm_init_builtins): Rename
10009 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
10010 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
10012 2017-05-04 Martin Liska <mliska@suse.cz>
10014 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
10015 variable cond_code.
10017 2017-05-04 Richard Biener <rguenther@suse.de>
10019 * tree.c (array_at_struct_end_p): Handle arrays at struct
10020 end with flexarrays more conservatively. Refactor and treat
10021 arrays of arrays or aggregates more strict. Fix
10022 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
10023 * tree.c (array_at_struct_end_p): Adjust prototype.
10024 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
10025 * gimple-fold.c (get_range_strlen): Likewise.
10026 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
10028 2017-05-04 Richard Biener <rguenther@suse.de>
10030 PR tree-optimization/31130
10031 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
10033 (supports_overflow_infinity): Likewise.
10034 (is_negative_overflow_infinity): Likewise.
10035 (is_positive_overflow_infinity): Likewise.
10036 (is_overflow_infinity): Likewise.
10037 (stmt_overflow_infinity): Likewise.
10038 (overflow_infinity_range_p): Likewise.
10039 (usable_range_p): Remove as always returning true.
10040 (make_overflow_infinity): Remove.
10041 (negative_overflow_infinity): Likewise.
10042 (positive_overflow_infinity): Likewise.
10043 (avoid_overflow_infinity): Likewise.
10044 (set_value_range): Adjust accordingly.
10045 (set_value_range_to_nonnegative): Likewise, remove now unused
10046 overflow_infinity arg.
10047 (vrp_operand_equal_p): Adjust.
10048 (update_value_range): Likewise.
10049 (range_int_cst_singleton_p): Likewise.
10050 (operand_less_p): Likewise.
10051 (compare_values_warnv): Likewise.
10052 (extract_range_for_var_from_comparison_expr): Likewise.
10053 (vrp_int_const_binop): Likewise.
10054 (zero_nonzero_bits_from_vr): Likewise.
10055 (extract_range_from_multiplicative_op_1): Likewise.
10056 (extract_range_from_binary_expr_1): Likewise.
10057 (extract_range_from_unary_expr): Likewise.
10058 (extract_range_from_comparison): Likewise.
10059 (extract_range_basic): Likewise.
10060 (adjust_range_with_scev): Likewise.
10061 (compare_ranges): Likewise.
10062 (compare_range_with_value): Likewise.
10063 (dump_value_range): Likewise.
10064 (test_for_singularity): Likewise, remove strict_overflow_p parameter
10066 (simplify_cond_using_ranges): Adjust.
10068 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
10070 * brig-builtins.def: Added a builtin for class_f64.
10071 * builtin-types.def: Added a builtin type needed by class_f64.
10073 2017-05-03 Jason Merrill <jason@redhat.com>
10075 * timevar.def: Add TV_CONSTEXPR.
10077 2017-05-03 David Malcolm <dmalcolm@redhat.com>
10079 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
10081 2017-05-03 Martin Jambor <mjambor@suse.cz>
10083 * ipa-prop.c (ipa_update_after_lto_read): Removed.
10084 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
10085 * ipa-cp.c (ipcp_propagate_stage): Do not call
10086 ipa_update_after_lto_read.
10087 * ipa-inline.c (ipa_inline): Likewise.
10089 2017-05-03 Martin Jambor <mjambor@suse.cz>
10091 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
10092 tag. Added a default constructor and a destructor.
10093 (ipa_edge_args_sum_t): New class;
10094 (ipa_edge_args_sum): Declare.
10095 (ipa_edge_args_vector): Remove declaration.
10096 (IPA_EDGE_REF): Use ipa_edge_args_sum.
10097 (ipa_free_edge_args_substructures): Remove declaration.
10098 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
10099 (ipa_edge_args_info_available_for_edge_p): Likewise.
10100 * ipa-prop.c (ipa_edge_args_vector): Removed.
10101 (edge_removal_hook_holder): Likewise.
10102 (edge_duplication_hook_holder): Likewise.
10103 (ipa_edge_args_sum): New variable.
10104 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
10105 ipa_edge_args_vector.
10106 (ipa_free_edge_args_substructures): Likewise.
10107 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
10108 ipa_edge_args_vector.
10109 (ipa_edge_removal_hook): Turned into method
10110 ipa_edge_args_sum_t::remove.
10111 (ipa_edge_duplication_hook): Turned into method
10112 ipa_edge_args_sum_t::duplicate.
10113 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
10114 registering edge hooks.
10115 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
10116 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
10117 ipa_edge_args_sum instead of ipa_edge_args_vector.
10118 * ipa-profile.c (ipa_profile): Likewise.
10120 2017-05-03 Martin Jambor <mjambor@suse.cz>
10122 * symbol-summary.h (function_summary): New method exists.
10123 (function_summary::symtab_removal): Deallocate through release.
10124 (call_summary): New class.
10125 (gt_ggc_mx): New overload.
10126 (gt_pch_nx): Likewise.
10127 (gt_pch_nx): Likewise.
10129 2017-05-03 Jeff Law <law@redhat.com>
10131 PR tree-optimization/78496
10132 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
10133 from simplify_cond_using_ranges. Split off code to walk
10134 backwards through casts into ...
10135 (simplify_cond_using_ranges_2): New function.
10136 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
10137 (execute_vrp): After identifying jump threads, call
10138 simplify_cond_using_ranges_2.
10140 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
10143 * ipa-inline.h (inline_summary): Add ctor.
10144 (create_ggc): Do not use ggc_cleared_alloc.
10146 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
10147 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10149 * gcc.c (handle_braces): Support escaping in switch matching
10151 * doc/invoke.texi (Spec Files): Document it.
10152 Remove superfluous @code markup in items.
10154 2017-05-03 David Malcolm <dmalcolm@redhat.com>
10156 * diagnostic-show-locus.c (struct column_range): New struct.
10157 (get_affected_columns): New function.
10158 (get_printed_columns): New function.
10159 (struct correction): New struct.
10160 (correction::ensure_capacity): New function.
10161 (correction::ensure_terminated): New function.
10162 (struct line_corrections): New struct.
10163 (line_corrections::~line_corrections): New dtor.
10164 (line_corrections::add_hint): New function.
10165 (layout::print_trailing_fixits): Reimplement in terms of the new
10167 (selftest::test_overlapped_fixit_printing): New function.
10168 (selftest::diagnostic_show_locus_c_tests): Call it.
10170 2017-05-03 Nathan Sidwell <nathan@acm.org>
10172 Canonicalize canonical type hashing
10173 * tree.h (type_hash_canon_hash): Declare.
10174 * tree.c (type_hash_list, attribute_hash_list): Move into
10175 type_hash_canon_hash.
10176 (build_type_attribute_qual_variant): Break out hash code calc into
10177 type_hash_canon_hash.
10178 (type_hash_canon_hash): New. Generic type hash computation.
10179 (build_range_type_1, build_array_type_1, build_function_type,
10180 build_method_type_directly, build_offset_type, build_complex_type,
10181 make_vector_type): Call it.
10183 2017-05-03 Richard Biener <rguenther@suse.de>
10185 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
10186 When all DRs have unknown misaligned do not always peel
10187 when there is a store but apply the same costing model as if
10188 there were only loads.
10190 2017-05-03 Richard Biener <rguenther@suse.de>
10193 PR tree-optimization/80492
10194 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
10195 compare_base_decls returning dont-know properly.
10197 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
10199 * config/arm/iterators.md (CCSI): New mode iterator.
10200 (arch): New mode attribute.
10201 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
10202 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
10203 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
10204 code iterator for success result mode.
10205 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
10206 the corresponding new insn generators.
10208 2017-05-03 Bin Cheng <bin.cheng@arm.com>
10211 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10212 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
10214 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
10216 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
10217 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
10218 (DDR_A): Wrap DDR argument in brackets.
10219 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
10220 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
10221 (DDR_REVERSED_P): Likewise.
10223 2017-05-03 Jakub Jelinek <jakub@redhat.com>
10225 PR tree-optimization/79472
10226 * tree-switch-conversion.c (struct switch_conv_info): Add
10227 contiguous_range and default_case_nonstandard fields.
10228 (collect_switch_conv_info): Compute contiguous_range and
10229 default_case_nonstandard fields, don't clear final_bb if
10230 contiguous_range and only the default case doesn't have the required
10232 (check_all_empty_except_final): Set default_case_nonstandard instead
10233 of failing if contiguous_range and the default case doesn't have empty
10235 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
10236 and only the default case doesn't have the required constants. Skip
10238 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
10239 if default_case_nonstandard.
10240 (build_constructors): Build constant 1 just once. Assert that default
10241 values aren't inserted in between cases if contiguous_range. Skip
10243 (build_arrays): Skip virtual phis.
10244 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
10245 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
10246 Handle virtual phis.
10247 (gen_inbound_check): Handle default_case_nonstandard case.
10248 (process_switch): Adjust check_final_bb caller. Call
10249 gather_default_values with the first non-default case instead of
10250 default case if default_case_nonstandard.
10252 2017-05-02 Nathan Sidwell <nathan@acm.org>
10254 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
10255 check. Fix formatting.
10257 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
10259 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
10260 errors when comparing specialized and unspecialized times.
10262 2017-05-02 David Malcolm <dmalcolm@redhat.com>
10264 * diagnostic-show-locus.c
10265 (layout::should_print_annotation_line_p): Make private.
10266 (layout::print_annotation_line): Make private.
10267 (layout::annotation_line_showed_range_p): Make private.
10268 (layout::show_ruler): Make private.
10269 (layout::print_source_line): Make private. Pass in line and
10270 line_width, rather than calling location_get_source_line. Drop
10272 (layout::print_leading_fixits): New method.
10273 (layout::print_any_fixits): Rename to...
10274 (layout::print_trailing_fixits): ...this, and make private.
10275 Don't print newline fixits.
10276 (diagnostic_show_locus): Move logic for printing one row into...
10277 (layout::print_line): ...this new function. Move the
10278 location_get_source_line call and error-handling from
10279 print_source_line to here. Call print_leading_fixits, and rename
10280 print_any_fixits to print_trailing_fixits.
10281 (selftest::test_fixit_insert_containing_newline): Update now that
10282 newlines are partially supported.
10283 (selftest::test_fixit_insert_containing_newline_2): New test.
10284 (selftest::test_fixit_replace_containing_newline): Update comments.
10285 (selftest::diagnostic_show_locus_c_tests): Call the new test.
10286 * edit-context.c (class added_line): New class.
10287 (class edited_line): Describe newline handling in comment.
10288 (edited_line::actually_edited_p): New method.
10289 (edited_line::print_content): Delete redundant decl.
10290 (edited_line::m_predecessors): New field.
10291 (edited_file::print_content): Call edited_line::print_content.
10292 (edited_file::print_diff): Update to support newlines.
10293 (edited_file::print_diff_hunk): Likewise.
10294 (edited_file::print_run_of_changed_lines): New function.
10295 (edited_file::print_diff_line): Convert to...
10296 (print_diff_line): ...this.
10297 (edited_file::get_effective_line_count): New function.
10298 (edited_line::edited_line): Initialize new field m_predecessors.
10299 (edited_line::~edited_line): Clean up m_predecessors.
10300 (edited_line::apply_fixit): Handle newlines.
10301 (edited_line::get_effective_line_count): New function.
10302 (edited_line::print_content): New function.
10303 (edited_line::print_diff_lines): New function.
10304 (selftest::test_applying_fixits_insert_containing_newline): New
10306 (selftest::test_applying_fixits_replace_containing_newline): New
10308 (selftest::insert_line): New function.
10309 (selftest::test_applying_fixits_multiple_lines): Add example of
10311 (selftest::edit_context_c_tests): Call the new tests.
10313 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10315 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
10316 parameter cand. Update dump information.
10317 (get_computation_cost): Update uses.
10319 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10321 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
10322 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
10323 (get_computation_at, rewrite_use_address): Update use of
10324 get_computation_aff.
10326 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10328 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
10329 (get_computation): Delete.
10330 (get_computation_cost): Implement like get_computation_cost_at.
10331 Use get_computation_at.
10332 (get_computation_cost_at): Delete.
10333 (rewrite_use_nonlinear_expr): Use get_computation_at.
10334 (rewrite_use_compare, remove_unused_ivs): Ditto.
10336 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10338 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
10340 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10342 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
10343 (ivopts_global_cost_for_size): Rename parameter and update uses.
10344 (iv_ca_recount_cost): Update uses.
10345 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
10346 candidates seperately in n_invs and n_cands.
10347 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
10349 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10351 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
10352 (find_inv_vars_cb): New.
10353 (find_depends): Renamed to ...
10354 (find_inv_vars): ... this.
10355 (add_candidate_1, force_var_cost): Call find_inv_vars.
10356 (split_address_cost, determine_group_iv_cost_cond): Ditto.
10358 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10360 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
10361 inv_vars. Add inv_exprs.
10362 (struct iv_cand): Rename depends_on to inv_vars.
10363 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
10364 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
10365 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
10366 (dump_cand): Dump inv_vars.
10367 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
10368 (record_invariant, find_depends, add_candidate_1): Ditto.
10369 (set_group_iv_cost, force_var_cost): Ditto.
10370 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
10371 (get_computation_cost_at, get_computation_cost): Ditto.
10372 (determine_group_iv_cost_generic): Ditto.
10373 (determine_group_iv_cost_address): Ditto.
10374 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
10375 (determine_group_iv_costs): Ditto.
10376 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
10377 (iv_ca_set_remove_invariants): Renamed to ...
10378 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
10379 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
10380 (iv_ca_set_add_invariants): Renamed to ...
10381 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
10382 (iv_ca_set_cp): Use iv_ca_set_add_invs.
10383 (iv_ca_has_deps): Support inv_vars and inv_exprs.
10384 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
10385 (create_new_ivs): Remove useless dump.
10387 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10389 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
10391 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
10392 (iv_ca_set_no_cp, create_new_iv): Ditto.
10394 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10396 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
10398 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10400 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
10401 function tree_check2.
10403 2017-05-02 Martin Liska <mliska@suse.cz>
10405 * doc/gcov.texi: Add missing preposition.
10406 * gcov.c (function_info::function_info): Properly fill up
10407 all member variables.
10409 2017-05-02 Tamar Christina <tamar.christina@arm.com>
10411 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
10413 2017-05-02 Tamar Christina <tamar.christina@arm.com>
10415 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
10417 2017-05-02 Martin Liska <mliska@suse.cz>
10420 * lto-streamer-in.c (lto_read_tree_1): Remove
10421 LTO_STREAMER_DEBUG.
10422 * lto-streamer.c (struct tree_hash_entry): Likewise.
10423 (struct tree_entry_hasher): Likewise.
10424 (tree_entry_hasher::hash): Likewise.
10425 (tree_entry_hasher::equal): Likewise.
10426 (lto_streamer_init): Likewise.
10427 (lto_orig_address_map): Likewise.
10428 (lto_orig_address_get): Likewise.
10429 (lto_orig_address_remove): Likewise.
10430 * lto-streamer.h: Likewise.
10431 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
10432 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
10434 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
10436 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
10437 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
10438 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
10439 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
10440 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
10441 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
10442 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
10443 (mm_maskz_sub_ss): New intrinsics.
10444 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
10445 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
10446 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
10447 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
10448 (__builtin_ia32_subss_mask_round): New builtins.
10449 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
10450 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
10451 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
10453 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
10454 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
10456 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
10459 2017-05-02 Martin Jambor <mjambor@suse.cz>
10461 PR tree-optimization/78687
10462 * tree-sra.c (access): New field parent.
10463 (process_subtree_disqualification): New function.
10464 (disqualify_candidate): Call it.
10465 (build_accesses_from_assign): Reset write flag if creating an
10467 (build_access_subtree): Fill in parent field and also prpagate
10468 down grp_write flag.
10469 (create_artificial_child_access): New parameter set_grp_write, set
10470 grp_write to its value.
10471 (propagate_subaccesses_across_link): Also propagate grp_write flag
10473 (propagate_all_subaccesses): Push the closest parent back to work
10474 queue if add_access_to_work_queue returned true.
10476 2017-05-02 Richard Biener <rguenther@suse.de>
10478 * common.opt (fstrict-overflow): Alias negative to fwrapv.
10479 * doc/invoke.texi (fstrict-overflow): Remove all traces of
10480 -fstrict-overflow documentation.
10481 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
10482 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
10483 flag_strict_overflow.
10484 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
10485 * lto-opts.c (lto_write_options): Do not stream it.
10486 * lto-wrapper.c (merge_and_complain): Do not handle it.
10487 * opts.c (default_options_table): Do not set -fstrict-overflow.
10488 (finish_options): Likewise do not clear it when sanitizing.
10489 * simplify-rtx.c (simplify_const_relational_operation): Do not
10490 test flag_strict_overflow.
10492 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10494 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
10495 using enabled attribute.
10496 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
10497 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
10498 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
10499 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
10500 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
10501 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
10502 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
10503 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
10504 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
10505 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
10507 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10509 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
10511 2017-05-02 Richard Biener <rguenther@suse.de>
10513 PR tree-optimization/80591
10515 2017-04-10 Richard Biener <rguenther@suse.de>
10517 * tree-ssa-structalias.c (find_func_aliases): Properly handle
10520 2017-05-02 Richard Biener <rguenther@suse.de>
10522 PR tree-optimization/80549
10523 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
10524 (cleanup_tree_cfg_noloop): Create forwarders to known loop
10525 headers if they do not have a preheader.
10527 2017-05-02 Martin Liska <mliska@suse.cz>
10530 * common.opt: Fix typo.
10531 * doc/invoke.texi: Likewise.
10533 2017-05-01 Jan Beulich <jbeulich@suse.com>
10535 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
10536 swapping, add (x,x,m,x,n) alternative.
10538 2017-05-01 Nathan Sidwell <nathan@acm.org>
10540 * calls.c (combine_pending_stack_adjustment_and_call): Remove
10541 unnecessary unadjusted_alignment check.
10543 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10546 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
10548 * gimplify.c (gimplify_cilk_detach): New function.
10549 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
10550 * tree-core.h: Document EXPR_CILK_SPAWN.
10551 * tree.h (EXPR_CILK_SPAWN): Define.
10553 2017-05-01 David Malcolm <dmalcolm@redhat.com>
10555 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
10556 to use new fixit_hint representation, using the "replace" logic.
10557 (get_line_span_for_fixit_hint): Likewise.
10558 (layout::print_any_fixits): Likewise.
10559 (selftest::test_one_liner_many_fixits): Rename to...
10560 (selftest::test_one_liner_many_fixits_1): ...this, and update
10561 comment and expected output to reflect that the multiple fix-it
10562 hints are now consolidated into one insertion.
10563 (selftest::test_one_liner_many_fixits_2): New test.
10564 (selftest::test_diagnostic_show_locus_one_liner): Update for
10566 (selftest::test_fixit_consolidation): Update for fix-it API
10568 * diagnostic.c (print_parseable_fixits): Likewise.
10569 * edit-context.c (edited_line::m_line_events): Convert from
10570 auto_vec <line_event *> to auto_vec <line_event>.
10571 (class line_event): Convert from abstract base class to a concrete
10572 class, taking over the role of replace_event.
10573 (class insert_event): Delete.
10574 (class replace_event): Rename to class line_event. Convert to
10576 (edit_context::add_fixits): Reimplement.
10577 (edit_context::apply_insert): Delete.
10578 (edit_context::apply_replace): Rename to...
10579 (edit_context::apply_fixit): ...this. Convert to half-open range.
10580 (edited_file::apply_insert): Delete.
10581 (edited_file::apply_replace): Rename to...
10582 (edited_file::apply_fixit): ...this.
10583 (edited_line::~edited_line): Drop deletion of events.
10584 (edited_line::apply_insert): Delete.
10585 (edited_line::apply_replace): Rename to...
10586 (edited_line::apply_fixit): ...this. Convert to half-open range.
10587 Update for change to type of m_line_events.
10588 * edit-context.h (edit_context::apply_insert): Delete.
10589 (edit_context::apply_replace): Rename to...
10590 (edit_context::apply_fixit): ...this.
10592 2017-05-01 Martin Sebor <msebor@redhat.com>
10594 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
10597 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
10600 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
10601 __get_cpuid_max returns 0.
10602 (__get_cpuid_count): Ditto.
10604 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
10606 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
10607 replacement expression is another instance of one of its arguments.
10609 2017-05-01 Jakub Jelinek <jakub@redhat.com>
10612 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
10613 check for stack push/pop autoinc.
10614 * config/i386/i386.c (ix86_agi_dependent): Return false
10615 if the only reason why modified_in_p returned true is that
10616 addr is SP based and set_insn is a push or pop.
10618 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10620 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
10623 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10626 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
10627 (account_size_time): Use two predicates - exec_pred and
10629 (evaluate_conditions_for_known_args): Compute both clause and
10631 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
10632 (inline_summary_t::duplicate): Update.
10633 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
10635 (compute_inline_parameters): Likewise.
10636 (estimate_edge_size_and_time): Update caluclation of time.
10637 (estimate_node_size_and_time): Compute both time and nonspecialized
10639 (estimate_ipcp_clone_size_and_time): Update.
10640 (inline_merge_summary): Update.
10641 (do_estimate_edge_time): Update.
10642 (do_estimate_edge_size): Update.
10643 (do_estimate_edge_hints): Update.
10644 (inline_read_section, inline_write_summary): Stream both new predicates.
10645 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
10647 (compute_inlined_call_time): Cleanup.
10648 (big_speedup_p): Update.
10649 (edge_badness): Update.
10650 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
10651 (size_time_entry): Replace predicate by exec_predicate and
10652 nonconst_predicate.
10653 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
10654 (estimate_edge_time): Return also nonspec_time.
10655 (reset_edge_growth_cache): Update.
10657 2017-04-29 Jakub Jelinek <jakub@redhat.com>
10659 PR rtl-optimization/80491
10660 * ifcvt.c (noce_process_if_block): When looking for x setter
10661 with missing else_bb, don't check only the insn right before
10662 cond_earliest, but look for the last insn that x is modified in
10663 within the same bb.
10665 PR rtl-optimization/80491
10666 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
10668 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
10670 PR tree-optimization/80487
10671 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
10673 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10675 PR tree-optimization/79697
10676 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
10677 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
10678 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
10680 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
10681 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
10683 2017-04-28 Martin Sebor <msebor@redhat.com>
10685 PR tree-optimization/80523
10686 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
10687 (init_target_to_host_charmap, target_to_host, target_strtol10): New
10689 (maybe_warn, format_directive, parse_directive): Use new functions.
10690 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
10692 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
10694 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
10696 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
10698 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
10699 target_header_dir): Set correctly.
10700 * configure: Regenerated.
10701 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
10702 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
10703 instead of SYSTEM_HEADER_DIR.
10705 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
10707 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
10708 (estimate_local_effects): Likewise.
10709 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
10710 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
10711 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
10712 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
10713 do_estimate_edge_time, estimate_edge_time): Likewise.
10714 * ipa-inline-analysis.c (estimate_node_size_and_time,
10715 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
10716 (estimate_time_after_inlining): Remove.
10718 2017-04-28 Martin Liska <mliska@suse.cz>
10720 * doc/gcov.texi: Enhance documentation of gcov.
10722 2017-04-28 Martin Liska <mliska@suse.cz>
10724 * doc/gcov.texi: Sort options in alphabetic order.
10725 * doc/gcov-dump.texi: Likewise.
10726 * doc/gcov-tool.texi: Likewise.
10727 * gcov.c (print_usage): Likewise.
10728 * gcov-dump.c (print_usage): Likewise.
10729 * gcov-tool.c (print_merge_usage_message): Likewise.
10730 (print_rewrite_usage_message): Likewise.
10731 (print_overlap_usage_message): Likewise.
10733 2017-04-28 Martin Liska <mliska@suse.cz>
10735 PR gcov-profile/53915
10736 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
10738 2017-04-28 Martin Liska <mliska@suse.cz>
10740 PR gcov-profile/79891
10741 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
10742 is marked by compiler as living on a line.
10743 (get_cycles_count): Remove usage of the union.
10744 (output_intermediate_file): Likewise.
10745 (find_source): Fix GNU coding style.
10746 (accumulate_line_counts): Remove old non-all block mode.
10747 (output_lines): Remove usage of the union.
10748 * profile.c (output_location): Include all BBs, even if
10749 belonging to a same line (and file) as a previous BB.
10751 2017-04-28 Martin Liska <mliska@suse.cz>
10753 * gcov.c (process_args): Handle new argument 'w'.
10754 (read_graph_file): Assign ID to BBs.
10755 (output_branch_count): Display BB # if verbose flag is set.
10756 (output_lines): Likewise for arcs.
10757 (print_usage): Add '--verbose' option help.
10758 * doc/gcov.texi: Document --verbose (-w) option.
10760 2017-04-28 Martin Liska <mliska@suse.cz>
10762 * gcov.c (struct block_location_info): New struct.
10763 (process_file): Fill up the new structure.
10764 (read_graph_file): Replace usage of encoding by the newly added
10766 (add_line_counts): Likewise.
10767 (accumulate_line_counts): Remove usage of the union.
10768 (function_info::function_info): New function.
10769 (function_info::~function_info): Likewise.
10770 (process_file): Call delete instead of release_function.
10771 (release_function): Release the function.
10772 (release_structures): Call delete instead of release_function.
10773 (solve_flow_graph): Replace usage of num_blocks.
10774 (find_exception_blocks): Likewise.
10775 (output_lines): Fix GNU coding style.
10777 2017-04-28 Martin Liska <mliska@suse.cz>
10780 * coverage.c (coverage_remove_note_file): New function.
10781 * coverage.h: Declare the function.
10782 * toplev.c (finalize): Clean if an error has been seen.
10784 2017-04-28 Martin Liska <mliska@suse.cz>
10786 PR gcov-profile/80031
10787 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
10788 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
10789 * gcov.c (read_graph_file): Read just number of blocks.
10790 * profile.c (branch_prob): Do not stream 0 flags per a basic
10793 2017-04-28 Martin Liska <mliska@suse.cz>
10795 * gcov-dump.c (tag_*): Add new argument to declarations.
10796 (dump_gcov_file): Likewise.
10797 (tag_blocks): Add and use new argument depth.
10798 (tag_arcs): Likewise.
10799 (tag_lines): Likewise.
10800 (tag_counters): Likewise.
10801 (tag_summary): Likewise.
10802 (dump_working_sets): Use depth to do a proper indentation.
10804 2017-04-28 Jakub Jelinek <jakub@redhat.com>
10807 * cgraph.h (symtab_node::debug_symtab): No longer inline.
10808 * symtab.c (symtab_node::debug_symtab): Move definition here.
10810 2017-04-28 Richard Biener <rguenther@suse.de>
10812 * lto-streamer.h (LTO_major_version): Bump to 7.
10814 2017-04-28 Richard Biener <rguenther@suse.de>
10816 * tree-vrp.c (assert_info): New struct.
10817 (add_assert_info): New helper.
10818 (register_edge_assert_for_2): Refactor to add asserts to a vector
10820 (register_edge_assert_for_1): Likewise.
10821 (register_edge_assert_for): Likewise.
10822 (finish_register_edge_assert_for): New helper actually registering
10823 asserts where live on edge.
10824 (find_conditional_asserts): Adjust.
10825 (find_switch_asserts): Likewise.
10826 (evrp_dom_walker::try_find_new_range): Generalize.
10827 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
10829 2017-04-27 Marek Polacek <polacek@redhat.com>
10832 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
10833 arg10 and arg11 to itype.
10835 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
10837 * doc/extend.texi (Object Size Checking): Improve grammar.
10839 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
10842 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
10843 that the logic for permitting reciprocal estimates matches that
10846 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10849 * tree.c (type_cache_hasher::equal): Only compare
10850 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
10851 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
10852 non-aggregate element types.
10853 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
10854 about the flag on ARRAY_TYPEs in the comment, formatting fix.
10856 2017-04-27 Richard Biener <rguenther@suse.de>
10858 PR middle-end/80533
10859 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
10860 stripping ARRAY_REFs from MEM_EXPR make sure we're not
10861 keeping a reference to a trailing array.
10863 2017-04-27 Richard Biener <rguenther@suse.de>
10865 PR middle-end/80539
10866 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
10867 being in loop-closed SSA form conservatively.
10868 (chrec_fold_multiply_poly_poly): Likewise.
10870 2017-04-27 Tamar Christina <tamar.christina@arm.com>
10872 PR middle-end/79665
10873 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
10874 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
10876 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10879 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
10880 (aarch64_function_arg_alignment): Return unsigned int again, but still
10881 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
10882 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
10883 Don't emit -Wpsabi note.
10884 (aarch64_function_arg_boundary): Likewise.
10885 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
10888 2017-04-26 Nathan Sidwell <nathan@acm.org>
10890 * tree.h (crc32_unsigned_n): Declare.
10891 (crc32_unsigned, crc32_unsigned): Make inline.
10892 * tree.c (crc32_unsigned_bits): Replace with ...
10893 (crc32_unsigned_n): ... this.
10894 (crc32_unsigned, crc32_byte): Remove.
10895 (crc32_string): Remove unnecessary braces.
10897 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
10899 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
10900 * ipa-inline-analysis.c (MAX_TIME): Remove.
10901 (account_size_time): Use sreal for time.
10902 (dump_inline_summary): Update.
10903 (estimate_function_body_sizes): Update.
10904 (estimate_edge_size_and_time): Update.
10905 (estimate_calls_size_and_time): Update.
10906 (estimate_node_size_and_time): Update.
10907 (inline_merge_summary): Update.
10908 (inline_update_overall_summary): Update.
10909 (estimate_time_after_inlining): Update.
10910 (inline_read_section): Update.
10911 (inline_write_summary): Update.
10912 * ipa-inline.c (compute_uninlined_call_time): Update.
10913 (compute_inlined_call_time): Update.
10914 (recursive_inlining): Update.
10915 (inline_small_functions): Update.
10916 (dump_overall_stats): Update.
10917 * ipa-inline.h: Include sreal.h.
10918 (size_time_entry): Turn time to sreal.
10919 (inline_summary): Turn self_time nad time to sreal.
10921 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
10923 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
10925 (sreal::stream_out, sreal::stream_in): New.
10926 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
10928 2017-04-25 Jakub Jelinek <jakub@redhat.com>
10930 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
10933 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
10936 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
10938 (dimode_scalar_chain::compute_convert_gain): Ditto.
10939 (dimode_scalar_chain::make_vector_copies): Ditto.
10940 (dimode_scalar_chain::convert_reg): Ditto.
10941 (dimode_scalar_chain::convert_insn): Ditto.
10942 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
10943 (VI248_AVX512BW_1): New mode iterator.
10944 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
10945 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
10948 2017-04-25 Martin Sebor <msebor@redhat.com>
10950 PR tree-optimization/80497
10951 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
10952 constants are representable in HOST_WIDE_INT.
10953 (parse_directive): Ditto.
10955 2017-04-25 Martin Sebor <msebor@redhat.com>
10958 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
10959 (new_zero_array): Adjust signature.
10960 (dom_info::dom_init): Used unsigned rather that size_t.
10961 (dom_info::dom_info): Same.
10963 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10964 Jakub Jelinek <jakub@redhat.com>
10967 * config/arm/arm.c: Include gimple.h.
10968 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
10969 returns negative, increment ncrn only if it returned positive.
10970 (arm_needs_doubleword_align): Return int instead of bool,
10971 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
10972 members, but if there is any such non-FIELD_DECL
10973 > PARM_BOUNDARY aligned decl, return -1 instead of false.
10974 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
10975 returns negative, increment nregs only if it returned positive.
10976 (arm_setup_incoming_varargs): Likewise.
10977 (arm_function_arg_boundary): Emit -Wpsabi note if
10978 arm_needs_doubleword_align returns negative, return
10979 DOUBLEWORD_ALIGNMENT only if it returned positive.
10981 2017-04-25 Marek Polacek <polacek@redhat.com>
10984 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
10985 first argument to type.
10987 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
10990 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
10991 type checks to test for compatibility instead of equality.
10993 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10994 Jakub Jelinek <jakub@redhat.com>
10997 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
10999 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
11000 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
11001 the alignment computation, but return their maximum in warn_alignment.
11002 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
11003 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
11005 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
11006 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
11009 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11011 * config/arc/simdext.md (dmpyh): Fix typo.
11013 2017-04-25 Richard Biener <rguenther@suse.de>
11015 PR tree-optimization/80492
11016 * alias.c (compare_base_decls): Handle registers with asm
11017 specification conservatively.
11018 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
11019 compare_base_decls returning dont-know properly.
11021 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11023 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
11024 (legitimate_offset_address_p): New function.
11025 (arc_legitimate_address_p): Use above function.
11027 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11029 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
11031 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11033 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
11034 ACCH registers whenever they are available.
11036 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11038 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
11039 double regs fix when not used.
11041 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11043 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
11045 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
11046 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
11048 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11050 * config/arc/arc.c (arc_output_addsi): Check for h-register class
11051 when emitting short ADD instructions.
11053 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11055 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
11057 (cmpsi_cc_c_insn): Likewise.
11058 (cbranchsi4_scratch): Compute proper instruction length using
11059 compact_hreg_operand.
11060 * config/arc/predicates.md (compact_hreg_operand): New predicate.
11062 2017-04-25 Richard Biener <rguenther@suse.de>
11064 PR middle-end/80509
11065 * passes.c (pass_manager::pass_manager): Initialize
11066 m_name_to_pass_map.
11068 2017-04-25 Richard Biener <rguenther@suse.de>
11070 PR tree-optimization/79201
11071 * tree-ssa-sink.c (statement_sink_location): Handle calls.
11073 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11076 * config/s390/vector.md: Split MEM->GPR vector moves for
11077 non-s_operand addresses.
11079 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11082 * config/s390/predicates.md (reload_const_wide_int_operand): New
11084 * config/s390/s390.md ("movti"): Remove d/P alternative.
11085 ("movti_bigconst"): New pattern definition.
11087 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11090 * s390-protos.h (s390_expand_cs_hqi): Removed.
11091 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
11092 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
11093 modes as well as CCZ1mode and CCZmode.
11094 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
11095 signature of s390_emit_compare_and_swap.
11096 (s390_expand_cs_hqi): Likewise, make static.
11097 (s390_expand_cs_tdsi): Generate an explicit compare before trying
11098 compare-and-swap, in some cases.
11099 (s390_expand_cs): Wrapper function.
11100 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
11102 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
11103 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
11104 patterns for small and large integers. Forbid symref memory operands.
11105 Move expander to s390.c. Require cc register.
11106 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
11107 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
11108 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
11109 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
11110 symref memory operands. Remove CC mode and call s390_match_ccmode
11112 ("atomic_exchange<mode>"): Allow and implement all integer modes.
11114 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11116 * config/s390/s390.md (define_peephole2): New peephole to help
11117 combining the load-and-test pattern with volatile memory.
11119 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11121 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
11122 with CCZmode for TARGET_Z196.
11124 2017-04-25 Jakub Jelinek <jakub@redhat.com>
11126 PR rtl-optimization/80501
11127 * combine.c (make_compound_operation_int): Set subreg_code to SET
11128 even for AND with mask of the sign bit of mode.
11130 PR rtl-optimization/80500
11131 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
11132 sum's initial value.
11134 2017-04-25 Julian Brown <julian@codesourcery.com>
11135 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11137 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
11139 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
11141 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
11143 2017-04-25 Julian Brown <julian@codesourcery.com>
11144 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11146 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
11147 (thunderx2t99_sha): New Reservation.
11149 2017-04-25 Julian Brown <julian@codesourcery.com>
11150 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11152 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
11153 type for 1-element load.
11155 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
11157 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
11159 2017-04-24 Martin Jambor <mjambor@suse.cz>
11161 PR tree-optimization/80293
11162 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
11163 char arrays not totally scalarizable if it is false.
11164 (analyze_all_variable_accesses): Pass correct value in the new
11165 parameter. Add a statistics counter.
11167 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
11169 PR middle-end/79931
11170 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
11172 2017-04-24 Richard Biener <rguenther@suse.de>
11174 PR tree-optimization/80494
11175 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
11176 out for complex types.
11178 2017-04-24 Richard Biener <rguenther@suse.de>
11180 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
11181 * tree-ssa-sccvn.c (print_scc): Print SCC size.
11182 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
11183 (DFS): Adjust and never fail.
11184 (sccvn_dom_walker::fail): Remove.
11185 (sccvn_dom_walker::before_dom_children): Adjust.
11186 (run_scc_vn): Likewise and never fail.
11187 * tree-ssa-pre.c (pass_pre::execute): Adjust.
11188 (pass_fre::execute): Likewise.
11190 2017-04-24 Richard Biener <rguenther@suse.de>
11192 PR tree-optimization/79725
11193 * tree-ssa-sink.c (statement_sink_location): Return whether
11194 failure reason was zero uses. Move that check later.
11195 (sink_code_in_bb): Deal with zero uses by removing the stmt
11198 2017-04-24 Richard Biener <rguenther@suse.de>
11201 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
11202 pointer-based references.
11204 2017-04-24 Richard Biener <rguenther@suse.de>
11207 * pass_manager.h (pass_manager::operator new): Remove.
11208 (pass_manager::operator delete): Likewise.
11209 * passes.c (pass_manager::operator new): Remove.
11210 (pass_manager::operator delete): Likewise.
11211 (pass_manager::pass_manager): Zero individual pass members.
11213 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
11216 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
11217 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
11218 Check "XEXP (src, 1)" operand here.
11219 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
11220 Check "XEXP (src, 1)" operand here.
11221 (dimode_scalar_chain::make_vector_copies): Detect count register
11222 of a shift instruction. Zero extend count register from QImode
11223 to DImode to satisfy vector shift pattern count operand predicate.
11224 Substitute vector shift count operand with a DImode copy.
11225 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
11228 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
11230 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
11231 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
11232 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
11233 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
11234 (UNSPEC_NOREX_MEM): Remove definition.
11236 2017-04-21 Richard Biener <rguenther@suse.de>
11238 PR tree-optimization/79547
11239 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
11240 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
11241 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
11242 without any constraints.
11244 2017-04-21 Richard Biener <rguenther@suse.de>
11246 PR tree-optimization/78847
11247 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
11249 2017-04-21 Richard Biener <rguenther@suse.de>
11251 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
11252 (build_distinct_type_copy): Likewise.
11253 (build_variant_type_copy): Likewise.
11254 * tree.c (build_qualified_type): Pass down mem-stat info.
11255 (build_distinct_type_copy): Likewise.
11256 (build_variant_type_copy): Likewise.
11258 2017-04-21 Richard Biener <rguenther@suse.de>
11260 PR tree-optimization/80237
11261 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
11263 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
11264 for a simplified result.
11266 2016-04-21 Richard Biener <rguenther@suse.de>
11268 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
11269 sth as strict as a simple_iv but a chrec without symbols and an
11270 operand defined in the loop we are peeling (and not some subloop).
11271 (propagate_constants_for_unrolling): Propagate all constants.
11273 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
11276 * config/i386/i386.c (print_reg): Remove assert for disalowed
11277 regno values, call output_operand_lossage instead.
11279 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
11282 * config/i386/constraints.md (Yc): New register constraint.
11283 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
11284 Use Yc constraint for alternative 2 of operand 0. Remove
11285 preferred_for_speed attribute.
11287 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
11289 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
11290 lastprivate clauses in SIMT case.
11292 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
11294 * doc/invoke.texi (-Wextra-semi): Document new warning option.
11296 2017-04-20 Richard Biener <rguenther@suse.de>
11298 PR tree-optimization/57796
11299 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
11300 as N scalar stores.
11301 (vect_model_load_cost): Cost gathers as N scalar loads.
11303 2017-04-20 Richard Biener <rguenther@suse.de>
11305 * ggc-page.c (ggc_allocated_p): Rename to ...
11306 (safe_lookup_page_table_entry): ... this and return the lookup
11308 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
11310 2017-04-20 Richard Biener <rguenther@suse.de>
11312 PR tree-optimization/80453
11313 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
11314 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
11315 from the conditions.
11316 (vn_phi_eq): Pass them down.
11317 (vn_phi_lookup): Record them.
11318 (vn_phi_insert): Likewise.
11320 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
11322 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
11323 uninitialized variable warning to avoid buffer overrun.
11325 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
11328 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
11329 is suppressed for '{ 0 }' in C.
11331 2017-04-20 Jakub Jelinek <jakub@redhat.com>
11333 * BASE-VER: Set to 8.0.0.
11335 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
11337 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
11338 priority .init_array and .fini_array section with SECTION_NOTYPE
11341 2017-04-20 Jakub Jelinek <jakub@redhat.com>
11343 PR middle-end/80423
11344 * tree.h (build_array_type): Add typeless_storage default argument.
11345 * tree.c (type_cache_hasher::equal): Also compare
11346 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
11347 (build_array_type): Add typeless_storage argument, set
11348 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
11350 (build_nonshared_array_type): Adjust build_array_type_1 caller.
11351 (build_array_type): Likewise. Add typeless_storage argument.
11353 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
11354 Jakub Jelinek <jakub@redhat.com>
11356 PR tree-optimization/80426
11357 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
11358 operation on symbolic operands, also compute the overflow for the
11359 invariant part when the operation degenerates into a negation.
11361 2017-04-19 Jakub Jelinek <jakub@redhat.com>
11364 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
11365 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
11368 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
11370 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
11373 * config/avr/avr.c (tree.h): Include it.
11374 (cgraph.h): Include it.
11375 (avr_encode_section_info): Don't warn for uninitialized progmem
11376 variable if it's just an alias.
11378 2017-04-19 Richard Biener <rguenther@suse.de>
11381 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
11382 when needed by AutoPGO.
11384 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
11387 * doc/lto.texi: Remove an extra 'that'.
11389 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
11391 PR rtl-optimization/80429
11392 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
11393 are only used in debug insns.
11395 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
11396 Vladimir Makarov <vmakarov@redhat.com>
11398 * config/sparc/predicates.md (input_operand): Add comment. Return
11399 true for any memory operand when LRA is in progress.
11400 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
11402 2017-04-18 Jeff Law <law@redhat.com>
11405 * mips.md ({return,simple_return}_internal): Do not overwrite
11408 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11410 PR tree-optimization/80443
11411 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
11412 instead of adding 1, subtract -1 and similarly instead of subtracting
11415 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
11417 PR rtl-optimization/80357
11418 * haifa-sched.c (tmp_bitmap): New variable.
11419 (model_recompute): Handle duplicate use records.
11420 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
11421 (free_global_sched_pressure_data): Free it.
11423 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
11426 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
11427 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
11428 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
11429 instead of SYSTEM_HEADER_DIR.
11431 2017-04-18 Jeff Law <law@redhat.com>
11433 PR middle-end/80422
11434 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
11435 predecessors after walking up the insn chain.
11437 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11440 * dwarf2out.c (modified_type_die): Try harder not to emit internal
11441 sizetype type into debug info.
11443 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
11446 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
11447 unneeded test for TARGET_UPPER_REGS_SF.
11448 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
11450 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11453 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
11454 instead of gsi_after_labels.
11456 2017-04-18 Jeff Law <law@redhat.com>
11458 * regcprop.c (maybe_mode_change): Avoid creating copies of the
11462 2017-04-13 Jeff Law <law@redhat.com>
11463 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11464 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11466 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
11469 * config/avr/avr.c (intl.h): Include it.
11470 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
11472 2017-04-18 Martin Liska <mliska@suse.cz>
11474 PR gcov-profile/78783
11475 * gcov-tool.c (gcov_output_files): Validate that destination
11476 file is either removed by the tool or by a user.
11478 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
11479 Guy Benyei <guybe@mellanox.com>
11481 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
11482 block, and do not negate it, the stored id is already negative.
11484 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
11486 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
11488 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
11491 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
11492 masks of options that should be turned off if the VSX vector
11493 options are turned off.
11494 (OTHER_P8_VECTOR_MASKS): Likewise.
11495 (OTHER_VSX_VECTOR_MASKS): Likewise.
11496 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
11497 rs6000_disable_incompatible_switches to validate no type switches
11499 (rs6000_incompatible_switch): New function to disallow turning on
11500 other vector options if -mno-vsx, -mno-power8-vector, or
11501 -mno-power9-vector are specified.
11503 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11505 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
11507 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11509 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
11510 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
11511 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
11512 (ARG_POINTER_CFA_OFFSET): Likewise.
11514 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11516 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
11517 conditions to take advantage of various optimizations.
11519 2017-04-13 Jeff Law <law@redhat.com>
11521 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11522 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11523 (zero_extendsidi2_dext): Likewise.
11525 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11528 * fold-const.c (fold_ternary_loc): Revert
11529 use op0 instead of fold_convert_loc (loc, type, arg0) part of
11532 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
11534 PR rtl-optimization/80343
11535 * lra-remat.c (update_scratch_ops): Assign original hard reg to
11536 new scratch pseudo.
11538 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
11541 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
11542 to ubsan_encode_value.
11544 2017-04-13 Jeff Law <law@redhat.com>
11546 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
11547 appearing in DEBUG_INSNs.
11549 2017-04-13 Martin Liska <mliska@suse.cz>
11551 PR gcov-profile/80413
11552 * gcov-io.c (gcov_write_string): Copy to buffer just when
11553 allocated size is greater than zero.
11555 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11558 * dwarf2out.c (decls_for_scope): Ignore declarations of
11559 current_function_decl in BLOCK_NONLOCALIZED_VARS.
11561 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
11564 * ipa-visibility.c (non_local_p): Fix typos.
11565 (localize_node): When localizing symbol in same comdat group,
11566 dissolve the group only when we know external symbols are going
11568 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
11570 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11572 PR tree-optimization/79390
11573 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
11574 order does not result in usable sequence, retry with reversed operand
11580 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
11581 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
11582 op0 instead of fold_convert_loc (loc, type, arg0).
11584 2017-04-12 Jeff Law <law@redhat.com>
11586 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
11587 has a delay slot in the generated code.
11589 * config/cris/cris.md (cris_preferred_reload_class): Return
11590 GENNONACR_REGS rather than GENERAL_REGS.
11592 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11595 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
11596 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
11597 signedness of the result type.
11599 2017-04-12 Richard Biener <rguenther@suse.de>
11600 Jeff Law <law@redhat.com>
11602 PR tree-optimization/80359
11603 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
11604 trim stores to TARGET_MEM_REFs.
11606 2017-04-12 Richard Biener <rguenther@suse.de>
11608 PR tree-optimization/79390
11609 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
11610 threading case even more.
11612 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
11615 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
11616 for quad_address_p for TImode, instead of just not indexed_address.
11618 2017-04-12 Richard Biener <rguenther@suse.de>
11619 Bernd Edlinger <bernd.edlinger@hotmail.de>
11621 PR middle-end/79671
11622 * alias.c (component_uses_parent_alias_set_from): Handle
11623 TYPE_TYPELESS_STORAGE.
11624 (get_alias_set): Likewise.
11625 * tree-core.h (tree_type_common): Add typeless_storage flag.
11626 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
11627 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
11628 for types containing members with TYPE_TYPELESS_STORAGE.
11629 (place_field): Likewise.
11630 (layout_type): Likewise for ARRAY_TYPE.
11631 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
11632 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
11633 TYPE_TYPELESS_STORAGE.
11634 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
11636 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11639 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
11640 first argument to type.
11642 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11646 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
11647 CONST0_RTX (mode) rather than const0_rtx where appropriate.
11648 (rs6000_expand_binop_builtin): Likewise.
11649 (rs6000_expand_ternop_builtin): Likewise; also add missing
11650 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
11651 vshasigma built-ins.
11652 * doc/extend.texi: Document that vec_xxpermdi's third argument
11653 must be a constant.
11655 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
11657 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
11658 Use shift_const cost parameter when calculating gain of STV shifts.
11660 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
11662 PR rtl-optimization/70478
11663 * lra-constraints.c (process_alt_operands): Check memory for
11664 disfavoring memory insn operand.
11666 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11668 PR middle-end/80100
11669 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
11670 left shift in unsigned HOST_WIDE_INT type.
11672 PR rtl-optimization/80385
11673 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
11674 (not (neg X)) into (plus X -1) for complex or non-integral modes.
11677 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
11678 if they have any depend clauses.
11680 2017-04-11 Martin Liska <mliska@suse.cz>
11683 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
11684 * ipa-split.c (split_function): Create a local comdat symbol
11685 if caller is in a comdat group.
11687 2017-04-11 Martin Liska <mliska@suse.cz>
11690 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
11693 2017-04-11 Martin Sebor <msebor@redhat.com>
11695 PR middle-end/80364
11696 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
11697 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
11699 (directive::set_width, directive::set_precision, format_character):
11701 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
11704 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
11707 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
11708 conflict, set target->arch_name instead of target->cpu_name.
11710 2017-04-11 Richard Biener <rguenther@suse.de>
11712 PR tree-optimization/80374
11713 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
11714 build_zero_cst, remove fold_convertible_p check again.
11716 2017-04-11 Martin Liska <mliska@suse.cz>
11719 * ubsan.c (instrument_object_size): Do not instrument register
11722 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11725 * config/i386/i386-builtin-types.def
11726 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
11727 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
11728 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
11729 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
11730 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
11731 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
11732 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
11733 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
11734 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
11735 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
11736 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
11737 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
11738 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
11739 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
11740 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
11741 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
11742 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
11743 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
11744 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
11745 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
11746 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
11747 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
11748 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
11749 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
11750 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
11751 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
11752 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
11753 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
11754 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
11755 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
11756 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
11757 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
11758 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
11759 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
11760 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
11761 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
11762 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
11763 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
11764 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
11765 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
11766 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
11767 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
11768 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
11769 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
11770 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
11771 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
11773 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
11774 flag to second_arg_count, handle 4 argument function type _COUNT
11775 aliases, handle second_arg_count on second argument rather than last.
11777 2017-04-10 Jeff Law <law@redhat.com>
11779 PR tree-optimization/80374
11780 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
11781 record anything if we can not convert integer_zero_node to the
11784 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
11787 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11788 Enhance special handling given to the TARGET_P9_MINMAX option in
11789 relation to certain other options.
11791 2017-04-10 Bin Cheng <bin.cheng@arm.com>
11793 PR tree-optimization/80153
11794 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
11795 remove POINTER_PLUS_EXPR's base part directly, rather than through
11798 2017-04-10 Richard Biener <rguenther@suse.de>
11799 Bin Cheng <bin.cheng@arm.com>
11801 PR tree-optimization/80153
11802 * tree-affine.c (aff_combination_to_tree): Get base pointer from
11803 the first element of pointer type aff_tree. Build result expr in
11805 (add_elt_to_tree): Convert to type unconditionally. Remove other
11806 fold_convert calls.
11807 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
11808 (rewrite_use_nonlinear_expr): Check invariant using iv information.
11810 2017-04-10 Richard Biener <rguenther@suse.de>
11812 * tree-ssa-structalias.c (find_func_aliases): Properly handle
11815 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
11817 PR rtl-optimization/70478
11818 * lra-constraints.c (curr_small_class_check): New.
11819 (update_and_check_small_class_inputs): New.
11820 (process_alt_operands): Update curr_small_class_check. Disfavor
11821 alternative insn memory operands. Check available regs for small
11824 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
11827 * config/mips/mips.opt (-mvirt): Update description.
11828 * doc/invoke.texi (-mvirt): Likewise.
11830 2017-04-10 Richard Biener <rguenther@suse.de>
11832 PR middle-end/80362
11833 * fold-const.c (fold_binary_loc): Look at unstripped ops when
11834 looking for NEGATE_EXPR in -A / -B to A / B folding.
11836 2017-04-10 Martin Liska <mliska@suse.cz>
11838 PR gcov-profile/80224
11839 * gcov.c (print_usage): Fix usage string.
11840 (get_gcov_intermediate_filename): Remove.
11841 (output_gcov_file): Use both for normal and intermediate format.
11842 (generate_results): Do not initialize special file for
11843 intermediate format.
11845 2017-04-10 Richard Biener <rguenther@suse.de>
11847 PR tree-optimization/80304
11848 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
11851 2017-04-10 Nathan Sidwell <nathan@acm.org>
11854 * config/rs6000/rs6000.c (rs6000_vector_type): New.
11855 (rs6000_init_builtins): Use it.
11857 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11859 * config/arm/arm.md (<mrc>): Add mode to SET source.
11860 (<mrrc>): Likewise.
11862 2017-04-10 Richard Biener <rguenther@suse.de>
11864 PR middle-end/80344
11865 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
11867 2017-04-10 Jakub Jelinek <jakub@redhat.com>
11870 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
11871 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
11872 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
11873 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
11874 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
11875 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
11876 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
11877 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
11878 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
11879 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
11880 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
11881 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
11882 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
11883 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
11884 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
11885 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
11886 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
11887 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
11888 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
11889 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
11890 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
11891 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
11892 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
11894 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
11896 PR rtl-optimization/70478
11897 * lra-constraints.c: Reverse the last patch.
11899 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
11901 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
11902 Add comment for WCHAR_T.
11904 2017-04-08 Martin Liska <mliska@suse.cz>
11907 2017-04-07 Martin Liska <mliska@suse.cz>
11910 * ipa-split.c (split_function): Add function part to a same comdat
11913 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11916 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
11918 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
11920 * rs6000/rs6000.c (vec_load_pendulum): Rename...
11921 (vec_pairing): ...to this.
11922 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
11923 (rs6000_sched_init): Adjust for name change.
11924 (struct rs6000_sched_context): Likewise.
11925 (rs6000_init_sched_context): Likewise.
11926 (rs6000_set_sched_context): Likewise.
11928 2017-04-07 Jakub Jelinek <jakub@redhat.com>
11934 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
11936 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
11937 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
11938 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
11940 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
11942 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
11944 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
11946 PR rtl-optimization/70703
11947 * ira-color.c (update_conflict_hard_regno_costs): Use
11948 int64_t instead of HOST_WIDE_INT.
11950 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
11952 PR rtl-optimization/70478
11953 * lra-constraints.c (process_alt_operands): Disfavor alternative
11954 insn memory operands.
11956 2017-04-07 Jeff Law <law@redhat.com>
11958 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
11959 CALL and NOTE_INSN_CALL_ARG_LOCATION.
11961 2017-04-07 Martin Liska <mliska@suse.cz>
11964 * config/aarch64/aarch64.c (aarch64_process_target_attr):
11965 Show error message instead of an ICE.
11967 2017-04-07 Martin Liska <mliska@suse.cz>
11970 * ipa-split.c (split_function): Add function part to a same comdat
11973 2017-04-07 Richard Biener <rguenther@suse.de>
11975 PR middle-end/80341
11976 * tree.c (get_unwidened): Also handle ! for_type case for
11978 * convert.c (do_narrow): Split out from ...
11979 (convert_to_integer_1): ... here. Do not pass final truncation
11980 type to get_unwidened for TRUNC_DIV_EXPR.
11982 2017-04-07 Richard Biener <rguenther@suse.de>
11984 * tree-affine.c (wide_int_ext_for_comb): Take type rather
11986 (aff_combination_const): Adjust.
11987 (aff_combination_scale): Likewise.
11988 (aff_combination_add_elt): Likewise.
11989 (aff_combination_add_cst): Likewise.
11990 (aff_combination_convert): Likewise.
11991 (add_elt_to_tree): Likewise. Remove unused argument.
11992 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
11994 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
11996 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
11998 * config/arm/arm.c (arm_default_short_enums): Use
11999 ARM_DEFAULT_SHORT_ENUMS.
12000 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
12002 2017-04-06 Jakub Jelinek <jakub@redhat.com>
12005 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
12006 members with redundant out-of-class redeclaration.
12008 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
12011 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
12012 * config/i386/i386.md (*zero_extendsidi2):
12013 Add (?*x,*x) and (?*v,*v) alternatives.
12015 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
12018 * config/i386/i386.c (ix86_expand_builtin)
12019 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
12020 mode from insn data. Convert operands to insn operand mode.
12021 Copy operands that don't satisfy insn predicate to a register.
12023 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
12025 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
12028 2017-04-06 Richard Biener <rguenther@suse.de>
12030 PR tree-optimization/80334
12031 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
12032 preserve alignment of accesses.
12034 2017-04-06 Richard Biener <rguenther@suse.de>
12036 PR tree-optimization/80262
12037 * tree-sra.c (build_ref_for_offset): Preserve address-space
12039 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
12040 Drop useless address-space information on MEM_REF offsets.
12042 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
12044 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
12046 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
12048 PR rtl-optimization/70703
12049 * ira-color.c (update_conflict_hard_regno_costs): Use
12050 HOST_WIDE_INT instead of long.
12052 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
12055 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
12056 not defined for x86_64 target. Add -mmmx target option when __SSE2__
12058 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
12059 for x86_64 target. Handle -m3dnowa option.
12061 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
12063 PR rtl-optimization/70703
12064 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
12065 (update_conflict_hard_regno_costs): Use long instead of unsigned
12066 arithmetic for cost calculation.
12068 2017-04-05 Jakub Jelinek <jakub@redhat.com>
12069 Bernd Edlinger <bernd.edlinger@hotmail.de>
12072 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
12075 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
12078 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
12079 ptr_mode with Pmode throughout.
12080 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
12081 into probe_stack_range and use DImode.
12083 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
12086 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
12087 call_eh_return is true.
12089 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12091 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
12092 Initialize last_match_fntype_index.
12094 2017-04-05 Jakub Jelinek <jakub@redhat.com>
12097 * tree-nvr.c: Include internal-fn.h.
12098 (pass_return_slot::execute): Ignore internal calls without
12101 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12102 Richard Biener <rguenther@suse.de>
12105 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
12106 captures used multiple times, except for the last use.
12107 * generic-match-head.c: Include gimplify.h.
12109 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12111 PR tree-optimization/79390
12112 * target.h (struct noce_if_info): Declare.
12113 * targhooks.h (default_noce_conversion_profitable_p): Declare.
12114 * target.def (noce_conversion_profitable_p): New target hook.
12115 * ifcvt.h (struct noce_if_info): New type, moved from ...
12116 * ifcvt.c (struct noce_if_info): ... here.
12117 (noce_conversion_profitable_p): Renamed to ...
12118 (default_noce_conversion_profitable_p): ... this. No longer
12120 (noce_try_store_flag_constants, noce_try_addcc,
12121 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
12122 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
12123 instead of noce_conversion_profitable_p.
12124 * config/i386/i386.c: Include ifcvt.h.
12125 (ix86_option_override_internal): Don't override
12126 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
12127 (ix86_noce_conversion_profitable_p): New function.
12128 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
12129 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
12130 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
12131 * doc/tm.texi: Regenerated.
12133 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12135 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
12138 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
12141 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
12142 instructions for small multiply cores.
12144 2017-04-04 Jeff Law <law@redhat.com>
12146 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
12148 (mips_expand_vec_perm_const): Initialize elements in orig_perm
12149 that are not set by the loop over the elements.
12151 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12154 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
12155 int mode, convert_modes it to mode as unsigned, otherwise use
12156 lowpart_subreg to mode rather than SImode.
12157 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
12158 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
12159 Use DImode instead of SImode for the shift count operand.
12160 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
12163 2017-04-04 Richard Biener <rguenther@suse.de>
12165 PR middle-end/80281
12166 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
12167 arithmetic done for the negate or the plus. Simplify.
12168 (A - (-B) -> A + B): Likewise.
12169 * fold-const.c (split_tree): Make sure to not negate pointers.
12171 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
12173 PR rtl-optimization/60818
12174 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
12175 a compare of comparisons with the thing compared if this results
12176 in a different machine mode.
12178 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
12180 * alias.c (base_alias_check): Fix typo in comment.
12181 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
12182 * cgraphunit.c (symbol_table::compile): Likewise.
12183 * collect2.c (maybe_run_lto_and_relink): Likewise.
12184 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
12185 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
12186 * config/avr/avr.c (avr_map_op_t): Likewise.
12187 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
12188 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
12189 * config/epiphany/epiphany.md (movcc): Likewise.
12190 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
12191 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
12193 * config/mips/mips.c (mips_save_restore_reg): Likewise.
12194 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
12195 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
12196 * config/sh/sh.c (sh_rtx_costs): Likewise.
12197 * fold-const.c (fold_truth_andor): Likewise.
12198 * genautomata.c (collapse_flag): Likewise.
12199 * gengtype.h (struct type::u::s): Likewise.
12200 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
12201 * input.c (FORMAT_AMOUNT): Likewise.
12202 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
12203 (known_aggs_to_agg_replacement_list): Likewise.
12204 * ipa-inline-analysis.c: Likewise.
12205 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
12206 * ipa-polymorphic-call.c
12207 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
12208 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
12209 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
12211 * modulo-sched.c (apply_reg_moves): Likewise.
12212 * omp-expand.c (build_omp_regions_1): Likewise.
12213 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
12214 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
12215 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
12216 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
12217 * value-prof.c: Likewise.
12218 * var-tracking.c (val_reset): Likewise.
12220 2017-04-03 Richard Biener <rguenther@suse.de>
12222 PR tree-optimization/80275
12223 * fold-const.c (split_address_to_core_and_offset): Handle
12226 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
12228 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
12229 descriptors is at least equal to that of functions.
12231 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
12233 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
12235 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
12238 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
12239 (mov<IMOD4:mode>): New expander.
12240 (*mov<IMOD4:mode>_internal): New insn and split pattern.
12242 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
12244 PR rtl-optimization/79405
12245 * fwprop.c (propagations_left): New variable.
12246 (forward_propagate_into): Decrement it.
12247 (fwprop_init): Initialize it.
12248 (fw_prop): If the variable has reached zero, stop propagating.
12249 (fwprop_addr): Ditto.
12251 2017-03-31 Jakub Jelinek <jakub@redhat.com>
12254 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
12255 a FUNCTION_DECL, pass it as decl instead of origin to
12258 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
12260 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
12263 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
12266 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
12267 TARGET_VSX_SMALL_INTEGER.
12269 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12271 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
12272 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
12274 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
12276 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
12277 extraction from odd-numbered MSA register.
12279 2017-03-31 Jakub Jelinek <jakub@redhat.com>
12281 PR middle-end/80173
12282 * expmed.c (store_bit_field_1): Don't attempt to create
12283 a word subreg out of hard registers wider than word if they
12284 have HARD_REGNO_NREGS of 1 for their mode.
12286 PR middle-end/80163
12287 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
12288 conversions to integer types wider than word and pointer.
12291 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
12292 (rtx_equal_for_cselib_p): Pass 0 to it.
12293 * cselib.c (cselib_hasher::equal): Likewise.
12294 (rtx_equal_for_cselib_1): Add depth argument. If depth
12295 is 128, don't look up VALUE locs and punt. Increment
12296 depth in recursive calls when walking VALUE locs.
12298 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
12300 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
12301 (make_gcov_file_name): Use the canonical path name for generating
12303 (read_line): Fix handling of files with ascii null bytes.
12305 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
12307 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
12308 to initialise a vector register instead
12309 of using a const_int.
12311 2017-03-30 Jakub Jelinek <jakub@redhat.com>
12313 PR translation/80189
12314 * gimplify.c (omp_default_clause): Use %qs instead of %s in
12315 diagnostic messages.
12317 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
12320 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
12321 (dfp_diex_<mode>): Update mode of operand 1.
12322 * doc/extend.texi (dxex, dxexq): Document change to return type.
12323 (diex, diexq): Document change to argument type.
12325 2017-03-30 Martin Jambor <mjambor@suse.cz>
12328 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
12329 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
12330 it reflects the signature changes performed at the callee side.
12331 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
12332 to cgraph_build_function_type_skip_args.
12333 (build_function_decl_skip_args): Adjust call to the above function.
12335 2017-03-30 Jakub Jelinek <jakub@redhat.com>
12338 * config/i386/sse.md
12339 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
12340 register as dest whenever it is a MEM not rtx_equal_p to the
12341 corresponding dup operand, and when forcing into reg move the
12342 reg into the memory afterwards.
12343 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
12344 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
12345 for the force_reg mode.
12346 (avx512vl_vextractf128<mode>): Use register as dest either
12347 always when a MEM, or when it is a MEM not rtx_equal_p to the
12348 corresponding dup operand, or even not when it is a CONST_VECTOR
12349 depending on the mode and lo vs. hi.
12350 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
12352 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
12353 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
12354 Likewise. Require that operands[2] is even.
12355 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
12356 Remove extraneous parens. Require that operands[2] is a multiple
12358 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
12359 operands[0] is a MEM if <mask_applied>, the predicates/constraints
12360 disallow memory then.
12362 2017-03-30 Richard Biener <rguenther@suse.de>
12364 PR tree-optimization/77498
12365 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
12366 to non-constants over backedges.
12368 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
12370 PR rtl-optimization/80233
12371 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
12372 as last_combined_insn. Do not test for BARRIER_P separately.
12374 2017-03-29 Andreas Schwab <schwab@suse.de>
12377 * calls.c (prepare_call_address): Convert funexp to Pmode before
12378 copying to temp reg.
12380 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12382 PR tree-optimization/80158
12383 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
12384 Handle possible future case of more than one alternate
12386 (replace_rhs_if_not_dup): Likewise.
12387 (replace_one_candidate): Likewise.
12389 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
12391 PR rtl-optimization/80193
12392 * ira.c (ira): Do not check allocation for LRA.
12394 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
12396 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
12397 (nvptx_output_simt_exit): Declare.
12398 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
12399 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
12400 (init_softstack_frame): Move initialization of crtl->is_leaf to...
12401 (nvptx_declare_function_name): ...here. Emit declaration of local
12402 memory space buffer for omp_simt_enter insn.
12403 (nvptx_output_unisimt_switch): New.
12404 (nvptx_output_softstack_switch): New.
12405 (nvptx_output_simt_enter): New.
12406 (nvptx_output_simt_exit): New.
12407 * config/nvptx/nvptx.h (struct machine_function): New fields
12408 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
12409 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
12410 (UNSPECV_SIMT_EXIT): Ditto.
12411 (omp_simt_enter_insn): New insn.
12412 (omp_simt_enter): New expansion.
12413 (omp_simt_exit): New insn.
12414 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
12416 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
12417 (expand_GOMP_SIMT_ENTER_ALLOC): New.
12418 (expand_GOMP_SIMT_EXIT): New.
12419 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
12420 (GOMP_SIMT_ENTER_ALLOC): Ditto.
12421 (GOMP_SIMT_EXIT): Ditto.
12422 * target-insns.def (omp_simt_enter): New insn.
12423 (omp_simt_exit): Ditto.
12424 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
12426 (lower_rec_simd_input_clauses): Implement SIMT privatization.
12427 (lower_rec_input_clauses): Likewise.
12428 (lower_lastprivate_clauses): Handle SIMT privatization.
12430 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
12431 (ompdevlow_adjust_simt_enter): New.
12432 (find_simtpriv_var_op): New.
12433 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
12434 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
12436 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
12437 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
12438 (copy_decl_for_dup_finish): Ditto.
12440 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
12442 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
12445 * config/i386/i386.c (ix86_option_override_internal): Always
12446 allow -mpreferred-stack-boundary=3 for 64-bit targets.
12448 2017-03-28 Bin Cheng <bin.cheng@arm.com>
12450 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
12452 2017-03-28 Bin Cheng <bin.cheng@arm.com>
12454 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
12455 mark new edge's irreducible flag accordign to it.
12456 (vect_do_peeling): Check loop preheader edge's irreducible flag
12457 and pass it to function slpeel_add_loop_guard.
12459 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
12461 PR tree-optimization/80218
12462 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
12463 Update block frequencies and counts.
12465 2017-03-28 Richard Biener <rguenther@suse.de>
12467 PR tree-optimization/78644
12468 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
12469 of a simplification result we may not use it at all.
12471 2017-03-28 Richard Biener <rguenther@suse.de>
12474 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
12475 without arguments, generate default definition of a SSA name.
12477 2017-03-28 Richard Biener <rguenther@suse.de>
12479 PR middle-end/80222
12480 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
12481 TYPE_REF_CAN_ALIAS_ALL references.
12482 * fold-const.c (fold_indirect_ref_1): Likewise.
12484 2017-03-28 Martin Liska <mliska@suse.cz>
12487 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
12488 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
12490 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12491 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
12493 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
12494 (EXTRA_SPECS): Define.
12495 (SUBTARGET_EXTRA_SPECS): Likewise.
12496 (SUBTARGET_CPP_SPEC): Likewise.
12497 * config/arc/elf.h (EXTRA_SPECS): Renamed to
12498 SUBTARGET_EXTRA_SPECS.
12499 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
12501 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12503 * config/arc/simdext.md (vst64_insn): Update pattern.
12504 (vld32wh_insn): Likewise.
12505 (vld32wl_insn): Likewise.
12506 (vld64_insn): Likewise.
12507 (vld32_insn): Likewise.
12509 2017-03-28 Marek Polacek <polacek@redhat.com>
12512 * fold-const.c (fold_comparison): Use protected_set_expr_location
12513 instead of SET_EXPR_LOCATION.
12515 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
12517 * tree.c (add_expr): Avoid name lookup warning.
12519 2017-03-27 Jeff Law <law@redhat.com>
12521 PR tree-optimization/80216
12522 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
12523 function name. Limit recursion depth.
12524 (record_temporary_equivalences): Corresponding changes.
12526 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
12528 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
12531 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12534 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
12536 * cfgcleanup.c (reg_note_cfa_p): New array.
12537 (insns_have_identical_cfa_notes): New function.
12538 (old_insns_match_p): Don't cross-jump in between /f
12539 and non-/f instructions. If both i1 and i2 are frame related,
12540 verify all CFA notes, their order and content.
12542 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
12545 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
12546 HImode and SImode with zero extend to DImode to one insn.
12547 (bswap<mode>2_extenddi): Likewise.
12548 (bswapsi2_extenddi): Likewise.
12549 (bswaphi2_extendsi): Likewise.
12550 (bswaphi2): Combine bswap HImode and SImode into one insn.
12551 Separate memory insns from swapping register.
12552 (bswapsi2): Likewise.
12553 (bswap<mode>2): Likewise.
12554 (bswaphi2_internal): Delete, no longer used.
12555 (bswapsi2_internal): Likewise.
12556 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
12557 store, and gpr<-gpr swap insns.
12558 (bswap<mode>2_store): Likewise.
12559 (bswaphi2_reg): Register only splitter, combine with the splitter.
12560 (bswaphi2 splitter): Likewise.
12561 (bswapsi2_reg): Likewise.
12562 (bswapsi2 splitter): Likewise.
12563 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
12564 the insns into load, store, and register/register insns.
12565 (bswapdi2_ldbrx): Likewise.
12566 (bswapdi2_load): Likewise.
12567 (bswapdi2_store): Likewise.
12568 (bswapdi2_reg): Likewise.
12570 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
12572 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
12573 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
12575 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12578 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
12580 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
12581 special handling for target option conflicts between dform
12582 options (-mpower9-dform, -mpower9-dform-vector,
12583 -mpower9-dform-scalar) and -mno-direct-move.
12585 2017-03-27 Richard Biener <rguenther@suse.de>
12587 PR tree-optimization/80181
12588 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
12590 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12592 * config/arc/predicates.md (move_double_src_operand): Replace the
12593 call to move_double_src_operand with a call to address_operand.
12595 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12597 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
12598 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
12599 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
12601 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12603 * config/arc/predicates.md (long_immediate_loadstore_operand):
12604 Consider scaled addresses cases.
12606 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12608 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
12609 restored when in interrupt.
12610 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
12611 doesn't have delay slot.
12613 2017-03-27 Richard Biener <rguenther@suse.de>
12616 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
12617 inlined thunk clones.
12619 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12622 * asan.c (instrument_derefs): Copy over last operand from
12623 original COMPONENT_REF to the new COMPONENT_REF with
12624 DECL_BIT_FIELD_REPRESENTATIVE.
12625 * ubsan.c (instrument_object_size): Likewise.
12627 2017-03-27 Richard Biener <rguenther@suse.de>
12629 PR tree-optimization/80170
12630 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
12631 sure DR/SCEV didnt fold in constants we do not see when looking
12632 at the reference base alignment.
12634 2017-03-27 Richard Biener <rguenther@suse.de>
12636 PR middle-end/80171
12637 * gimple-fold.c (fold_ctor_reference): Properly guard against
12638 NULL return value from canonicalize_constructor_val.
12640 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
12643 * config/i386/i386.c (ix86_expand_builtin)
12644 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
12645 flags reg setting and flags reg using instructions.
12646 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
12647 clobbering instructions to zero extend op2.
12649 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
12651 * doc/install.texi (Configuration) <--with-aix-soname>:
12652 Update link to AIX ld.
12654 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
12656 PR rtl-optimization/80160
12657 PR rtl-optimization/80159
12658 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
12659 reg_alternate_class into account.
12661 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
12664 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
12665 to consider in curr_insn_transform.
12667 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12669 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
12670 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
12671 and emit_mode_inner.
12673 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12675 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
12676 argument to the overloaded builtin variants. Use the new flag to
12677 deprecate certain builtin variants.
12678 * config/s390/s390-builtin-types.def: Add new builtin types.
12679 * config/s390/s390-builtins.h: Support new flags field for
12680 overloaded builtins.
12681 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
12682 (s390_macro_to_expand): Enable vector float data type.
12683 (s390_cpu_cpp_builtins_internal): Indicate support of the new
12684 builtins by incrementing the __VEC__ version number.
12685 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
12687 (s390_resolve_overloaded_builtin): Emit error messages depending
12688 on the builtin flags.
12689 * config/s390/s390.c (s390_expand_builtin): Support additional
12690 flags argument. Change error message to match the messages
12691 emitted in s390-c.c.
12692 * config/s390/s390.md: New UNSPEC_* constants.
12693 (op_type): Add new instruction types.
12694 * config/s390/vecintrin.h: Add new builtins and test data class
12696 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
12697 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
12698 (VEC_INEXACT, VEC_NOINEXACT): New constants.
12699 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
12700 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
12701 ("vec_mergel<mode>"): V_HW -> VEC_HW.
12703 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
12704 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
12705 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
12706 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
12708 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
12709 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
12710 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
12711 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
12713 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
12714 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
12715 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
12716 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
12717 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
12718 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
12719 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
12721 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
12722 ("vec_scatter_element<V_HW_4:mode>_DI")
12723 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
12724 ("vec_fpint<mode>", "vflls")
12725 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
12726 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
12727 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
12728 ("*vec_cmphe<mode>_cc"): ... these.
12730 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
12731 mode constant instead of magic value.
12733 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12735 * config/s390/s390.c (s390_expand_vec_compare): Support other
12736 vector floating point modes than just V2DF.
12737 (s390_expand_vcond): Likewise.
12738 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
12739 (s390_cannot_change_mode_class): Prevent mode changes between TF
12740 and V1TF in vector registers.
12741 * config/s390/s390.md (DF, SF): New mode attributes.
12742 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
12743 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
12744 SFmode support for VRs.
12745 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
12747 (VFT, VF_HW): New mode iterators.
12748 (vw, sdx): New mode attributes.
12749 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
12750 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
12751 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
12752 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
12753 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
12754 also the new vector floating point modes. Renaming to ...
12756 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
12757 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
12758 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
12759 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
12760 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
12761 ("vec_unordered<mode>"): ... these.
12763 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
12764 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
12765 ("*vec_extendv2df"): New insn definitions.
12767 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12769 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
12770 ("mulditi3_2", "*muldi3_sign"): New patterns.
12771 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
12772 rename the pattern definition.
12774 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12776 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
12778 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
12780 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12782 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
12783 instruction if possible.
12784 * config/s390/vector.md (vec_halfnumelts): New mode
12786 ("*vec_vllezlf<mode>"): New pattern.
12788 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12790 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
12791 ("popcountv4si2", "popcountv2di2"): Rename to ...
12792 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
12793 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
12795 ("popcount<mode>2_vxe"): New pattern.
12797 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12799 * common/config/s390/s390-common.c (processor_flags_table): Add
12801 * config.gcc: Add arch12.
12802 * config/s390/driver-native.c (s390_host_detect_local_cpu):
12803 Default to arch12 for unknown CPU model numbers.
12804 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
12805 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
12806 PROCESSOR_max sanity check.
12807 * config/s390/s390-opts.h (enum processor_type): Add
12809 * config/s390/s390.c (processor_table): Add arch12.
12810 (s390_expand_builtin): Add check for B_VXE flag.
12811 (s390_issue_rate): Add PROCESSOR_ARCH12.
12812 (s390_get_sched_attrmask): Likewise.
12813 (s390_get_unit_mask): Likewise.
12814 (s390_sched_score): Enable z13 scheduling for arch12.
12815 (s390_sched_reorder): Likewise.
12816 (s390_sched_variable_issue): Likewise.
12817 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
12819 (s390_tune_attr): Use z13 scheduling also for arch12.
12820 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
12821 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
12822 (TARGET_VXE_P): New macros.
12823 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
12824 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
12825 * config/s390/s390.opt: Add arch12 as processor_type.
12827 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12829 * config/s390/s390.md
12830 ("fixuns_truncdddi2", "fixuns_trunctddi2")
12831 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
12832 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
12834 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
12835 Rename expanders to ...
12837 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
12838 ("fixuns_truncdddi2_emu"): ... these.
12840 ("fixuns_trunc<mode>si2_emu"): New expander.
12842 ("*fixuns_truncdfdi2_z13"): Rename to ...
12843 ("*fixuns_truncdfdi2_vx"): ... this.
12845 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12847 * config/s390/2964.md: Remove the single element vector compare
12848 instructions which are no longer used.
12849 * config/s390/s390.c (s390_select_ccmode): Remove handling of
12851 (s390_canonicalize_comparison): Remove handling of DFmode
12853 (s390_expand_vec_compare_scalar): Remove function.
12854 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
12855 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
12857 ("*cmp<mode>_ccs"): Add wfcdb instruction.
12859 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12861 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
12863 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
12864 will anyway by matched by mov<mode>_64dfp.
12866 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12868 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
12869 vlef/vstef. Add missing operand to vleif.
12871 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12873 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
12874 pair for all vector types with 64 bit elements.
12875 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
12876 * config/s390/vector.md (V_HW_64): ... here.
12877 (V_128_NOSINGLE): New mode iterator.
12878 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
12879 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
12880 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
12881 ("*vec_load_pairv2di"): Change to ...
12882 ("*vec_load_pair<mode>"): ... this one.
12884 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12886 * config/s390/constraints.md: Add comments.
12887 (jKK): Reject element sizes > 8 bytes.
12888 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
12890 * config/s390/s390.md: Add the s_operand checks formerly in
12891 s390_split_ok_p to various splitters where they are still
12893 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
12894 for 128 bit vectors. Plus two splitters.
12896 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12898 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
12901 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12904 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
12905 error if the boundary argument is not constant.
12907 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12909 PR rtl-optimization/80112
12910 * loop-doloop.c (doloop_condition_get): Don't check condition
12911 if cmp isn't SET with IF_THEN_ELSE src.
12913 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12915 PR tree-optimization/80158
12916 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
12917 replacing a candidate statement, also replace it for the
12918 candidate's alternate interpretation.
12919 (replace_rhs_if_not_dup): Likewise.
12920 (replace_one_candidate): Likewise.
12922 2017-03-24 Richard Biener <rguenther@suse.de>
12924 PR tree-optimization/80167
12925 * graphite-isl-ast-to-gimple.c
12926 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
12928 (translate_isl_ast_to_gimple::get_rename): Likewise.
12930 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
12932 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
12933 handling of certain combinations of target options, including the
12934 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
12935 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
12937 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12940 * config/arm/arm.md (*load_multiple): Add reload_completed to
12941 matching condition.
12943 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12944 Richard Biener <rguenth@suse.de>
12946 PR tree-optimization/79908
12947 PR tree-optimization/80136
12948 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
12949 been cast away, gimplify_and_add suffices.
12951 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
12953 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
12955 2017-03-23 Richard Biener <rguenther@suse.de>
12957 PR tree-optimization/80032
12958 * gimplify.c (gimple_push_cleanup): Forced unconditional
12959 cleanups still have to go to the conditional_cleanups
12962 2017-03-22 Jakub Jelinek <jakub@redhat.com>
12964 PR tree-optimization/80072
12965 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
12967 (next_operand_entry_id): Change type to unsigned int.
12968 (sort_by_operand_rank): Make sure to return the right return value
12969 even if unsigned fields are bigger than INT_MAX.
12970 (struct oecount): Change cnt and id type to unsigned int.
12971 (oecount_hasher::equal): Formatting fix.
12972 (oecount_cmp): Make sure to return the right return value
12973 even if unsigned fields are bigger than INT_MAX.
12974 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
12977 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
12978 TREE_READONLY on result if writing it more than once.
12981 * doc/invoke.texi (-fsanitize=thread): Document that with
12982 -fnon-call-exceptions atomics are not able to throw
12986 * tsan.c: Include tree-eh.h.
12987 (instrument_builtin_call): Call maybe_clean_eh_stmt or
12988 maybe_clean_or_replace_eh_stmt where needed.
12989 (instrument_memory_accesses): Add cfg_changed argument.
12990 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
12991 if it returned true.
12992 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
12994 PR rtl-optimization/63191
12995 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
12996 wrapper function, moved the whole old content into ...
12997 (ix86_delegitimize_address_1): ... this. New inline function.
12998 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
12999 true as last argument instead of ix86_delegitimize_address.
13001 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
13003 * config/aarch64/aarch64.c (generic_branch_cost): Copy
13004 cortexa57_branch_cost.
13006 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
13008 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
13010 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13013 * doc/md.texi (Constraints): Document wA constraint.
13014 * config/rs6000/constraints.md (wA): New.
13015 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
13016 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
13017 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
13018 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
13020 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
13023 * gimplify.c (is_oacc_declared): New function.
13024 (oacc_default_clause): Use it to set default flags for acc declared
13025 variables inside parallel regions.
13026 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
13027 declared variables.
13028 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
13029 declare attribute to any decl as necessary.
13031 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
13034 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
13035 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
13036 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
13037 (arm_arch_lpae): This.
13038 * config/arm/arm.c (arm_arch7ve): Rename into ...
13039 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
13040 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
13043 2017-03-22 Martin Liska <mliska@suse.cz>
13046 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
13047 error message instead of an ICE.
13049 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13051 * doc/extend.texi (6.11 Additional Floating Types): Revise.
13053 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
13055 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
13057 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
13060 2017-03-21 Martin Sebor <msebor@redhat.com>
13062 * doc/extend.texi: Use "cannot" instead of "can't."
13063 * doc/hostconfig.texi: Same.
13064 * doc/install.texi: Same.
13065 * doc/invoke.texi: Same.
13066 * doc/loop.texi: Same.
13067 * doc/md.texi: Same.
13068 * doc/objc.texi: Same.
13069 * doc/rtl.texi: Same.
13070 * doc/tm.texi: Same.
13071 * doc/tm.texi.in: Same.
13072 * doc/trouble.texi: Same.
13074 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
13077 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
13078 (collect_checksum_attributes): Set it.
13079 (die_checksum_ordered): Use it.
13081 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13083 PR tree-optimization/79908
13084 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
13085 change: For a VA_ARG whose LHS has been cast away, use
13086 force_gimple_operand to construct the side effects.
13088 2017-03-21 David Malcolm <dmalcolm@redhat.com>
13090 PR translation/80001
13091 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
13092 more amenable to translation.
13093 (oacc_loop_auto_partitions): Likewise.
13095 2017-03-21 Marek Polacek <polacek@redhat.com>
13096 Martin Sebor <msebor@redhat.com>
13098 PR tree-optimization/80109
13099 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
13100 on INTEGRAL_TYPE_P.
13102 2017-03-21 Jakub Jelinek <jakub@redhat.com>
13103 Segher Boessenkool <segher@kernel.crashing.org>
13106 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
13107 check reg_used_between_p between insn and one of succ or succ2
13108 depending on if succ is artificial insn not inserted into insn
13111 2017-03-21 Martin Liska <mliska@suse.cz>
13113 PR gcov-profile/80081
13114 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
13115 * doc/gcc.texi: Include gcov-dump stuff.
13116 * doc/gcov-dump.texi: New file.
13118 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
13120 PR rtl-optimization/79150
13121 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
13122 conditional jump, if the jump is the last insn of the loop.
13124 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13125 Richard Biener <rguenth@suse.de>
13127 PR tree-optimization/79908
13128 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
13129 been cast away, use force_gimple_operand to construct the side
13132 2017-03-21 Martin Liska <mliska@suse.cz>
13134 PR libfortran/79956
13135 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
13138 2017-03-21 Brad Spengler <spender@grsecurity.net>
13141 * plugin.c (htab_hash_plugin): New function.
13142 (add_new_plugin): Use it and adjust.
13143 (parse_plugin_arg_opt): Adjust.
13144 (init_one_plugin): Likewise.
13146 2017-03-21 Richard Biener <rguenther@suse.de>
13148 PR tree-optimization/80032
13149 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
13150 if set force the cleanup to happen unconditionally.
13151 (gimplify_target_expr): Push inserted clobbers with force_uncond
13152 to avoid them being removed by control-dependent DCE.
13154 2017-03-21 Richard Biener <rguenther@suse.de>
13156 PR tree-optimization/80122
13157 * tree-inline.c (copy_bb): Do not expans va-arg packs or
13158 va_arg_pack_len when the inlined call stmt requires pack
13160 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
13162 2017-03-21 Jakub Jelinek <jakub@redhat.com>
13165 * tsan.c (instrument_builtin_call): If the memory model argument
13166 is not a constant, assume it is valid.
13169 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
13172 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
13174 PR rtl-optimization/79910
13175 * combine.c (can_combine_p): Do not allow combining an I0 or I1
13176 if its dest is used by an insn before I2 (other than the combined
13177 insns themselves, which are properly handled already).
13179 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
13182 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
13184 * combine.c (record_used_regs): New static function.
13185 (try_combine): Handle situations where there is an additional
13186 instruction between I2 and I3 which needs to have a LOG_LINK
13190 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
13192 * combine.c (try_combine): Delete redundant i1 test. Call
13193 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
13195 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13198 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
13199 alternatives 13/14.
13201 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13203 PR tree-optimization/80054
13204 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
13205 the optimization if a PHI or any of its arguments is not dominated
13206 by the candidate's basis. Use gphi* rather than gimple* as
13208 (replace_profitable_candidates): Clean up a gimple* variable that
13209 should be a gphi* variable.
13211 2017-03-20 Martin Sebor <msebor@redhat.com>
13214 * doc/extend.texi (attribute constructor): Document present limitation.
13216 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
13219 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
13220 __POWER9_VECTOR__ #ifdef control, change template definition to
13221 use Power9-specific built-in function.
13222 (vec_any_eq): Likewise.
13223 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
13224 to control outcomes from this test.
13225 (vector_ae_<mode>p): For VEC_F modes, likewise.
13227 2017-03-20 Ian Lance Taylor <iant@google.com>
13229 * config/i386/i386.c (ix86_function_regparm): Save an extra
13230 register for -fsplit-stack with DECL_STATIC_CHAIN.
13232 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
13235 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
13236 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
13238 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
13240 * config/riscv/riscv.c (riscv_print_operand): Use "fence
13242 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
13245 2017-03-20 Marek Polacek <polacek@redhat.com>
13248 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
13250 2017-03-20 Richard Biener <rguenther@suse.de>
13252 PR tree-optimization/80113
13253 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
13254 allocate extra SSA name for PHI def.
13255 (add_close_phis_to_outer_loops): Likewise.
13256 (add_close_phis_to_merge_points): Likewise.
13257 (copy_loop_close_phi_args): Likewise.
13258 (copy_cond_phi_nodes): Likewise.
13260 2017-03-20 Martin Liska <mliska@suse.cz>
13262 PR middle-end/79753
13263 * tree-chkp.c (chkp_build_returned_bound): Do not build
13264 returned bounds for a LHS that's not a BOUNDED_P type.
13266 2017-03-20 Martin Liska <mliska@suse.cz>
13270 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
13271 COMPLEX_CST and VECTOR_CST.
13273 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13276 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
13277 target operand. A new splitter adds the clobber statement in case
13278 the target operand is dead anyway.
13280 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
13282 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
13283 to age-old versions of binutils and glibc.
13285 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
13287 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
13289 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
13291 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
13293 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
13295 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
13296 requirement for binutils 2.13.
13298 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
13300 * combine.c (try_combine): Delete redundant i1 test. Call
13301 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
13303 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
13305 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
13306 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
13308 <riscv64-*-elf>: Re-arrange section
13309 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
13310 <riscv32-*-linux>: Likewise.
13311 <riscv64-*-elf>: Likewise
13312 <riscv64-*-linux>: Likewise.
13314 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
13317 * aarch64.opt(verbose-cost-dump): Fix typo.
13319 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
13322 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
13323 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
13325 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
13327 * reload.c (find_reloads): When reloading a nonoffsettable address,
13328 use RELOAD_OTHER for it and its address reloads.
13330 PR rtl-optimization/79910
13331 * combine.c (record_used_regs): New static function.
13332 (try_combine): Handle situations where there is an additional
13333 instruction between I2 and I3 which needs to have a LOG_LINK
13336 2017-03-17 Jeff Law <law@redhat.com>
13338 PR tree-optimization/71437
13339 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
13340 conditional in the hash table first.
13341 (vrp_dom_walker::before_dom_children): Extract condition from
13342 ASSERT_EXPR. Record condition, its inverion and any implied
13343 conditions as well.
13345 2017-03-17 Marek Polacek <polacek@redhat.com>
13346 Markus Trippelsdorf <markus@trippelsdorf.de>
13348 PR tree-optimization/80079
13349 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
13352 2017-03-17 Richard Biener <rguenther@suse.de>
13354 PR middle-end/80075
13355 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
13356 Properly verify the LHS before the RHS possibly claims to be
13358 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
13361 2017-03-17 Martin Jambor <mjambor@suse.cz>
13363 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
13364 (List of -O2 options): Likewise.
13365 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
13368 2017-03-17 Tom de Vries <tom@codesourcery.com>
13370 * gcov-dump.c (print_usage): Print bug_report_url.
13372 2017-03-17 Richard Biener <rguenther@suse.de>
13374 PR middle-end/80050
13375 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
13376 (parser::peek): Likewise.
13378 2017-03-17 Richard Biener <rguenther@suse.de>
13380 PR tree-optimization/80048
13381 * sese.c (free_sese_info): Properly release rename_map and
13382 copied_bb_map elements.
13384 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
13386 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
13387 Add linked-list forward and backlinks. Insert on
13388 construction, remove on destruction.
13389 (class pass_store_merging): Add m_stores_head field.
13390 (pass_store_merging::terminate_and_process_all_chains):
13391 Iterate over m_stores_head list.
13392 (pass_store_merging::terminate_all_aliasing_chains):
13394 (pass_store_merging::execute): Check for debug stmts first.
13395 Push new chains onto the m_stores_head stack.
13397 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
13400 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
13401 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
13402 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
13404 2017-03-16 Jeff Law <law@redhat.com>
13406 PR tree-optimization/71437
13407 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
13408 member function. Implementation moved into after_dom_children
13409 member function and into the threader's thread_outgoing_edges
13411 (dom_opt_dom_walker::after_dom_children): Simplify by moving
13412 some code into new thread_outgoing_edges.
13413 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
13414 definition. Simplify marker handling (do it here). Assume we always
13415 have the available expression and the const/copies tables.
13416 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
13418 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
13419 * tree-vrp.c (equiv_stack): No longer file scoped.
13420 (vrp_dom_walker): New class.
13421 (vrp_dom_walker::before_dom_children): New member function.
13422 (vrp_dom_walker::after_dom_children): Likewise.
13423 (identify_jump_threads): Setup domwalker. Use it rather than
13424 walking edges in a random order by hand. Simplify setup/finalization.
13425 (finalize_jump_threads): Remove.
13426 (vrp_finalize): Do not call identify_jump_threads here.
13427 (execute_vrp): Do it here instead and call thread_through_all_blocks
13430 PR tree-optimization/71437
13431 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
13433 (simplify_stmt_for_jump_threading): Add basic_block argument. All
13435 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
13436 (dom_opt_dom_walker::thread_across_edge): Remove
13437 handle_dominating_asserts argument. All callers changed.
13438 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
13439 changes. Remove calls to lhs_of_dominating_assert. Other
13440 uses of handle_dominating_asserts turn into unconditional code
13441 (simplify_control_stmt_condition_1): Likewise.
13442 (simplify_control_stmt_condition): Likewise.
13443 (thread_through_normal_block, thread_across_edge): Likewise.
13444 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
13445 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
13446 object if it is not an SSA_NAME.
13447 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
13448 before calling into the VRP specific simplifiers.
13449 (identify_jump_threads): Remove handle_dominating_asserts
13452 2017-03-16 Jakub Jelinek <jakub@redhat.com>
13455 * tree-diagnostic.c (default_tree_printer): No longer static.
13456 * tree-diagnostic.h (default_tree_printer): New prototype.
13458 2017-03-16 Tamar Christina <tamar.christina@arm.com>
13460 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
13461 Change ins into fmov.
13463 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13465 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
13466 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
13467 Use h_con constraint for operand 1.
13468 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
13469 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
13471 2017-03-15 Jeff Law <law@redhat.com>
13473 PR tree-optimization/71437
13474 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
13475 (record_temporary_equivalences): Use it.
13477 PR tree-optimization/71437
13478 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
13479 tree-ssa-scopedtables.
13480 (lookup_avail_expr, build_and_record_new_cond): Likewise.
13481 (record_conditions, record_cond, vuse_eq): Likewise.
13482 (record_edge_info): Adjust to API tweak of record_conditions.
13483 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
13484 (record_temporary_equivalences, optimize_stmt): Likewise.
13485 (eliminate_redundant_computations): Likewise.
13486 (record_equivalences_from_stmt): Likewise.
13487 * tree-ssa-scopedtables.c: Include options.h and params.h.
13488 (vuse_eq): New function, moved from tree-ssa-dom.c
13489 (build_and_record_new_cond): Likewise.
13490 (record_conditions): Likewise. Accept vector of conditions rather
13491 than edge_equivalence structure for first argument.
13492 for the first argument.
13493 (avail_exprs_stack::lookup_avail_expr): New member function, moved
13494 from tree-ssa-dom.c.
13495 (avail_exprs_stack::record_cond): Likewise.
13496 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
13497 from tree-ssa-dom.c.
13498 (avail_exprs_stack): Add new member functions lookup_avail_expr
13500 (record_conditions): Declare.
13502 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
13505 * lra-constraints.c (process_alt_operands): Increase reject for
13506 reloading an input/output operand.
13508 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
13511 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
13512 insns to convert from signed/unsigned char/short to IEEE 128-bit
13514 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
13516 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
13519 * config/i386/i386.c (ix86_vector_duplicate_value): Create
13520 subreg of inner mode for values already in registers.
13522 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
13524 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
13525 iteration reg is used after the loop.
13527 2017-03-14 Martin Sebor <msebor@redhat.com>
13529 PR tree-optimization/79800
13530 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
13531 precision in negative-positive range.
13532 (format_floating): Call non-const overload with adjusted precision.
13534 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
13537 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
13540 2017-03-14 Martin Sebor <msebor@redhat.com>
13542 PR middle-end/80020
13543 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
13544 * builtins.def (aligned_alloc): Use it.
13547 * Makefile.in (GTFILES): Add calls.c.
13548 * calls.c: Include "gt-calls.h".
13550 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
13552 PR rtl-optimization/79728
13553 * regs.h (struct target_regs): New field
13554 x_contains_allocatable_regs_of_mode.
13555 (contains_allocatable_regs_of_mode): New macro.
13556 * reginfo.c (init_reg_sets_1): Initialize it, and change
13557 contains_reg_of_mode so it includes global regs as well.
13558 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
13559 rather than contains_regs_of_mode.
13561 2017-03-14 Martin Liska <mliska@suse.cz>
13563 * doc/invoke.texi: Document options that can't be combined with
13564 -fcheck-pointer-bounds.
13566 2017-03-14 Martin Liska <mliska@suse.cz>
13568 PR middle-end/79831
13569 * doc/invoke.texi (-Wchkp): Document the option.
13571 2017-03-14 Martin Liska <mliska@suse.cz>
13573 * Makefile.in: Install gcov-dump.
13575 2017-03-14 Martin Liska <mliska@suse.cz>
13577 * multiple_target.c (expand_target_clones): Bail out for
13578 an invalid attribute.
13580 2017-03-14 Richard Biener <rguenther@suse.de>
13582 * alias.c (struct alias_set_entry): Pack properly.
13583 * cfgloop.h (struct loop): Likewise.
13584 * cse.c (struct set): Likewise.
13585 * ipa-utils.c (struct searchc_env): Likewise.
13586 * loop-invariant.c (struct invariant): Likewise.
13587 * lra-remat.c (struct cand): Likewise.
13588 * recog.c (struct change_t): Likewise.
13589 * rtl.h (struct address_info): Likewise.
13590 * symbol-summary.h (function_summary): Likewise.
13591 * tree-loop-distribution.c (struct partition): Likewise.
13592 * tree-object-size.c (struct object_size_info): Likewise.
13593 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
13594 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
13595 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
13596 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
13597 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
13598 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
13599 (struct _stmt_vec_info): Likewise.
13601 2017-03-14 Martin Liska <mliska@suse.cz>
13604 * multiple_target.c (create_dispatcher_calls): Check that
13605 a target can create a function dispatcher.
13607 2017-03-14 Martin Liska <mliska@suse.cz>
13610 * multiple_target.c (expand_target_clones): Drop local.local
13611 flag for default implementation.
13613 2017-03-14 Richard Biener <rguenther@suse.de>
13615 PR tree-optimization/80030
13616 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
13618 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
13620 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
13621 gcc_fallthrough() instead of __attribute__((fallthrough));
13623 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13625 * doc/gcc.texi: Remove "up" link to (DIR).
13626 * doc/gccint.texi: Ditto.
13628 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13630 * doc/install.texi (Specific) <avr>: Remove reference to
13633 2017-03-13 Jeff Law <law@redhat.com>
13635 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
13636 attribute rather than comments.
13638 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
13639 match_scratch operand is highest.
13641 2017-03-13 Martin Liska <mliska@suse.cz>
13643 PR middle-end/78339
13644 * ipa-pure-const.c (warn_function_noreturn): If the declarations
13645 is a CHKP clone, use original declaration.
13647 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13649 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
13650 (arc_conditional_register_usage): Use a different allocation order
13651 when optimizing for size.
13652 * common/config/arc/arc-common.c (arc_option_optimization_table):
13653 Section anchors default on when optimizing for size.
13655 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13657 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
13659 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13661 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
13662 * config/arc/arc.md (cpu_facility): Add cd variant.
13663 (*movqi_insn): Add code density variant.
13664 (*movhi_insn): Likewise.
13665 (*movqi_insn): Likewise.
13666 (*addsi3_mixed): Likewise.
13667 (subsi3_insn): Likewise.
13669 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13671 * config/arc/arc.md (movsi_cond_exec): Update constraint.
13673 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13675 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
13676 expressions with MINUS and UNARY ops.
13678 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13681 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
13683 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
13684 between vec_select and vector argument.
13685 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
13686 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
13687 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
13688 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
13689 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
13690 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
13692 2017-03-13 Richard Biener <rguenther@suse.de>
13695 * params.def (vect-max-peeling-for-alignment): Fix typo.
13697 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13699 * doc/install.texi (Specific) <mips-*-*>: Remove description of
13700 issue that only occurred with binutils below 2.18.
13702 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13704 * doc/install.texi (Specific) <cris-axis-elf>: No longer
13705 refer to binutils 2.11/2.12 minimum.
13707 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13709 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
13710 ftp.kernel.org and simplify binutils requirement.
13712 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
13714 * doc/invoke.texi (Warning Options): Fix spelling of link-time
13716 (Optimize Options): Ditto. Also remove redundancy.
13718 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13720 PR translation/79848
13721 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
13723 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
13724 to G_ to avoid double translation.
13726 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13728 PR translation/79923
13729 * auto-profile.c (get_combined_location): Convert leading
13730 character of diagnostics to lower case and remove trailing period.
13731 (read_profile): Likewise for various diagnostics.
13732 * config/arm/arm.c (arm_option_override): Remove trailing period
13733 from various diagnostics.
13734 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
13735 (msp430_expand_delay_cycles): Likewise.
13737 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13740 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
13741 full command-line argument, rather than just "str".
13742 (aarch64_validate_march): Likewise.
13743 (aarch64_validate_mtune): Likewise.
13745 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
13747 PR rtl-optimization/78911
13748 * lra-assigns.c (must_not_spill_p): New function.
13749 (spill_for): Use it.
13751 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13753 PR tree-optimization/79981
13754 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
13755 ATOMIC_COMPARE_EXCHANGE ifn result.
13756 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
13757 IFN_ATOMIC_COMPARE_EXCHANGE.
13759 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13762 * opts.c (parse_sanitizer_options): Add missing question mark to
13763 "did you mean" message.
13765 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13767 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
13769 (VMULEUH_UNS): Likewise.
13770 (VMULOUB_UNS): Likewise.
13771 (VMULOUH_UNS): Likewise.
13772 * config/rs6000/rs6000.c (builtin_function_type): Remove
13773 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
13775 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13778 * read-rtl-function.c (function_reader::read_rtx_operand): Update
13779 x with result of extra_parsing_for_operand_code_0.
13780 (function_reader::extra_parsing_for_operand_code_0): Convert
13781 return type from void to rtx, returning x. When reading
13782 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
13783 larger size containing struct block_symbol.
13785 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
13787 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
13788 -mfloat128-hardware without -m64.
13790 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
13793 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
13794 entries to the case statement that marks unsigned arguments to
13795 overloaded functions.
13797 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
13799 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
13800 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
13802 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
13805 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
13806 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
13808 2017-03-10 Martin Liska <mliska@suse.cz>
13812 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
13813 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
13816 2017-03-10 Olivier Hainque <hainque@adacore.com>
13818 * tree-switch-conversion (array_value_type): Start by resetting
13819 candidate type to it's main variant.
13821 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13823 PR rtl-optimization/79909
13824 * combine.c (try_combine): Use simplify_replace_rtx on individual
13825 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
13826 of the whole CALL_INSN_FUNCTION_USAGE.
13828 PR tree-optimization/79972
13829 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
13830 get_range_info on SSA_NAMEs. Formatting fixes.
13832 2017-03-10 Richard Biener <rguenther@suse.de>
13833 Jakub Jelinek <jakub@redhat.com>
13835 PR tree-optimization/77975
13836 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
13837 edge to be constant.
13838 (get_val_for): For constant x return it. Formatting fix.
13839 (loop_niter_by_eval): Avoid pointless looping if the next iteration
13840 would use the same bases as the current one.
13842 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13844 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
13845 instead of vec_select for V1TImode.
13846 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
13848 (VSX_LE_128): Add V1TI to this mode iterator.
13849 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
13850 (*vsx_le_perm_store_<mode>): Likewise.
13851 (pre-reload splitter for VSX stores): Likewise.
13852 (post-reload splitter for VSX stores): Likewise.
13853 (*vsx_xxpermdi2_le_<mode>): Likewise.
13854 (*vsx_lxvd2x2_le_<mode>): Likewise.
13855 (*vsx_stxvd2x2_le_<mode>): Likewise.
13857 2017-03-09 Michael Eager <eager@eagercon.com>
13859 Correct failures with --enable-checking=yes,rtl.
13861 * config/microblaze/microblaze.c (microblaze_expand_shift):
13862 Replace GET_CODE test with CONST_INT_P and INTVAL test with
13863 test for const0_rtx.
13864 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
13865 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
13867 2017-03-09 Richard Biener <rguenther@suse.de>
13869 PR tree-optimization/79977
13870 * graphite-scop-detection.c (scop_detection::merge_sese):
13871 Handle the case of extra exits to blocks dominating the entry.
13873 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
13875 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13878 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
13880 PR rtl-optimization/79949
13881 * lra-constraints.c (process_alt_operands): Check memory when
13882 trying to predict a cycle. Print about the overall increase.
13884 2017-03-09 Richard Biener <rguenther@suse.de>
13886 PR middle-end/79971
13887 * gimple-expr.c (useless_type_conversion_p): Preserve
13888 TYPE_SATURATING for fixed-point types.
13890 2017-03-09 Richard Biener <rguenther@suse.de>
13893 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
13894 alignment of BLKmode params.
13896 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13899 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
13900 (VALL_NO_V2Q): Likewise.
13902 * config/aarch64/aarch64-simd.md
13903 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
13905 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
13906 VALL_NO_V2Q mode iterator.
13907 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
13909 2017-03-09 Martin Liska <mliska@suse.cz>
13911 PR tree-optimization/79631
13912 * tree-chkp-opt.c (chkp_is_constant_addr): Call
13913 tree_int_cst_sign_bit just for INTEGER constants.
13915 2017-03-09 Martin Liska <mliska@suse.cz>
13919 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
13922 2017-03-09 Marek Polacek <polacek@redhat.com>
13925 * tree.c (inchash::add_expr): Handle TREE_VEC.
13927 2017-03-09 Martin Liska <mliska@suse.cz>
13930 (chkp_narrow_size_and_offset): New function.
13931 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
13932 (void chkp_parse_bit_field_ref): New function.
13933 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
13934 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
13936 2017-03-09 Martin Liska <mliska@suse.cz>
13939 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
13940 (chkp_find_bounds_1): Remove gcc_unreachable.
13942 2017-03-09 Jakub Jelinek <jakub@redhat.com>
13945 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
13946 BUILT_IN_SYNC*, determine the access type from the size suffix and
13947 always build a MEM_REF with that type. Handle forgotten
13948 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
13951 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
13952 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
13953 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
13954 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
13955 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
13956 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
13957 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
13958 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
13959 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
13960 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
13961 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
13962 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
13963 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
13964 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
13965 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
13966 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
13967 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
13968 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
13969 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
13970 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
13971 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
13972 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
13973 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
13974 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
13975 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
13976 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
13977 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
13978 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
13979 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
13980 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
13981 definitions outside of __OPTIMIZE__ guarded section.
13984 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
13985 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
13986 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
13987 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
13990 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13992 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
13993 ("vfenez<mode>"): Add missing constraints.
13995 2017-03-08 Martin Sebor <msebor@redhat.com>
13998 * config/nds32/nds32.c (nds32_option_override):
13999 Fix misspelled diagnostic.
14001 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14004 * gimplify.c (gimplify_omp_for): Replace index var in outer
14005 taskloop statement with an artificial variable and add
14006 OMP_CLAUSE_PRIVATE clause for it.
14008 2017-03-08 Richard Biener <rguenther@suse.de>
14010 PR tree-optimization/79955
14011 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
14012 for accesses that are completely outside of the variable.
14014 2017-03-08 Andrew Haley <aph@redhat.com>
14016 PR tree-optimization/79943
14017 * tree-ssa-loop-split.c (compute_new_first_bound): When
14018 calculating the new upper bound, (END-BEG) should be added, not
14021 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14023 * config/avr/avr.md (setmemhi): Make sure match_dup
14024 operand number comes before match_scratch.
14026 2017-03-08 Richard Biener <rguenther@suse.de>
14028 PR tree-optimization/79920
14029 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
14030 with ncopies == 1 to ...
14031 (vect_transform_slp_perm_load): ... here. Properly compute
14032 all element loads by iterating VF times over the group. Do
14033 not handle ncopies (computed in a broken way) in
14034 vect_create_mask_and_perm.
14036 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14039 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
14040 is a uniform vector, use uniform_vector_p return value instead of
14041 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
14043 2017-03-07 Marek Polacek <polacek@redhat.com>
14045 PR middle-end/79809
14046 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
14047 (alloca_call_type): Likewise.
14049 2017-03-07 Martin Liska <mliska@suse.cz>
14051 * gcov.c (process_args): Put comment to correct location.
14053 2017-03-07 Martin Liska <mliska@suse.cz>
14055 PR middle-end/68270
14056 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
14057 Use array_at_struct_end_p instead of DECL_CHAIN (field).
14058 (chkp_narrow_bounds_for_field): Likewise.
14059 (chkp_parse_array_and_component_ref): Pass one more argument to
14062 2017-03-07 Richard Biener <rguenther@suse.de>
14064 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
14067 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
14069 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
14070 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
14072 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14075 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
14076 to end of description.
14077 (PARAM_MAX_STORES_TO_MERGE): Likewise.
14079 2017-03-07 Jakub Jelinek <jakub@redhat.com>
14081 PR rtl-optimization/79901
14082 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
14084 (*avx512f_<code><mode>3<mask_name>): ... this.
14085 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
14086 iterator instead of VI8_AVX2_AVX512BW.
14088 PR rtl-optimization/79901
14089 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
14090 min/max expander, expand it using expand_vec_cond_expr.
14093 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
14096 2017-03-06 Jakub Jelinek <jakub@redhat.com>
14099 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
14100 to void * for PCH reasons.
14101 * dwarf2out.c (output_loc_operands, output_die): Cast
14102 v.val_vec.array to unsigned char *.
14104 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
14107 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
14110 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
14112 PR rtl-optimization/79571
14113 * lra-constraints.c (process_alt_operands): Calculate static
14114 reject and subtract it from overall when only addresses will be
14117 2017-03-06 Julia Koval <julia.koval@intel.com>
14120 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
14121 incoming stack boundary to 128 for 64-bit targets.
14123 2017-03-06 Richard Biener <rguenther@suse.de>
14125 PR tree-optimization/79894
14126 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
14127 to NULL after folding it.
14129 2017-03-06 Richard Biener <rguenther@suse.de>
14131 PR tree-optimization/79824
14132 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
14133 check disabling peeling for gaps.
14135 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
14137 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
14138 attributes): Document gettimeofday.
14140 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
14142 * config/s390/s390.c (s390_option_override_internal): Set
14143 PARAM_MIN_VECT_LOOP_BOUND
14145 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
14147 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
14148 * config/s390/s390.md: Likewise.
14150 2017-03-06 Jakub Jelinek <jakub@redhat.com>
14153 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
14154 (<avx2_avx512>_perm<mode>): Rename to ...
14155 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
14157 (<avx512>_perm<mode>_mask): Rename to ...
14158 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
14160 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
14161 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
14162 instead of VI8F_256_512.
14163 (avx512f_perm<mode>): New define_expand.
14164 (avx512f_perm<mode>_mask): Likewise.
14165 (avx512f_perm<mode>_1<mask_name>): New define_insn.
14166 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
14168 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14170 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
14171 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
14173 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
14175 2017-03-06 Martin Liska <mliska@suse.cz>
14178 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
14179 when having a SSA NAME w/o VAR_DECL assigned to it.
14181 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14183 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
14184 msa_dpsub_<su>_d): Fix MODE for vec_select.
14186 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14188 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
14190 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
14192 2017-03-06 Richard Biener <rguenther@suse.de>
14194 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
14195 * plugin.c (register_plugin_info): Likewise.
14196 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
14198 2017-03-05 Jakub Jelinek <jakub@redhat.com>
14200 * config/i386/sse.md (sse_storehps, sse_storelps,
14201 avx_<castmode><avxsizesuffix>_<castmode>,
14202 avx512f_<castmode><avxsizesuffix>_<castmode>,
14203 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
14204 in condition that at least one operand is not a MEM.
14206 2017-03-03 Jakub Jelinek <jakub@redhat.com>
14208 PR middle-end/79805
14209 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
14210 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
14212 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
14213 gimple_call_nothrow_p flag based on whether original builtin can throw.
14214 If it can, emit following stmts on the fallthrough edge.
14215 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
14216 don't create new bb if inserting just debug stmts on the edge, try to
14217 insert them on the fallthru bb or just reset debug stmts.
14219 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
14222 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
14223 restore recog_data (including the operand rtxes inside it) around
14224 the call to get_insn_template.
14226 2017-03-03 Martin Sebor <msebor@redhat.com>
14228 PR tree-optimization/79699
14229 * context.c (context::~context): Free MPFR caches to avoid
14230 a memory leak on program exit.
14232 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14234 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
14235 Use wide_int::ulow () instead of .elt (0).
14237 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
14239 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
14240 (*pushxf): Limit oF constraint to 32bit targets and add oC
14241 constraint for 64bit targets.
14242 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
14243 (*pushdf): Change rmF constraint to rmC.
14245 2017-03-03 Martin Liska <mliska@suse.cz>
14247 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
14248 Remove unused variable.
14250 2017-03-03 Jakub Jelinek <jakub@redhat.com>
14253 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
14254 is a memory operand, increase num_memory.
14255 (ix86_expand_args_builtin): Likewise.
14257 2017-03-03 Jan Hubicka <jh@suse.cz>
14260 * ipa-devirt.c (maybe_record_node): Properly handle
14261 __cxa_pure_virtual visibility.
14263 2017-03-03 Martin Liska <mliska@suse.cz>
14265 PR tree-optimization/79803
14266 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
14268 (pass_loop_prefetch::execute): Disabled optimization if an
14269 assumption about L1 cache size is not met.
14271 2017-03-03 Martin Liska <mliska@suse.cz>
14273 PR rtl-optimization/79574
14274 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
14275 (hash_scan_set): Likewise.
14276 (dump_hash_table): Likewise.
14277 (hoist_code): Likewise.
14279 2017-03-03 Richard Biener <rguenther@suse.de>
14281 * fixed-value.c (fixed_from_string): Restore use of elt (1)
14282 in place of uhigh ().
14283 (fixed_convert_from_real): Likewise.
14285 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
14288 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
14290 2017-03-03 Richard Biener <rguenther@suse.de>
14292 PR middle-end/79818
14293 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
14294 TYPE_OVERFLOW_UNDEFINED check.
14296 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14298 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
14300 (vector_ae_<mode>_p): Likewise.
14301 (vector_nez_<mode>_p): Likewise.
14302 (vector_ne_v2di_p): Likewise.
14303 (vector_ae_v2di_p): Likewise.
14304 (vector_ne_<mode>_p): Likewise.
14305 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
14307 (vsx_tsqrt<mode>2_fe): Likewise.
14309 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
14312 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
14314 2017-03-02 Jakub Jelinek <jakub@redhat.com>
14316 PR rtl-optimization/79780
14317 * cprop.c (one_cprop_pass): When second and further conditional trap
14318 in a single basic block is turned into an unconditional trap, turn it
14319 into a deleted note to avoid RTL verification failures.
14321 2017-03-02 Richard Biener <rguenther@suse.de>
14323 * fold-const.c (const_binop): Use ulow () instead of elt (0).
14325 2017-03-02 Richard Biener <rguenther@suse.de>
14327 PR tree-optimization/79345
14329 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
14330 param and abort the walk, returning -1 if it is hit.
14331 (walk_aliased_vdefs): Take a limit param and pass it on.
14332 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
14333 defaulting to 0 and return a signed int.
14334 * tree-ssa-uninit.c (struct check_defs_data): New struct.
14335 (check_defs): New helper.
14336 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
14337 about uninitialized memory.
14338 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
14339 bogus uninitialized warning.
14340 (fixed_convert_from_real): Likewise.
14342 2017-03-02 Bin Cheng <bin.cheng@arm.com>
14344 PR tree-optimization/66768
14345 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
14346 iv_use if base object can't be determined.
14348 2017-03-02 Jakub Jelinek <jakub@redhat.com>
14350 PR tree-optimization/79345
14351 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
14352 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
14353 (get_pattern_stats): Initialize it.
14354 * genemit.c (gen_expand): Verify match_scratch numbers come after
14355 match_operand/match_dup numbers.
14356 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
14357 match_scratch numbers.
14358 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
14360 * config/s390/s390.md (trunctdsd2): Likewise.
14362 2017-03-02 Richard Biener <rguenther@suse.de>
14364 * wide-int.h (wide_int_storage::operator=): Implement in terms
14367 2017-03-02 Richard Biener <rguenther@suse.de>
14369 PR tree-optimization/79777
14370 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
14371 the to insert expression to sth existing.
14373 2017-03-01 Martin Sebor <msebor@redhat.com>
14375 PR middle-end/79692
14376 * gimple-ssa-sprintf.c
14377 (directive::known_width_and_precision): New function.
14378 (format_integer): Use it.
14379 (get_mpfr_format_length): Consider the full range of precision
14380 when computing %g output with the # flag. Set the likely byte
14381 count to 3 rather than 1 when precision is indeterminate.
14382 (format_floating): Correct the lower bound of precision.
14384 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14386 * doc/invoke.texi: Document default code model for 64-bit Linux.
14388 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14391 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
14392 udiv rather than div since input pattern is unsigned.
14394 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
14396 * config/i386/i386.c (print_reg): Warn for values of
14397 unsupported size in integer register.
14399 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
14402 * config/rs6000/predicates.md (current_file_function_operand): Do
14403 not allow self calls to be local if the function is replaceable.
14405 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
14408 * config/rs6000/altivec.h (vec_ctz and others): Change the
14409 preprocessor macro that controls conditional compilation from
14410 _ARCH_PWR9 to __POWER9_VECTOR__.
14411 (vec_all_ne): Change parameterization of __altivec_scalar_pred
14412 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
14413 control (instead of _ARCH_PWR9 control) so that template
14414 definition uses power9-specific function.
14415 (vec_any_eq): Likewise.
14416 (vec_all_ne): Change macro definition to use a power9-specific
14417 expansion under #ifdef __POWER9_VECTOR__ control (instead of
14418 _ARCH_PWR9 control).
14419 (vec_any_eq) Likewise.
14420 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
14421 expansion for CMPNEF to remove support for xvcmpnesp instruction.
14422 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
14423 support for xvcmpnedp instruction.
14424 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
14425 macro expansion so that Power9 implementation of vec_all_ne does
14426 not use the AltiVec predicate framework.
14427 (VCMPNEH_P): Likewise.
14428 (VCMPNEW_P): Likewise.
14429 (VCMPNED_P): Likewise.
14430 (VCMPNEFP_P): Likewise.
14431 (VCMPNEDP_P): Likewise.
14432 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
14433 implementation of vec_any_eq to not use AltiVec predicate
14435 (VCMPAEH_P): Likewise.
14436 (VCMPAEW_P): Likewise.
14437 (VCMPAED_P): Likewise.
14438 (VCMPAEFP_P): Likewise.
14439 (VCMPAEDP_P): Likewise.
14440 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
14441 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
14442 not use the AltiVec predicate framework.
14443 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
14444 of vec_any_eq to not use AltiVec predicate framework.
14445 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
14446 support for predefined __POWER9_VECTOR__ macro to indicate that
14447 Power9 instruction selection is enabled.
14448 (altivec_overloaded_builtins): Remove extraneous
14449 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
14450 function argument types RS6000_BTI_bool_V16QI and
14451 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
14452 entry for overloaded function argument types RS6000_BTI_bool_V4SI
14453 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
14454 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
14455 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
14456 Power9 for implementations of vec_cmpne. Change the signature for
14457 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
14458 (representing vec_all_ne) to remove the previously described first
14459 argument of type RS6000_BTI_INTSI, as this was an artifact of
14460 reliance on the AltiVec predicate framework, which is no longer
14461 used in the implementation of these functions. Add
14462 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
14463 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
14464 since, unlike the AltiVec predicate framework implementation, we
14465 do not share function descriptors between vec_alle and vec_anyeq.
14466 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
14467 set of modes that receive special treatment even when
14468 TARGET_P9_VECTOR is true. The special treatment emits code that
14469 does not depend on Power9 instructions.
14470 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
14471 define_expand to not rely on AltiVec predicate framework.
14472 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
14474 (vector_ne_v2di_p): Change this define_expand to not rely on
14475 AltiVec predicate framework.
14476 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
14478 (vector_ne_<mode>_p): Change this define_expand to not rely on
14479 AltiVec predicate framework.
14480 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
14482 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
14483 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
14484 define_insn pattern.
14485 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
14486 define_insn pattern because the xvcmpne<VSs>. instruction is not
14488 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
14489 instruction is not supported.
14491 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14493 * config/nvptx/nvptx.c: Include intl.h.
14495 2017-03-01 Martin Jambor <mjambor@suse.cz>
14498 * ipa-prop.h (ipa_bits): Removed field known.
14499 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
14500 to pointers. Adjusted their comments to warn about their sharing.
14501 (ipcp_transformation_summary): Change bits to a vector of pointers.
14502 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
14503 (ipa_get_ipa_bits_for_value): Declare.
14504 * tree-vrp.h (value_range): Mark as GTY((for_user)).
14505 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
14506 (ipa_bits_hash_table): Likewise.
14507 (ipa_vr_ggc_hash_traits): Likewise.
14508 (ipa_vr_hash_table): Likewise.
14509 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
14510 being pointers and vr_known being removed.
14511 (ipa_set_jf_unknown): Likewise.
14512 (ipa_get_ipa_bits_for_value): New function.
14513 (ipa_set_jfunc_bits): Likewise.
14514 (ipa_get_value_range): New overloaded functions.
14515 (ipa_set_jfunc_vr): Likewise.
14516 (ipa_compute_jump_functions_for_edge): Use the above functions to
14517 construct bits and vr parts of jump functions.
14518 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
14519 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14521 (ipcp_grow_transformations_if_necessary): Also allocate
14522 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14524 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
14525 them. Fix too long lines.
14526 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
14527 vr_known being removed.
14528 (ipa_read_jump_function): Use new setter functions to construct bits
14529 and vr parts of jump functions or set them to NULL.
14530 (write_ipcp_transformation_info): Adjust for bits being pointers.
14531 (read_ipcp_transformation_info): Likewise.
14532 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
14534 Include gt-ipa-prop.h.
14535 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
14537 (ipcp_store_bits_results): Likewise.
14538 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
14539 Do not write to existing jump functions but use a temporary instead.
14541 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14544 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
14545 attempt to use its first operand as BIT_FIELD_REF base.
14547 2017-03-01 Richard Biener <rguenther@suse.de>
14549 PR middle-end/79721
14550 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
14551 interpolating formula in wrapping arithmetic.
14552 (chrec_apply): Convert chrec_evaluate return value to wanted type.
14554 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14556 PR tree-optimization/79734
14557 * tree-vect-generic.c (expand_vector_condition): Optimize
14558 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
14559 Handle VEC_COND_EXPR where comparison has different inner width from
14560 type's inner width.
14562 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
14564 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
14565 markup, and similar issues. Remove @opindex entries for things
14566 that aren't options. Add missing -mmpy-option entries.
14568 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14570 PR tree-optimization/79737
14571 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
14572 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
14573 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
14574 instead of byte_size. Formatting fix.
14575 (shift_bytes_in_array_right): Formatting fix.
14577 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
14580 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
14581 condition on optimize for the leaf function test.
14583 2017-02-28 Martin Liska <mliska@suse.cz>
14586 * read-rtl-function.c (function_reader::handle_unknown_directive):
14587 Bail out when one uses -flto.
14589 2017-02-28 Martin Liska <mliska@suse.cz>
14591 * common.opt: Replace space with tabular for options of <number>
14593 * config/i386/i386.opt: Show <number> value for
14594 -mlarge-data-threshold.
14595 * opts.c (print_filtered_help): Do not display number in hexadecimal
14598 2017-02-28 Martin Liska <mliska@suse.cz>
14600 * common.opt: Fix --help=option -Q for options which are of
14603 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
14605 * config/i386/i386.c (print_reg): Error out for values
14606 of 8-bit size in invalid integer register.
14608 2017-02-28 Martin Sebor <msebor@redhat.com>
14610 PR tree-optimization/79691
14611 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
14613 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14616 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
14617 gcc_unreachable with output_operand_lossage.
14619 2017-02-28 Richard Biener <rguenther@suse.de>
14621 PR tree-optimization/79740
14622 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
14624 (visit_nary_op): Insert the nary into the hashtable if we
14625 pattern-matched sth.
14626 * tree-ssa-pre.c (eliminate_insert): Robustify.
14628 2017-02-28 Richard Biener <rguenther@suse.de>
14630 PR middle-end/79731
14631 * fold-const.c (decode_field_reference): Reject out-of-bound
14634 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14636 * config/i386/i386.c: Include intl.h.
14637 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
14638 instead of just cond ? "..." : "...".
14639 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
14640 * coverage.c (read_counts_file): Likewise.
14641 * omp-offload.c: Include intl.h.
14642 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
14643 of just cond ? "..." : "...".
14644 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
14645 of just cond ? "..." : "...".
14647 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
14650 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
14652 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
14653 'tune for' CPU name.
14654 * config/arm/arm-cpu-data.h: Regenerated.
14656 2017-02-28 Richard Biener <rguenther@suse.de>
14658 PR tree-optimization/79732
14659 * tree-inline.c (expand_call_inline): Do not shadow var.
14661 2017-02-28 Richard Biener <rguenther@suse.de>
14663 PR tree-optimization/79723
14664 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
14665 address-space properly.
14667 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
14669 * doc/optinfo.texi (Optimization groups): Fix option used for
14671 * doc/invoke.texi (-fopt-info): Document "omp".
14672 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
14673 (OPTGROUP_ALL): Add OPTGROUP_OMP.
14674 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
14675 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
14676 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
14678 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
14680 * dumpfile.c (optgroup_options): Instead of "openmp", associate
14681 OPTGROUP_OMP with "omp".
14683 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
14686 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
14687 for arithmetic shift of unsigned V2DI.
14689 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
14691 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
14692 arc/linux.h headers.
14693 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
14694 (LINK_SPEC): Likewise.
14695 (ARC_TLS_EXTRA_START_SPEC): Likewise.
14696 (EXTRA_SPECS): Likewise.
14697 (STARTFILE_SPEC): Likewise.
14698 (ENDFILE_SPEC): Likewise.
14699 (LIB_SPEC): Likewise.
14700 (TARGET_SDATA_DEFAULT): Likewise.
14701 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
14702 (MULTILIB_DEFAULTS): Likewise.
14703 (DWARF2_UNWIND_INFO): Likewise.
14704 * config/arc/big.h: New file.
14705 * config/arc/elf.h: Likewise.
14706 * config/arc/linux.h: Likewise.
14707 * config/arc/t-uClibc: Remove.
14709 2017-02-27 Bin Cheng <bin.cheng@arm.com>
14711 PR tree-optimization/77536
14712 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
14713 (tree_transform_and_unroll_loop): Use above function to compute the
14714 estimated niter of unrolled loop and use it when scaling profile.
14715 Also use count info rather than frequency if it's non-zero.
14716 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
14717 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
14718 (vect_transform_loop): Call above function.
14720 2017-02-27 Richard Biener <rguenther@suse.de>
14722 PR tree-optimization/45397
14723 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
14724 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
14725 (visit_nary_op): Add pattern matching for CSEing sign-changed
14726 or truncated operations with wider ones.
14728 2017-02-27 Richard Biener <rguenther@suse.de>
14730 PR tree-optimization/79690
14731 * tree-vect-stmts.c (vectorizable_store): Use vector type
14732 built from the DR with address-space.
14734 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
14736 * doc/invoke.texi (Optimize Options): Refine the description
14737 of asan-use-after-return.
14739 2017-02-25 Alan Modra <amodra@gmail.com>
14741 PR rtl-optimization/79584
14742 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
14743 base, not ad->base_term, the reg within base. Remove assertion
14744 that ad->base == ad->base_term. Replace gen_int_mode using
14745 bogus mode with const0_rtx.
14747 2017-02-25 Jakub Jelinek <jakub@redhat.com>
14749 PR middle-end/79396
14750 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
14751 FMA_EXPR like tcc_binary or tcc_unary.
14753 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
14756 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
14758 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
14759 (output_loc_operands): Handle DW_OP_call_ref and
14760 DW_OP_GNU_variable_value.
14761 (struct variable_value_struct): New type.
14762 (struct variable_value_hasher): Likewise.
14763 (variable_value_hash): New variable.
14764 (string_types): Remove.
14765 (copy_loc_descr): New function.
14766 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
14767 (prepend_loc_descr_to_each): New function.
14768 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
14769 instead of add_loc_descr_to_each if the first argument is single
14770 location list and the second has multiple.
14771 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
14772 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
14773 when looking for variable value which doesn't have other location info.
14774 (loc_list_from_tree): Formatting fix.
14775 (gen_array_type_die): Simplify DW_AT_string_length handling.
14776 (adjust_string_types): Remove.
14777 (gen_subprogram_die): Don't call adjust_string_types nor test/set
14778 string_types. Call resolve_variable_values.
14779 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
14780 (resolve_addr_in_expr): Likewise. Add A argument.
14781 (copy_deref_exprloc): Remove deref argument. Adjust for the
14782 original expression being DW_OP_GNU_variable_value with optionally
14783 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
14784 optionally after it.
14785 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
14786 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
14787 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
14788 (variable_value_hasher::hash, variable_value_hasher::equal): New
14790 (resolve_variable_value_in_expr, resolve_variable_value,
14791 resolve_variable_values, note_variable_value_in_expr,
14792 note_variable_value): New functions.
14793 (dwarf2out_early_finish): Call note_variable_value on all toplevel
14796 2017-02-24 Jakub Jelinek <jakub@redhat.com>
14799 * opts.h (handle_generated_option): Add GENERATED_P argument.
14800 * opts-common.c (handle_option): Adjust function comment.
14801 (handle_generated_option): Add GENERATED_P argument, pass it to
14803 (control_warning_option): Pass false to handle_generated_option
14805 * opts.c (maybe_default_option): Pass true to handle_generated_option
14807 * optc-gen.awk: Likewise.
14809 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14811 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
14812 a REG, look at the REG it is a SUBREG of.
14813 (splitter for cmpeqsi_t): Ditto.
14815 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14817 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
14818 the special USEs with the pattern of the insn, not the insn itself.
14820 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
14823 * doc/invoke.texi: Document -mload-store-pairs.
14825 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14826 Sandra Loosemore <sandra@codesourcery.com>
14828 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
14829 argument isn't a CONST_INT.
14830 (nios2_alternate_compare_const): Assert op is a CONST_INT.
14831 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
14832 (nios2_validate_compare): Bypass alternate compare logic if *op2
14833 is not a CONST_INT.
14834 (ldstwm_operation_p): Return false if first_base is not a REG or
14835 if first_offset is not a CONST_INT.
14837 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14839 * config/cris/cris.md: Use correct operand in a define_peephole2.
14841 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14843 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
14845 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14847 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
14848 this_insn if it is an INSN or JUMP_INSN.
14849 (force_offsettable): Look at base, not at addr.
14850 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
14851 on things that aren't necessarily CONST_INTs.
14853 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
14855 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
14856 -mfpmath=sse is the default also for x86-32 targets with SSE2
14857 instruction set when @option{-ffast-math} is enabled
14859 2017-02-24 Jeff Law <law@redhat.com>
14861 PR rtl-optimizatoin/79286
14862 * ira.c (update_equiv_regs): Drop may_trap_p exception to
14865 2017-02-24 Richard Biener <rguenther@suse.de>
14867 PR tree-optimization/79389
14868 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
14871 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
14873 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
14874 function comment to reflect reality.
14875 (loop_exits_before_overflow): Fix typo in function description.
14877 2017-02-24 Richard Biener <rguenther@suse.de>
14879 PR tree-optimization/79389
14880 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
14881 properly that a threading opportunity exists. Detect conditional
14882 copy/constant propagation opportunities.
14884 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
14886 * config/visium/visium.md (type): Add trap.
14887 (b): New mode attribute.
14888 (*btst): Rename into...
14889 (*btst<mode>): ...this and adjust.
14890 (*cbranchsi4_btst_insn): Rename into...
14891 (*cbranch<mode>4_btst_insn): ...this and adjust.
14892 (trap): New define_insn.
14894 2017-02-23 Jakub Jelinek <jakub@redhat.com>
14896 PR tree-optimization/79389
14897 * ifcvt.c (struct noce_if_info): Add rev_cond field.
14898 (noce_reversed_cond_code): New function.
14899 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
14900 reversed_comparison_code. Formatting fix.
14901 (noce_try_store_flag): Test rev_cond != NULL in addition to
14902 reversed_comparison_code.
14903 (noce_try_store_flag_constants): Likewise.
14904 (noce_try_store_flag_mask): Likewise.
14905 (noce_try_addcc): Use rev_cond if non-NULL instead of
14906 reversed_comparison_code.
14907 (noce_try_cmove_arith): Likewise. Formatting fixes.
14908 (noce_try_minmax, noce_try_abs): Clear rev_cond.
14909 (noce_find_if_block): Initialize rev_cond.
14910 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
14911 instead of false as last argument never attempt to reverse it
14914 2017-02-23 Bin Cheng <bin.cheng@arm.com>
14916 PR tree-optimization/79663
14917 * tree-predcom.c (combine_chains): Process refs in reverse order
14918 only for ZERO length chains, and add explaining comment.
14920 2017-02-23 Jeff Law <law@redhat.com>
14922 PR tree-optimization/79578
14923 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
14924 in call to operand_equal_p.
14926 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
14929 * config/i386/cpuid.h: Fix another undefined behavior.
14931 2017-02-23 Richard Biener <rguenther@suse.de>
14933 PR tree-optimization/79683
14934 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
14935 vector types for data-refs.
14937 2017-02-23 Martin Liska <mliska@suse.cz>
14939 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
14941 2017-02-23 Jakub Jelinek <jakub@redhat.com>
14943 PR middle-end/79665
14944 * internal-fn.c (get_range_pos_neg): Moved to ...
14945 * tree.c (get_range_pos_neg): ... here. No longer static.
14946 * tree.h (get_range_pos_neg): New prototype.
14947 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
14948 are known to be in between 0 and signed maximum inclusive, try to
14949 expand both unsigned and signed divmod and use the cheaper one from
14952 2017-02-22 Jeff Law <law@redhat.com>
14954 PR tree-optimization/79578
14955 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
14956 to compare base operands.
14958 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
14961 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
14962 gpc_reg_operand instead of fpr_reg_operand.
14964 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
14966 * config/mips/mips.c (mips_return_in_memory): Force FP
14967 vector types to be returned in memory for o32 ABI.
14969 2017-02-22 Jakub Jelinek <jakub@redhat.com>
14971 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
14972 instead of DW_TAG_member for static data member declarations and don't
14973 set no_linkage_name for static inline data members.
14974 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
14977 2017-02-22 Martin Liska <mliska@suse.cz>
14979 * doc/invoke.texi: Replace inequality signs with square brackets
14982 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14984 PR tree-optimization/68644
14985 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
14987 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
14990 * lra-constraints.c (simplify_operand_subreg): Handle
14991 WORD_REGISTER_OPERATIONS targets.
14993 2017-02-22 Jakub Jelinek <jakub@redhat.com>
14996 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
14997 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
14998 elimination by swapping fld*.
15000 2017-02-22 Richard Biener <rguenther@suse.de>
15002 PR tree-optimization/79673
15003 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
15004 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
15005 irrelevant address-space qualifiers and avoiding a
15006 ADDR_SPACE_CONVERT_EXPR from fold_convert.
15008 2017-02-22 Richard Biener <rguenther@suse.de>
15010 PR tree-optimization/79666
15011 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
15012 to not symbolically negate if that may introduce undefined
15015 2017-02-22 Martin Liska <mliska@suse.cz>
15018 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
15019 * data-streamer-out.c (streamer_write_gcov_count_stream):
15021 * value-prof.c (stream_out_histogram_value): Make assert more
15022 precise based on type of counter.
15024 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
15027 * config/i386/i386.md (standard_x87sse_constant_load splitter):
15028 Use nonimmediate_operand instead of memory_operand for operand 1.
15029 (float-extend standard_x87sse_constant_load splitter): Ditto.
15031 2017-02-21 Jeff Law <law@redhat.com>
15033 PR tree-optimization/79621
15034 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
15035 blocks with edges to themselves.
15037 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15040 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
15041 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
15042 Use gimple_call_builtin_p.
15045 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
15046 on temporarily removed DEBUG_INSNs.
15048 PR tree-optimization/79649
15049 * tree-loop-distribution.c (classify_partition): Give up on
15050 non-generic address space loads/stores.
15052 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
15054 * doc/loop.texi (Loop manipulation): Remove nonexistent
15055 tree_ssa_loop_version from the documentation.
15056 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
15058 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15061 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
15062 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
15063 * config/rs6000/rs6000.c: Include except.h.
15064 (rs6000_expand_split_stack_prologue): Call
15065 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
15067 2017-02-21 Martin Jambor <mjambor@suse.cz>
15070 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
15071 have been analyzed.
15073 2017-02-21 Martin Jambor <mjambor@suse.cz>
15075 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
15076 for backward compatibility only.
15077 * doc/invoke.texi (Option Summary): Remove all references to
15078 -fipa-cp-alignment.
15080 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
15084 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15086 * lra-constraints.c (curr_insn_transform): Handle
15087 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
15089 2017-02-21 Martin Liska <mliska@suse.cz>
15091 * config/i386/i386.opt: Replace -masm-dialect with -masm.
15093 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
15095 PR translation/79638
15096 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
15098 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
15101 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
15102 (arm_function_ok_for_sibcall): Return false for an indirect call by
15103 descriptor if all the argument registers are used.
15104 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
15105 alignment of the function.
15107 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15109 PR tree-optimization/61441
15110 * simplify-rtx.c (simplify_const_unary_operation): For
15111 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
15112 the sNaN unmodified.
15114 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
15116 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
15117 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
15118 instead of SYSTEM_HEADER_DIR.
15120 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
15121 Martin Liška <mliska@suse.cz>
15123 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
15124 Fix typos and grammar, use active voice, and clarify.
15126 2017-02-20 Marek Polacek <polacek@redhat.com>
15128 PR middle-end/79537
15129 * gimplify.c (gimplify_expr): Handle unused *&&L;.
15132 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
15134 2017-02-20 Jakub Jelinek <jakub@redhat.com>
15137 * config/i386/i386.c (ix86_expand_builtin): Handle
15138 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
15139 ix86_builtins_isa[fcode].isa as a requirement of those
15140 flags and any other flag in the bitmask.
15141 (ix86_init_mmx_sse_builtins): Use 0 instead of
15142 ~OPTION_MASK_ISA_64BIT as mask.
15143 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
15144 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
15145 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
15146 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
15148 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15151 * lra-constraints.c (split_reg): Check requested split mode
15152 is supported by the register.
15154 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15156 * lra-constraints.c (simplify_operand_subreg): Remove early
15159 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15162 * lra-constraints.c (curr_insn_transform): Tighten condition
15163 for converting SUBREG reloads from OP_OUT to OP_INOUT.
15165 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15168 * lra-constraints.c (curr_insn_transform): Handle
15169 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
15171 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
15174 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
15176 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
15178 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
15181 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
15184 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15186 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
15189 2017-02-18 Jakub Jelinek <jakub@redhat.com>
15191 * final.c (last_columnnum, override_columnnum): New variables.
15192 (final_start_function): Set last_columnnum, pass it to begin_prologue
15193 hook and pass 0 to dwarf2out_begin_prologue.
15194 (final_scan_insn): Update override_columnnum. Pass last_columnnum
15195 to source_line debug hook.
15196 (notice_source_line): Compute last_columnnum and for debug_column_info
15197 return true on column changes.
15198 * debug.h (struct gcc_debug_hooks): Add column argument to
15199 source_line and begin_prologue hooks.
15200 (debug_nothing_int_charstar_int_bool): Remove prototype.
15201 (debug_nothing_int_int_charstar,
15202 debug_nothing_int_int_charstar_int_bool): New prototypes.
15203 (dwarf2out_begin_prologue): Add column argument.
15204 * debug.c (do_nothing_debug_hooks): Adjust source_line and
15205 begin_prologue hooks.
15206 (debug_nothing_int_charstar_int_bool): Remove.
15207 (debug_nothing_int_int_charstar,
15208 debug_nothing_int_int_charstar_int_bool): New functions.
15209 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
15210 through to dwarf2out_source_line.
15211 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
15212 (dwarf2out_source_line): Add column argument, emit it if requested.
15213 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
15215 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
15216 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
15217 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
15218 through to dwarf2out_begin_prologue.
15219 (vmsdbgout_source_line): Add column argument, pass it through to
15220 dwarf2out_source_line.
15221 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
15222 dbxout_source_line caller.
15223 (dbxout_source_line): Add column argument.
15225 * common.opt (gno-column-info, gcolumn-info): New options.
15226 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
15227 (check_die): Also test for multiple DW_AT_decl_column attributes.
15228 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
15229 DW_AT_decl_column if requested.
15230 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
15232 (gen_variable_die): Likewise.
15233 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
15234 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
15237 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
15238 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
15239 (ix86_handle_option): Handle OPT_m3dnowa.
15240 * doc/invoke.texi (-m3dnowa): Document.
15241 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
15242 -m3dnowa instead of -m3dnow -march=athlon.
15245 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
15246 instead of gcc_assert for K, r and R code checks. Formatting fixes.
15248 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15251 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
15252 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
15253 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
15254 generator for vsx_xxpermdi_<mode>_be.
15255 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
15256 force big-endian semantics.
15257 (vsx_xxpermdi_<mode>_be): New define_expand with same
15258 implementation as previous version of vsx_xxpermdi_<mode>.
15260 2017-02-17 Jakub Jelinek <jakub@redhat.com>
15262 PR tree-optimization/79327
15263 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
15264 variable, its initialization and use.
15266 2017-02-17 Julia Koval <julia.koval@intel.com>
15268 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
15269 (OPTION_MASK_ISA_PKU_UNSET): New.
15270 (ix86_handle_option): Handle -mrdpid.
15271 * config/i386/cpuid.h (bit_RDPID): New.
15272 * config/i386/driver-i386.c (host_detect_local_cpu):
15273 Detect RDPID feature.
15274 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
15275 * config/i386/i386-c.c (ix86_target_macros_internal):
15277 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
15278 (ix86_valid_target_attribute_inner_p): Add "rdpid".
15279 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
15280 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
15281 * config/i386/i386.md (define_insn "rdpid"): New.
15282 * config/i386/i386.opt Add -mrdpid.
15283 * config/i386/immintrin.h (_rdpid_u32): New.
15285 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
15287 PR rtl-optimization/79541
15288 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
15289 instead of transforming it into USE.
15291 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
15293 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
15294 If HONOR_SNANS (SFmode) force the input to a register.
15295 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
15296 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
15297 an frsp or similar insn.
15299 2017-02-17 Martin Liska <mliska@suse.cz>
15301 PR rtl-optimization/79577
15302 * params.def (selsched-max-sched-times): Increase minimum to 1.
15304 2017-02-17 Martin Liska <mliska@suse.cz>
15306 PR rtl-optimization/79574
15307 * gcse.c (want_to_gcse_p): Prevent integer overflow.
15309 2017-02-17 Martin Liska <mliska@suse.cz>
15311 PR tree-optimization/79529
15312 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
15313 ssa_defined_default_def_p to handle cases which are implicitly
15315 * tree-ssa.c (ssa_defined_default_def_p): New function.
15316 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
15317 which are implicitly defined.
15318 * tree-ssa.h (ssa_defined_default_def_p): Declare.
15320 2017-02-17 Richard Biener <rguenther@suse.de>
15322 PR middle-end/79576
15323 * params.def (max-ssa-name-query-depth): Limit to 10.
15325 2017-02-17 Richard Biener <rguenther@suse.de>
15327 PR tree-optimization/79552
15328 * tree-ssa-structalias.c (visit_loadstore): Properly verify
15331 2017-02-17 Richard Biener <rguenther@suse.de>
15334 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
15336 2017-02-17 Marek Polacek <polacek@redhat.com>
15338 PR middle-end/79536
15339 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
15340 (fold_negate_expr): New wrapper.
15342 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
15344 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
15345 Correct terminology and de-emphasize pre-standard behavior.
15347 2017-02-16 Alan Modra <amodra@gmail.com>
15349 PR rtl-optimization/79286
15350 * ira.c (def_dominates_uses): New function.
15351 (update_equiv_regs): Don't create an equivalence for insns that
15352 may trap where the register def does not dominate the use.
15354 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
15356 PR rtl-optimization/78127
15357 * lra.c (lra): Call lra_eliminate before finish the loop after
15360 2017-02-16 Richard Biener <rguenther@suse.de>
15362 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
15364 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
15365 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
15366 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
15367 (isl_val_int_from_wi): New function.
15368 (extract_affine_gmp): Rename to ...
15369 (extract_affine_wi): ... this, take a widest_int.
15370 (extract_affine_int): Just wrap extract_affine_wi.
15371 (add_param_constraints): Use isl_val_int_from_wi.
15372 (add_loop_constraints): Likewise, and extract_affine_wi.
15374 2017-02-15 Jeff Law <law@redhat.com>
15376 PR middle-end/79521
15377 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
15378 ira_init_register_move_cost_if_necessary.
15380 2017-02-15 Martin Sebor <msebor@redhat.com>
15382 PR middle-end/32003
15383 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
15384 removed in a prior commit.
15386 2017-02-15 Bin Cheng <bin.cheng@arm.com>
15388 PR tree-optimization/79347
15389 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
15390 counters during peeling.
15392 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
15394 * Makefile.in (site.exp): Remove "set ISLVER".
15396 2017-02-15 Jakub Jelinek <jakub@redhat.com>
15399 * real.c (real_from_integer): Call real_convert even for decimal.
15401 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
15404 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
15406 2017-02-14 Andrew Pinski <apinski@cavium.com>
15408 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
15409 cores and change the partno/implementer to be correct.
15410 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
15411 the 'B" as the implementer.
15412 * config/aarch64/aarch64-tune.md: Regenerate.
15414 2017-02-14 Carl Love <cel@us.ibm.com>
15416 * config/rs6000/rs6000.c: Add case statement entry to make the
15417 xvcvuxdsp built-in argument unsigned.
15418 * config/rs6000/vsx.md: Fix the source and return operand types so they
15419 match the instruction definitions from the ISA document. Fix typo
15420 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
15423 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
15426 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
15427 member early_clobber_alts.
15428 * lra-lives.c (reg_early_clobber_p): New.
15429 (process_bb_lives): Use it.
15430 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
15431 (debug_operand_data): Initialize early_clobber_alts.
15432 (setup_operand_alternative): Set up early_clobber_alts.
15433 (collect_non_operand_hard_regs): Ditto. Pass early clobber
15434 alternatives to new_insn_reg.
15435 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
15437 (lra_update_insn_regno_info): Pass the new arg.
15439 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15441 PR middle-end/79505
15442 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
15443 (new_oacc_loop_raw): Don't clear already cleared fields.
15446 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
15447 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
15448 _mm512_prefetch_i64gather_ps): New inline functions and macros.
15450 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
15453 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
15455 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
15458 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
15459 the extra instruction to the right place to store 128-bit constant
15462 2017-02-14 Martin Sebor <msebor@redhat.com>
15464 PR middle-end/79448
15465 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
15466 warning for strings of unknown length.
15468 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
15470 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
15472 2017-02-14 Jeff Law <law@redhat.com>
15475 * ira-costs.c (scan_one_insn): Initialize register move costs
15476 for pseudos seen in USE/CLOBBER insns.
15478 PR tree-optimization/79095
15479 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
15480 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
15481 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
15482 if the operands are known to be not equal, then the resulting range
15484 (intersect_ranges): If the new range is ~[0,0] and the old range is
15485 wide, then prefer ~[0,0].
15486 * tree-vrp.c (overflow_comparison_p_1): New function.
15487 (overflow_comparison_p): New function.
15488 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
15489 if NAME is used in an overflow test.
15490 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
15491 overflow check that can be expressed as an equality test, then adjust
15492 ops to be that equality test.
15494 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15496 * config/s390/s390-builtin-types.def: Remove flags argument.
15497 * config/s390/s390.c (s390_init_builtins): Likewise.
15499 2017-02-14 Martin Liska <mliska@suse.cz>
15501 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
15502 vector. Fix trailing white spaces.
15504 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
15506 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
15509 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15511 PR rtl-optimization/68664
15512 * config/arm/arm.c (arm_sched_can_speculate_insn):
15513 New function. Declare prototype.
15514 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15516 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15518 PR rtl-optimization/68664
15519 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
15521 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15523 2017-02-14 Amit Pawar <amit.pawar@amd.com>
15525 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
15526 max skip bytes for function, loop and jump.
15528 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15530 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
15531 ABS_EXPR for gimple dump.
15533 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15536 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
15538 PR tree-optimization/79408
15539 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
15540 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
15541 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
15542 also if rhs1 is INTEGER_CST.
15544 2017-02-14 Richard Biener <rguenther@suse.de>
15546 PR middle-end/79432
15547 * tree-into-ssa.c (insert_phi_nodes): When the function can
15548 have abnormal edges rewrite SSA names with broken use-def
15549 dominance out of SSA and register them for PHI insertion.
15551 2017-02-13 Martin Sebor <msebor@redhat.com>
15553 PR middle-end/79496
15554 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
15555 clearing info.nowrite flag when snprintf size argument is a range.
15557 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15559 * cprop.c (cprop_jump): Add missing space in string literal.
15560 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
15561 (get_constraint_for_component_ref): Likewise.
15562 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
15563 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
15564 * lra-constraints.c (process_alt_operands): Likewise.
15565 * ipa-inline.c (inline_small_functions): Likewise.
15566 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
15567 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15568 * trans-mem.c (diagnose_tm_1_op): Likewise.
15569 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
15570 (grid_parallel_clauses_gridifiable): Likewise.
15572 * config/nvptx/mkoffload.c (process): Add space in between
15575 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
15576 "MOD4_SSE_REGS" and "ALL_REGS".
15578 * spellcheck.c (test_data): Add , in between "foo" and "food".
15580 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15583 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
15584 boundary crossing check and subsequent code generation agree.
15586 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15588 * config/aarch64/aarch64.c (has_memory_op): Delete.
15589 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
15592 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15594 PR rtl-optimization/79388
15595 PR rtl-optimization/79450
15596 * combine.c (distribute_notes): When removing TEM_INSN for which
15597 corresponding dest has last value recorded, invalidate that last
15600 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15602 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
15603 of explicit '@'. Add missing assembly comment marker on branch costs
15606 2017-02-13 Nathan Sidwell <nathan@acm.org>
15608 * gengtype-lex.l (<in_struct>): Add '/'.
15610 2017-02-13 Martin Liska <mliska@suse.cz>
15613 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
15615 2017-02-13 Richard Biener <rguenther@suse.de>
15617 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
15619 * configure: Re-generate.
15620 * config.in: Likewise.
15621 * graphite-dependences.c: Simplify as if
15622 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
15623 * graphite-isl-ast-to-gimple.c: Likewise.
15624 * graphite-optimize-isl.c: Likewise.
15625 * graphite-poly.c: Likewise.
15626 * graphite-sese-to-poly.c: Likewise.
15627 * graphite.h: Likewise.
15628 * toplev.c: Include isl/version.h and use isl_version () for
15629 printing the ISL version.
15630 * doc/install.texi: Update ISL requirement.
15632 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15634 * doc/standards.texi (Standards): Update reference to
15637 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15639 * doc/extend.texi (Named Address Spaces): sourceware.org now
15641 * doc/install.texi (Binaries): Ditto.
15644 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15646 * doc/cpp.texi: Replace "stringify"/"stringification" with C
15647 standard terminology "stringize"/"stringizing" throughout.
15648 * doc/cppinternals.texi: Likewise.
15650 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15652 * doc/extend.texi: Fix some spelling mistakes and typos.
15653 * doc/invoke.texi: Likewise.
15655 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15658 * params.def (inline-min-speedup) Change from 10 to 8.
15660 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15662 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
15665 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15668 * ipa-inline-analysis.c (get_minimal_bb): New function.
15669 (record_modified): Use it.
15670 (remap_edge_change_prob): Handle also ancestor functions.
15672 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
15674 * doc/contrib.texi (Contributors): Remove broken link into
15675 the Mauve CVS repository.
15677 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15679 PR middle-end/79454
15680 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
15681 result computation whenever lhs doesn't have vector mode, not
15682 just when it has BLKmode.
15684 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15686 * doc/makefile.texi (profiledbootstrap): Refer to the
15687 installation instructions only in textual form.
15689 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15692 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
15694 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15696 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
15697 (Specific): Update mingw-w64 reference.
15699 (Specific): Remove broken link to Renesas RX processor.
15701 2017-02-10 Richard Biener <rguenther@suse.de>
15703 * toplev.c (process_options): Do not mention obsolete graphite
15704 options when printing sorry message about missing graphite support.
15705 Mention -floop-nest-optimize.
15707 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
15709 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
15710 (vtst_p16): Likewise.
15711 (vtstq_p8): Likewise.
15712 (vtstq_p16): Likewise.
15714 (vtstq_p64): Likewise.
15715 * config/arm/arm_neon.h (vgetq_lane_p64): New.
15716 (vset_lane_p64): New.
15717 (vsetq_lane_p64): New.
15719 2017-02-10 Jakub Jelinek <jakub@redhat.com>
15721 PR tree-optimization/79411
15722 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
15723 stmt operands are SSA_NAMEs used in abnormal phis.
15724 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
15727 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15730 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
15733 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15735 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
15737 2017-02-09 Jakub Jelinek <jakub@redhat.com>
15739 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
15743 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
15744 not arbitrary TREE_CONSTANT.
15747 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
15748 "omp declare target link" attribute unless is_global_var.
15749 * omp-offload.c (find_link_var_op): Likewise.
15751 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
15752 Chung-Lin Tang <cltang@codesourcery.com>
15754 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
15756 (gimplify_adjust_omp_clauses): Don't delete TILE.
15757 (gimplify_omp_for): Deal with TILE.
15758 * internal-fn.c (expand_GOACC_TILE): New function.
15759 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
15761 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
15762 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
15764 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
15765 avoid DIV for outermost collapse var.
15766 (expand_oacc_for): Insert tile element loop as needed. Adjust.
15767 Remove out of date comments, fix whitespace.
15768 * omp-general.c (omp_extract_for_data): Deal with tiling.
15769 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
15770 adjust OLF_DIM_BASE value.
15771 (struct omp_for_data): Add tiling field.
15772 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
15773 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
15774 for auto loops. Remove default auto determining, moved to
15775 oacc_loop_fixed_partitions.
15776 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
15777 stmts, add e_mask field.
15778 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
15779 (oacc_thread_numbers): Use oacc_dim_call.
15780 (oacc_xform_tile): New.
15781 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
15782 (finish_oacc_loop): Adjust for ifns vector.
15783 (oacc_loop_discover_walk): Append loop abstraction sites to list,
15784 add case for GOACC_TILE fns.
15785 (oacc_loop_xform_loop): Delete.
15786 (oacc_loop_process): Iterate over call list directly, and add
15787 handling for GOACC_TILE fns.
15788 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
15790 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
15791 vector partitioning to outer loops. Assign 2 partitions to loops
15792 when available. Add TILE handling.
15793 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
15794 (execite_oacc_device_lower): Process GOACC_TILE fns,
15795 ignore unknown specs.
15796 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
15797 * tree.c (omp_clause_num_ops): Adjust TILE ops.
15798 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
15800 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
15802 * configure.ac (ACX_BUGURL): Update.
15803 * configure: Regenerate.
15805 2017-02-09 Richard Biener <rguenther@suse.de>
15807 PR tree-optimization/69823
15808 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
15809 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
15811 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15813 * config/arc/arc-c.def: Add __NPS400__ definition.
15814 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
15815 (TARGET_NPS400): Define.
15817 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15819 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
15821 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
15822 pointer, arch_info.
15823 (arc_cpu_types): Fill the arch_info field with a pointer into the
15824 arc_arch_types table.
15825 (arc_selected_cpu): Declare.
15826 * config/arc/arc.c (arc_selected_cpu): Make global.
15827 (arc_selected_arch): Delete.
15828 (arc_base_cpu): Delete.
15829 (arc_override_options): Remove references to deleted variables,
15830 update access to arch information.
15831 (ARC_OPT): Update access to arch information.
15832 (ARC_OPTX): Likewise.
15833 * config/arc/arc.h (arc_base_cpu): Remove declaration.
15834 (TARGET_ARC600): Update access to arch information.
15835 (TARGET_ARC601): Likewise.
15836 (TARGET_ARC700): Likewise.
15837 (TARGET_EM): Likewise.
15838 (TARGET_HS): Likewise.
15839 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
15842 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
15845 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
15846 condition/operands for integer GE/LE/GEU/LEU operations.
15848 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
15850 PR translation/79397
15851 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
15854 2017-02-08 Martin Jambor <mjambor@suse.cz>
15857 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
15858 whether allocation happened.
15859 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
15860 nothing was allocated.
15862 2017-02-08 Jakub Jelinek <jakub@redhat.com>
15864 PR tree-optimization/79408
15865 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
15866 constant, but SSA_NAME with a known integer range, use the minimum
15867 of that range instead of op1 to determine if modulo can be replaced
15868 with its first operand.
15870 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15872 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
15874 2017-02-08 Richard Biener <rguenther@suse.de>
15876 PR tree-optimization/71824
15877 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
15878 Check all loops contained in the merged region.
15880 2017-02-07 Andrew Pinski <apinski@cavium.com>
15882 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
15884 2017-02-07 Andrew Pinski <apinski@cavium.com>
15886 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
15887 (thunderxt88): Likewise.
15888 (thunderxt81): Disable LSE and change v8.1 to v8.
15889 (thunderxt83): Likewise.
15891 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15892 Richard Biener <rguenther@suse.de>
15894 PR middle-end/79399
15895 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
15896 type from int to size_t.
15897 * ira-costs.c (struct_costs_size): Change type from int to size_t.
15899 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15901 PR rtl-optimization/79386
15902 * cprop.c (bypass_conditional_jumps): Initialize
15903 bypass_last_basic_block already before splitting bbs after
15904 unconditional traps...
15905 (bypass_conditional_jumps): ... rather than here.
15908 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
15909 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
15910 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
15911 fix -masm=intel patterns.
15913 2017-02-07 Richard Biener <rguenther@suse.de>
15915 PR tree-optimization/79256
15916 PR middle-end/79278
15917 * builtins.c (get_object_alignment_2): Use min_align_of_type
15918 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
15919 and ADJUST_FIELD_ALIGN.
15921 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
15923 * doc/tm.texi: Regenerate.
15924 * stor-layout.c (layout_decl): Adjust.
15925 (update_alignment_for_field): Likewise.
15926 (place_field): Likewise.
15927 (min_align_of_type): Likewise.
15928 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
15929 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
15930 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
15931 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
15932 * config/frv/frv.c (frv_adjust_field_align): Likewise.
15933 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
15934 * config/i386/i386.c (x86_field_alignment): Likewise.
15935 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
15936 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
15937 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
15938 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
15939 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
15940 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
15944 2017-01-30 Richard Biener <rguenther@suse.de>
15946 PR tree-optimization/79256
15947 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
15948 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
15951 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
15953 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
15954 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
15955 builtins to SImode and emit a zero-extend, if necessary.
15957 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15959 * docs/invoke.texi (RISC-V Options): Alphabetize.
15961 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15963 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
15966 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15968 * config/riscv/riscv.c: New file.
15969 * common/config/riscv/riscv-common.c: Likewise.
15970 * config.gcc: Likewise.
15971 * config/riscv/constraints.md: Likewise.
15972 * config/riscv/elf.h: Likewise.
15973 * config/riscv/generic.md: Likewise.
15974 * config/riscv/linux.h: Likewise.
15975 * config/riscv/multilib-generator: Likewise.
15976 * config/riscv/peephole.md: Likewise.
15977 * config/riscv/pic.md: Likewise.
15978 * config/riscv/predicates.md: Likewise.
15979 * config/riscv/riscv-builtins.c: Likewise.
15980 * config/riscv/riscv-c.c: Likewise.
15981 * config/riscv/riscv-ftypes.def: Likewise.
15982 * config/riscv/riscv-modes.def: Likewise.
15983 * config/riscv/riscv-opts.h: Likewise.
15984 * config/riscv/riscv-protos.h: Likewise.
15985 * config/riscv/riscv.h: Likewise.
15986 * config/riscv/riscv.md: Likewise.
15987 * config/riscv/riscv.opt: Likewise.
15988 * config/riscv/sync.md: Likewise.
15989 * config/riscv/t-elf-multilib: Likewise.
15990 * config/riscv/t-linux: Likewise.
15991 * config/riscv/t-linux-multilib: Likewise.
15992 * config/riscv/t-riscv: Likewise.
15993 * configure.ac: Likewise.
15994 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
15995 Waterman as RISC-V maintainers.
15996 * doc/install.texi: Add RISC-V entries.
15997 * doc/invoke.texi: Add RISC-V options section.
15998 * doc/md.texi: Add RISC-V constraints section.
15999 * configure: Regenerated.
16001 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
16004 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
16005 false values to be constant vectors with all 0 or all 1 bits set.
16006 (vcondu<mode><mode>): Likewise.
16007 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
16009 (fpmask_comparison_operator): Update comment.
16010 (vecint_comparison_operator): New predicate.
16011 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
16012 vector conditionals when the true and false values are constant
16013 vectors with all 0 bits or all 1 bits set.
16015 2017-02-06 Martin Sebor <msebor@redhat.com>
16017 PR tree-optimization/79376
16018 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
16020 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
16022 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
16023 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
16024 to simplify split condition.
16026 2017-02-06 Jakub Jelinek <jakub@redhat.com>
16028 * omp-expand.c (oxpand_omp_atomic_fetch_op,
16029 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
16032 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
16034 PR rtl-optimization/68664
16035 * target.def (can_speculate_insn): New hook.
16036 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
16037 * doc/tm.texi: Regenerate.
16038 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
16039 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
16040 (rs6000_sched_can_speculate_insn): New function.
16042 2017-02-06 Jakub Jelinek <jakub@redhat.com>
16044 PR tree-optimization/79284
16045 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
16046 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
16047 vectorizable_mask_load_store, vectorizable_operation,
16048 vect_is_simple_cond, get_same_sized_vectype): Use it instead
16049 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
16050 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
16051 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
16052 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
16053 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
16054 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
16055 is_gimple_assign (stmt). Replace another such test with
16056 is_gimple_assign (stmt).
16058 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
16061 * config/avr/avr.c (rtl-iter.h): Include it.
16062 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
16063 (avr_legitimate_combined_insn): ...and implementation.
16065 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
16067 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
16068 * config/s390/s390.c (s390_const_operand_ok)
16069 (s390_canonicalize_comparison, s390_extract_part)
16070 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
16071 (s390_contiguous_bitmask_p, s390_rtx_costs)
16072 (legitimize_pic_address): Likewise.
16073 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
16074 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
16075 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
16076 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
16077 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
16079 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
16081 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
16082 REGNO($0) == REGNO($1).
16084 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16086 * config/s390/linux.h(SIZE_TYPE): Add comment.
16088 2017-02-06 Julian Brown <julian@codesourcery.com>
16089 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16090 Virendra Pathak <virendra.pathak@broadcom.com>
16092 * config/aarch64/aarch64-cores.def: Change the scheduler
16094 * config/aarch64/aarch64.md: Include thunderx2t99.md.
16095 * config/aarch64/thunderx2t99.md: New file.
16097 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16099 * doc/standards.texi (Go Language): Update link to language
16102 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
16104 * tree-eh.c (lower_resx): Sanitize profile.
16105 (cleanup_empty_eh_move_lp): Likewise.
16107 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
16110 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
16112 * cfgloopmanip.h (loop_version): Update prototype.
16113 * modulo-sched.c (sms_schedule): Update call of loop_version.
16114 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
16115 * tree-parloops.c (gen_parallel_loop): Likewise.
16116 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
16117 * tree-ssa-loop-split.c (split_loop): Likewise.
16118 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
16119 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
16121 2017-02-05 Martin Liska <mliska@suse.cz>
16124 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
16126 (print_operand_address): Initialize a struct to zero.
16128 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16130 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
16131 garbage collector only in textual form.
16133 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16135 * doc/extend.texi (x86 specific memory model extensions for
16136 transactional memory): Simplify a phrase.
16138 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
16141 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
16142 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
16143 (atomic_storedi_1): Likewise.
16145 2017-02-04 Jakub Jelinek <jakub@redhat.com>
16147 PR tree-optimization/79338
16148 * tree-parloops.c (gather_scalar_reductions): Don't call
16149 vect_analyze_loop_form for loop->inner before destroying loop's
16152 2017-02-03 Martin Sebor <msebor@redhat.com>
16154 PR tree-optimization/79327
16155 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
16156 when precision has resulted in leading zeros.
16157 (format_integer): Adjust the likely counter to assume an unknown
16158 argument that may be zero is non-zero.
16160 2017-02-03 Jason Merrill <jason@redhat.com>
16163 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
16164 avoid copying non-taken branch.
16166 2017-02-03 Jakub Jelinek <jakub@redhat.com>
16168 PR tree-optimization/79340
16169 * tree-vect-loop.c (vectorizable_reduction): Release
16170 vec_defs elements after safe_splicing them into other vectors.
16173 PR tree-optimization/79327
16174 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
16175 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
16177 (format_integer): Use wide_int_to_tree instead of build_int_cst
16178 + to_?hwi. If argmin is NULL, just set argmin and argmax to
16179 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
16180 of shortest and longest sequence.
16182 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
16184 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
16185 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
16187 2017-02-03 Walter Lee <walt@tilera.com>
16190 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
16191 after initial stackframe link reg save.
16192 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
16194 2017-02-03 Jakub Jelinek <jakub@redhat.com>
16197 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
16198 wu for stxssp alternative.
16200 2017-02-03 Martin Sebor <msebor@redhat.com>
16202 PR tree-optimization/79352
16203 * gimple-fold.c (get_range_strlen): Add argument.
16204 (get_range_strlen): Change return type to bool.
16205 (get_maxval_strlen): Pass in a dummy argument.
16206 * gimple-fold.h (get_range_strlen): Change return type to bool.
16207 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
16208 * tree.h (array_at_struct_end_p): Add argument.
16209 * tree.c (array_at_struct_end_p): Handle it.
16211 2017-02-03 Martin Liska <mliska@suse.cz>
16214 * multiple_target.c (create_dispatcher_calls): Redirect edge
16215 from a caller of a dispatcher.
16216 (expand_target_clones): Make the clones local.
16217 (ipa_target_clone): Do both target clones and resolvers.
16218 (ipa_dispatcher_calls): Remove the pass.
16219 (pass_dispatcher_calls::gate): Likewise.
16220 (make_pass_dispatcher_calls): Likewise.
16221 * passes.def (pass_target_clone): Put as very first IPA early
16224 2017-02-03 Martin Liska <mliska@suse.cz>
16226 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
16227 in case of a function with ifunc attribute.
16229 2017-02-03 Martin Liska <mliska@suse.cz>
16231 * cgraph.c (cgraph_node::dump): Dump function version info.
16232 * symtab.c (symtab_node::dump_base): Add missing new line.
16234 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
16236 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
16237 (ifcombine_ifandif): Use it.
16239 2017-02-03 Martin Liska <mliska@suse.cz>
16241 * doc/invoke.texi: Document default value for
16242 use-after-scope-direct-emission-threshold.
16244 2017-02-03 Martin Liska <mliska@suse.cz>
16246 PR tree-optimization/79339
16247 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
16248 (format_floating): Likewise.
16250 2017-02-03 Martin Liska <mliska@suse.cz>
16253 * ipa-prop.c (ipa_node_params_t::insert): Remove current
16255 (ipa_node_params_t::remove): Likewise.
16256 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
16257 initialization from removed ipa_node_params_t::insert.
16258 (ipa_node_params::~ipa_node_params): Move from removed
16259 ipa_node_params_t::release.
16260 * symbol-summary.h (symbol_summary::m_released): New member.
16261 Do not release a summary twice. Do not allow to call finalizer
16262 for types of a summary that live in GGC memory.
16264 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16266 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
16269 2017-02-02 Martin Sebor <msebor@redhat.com>
16271 PR middle-end/79275
16272 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
16273 (format_string): Tighten up the range of output for non-constant
16274 strings and correct the expected range for wide non-constant strings.
16276 2017-02-02 Martin Sebor <msebor@redhat.com>
16278 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
16280 PR middle-end/32003
16281 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
16283 (-fdump-tree-@var): Add to index and document how to come up
16284 with pass-specific option and dump file names.
16285 (-fdump-passes): Clarify where to look for output.
16287 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
16289 PR middle-end/77445
16290 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
16291 statistics of the analyzed path; allow threading for speed when
16292 any of BBs along the path are optimized for speed.
16294 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
16296 PR middle-end/78468
16297 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
16298 settings of the virtual registers.
16301 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
16303 * explow.c (get_dynamic_stack_size): Take known alignment of stack
16304 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
16307 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16309 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
16310 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
16312 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16314 * config/s390/s390.md: Add missing comments with the expanded
16316 * config/s390/vector.md: Likewise.
16317 * config/s390/vx-builtins.md: Likewise.
16319 2017-02-02 Jakub Jelinek <jakub@redhat.com>
16322 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
16323 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
16324 conditions on a single line.
16326 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16328 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
16329 __S390_VX__ to __VX__.
16331 2017-02-01 Andrew Pinski <apinski@cavium.com>
16333 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
16334 stmt_info to record_stmt_cost.
16335 (vect_get_known_peeling_cost): Pass stmt_info if known to
16337 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
16338 cpu_vector_cost field into
16339 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
16340 field into vec_int_stmt_cost and vec_fp_stmt_cost.
16341 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
16342 splitting of scalar_stmt_cost and vec_stmt_cost.
16343 (thunderx_vector_cost): Likewise.
16344 (cortexa57_vector_cost): LIkewise.
16345 (exynosm1_vector_cost): Likewise.
16346 (xgene1_vector_cost): Likewise.
16347 (thunderx2t99_vector_cost): Improve after the splitting of the two
16349 (aarch64_builtin_vectorization_cost): Update for the splitting of
16350 scalar_stmt_cost and vec_stmt_cost.
16352 2017-02-01 Torvald Riegel <triegel@redhat.com>
16353 Richard Henderson <rth@redhat.com>
16355 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
16356 conditional on existance of a fast atomic load.
16357 * optabs-query.c (can_atomic_load_p): New function.
16358 * optabs-query.h (can_atomic_load_p): Declare it.
16359 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
16360 no fast atomic load is available for the particular size of access.
16361 (expand_atomic_compare_and_swap): Likewise.
16362 (expand_atomic_load): Likewise.
16363 (expand_atomic_store): Likewise.
16364 (expand_atomic_fetch_op): Likewise.
16365 * testsuite/lib/target-supports.exp
16366 (check_effective_target_sync_int_128): Remove x86 because it provides
16367 no fast atomic load.
16368 (check_effective_target_sync_int_128_runtime): Likewise.
16370 2017-02-01 Richard Biener <rguenther@suse.de>
16372 * graphite.c: Include tree-vectorizer.h for find_loop_location.
16373 (graphite_transform_loops): Provide opt-info for optimized nests.
16374 * tree-parloop.c (parallelize_loops): Provide opt-info for
16375 parallelized loops.
16377 2017-02-01 Richard Biener <rguenther@suse.de>
16379 PR middle-end/79315
16380 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
16381 was not set before.
16383 2017-02-01 Richard Biener <rguenther@suse.de>
16385 PR tree-optimization/71824
16386 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
16387 Verify the loops are valid in the merged SESE region.
16388 (scop_detection::can_represent_loop_1): Check analyzing the
16389 evolution of the number of iterations in the region succeeds.
16391 2017-01-31 Ian Lance Taylor <iant@golang.org>
16393 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
16394 REG_ARGS_SIZE note to 32-bit push insns and call insn.
16396 2017-01-31 David Malcolm <dmalcolm@redhat.com>
16398 PR preprocessor/79210
16399 * input.c (get_substring_ranges_for_loc): Replace line_width
16400 assertion with error-handling.
16402 2017-01-31 Richard Biener <rguenther@suse.de>
16404 PR tree-optimization/77318
16405 * graphite-sese-to-poly.c (extract_affine): Fix assert.
16406 (create_pw_aff_from_tree): Take loop parameter.
16407 (add_condition_to_pbb): Pass loop of the condition to
16408 create_pw_aff_from_tree.
16410 2017-01-31 Jakub Jelinek <jakub@redhat.com>
16412 * config/s390/s390.c (s390_asan_shadow_offset): New function.
16413 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
16415 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
16419 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
16421 (convert_int_to_float128): Likewise.
16422 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
16423 (convert_int_to_float128): Likewise.
16424 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
16425 (UNSPEC_IEEE128_CONVERT): Likewise.
16426 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
16427 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
16428 Use local variables for IBM extended format.
16429 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
16430 (fix_trunc<mode>si2_fprs): Likewise.
16431 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
16432 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
16433 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
16434 to know that we can now have integers of all sizes in vector
16436 (fix<uns>_<mode>di2_hw): Likewise.
16437 (float<uns>_<mode>si2_hw): Likewise.
16438 (fix_<mode>si2_hw): Likewise.
16439 (fixuns_<mode>si2_hw): Likewise.
16440 (float<uns>_<mode>di2_hw): Likewise.
16441 (float_<mode>di2_hw): Likewise.
16442 (float_<mode>si2_hw): Likewise.
16443 (floatuns_<mode>di2_hw): Likewise.
16444 (floatuns_<mode>si2_hw): Likewise.
16445 (xscvqp<su>wz_<mode>): Delete, no longer used.
16446 (xscvqp<su>dz_<mode>): Likewise.
16447 (xscv<su>dqp_<mode>): Likewise.
16448 (ieee128_mfvsrd_64bit): Likewise.
16449 (ieee128_mfvsrd_32bit): Likewise.
16450 (ieee128_mfvsrwz): Likewise.
16451 (ieee128_mtvsrw): Likewise.
16452 (ieee128_mtvsrd_64bit): Likewise.
16453 (ieee128_mtvsrd_32bit): Likewise.
16455 2017-01-31 Martin Liska <mliska@suse.cz>
16458 * ipa-prop.c (ipa_free_all_node_params): Call release method
16459 instead of ~sumbol_summary to not to trigger double times
16462 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
16464 PR tree-optimization/71691
16465 * bitmap.h (class auto_bitmap): New.
16466 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
16467 is_maybe_undefined instead of ssa_undefined_value_p.
16469 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16471 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
16472 __S390_ARCH_LEVEL__ to __ARCH__.
16474 2017-01-31 Jakub Jelinek <jakub@redhat.com>
16476 PR tree-optimization/79267
16477 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
16478 if should_remove_lhs_p is true.
16480 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
16483 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
16484 (add_alignment_attribute): New.
16485 (base_type_die): Add alignment attribute.
16486 (subrange_type_die): Likewise.
16487 (modified_type_die): Likewise.
16488 (gen_array_type_die): Likewise.
16489 (gen_descr_array_type_die: Likewise.
16490 (gen_enumeration_type_die): Likewise.
16491 (gen_subprogram_die): Likewise.
16492 (gen_variable_die): Likewise.
16493 (gen_field_die): Likewise.
16494 (gen_ptr_to_mbr_type_die): Likewise.
16495 (gen_struct_or_union_type_die): Likewise.
16496 (gen_subroutine_type_die): Likewise.
16497 (gen_typedef_die): Likewise.
16498 (base_type_cmp): Compare alignment attribute.
16500 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16503 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
16504 (setb_unsigned) New pattern for setb with CCUNS.
16505 * config/rs6000/rs6000.c (expand_block_compare): Use a different
16506 subfc./subfe sequence to avoid overflow problems. Generate a
16507 shorter sequence with cmpld/setb for power9.
16508 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
16509 for generating subfc. instruction.
16510 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
16511 now uses this instruction.
16513 2017-01-30 Ian Lance Taylor <iant@google.com>
16516 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
16517 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
16519 2017-01-30 Martin Sebor <msebor@redhat.com>
16521 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
16522 Move constant to the right of a relational operator.
16523 (get_mpfr_format_length, format_character, format_string): Ditto.
16524 (should_warn_p, maybe_warn): Same.
16526 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
16528 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
16531 * asan.c (get_translation_unit_decl): Remove function.
16532 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
16534 2017-01-30 Martin Liska <mliska@suse.cz>
16536 PR gcov-profile/79259
16537 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
16538 -fprofile-generate.
16540 2017-01-30 Martin Liska <mliska@suse.cz>
16543 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
16544 Initialize variables with NULL value.
16546 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
16549 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
16551 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
16553 2017-01-30 Richard Biener <rguenther@suse.de>
16555 PR tree-optimization/79276
16556 * tree-vrp.c (process_assert_insertions): Properly adjust common
16557 when removing a duplicate.
16559 * gcc.dg/torture/pr79276.c: New testcase.
16561 2017-01-30 Richard Biener <rguenther@suse.de>
16563 PR tree-optimization/79256
16564 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
16565 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
16567 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
16569 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
16572 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
16573 ("*r<noxa>sbg_<mode>_sll_bitmask")
16574 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
16575 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
16576 Use contiguous_bitmask_nowrap_operand.
16578 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16581 * config/rs6000/altivec.h (vec_xl): Revise #define.
16582 (vec_xst): Likewise.
16584 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
16586 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
16588 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
16590 PR rtl-optimization/79194
16591 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
16592 traps before call to bypass_conditional_jumps.
16594 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16596 PR tree-optimization/71374
16597 * lra-constraints.c (check_conflict_input_operands): New.
16598 (match_reload): Use it.
16600 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16603 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
16604 account to calculate conflict_set.
16606 2017-01-27 Bin Cheng <bin.cheng@arm.com>
16608 PR rtl-optimization/78559
16609 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
16610 other_insn in combine.
16612 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16614 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
16615 uint16_type_node for BT_UINT16.
16617 2017-01-27 David Malcolm <dmalcolm@redhat.com>
16619 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
16620 "RTL Tests" to menu.
16621 (GIMPLE Tests): New node.
16622 (RTL Tests): New node.
16624 2017-01-27 Richard Biener <rguenther@suse.de>
16626 PR tree-optimization/79245
16627 * tree-loop-distribution.c (distribute_loop): Apply cost
16628 modeling also to detected patterns.
16630 2017-01-27 Richard Biener <rguenther@suse.de>
16632 PR tree-optimization/71433
16633 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
16634 (compare_assert_loc): New function.
16635 (process_assert_insertions): Sort and optimize assert locations
16636 to remove duplicates and push down identical assertions on
16637 edges to their destination block.
16639 2017-01-27 Richard Biener <rguenther@suse.de>
16641 PR tree-optimization/79244
16642 * tree-vrp.c (remove_range_assertions): Forcefully propagate
16643 out SSA names even if abnormal.
16645 2017-01-27 Jakub Jelinek <jakub@redhat.com>
16647 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
16648 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
16649 instead of MPFR_RNDN.
16651 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
16654 * arm.c (arm_option_override): Don't call build_target_option_node
16655 until after doing all option overrides.
16656 (arm_valid_target_attribute_tree): Likewise.
16658 2017-01-27 Martin Liska <mliska@suse.cz>
16660 * doc/invoke.texi (-fprofile-arcs): Document profiling support
16661 for {cd}tors and C++ {cd}tors.
16663 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
16665 * config/s390/s390.md ("*setmem_long_and")
16666 ("*setmem_long_and_31z"): Use zero_extend instead of and.
16668 2017-01-26 Martin Sebor <msebor@redhat.com>
16670 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
16673 2017-01-26 Martin Sebor <msebor@redhat.com>
16675 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
16676 HAVE_DFmode before using XFmode or DFmode.
16677 (parse_directive): Avoid using the z length modifier to avoid
16678 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
16680 PR middle-end/78703
16681 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
16682 to accept adjustment as an array.
16683 (get_int_range): New function.
16684 (struct directive): Make width and prec arrays.
16685 (directive::set_width, directive::set_precision): Call get_int_range.
16686 (format_integer, format_floating): Handle width and precision ranges.
16687 (format_string, parse_directive): Same.
16689 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16692 * dwarf2out.c (generate_skeleton_bottom_up): For children with
16693 comdat_type_p set, just clone them, but keep the children in the
16697 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
16698 which have direct callers with -fvar-tracking-assignments enabled
16700 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
16701 inside of type units.
16703 2017-01-26 Martin Sebor <msebor@redhat.com>
16705 PR middle-end/78703
16706 * gimple-ssa-sprintf.c (struct result_range): Add likely and
16708 (struct format_result): Replace number_chars, number_chars_min,
16709 and number_chars_max with a single member of struct result_range.
16711 (format_result::operator+=): Adjust.
16712 (struct fmtresult): Remove bounded. Handle likely and unlikely
16714 (fmtresult::adjust_for_width_or_precision): New function.
16715 (fmtresult:type_max_digits): New function.
16716 (bytes_remaining): Handle likely and unlikely counters.
16717 (min_bytes_remaining): Remove.
16718 (format_percent): Simplify.
16719 (format_integer, format_floating): Set likely and unlikely counters.
16720 (get_string_length, format_character, format_string): Same.
16721 (format_plain, should_warn_p): New function.
16722 (maybe_warn): Call should_warn_p. Update diagnostic messages
16723 and handle those for all directives, including plain strings.
16724 (format_directive): Handle likely and unlikely counters.
16725 Remove unnecessary quoting from diagnostics. Add an informational
16727 (add_bytes): Remove.
16728 (pass_sprintf_length::compute_format_length): Simplify.
16729 (try_substitute_return_value): Handle likely and unlikely counters.
16731 2017-01-26 Carl Love <cel@us.ibm.com>
16733 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
16734 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
16736 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
16739 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
16740 endianess for subregs into account.
16741 * lra-constraints.c (lra_constraints): Do risky transformations
16742 always on the first iteration.
16743 * lra-lives.c (check_pseudos_live_through_calls): Add arg
16744 last_call_used_reg_set.
16745 (process_bb_lives): Define and use last_call_used_reg_set.
16746 * lra.c (lra): Always continue after lra_constraints on the first
16749 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
16751 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
16753 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
16755 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16757 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
16758 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
16759 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
16760 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
16761 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
16762 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
16763 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
16764 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
16765 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
16767 2017-01-26 Marek Polacek <polacek@redhat.com>
16770 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
16771 for the third operand.
16773 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16775 PR middle-end/79236
16776 * omp-low.c (struct omp_context): Add simt_stmt field.
16777 (scan_omp_for): Return omp_context *.
16778 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
16779 context to the _simt_ SIMD stmt.
16780 (lower_omp_for): For combined SIMD with sibling _simt_
16781 SIMD, make sure to use the same decls in _looptemp_
16782 clauses as in the sibling.
16784 2017-01-26 David Sherwood <david.sherwood@arm.com>
16786 PR middle-end/79212
16787 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
16790 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16793 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
16794 emit fld b; fld a; if possible.
16796 * brig-builtins.def: Update copyright years.
16797 * config/arm/arm_acle_builtins.def: Update copyright years.
16799 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16802 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
16803 constraint instead of o for the stxsd instruction.
16805 2017-01-25 Carl Love <cel@us.ibm.com>
16807 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
16808 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
16810 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
16812 * doc/invoke.texi (C++ Dialect Options): Fix typo.
16814 2017-01-25 Richard Biener <rguenther@suse.de>
16816 PR tree-optimization/69264
16817 * target.def (vector_alignment_reachable): Improve documentation.
16818 * doc/tm.texi: Regenerate.
16819 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
16821 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
16822 earlier changes with respect to TYPE_USER_ALIGN.
16823 (vector_alignment_reachable_p): Likewise. Improve dumping.
16825 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16828 * config/arm/arm.md (xordi3): Force constant operand into a register
16831 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16833 * doc/invoke.texi (-fstore-merging): Correct default optimization
16834 levels at which it is enabled.
16835 (-O): Move -fstore-merging from list to...
16838 2017-01-25 Richard Biener <rguenther@suse.de>
16841 * omp-expand.c: Include debug.h.
16842 (expand_omp_taskreg): Make sure to generate early debug before
16843 outlining anything from a function.
16844 (expand_omp_target): Likewise.
16845 (grid_expand_target_grid_body): Likewise.
16847 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
16850 * asan.c (get_translation_unit_decl): New function.
16851 (asan_add_global): Extract modules file name from globals
16852 TRANSLATION_UNIT_DECL name.
16854 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
16857 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
16858 for long calls with APCS frame and VFP.
16860 2017-01-24 David Malcolm <dmalcolm@redhat.com>
16862 * cfg.c (original_copy_tables_initialized_p): New function.
16863 * cfg.h (original_copy_tables_initialized_p): New decl.
16864 * cfgrtl.c (relink_block_chain): Guard the call to
16865 free_original_copy_tables with a call to
16866 original_copy_tables_initialized_p.
16867 * cgraph.h (symtab_node::native_rtl_p): New decl.
16868 * cgraphunit.c (symtab_node::native_rtl_p): New function.
16869 (symtab_node::needed_p): Don't assert for early assembly output
16870 for __RTL functions.
16871 (cgraph_node::finalize_function): Set "force_output" for __RTL
16873 (cgraph_node::analyze): Bail out early for __RTL functions.
16874 (analyze_functions): Update assertion to support __RTL functions.
16875 (cgraph_node::expand): Bail out early for __RTL functions.
16876 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
16878 * function.h (struct function): Update comment for field
16880 * gimple-expr.c: Include "tree-pass.h".
16881 (gimple_has_body_p): Return false for __RTL functions.
16882 * Makefile.in (OBJS): Add run-rtl-passes.o.
16883 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
16885 (gcc::pass_manager::get_clean_slate): New accessor.
16886 * passes.c: Include "insn-addr.h".
16887 (should_skip_pass_p): Add logging. Update logic for running
16888 "expand" to be compatible with both __GIMPLE and __RTL. Guard
16889 property-provider override so it is only done for gimple passes.
16891 (skip_pass): New function.
16892 (execute_one_pass): Call skip_pass when skipping passes.
16893 * read-md.c (md_reader::read_char): Support filtering
16894 the input to a subset of line numbers.
16895 (md_reader::md_reader): Initialize fields
16896 m_first_line and m_last_line.
16897 (md_reader::read_file_fragment): New function.
16898 * read-md.h (md_reader::read_file_fragment): New decl.
16899 (md_reader::m_first_line): New field.
16900 (md_reader::m_last_line): New field.
16901 * read-rtl-function.c (function_reader::create_function): Only
16902 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
16903 curr_properties. Set DECL_INITIAL to a dummy block.
16904 (read_rtl_function_body_from_file_range): New function.
16905 * read-rtl-function.h (read_rtl_function_body_from_file_range):
16907 * run-rtl-passes.c: New file.
16908 * run-rtl-passes.h: New file.
16910 2017-01-24 Jeff Law <law@redhat.com>
16912 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
16915 2017-01-24 Bin Cheng <bin.cheng@arm.com>
16917 PR tree-optimization/79159
16918 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
16919 (record_nonwrapping_iv): Improve boundary using above function if no
16920 value range information.
16922 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
16923 Martin Jambor <mjambor@suse.cz>
16925 * brig-builtins.def: New file.
16926 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
16927 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
16928 (DEF_HSAIL_SAT_BUILTIN): Likewise.
16929 (DEF_HSAIL_INTR_BUILTIN): Likewise.
16930 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
16931 * builtin-types.def (BT_INT8): New.
16932 (BT_INT16): Likewise.
16933 (BT_UINT8): Likewise.
16934 (BT_UINT16): Likewise.
16935 (BT_FN_ULONG): Likewise.
16936 (BT_FN_UINT_INT): Likewise.
16937 (BT_FN_UINT_ULONG): Likewise.
16938 (BT_FN_UINT_LONG): Likewise.
16939 (BT_FN_UINT_PTR): Likewise.
16940 (BT_FN_ULONG_PTR): Likewise.
16941 (BT_FN_INT8_FLOAT): Likewise.
16942 (BT_FN_INT16_FLOAT): Likewise.
16943 (BT_FN_UINT32_FLOAT): Likewise.
16944 (BT_FN_UINT16_FLOAT): Likewise.
16945 (BT_FN_UINT8_FLOAT): Likewise.
16946 (BT_FN_UINT64_FLOAT): Likewise.
16947 (BT_FN_UINT16_UINT32): Likewise.
16948 (BT_FN_UINT32_UINT16): Likewise.
16949 (BT_FN_UINT16_UINT16_UINT16): Likewise.
16950 (BT_FN_INT_PTR_INT): Likewise.
16951 (BT_FN_UINT_PTR_UINT): Likewise.
16952 (BT_FN_LONG_PTR_LONG): Likewise.
16953 (BT_FN_ULONG_PTR_ULONG): Likewise.
16954 (BT_FN_VOID_UINT64_UINT64): Likewise.
16955 (BT_FN_UINT8_UINT8_UINT8): Likewise.
16956 (BT_FN_INT8_INT8_INT8): Likewise.
16957 (BT_FN_INT16_INT16_INT16): Likewise.
16958 (BT_FN_INT_INT_INT): Likewise.
16959 (BT_FN_UINT_FLOAT_UINT): Likewise.
16960 (BT_FN_FLOAT_UINT_UINT): Likewise.
16961 (BT_FN_ULONG_UINT_UINT): Likewise.
16962 (BT_FN_ULONG_UINT_PTR): Likewise.
16963 (BT_FN_ULONG_ULONG_ULONG): Likewise.
16964 (BT_FN_UINT_UINT_UINT): Likewise.
16965 (BT_FN_VOID_UINT_PTR): Likewise.
16966 (BT_FN_UINT_UINT_PTR: Likewise.
16967 (BT_FN_UINT32_UINT64_PTR): Likewise.
16968 (BT_FN_INT_INT_UINT_UINT): Likewise.
16969 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
16970 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
16971 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
16972 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
16973 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
16974 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
16975 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
16976 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
16977 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
16978 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
16979 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
16980 * doc/frontends.texi: List BRIG FE.
16981 * doc/install.texi (Testing): Add BRIG tesring requirements.
16982 * doc/invoke.texi (Overall Options): Mention BRIG.
16983 * doc/standards.texi (Standards): Doucment BRIG HSA version.
16985 2017-01-24 Richard Biener <rguenther@suse.de>
16987 PR translation/79208
16988 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
16990 2017-01-24 Martin Jambor <mjambor@suse.cz>
16993 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
16994 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
16995 and known_contexts.
16997 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
16999 PR middle-end/79123
17000 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
17001 casts from signed to unsigned really don't have a range.
17003 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
17005 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
17006 GMP_RNDx for compatiblity.
17008 2017-01-24 Martin Liska <mliska@suse.cz>
17011 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
17012 that would prevent us to call alloca with -1 as argument.
17014 2017-01-24 Jakub Jelinek <jakub@redhat.com>
17016 * dwarf2out.c (output_compilation_unit_header, output_file_names):
17017 Avoid -Wformat-security warning.
17019 2017-01-23 Andrew Pinski <apinski@cavium.com>
17021 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
17024 2017-01-23 Martin Sebor <msebor@redhat.com>
17026 PR middle-end/78703
17027 * gimple-ssa-sprintf.c (warn_level): New global.
17028 (format_integer): Use it here and throughout the rest of the file.
17029 Use the same switch to compute sign as base.
17030 (maybe_warn): New function.
17031 (format_directive): Factor out warnings into maybe_warn.
17032 Add debugging output. Use warn_level.
17033 (add_bytes): Use warn_level.
17034 (pass_sprintf_length::compute_format_length): Add debugging output.
17035 (try_substitute_return_value): Same.
17036 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
17038 PR middle-end/78703
17039 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
17040 (struct fmtresult, format_integer, format_floating): Adjust.
17041 (fmtresult::fmtresult): Set max correctly in two argument ctor.
17042 (get_string_length, format_string,format_directive): Same.
17043 (pass_sprintf_length::compute_format_length): Same.
17044 (try_substitute_return_value): Simplify slightly.
17046 PR middle-end/78703
17047 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
17048 (fmtresult::operator+=): Outlined.
17049 (struct fmtresult): Add ctors.
17050 (struct conversion_spec): Rename...
17051 (struct directive): ...to this. Add and remove data members.
17052 (directive::set_width, directive::set_precision): New functions.
17053 (format_percent): Use fmtresult ctor.
17054 (get_width_and_precision): Remove.
17055 (format_integer): Make naming changes. Avoid computing width and
17057 (format_floating): Same. Adjust indentation.
17058 (format_character, format_none): New functions.
17059 (format_string): Moved character handling to format_character.
17060 (format_directive): Remove arguments, change return type.
17061 (parse_directive): New function.
17062 (pass_sprintf_length::compute_format_length): Move directive
17063 parsing to parse_directive.
17065 2017-01-23 Jakub Jelinek <jakub@redhat.com>
17067 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
17068 (assign_assembler_name_if_needed): ... this.
17069 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
17070 (assign_assembler_name_if_needed): ... this.
17071 (free_lang_data_in_cgraph): Adjust callers.
17072 * cgraphunit.c (cgraph_node::analyze): Likewise.
17073 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
17076 2017-01-23 Richard Biener <rguenther@suse.de>
17078 PR tree-optimization/79088
17079 PR tree-optimization/79188
17080 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
17081 resetting loop bounds after last path deletion. Reset loop
17082 bounds of the target loop, make code match the comments.
17083 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
17084 Make sure loops need no fixups.
17086 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
17088 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
17089 exponent support with double type for first argument.
17090 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
17091 type returned by __builtin_vec_extract_sig,
17092 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
17093 functions from "vector int" to "vector unsigned int" or from
17094 "vector long long int" to "vector unsigned long long int".
17095 Changed type returned by __builtin_vec_extract_exp,
17096 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
17097 functions from "vector int" to "vector unsigned int" or from
17098 "vector long long int" to "vector unsigned long long int".
17099 Changed return type of __builtin_vec_test_data_class,
17100 __builtin_vec_test_data_class_sp, and
17101 __builtin_vec_test_data_class_dp from "vector int" to
17102 "vector bool int" or from "vector long long int" to "vector bool
17103 long long int" and changed second argument type from "unsigned
17104 int" to "int". Added new overloaded function forms "vector float
17105 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
17106 "vector float __builtin_vec_insert_exp_sp (vector float, vector
17107 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
17108 double, vector unsigned long long int)" and "vector double
17109 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
17110 long int)". Changed return type of
17111 __builtin_scalar_test_data_class and
17112 __builtin_scalar_test_data_class_sp and
17113 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
17114 int" and changed second argument from "unsigned int" to "int".
17115 Changed type returned by __builtin_scalar_test_neg,
17116 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
17117 from "int" to "bool int". Added new overloaded function form
17118 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
17119 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
17120 exponent double-precision with floating point first argument.
17121 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
17122 documentation of scalar_test_data_class, scalar_test_neg,
17123 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
17124 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
17125 vec_test_data_class built-in functions to reflect refinements in
17126 their type signatures.
17128 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
17130 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
17132 (aarch64_elf_asm_destructor): Likewise.
17134 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
17136 PR rtl-optimization/78634
17137 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
17138 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
17139 * ifcvt.c (noce_try_cmove): Add missing cost check.
17141 PR rtl-optimization/71724
17142 * combine.c (if_then_else_cond): Look for situations where it is
17143 beneficial to undo the work of one of the recursive calls.
17145 2017-01-23 Bin Cheng <bin.cheng@arm.com>
17147 PR tree-optimization/70754
17148 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
17149 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
17150 combined stmt before it if not NULL.
17151 (combine_chains): Process refs reversely and compute dominance point
17154 2017-01-23 Martin Liska <mliska@suse.cz>
17156 PR tree-optimization/79196
17157 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
17158 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
17160 (strlen_optimize_stmt): Call the renamed function.
17162 2017-01-23 Michael Matz <matz@suse.de>
17164 PR tree-optimization/78384
17165 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
17167 2017-01-23 Richard Biener <rguenther@suse.de>
17169 PR tree-optimization/79186
17170 * tree-vrp.c (register_new_assert_for): Make sure we've seen
17171 both incoming edges before moving an assert.
17173 2017-01-23 Martin Jambor <mjambor@suse.cz>
17175 * ipa-prop.c (load_from_param_1): Removed.
17176 (load_from_unmodified_param): Bits from load_from_param_1 put back
17178 (load_from_param): Removed.
17179 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
17180 with stmt. Reverted back to use of load_from_unmodified_param.
17182 2017-01-23 Martin Jambor <mjambor@suse.cz>
17185 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
17186 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
17187 field a pointer to garbage collected vector, mark lattices and
17188 ipcp_orig_node with GTY((skip)).
17189 (ipa_get_param_count): Adjust to descriptors being a pointer.
17190 (ipa_get_param): Likewise.
17191 (ipa_get_type): Likewise.
17192 (ipa_get_param_move_cost): Likewise.
17193 (ipa_set_param_used): Likewise.
17194 (ipa_get_controlled_uses): Likewise.
17195 (ipa_set_controlled_uses): Likewise.
17196 (ipa_is_param_used): Likewise.
17197 (ipa_node_params_t): Move into garbage collector. New methods insert
17199 (ipa_node_params_sum): Annotate wth GTY(()).
17200 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
17202 (ipa_load_from_parm_agg): Adjust declaration.
17203 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
17204 * ipa-profile.c (ipa_profile): Likewise.
17205 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
17206 (ipa_populate_param_decls): Make descriptors parameter garbage
17208 (ipa_dump_param): Adjust to descriptors being a pointer.
17209 (ipa_alloc_node_params): Likewise.
17210 (ipa_initialize_node_params): Likewise.
17211 (load_from_param_1): Make descriptors parameter garbage collected.
17212 (load_from_unmodified_param): Likewise.
17213 (load_from_param): Likewise.
17214 (ipa_load_from_parm_agg): Likewise.
17215 (ipa_node_params::~ipa_node_params): Removed.
17216 (ipa_free_all_node_params): Remove call to delete operator.
17217 (ipa_node_params_t::insert): New.
17218 (ipa_node_params_t::remove): Likewise.
17219 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
17220 copy known_csts and known_contexts vectors.
17221 (ipa_read_node_info): Adjust to descriptors being a pointer.
17222 (ipcp_modif_dom_walker): Make m_descriptors field garbage
17224 (ipcp_transform_function): Make descriptors variable garbage
17227 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
17229 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
17230 * config/i386/avx512dqintrin.h: Ditto.
17231 * config/i386/avx512fintrin.h: Ditto.
17232 * config/i386/i386.c: Handle new builtins.
17233 * config/i386/i386-builtin.def: Add new builtins.
17234 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
17235 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
17237 2017-01-23 Jakub Jelinek <jakub@redhat.com>
17238 Martin Liska <mliska@suse.cz>
17240 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
17241 * asan.c (asan_expand_poison_ifn): Support stores and use
17242 appropriate ASAN report function.
17243 * internal-fn.c (expand_ASAN_POISON_USE): New function.
17244 * internal-fn.def (ASAN_POISON_USE): Declare.
17245 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
17246 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
17247 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
17248 ASAN_POISON calls w/o LHS.
17249 * tree-ssa.c (execute_update_addresses_taken): Create clobber
17250 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
17251 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
17252 * gimplify.c (asan_poison_variables): Add attribute
17253 use_after_scope_memory to variables that really needs to live
17255 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
17256 having the attribute.
17258 2017-01-23 Martin Liska <mliska@suse.cz>
17260 * asan.c (create_asan_shadow_var): New function.
17261 (asan_expand_poison_ifn): Likewise.
17262 * asan.h (asan_expand_poison_ifn): New declaration.
17263 * internal-fn.c (expand_ASAN_POISON): Likewise.
17264 * internal-fn.def (ASAN_POISON): New builtin.
17265 * sanopt.c (pass_sanopt::execute): Expand
17266 asan_expand_poison_ifn.
17267 * tree-inline.c (copy_decl_for_dup_finish): Make function
17269 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
17270 * tree-ssa.c (is_asan_mark_p): New function.
17271 (execute_update_addresses_taken): Rewrite local variables
17272 (identified just by use-after-scope as addressable) into SSA.
17274 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
17276 * doc/install.texi (Specific): opensource.apple.com uses https
17277 now. Remove trailing slash.
17279 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
17281 * README.Portability: Remove note on an Irix compatibility issue.
17283 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
17285 * gcov.c (INCLUDE_ALGORITHM): Define.
17286 (INCLUDE_VECTOR): Define.
17287 No longer include <vector> and <algorithm> directly.
17289 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
17291 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
17293 * doc/invoke.texi (Code Gen Options): Ditto.
17295 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
17298 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
17300 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
17302 rtl-optimization/79125
17303 * cprop.c (local_cprop_pass): Handle cases where we make an
17304 unconditional trap.
17306 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
17310 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
17311 read from, for big endian.
17313 2017-01-20 Jiong Wang <jiong.wang@arm.com>
17315 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
17316 register pauth builtins for LP64 only.
17318 2017-01-20 Marek Polacek <polacek@redhat.com>
17321 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
17324 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
17326 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
17328 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
17329 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
17330 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
17332 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17335 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
17336 in big-endian mode when they are not a single duplicated value.
17338 2017-01-20 Richard Biener <rguenther@suse.de>
17340 * BASE-VER: Bump to 7.0.1.
17342 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
17344 * omp-low.c (omplow_simd_context): New struct. Use it...
17345 (lower_rec_simd_input_clauses): ...here and...
17346 (lower_rec_input_clauses): ...here to hold common data. Adjust all
17347 references to idx, lane, max_vf, is_simt.
17349 2017-01-20 Graham Markall <graham.markall@embecosm.com>
17351 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
17354 2017-01-20 Martin Jambor <mjambor@suse.cz>
17356 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
17357 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
17359 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
17360 (GTFILES): Rename hsa.c to hsa-common.c.
17361 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
17362 * hsa-dump.c: Likewise.
17363 * hsa-gen.c: Likewise.
17364 * hsa-regalloc.c: Likewise.
17365 * ipa-hsa.c: Likewise.
17366 * omp-expand.c: Likewise.
17367 * omp-low.c: Likewise.
17368 * toplev.c: Likewise.
17370 2017-01-20 Marek Polacek <polacek@redhat.com>
17373 * doc/invoke.texi: Document -Wduplicated-branches.
17374 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
17375 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
17376 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
17377 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
17378 return 0 only when not OEP_LEXICOGRAPHIC.
17379 (fold_build_cleanup_point_expr): Use the expression
17380 location when building CLEANUP_POINT_EXPR.
17381 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
17382 * tree.c (add_expr): Handle error_mark_node.
17384 2017-01-20 Martin Liska <mliska@suse.cz>
17387 * tree-profile.c (init_ic_make_global_vars): Do not call
17389 (gimple_init_gcov_profiler): Likewise.
17391 2017-01-20 Martin Liska <mliska@suse.cz>
17394 * cgraph.h (maybe_create_reference): Remove argument and
17396 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
17398 * ipa-cp.c (create_specialized_node): Likewise.
17399 * symtab.c (symtab_node::maybe_create_reference): Handle
17400 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
17402 2017-01-20 Martin Liska <mliska@suse.cz>
17404 * read-rtl-function.c (function_reader::create_function): Use
17405 build_decl instread of build_decl_stat.
17407 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
17409 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
17410 * config/i386/avx512dqintrin.h: Ditto.
17411 * config/i386/avx512fintrin.h: Ditto.
17412 * config/i386/i386-builtin-types.def: Add new types.
17413 * config/i386/i386.c: Handle new types.
17414 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
17415 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
17416 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
17417 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
17418 (__builtin_ia32_kshiftridi): New.
17419 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
17421 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
17425 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
17426 define to rs6000_init_stack_protect_guard.
17427 (rs6000_init_stack_protect_guard): New function.
17429 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
17430 Yunqiang Su <yunqiang.su@imgtec.com>
17432 * config.gcc (supported_defaults): Add madd4.
17433 (with_madd4): Add validation.
17434 (all_defaults): Add madd4.
17435 * config/mips/mips.opt (mmadd4): New option.
17436 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17438 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17440 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
17441 (ISA_HAS_FUSED_MADD4): Likewise.
17442 * doc/invoke.texi (-mmadd4): Document the new option.
17443 * doc/install.texi (--with-madd4): Document the new option.
17445 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17447 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
17448 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
17449 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
17450 (aarch64_init_pauth_hint_builtins): New.
17451 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
17452 (aarch64_expand_builtin): Expand new builtins.
17454 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17456 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
17457 * combine-stack-adj.c (no_unhandled_cfa): Handle
17458 REG_CFA_TOGGLE_RA_MANGLE.
17459 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
17460 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
17461 info for return address signing.
17462 (aarch64_expand_epilogue): Likewise.
17464 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17466 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
17467 * config/aarch64/aarch64-protos.h
17468 (aarch64_return_address_signing_enabled): New declaration.
17469 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
17471 (aarch64_expand_prologue): Sign return address before it's pushed onto
17473 (aarch64_expand_epilogue): Authenticate return address fetched from
17475 (aarch64_override_options): Sanity check for ILP32 and ISA level.
17476 (aarch64_attributes): New function attributes for "sign-return-address".
17477 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
17478 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
17479 ("*do_return"): Generate combined instructions according to key index.
17480 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
17481 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
17483 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
17484 * config/aarch64/aarch64.opt (msign-return-address=): New.
17485 * doc/extend.texi (AArch64 Function Attributes): Documents
17486 "sign-return-address=".
17487 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
17489 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
17491 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
17492 overall option summary.
17494 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17496 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
17497 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
17498 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
17499 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
17501 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
17503 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
17504 -mpower9-minmax by default for -mcpu=power9.
17505 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
17506 128-bit floating point.
17508 2017-01-20 Alan Modra <amodra@gmail.com>
17510 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
17511 optimizing for size.
17513 2017-01-20 Alan Modra <amodra@gmail.com>
17516 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
17517 for strcmp and strncmp from corresponding builtin decl.
17519 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17521 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
17522 instead of i386/rtems-64.h.
17523 * config/i386/rtems-64.h: Remove.
17525 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17529 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
17531 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
17533 2017-01-19 Tamar Christina <tamar.christina@arm.com>
17535 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
17536 Change int to HOST_WIDE_INT.
17537 * config/aarch64/aarch64-protos.h
17538 (aarch64_simd_gen_const_vector_dup): Likewise.
17539 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
17541 2017-01-19 David Malcolm <dmalcolm@redhat.com>
17543 * langhooks-def.h (lhd_type_for_size): New decl.
17544 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
17545 * langhooks.c (lhd_type_for_size): New function, taken from
17548 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
17550 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
17551 define_bypass for CR latency.
17552 (power9-cracked-alu): Update bypass latency and remove power9-branch.
17553 (power9-alu2): Add define_bypass for CR latency.
17555 (power9-mul): Update insn latency.
17556 (power9-mul-compare): Update insn latency, bypass latency and remove
17559 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17561 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
17563 * config/aarch64/aarch64.md
17564 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
17565 aarch64_nopcrelative_literal_loads.
17566 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
17568 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
17570 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
17571 TARGET_LOONGSON_3A.
17572 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
17574 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
17577 * config.gcc (supported_defaults): Add lxc1-sxc1.
17578 (with_lxc1_sxc1): Add validation.
17579 (all_defaults): Add lxc1-sxc1.
17580 * config/mips/mips.opt (mlxc1-sxc1): New option.
17581 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17583 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17584 __mips_no_lxc1_sxc1.
17585 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
17586 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
17587 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
17589 2017-01-19 Richard Biener <rguenther@suse.de>
17591 PR tree-optimization/72488
17592 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
17593 sure to restore SSA info.
17594 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
17596 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
17598 PR rtl-optimization/79121
17599 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
17600 of the inner type when shifting an extended value.
17602 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
17605 * symtab.c (symtab_node::equal_address_to): Fix comparing of
17606 interposable aliases.
17608 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
17611 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
17612 Use the evmergelohi instruction.
17613 (mov_si<mode>_e500_subreg4_2_le): Likewise.
17614 (mov_sitf_e500_subreg8_2_be): Likewise.
17615 (mov_sitf_e500_subreg12_2_le): Likewise.
17616 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
17617 (mov_si<mode>_e500_subreg4_2_be): Likewise.
17618 (mov_sitf_e500_subreg8_2_le): Likewise.
17619 (mov_sitf_e500_subreg12_2_be): Likewise.
17621 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17623 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
17624 attribute from vecsimple to vecperm.
17625 (altivec_vbpermq2): Likewise.
17627 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17630 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
17632 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17633 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
17634 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
17635 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
17636 case where N arg is SIZE_MAX.
17637 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
17638 (cmpstrsi): Add pattern.
17640 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17642 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17643 __builtin_vec_revb builtins.
17644 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
17645 built-in functions to support generation of the ISA 3.0 XXBR<x>
17646 vector byte reverse instructions.
17647 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
17648 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
17649 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
17650 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
17651 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
17652 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
17653 (P9V_BUILTIN_VEC_REVB): Likewise.
17654 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
17655 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
17656 (p9_xxbrq_v16qi): Likewise.
17657 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
17658 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
17659 (p9_xxbrh_v8hi): Likewise.
17660 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
17661 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
17662 vec_revb built-in functions.
17664 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
17666 PR rtl-optimization/78952
17667 * config/i386/i386.md (any_extract): New code iterator.
17668 (*insvqi_2): Use any_extract for source operand.
17669 (*insvqi_3): Use any_shiftrt for source operand.
17671 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
17673 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
17675 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
17677 2017-01-18 Matthias Klose <doko@ubuntu.com>
17679 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
17681 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17683 * config/rs6000/altivec.h (vec_bperm): Change #define.
17684 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
17685 (altivec_vbpermq2): New define_insn.
17686 (altivec_vbpermd): Likewise.
17687 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
17688 function interface.
17689 (VBPERMD): Likewise.
17690 (VBPERM): New polymorphic function interface.
17691 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
17692 Add entries for P9V_BUILTIN_VEC_VBPERM.
17693 * doc/extend.texi: Add interfaces for vec_bperm.
17695 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17697 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
17698 first letter of error messages.
17699 (s390_resolve_overloaded_builtin): Likewise.
17700 * config/s390/s390.c (s390_expand_builtin): Likewise.
17701 (s390_invalid_arg_for_unprototyped_fn): Likewise.
17702 (s390_valid_target_attribute_inner_p): Likewise.
17703 * config/s390/s390.md ("tabort"): Likewise.
17705 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
17707 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
17708 (ISA_AVOID_DIV_HILO): New macro.
17709 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
17710 (ISA_HAS_DDIV): Likewise.
17712 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17714 * doc/invoke.texi (fabi-version): Correct number of occurrences.
17716 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17718 * doc/invoke.texi (fabi-version): Spelling fix.
17720 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17723 * doc/invoke.texi (fabi-version): Mention mangling fix for
17726 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17729 * doc/invoke.texi (fabi-version): Document discriminator mangling.
17731 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
17734 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
17735 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
17737 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
17739 (stack_protect_test): Ditto.
17740 * config/rs6000/rs6000.opt (mstack-protector-guard=,
17741 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
17743 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
17744 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
17745 -mstack-protector-guard-offset=.
17746 (RS/6000 and PowerPC Options): Ditto.
17748 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
17750 * config/i386/i386.h (MASK_CLASS_P): New define.
17751 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
17752 there are no registers from different register sets also when
17753 mask registers are used. Update function comment.
17754 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
17755 to (*k/*r) and (*k/*km) alternatives.
17757 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
17759 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
17760 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
17761 (EH_RETURN_HANDLER_RTX): New define.
17762 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
17763 Force frame pointer in EH return functions.
17764 (aarch64_expand_epilogue): Add barrier for eh_return.
17765 (aarch64_final_eh_return_addr): Remove.
17766 (aarch64_eh_return_handler_rtx): New function.
17767 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
17769 (aarch64_eh_return_handler_rtx): New prototype.
17771 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17773 * config/rs6000/altivec.h (vec_rlmi): New #define.
17774 (vec_vrlnm): Likewise.
17775 (vec_rlnm): Likewise.
17776 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
17777 (UNSPEC_VRLNM): Likewise.
17778 (VIlong): New mode iterator.
17779 (altivec_vrl<VI_char>mi): New define_insn.
17780 (altivec_vrl<VI_char>nm): Likewise.
17781 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
17783 (VRLDNM): Likewise.
17784 (RLNM): New polymorphic function entry.
17785 (VRLWMI): New monomorphic function entry.
17786 (VRLDMI): Likewise.
17787 (RLMI): New polymorphic function entry.
17788 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
17789 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
17790 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
17793 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17796 * dwarf2out.c (field_byte_offset): Restore the
17797 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
17798 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
17799 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
17802 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17805 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
17807 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17810 * dwarf2out.c (add_data_member_location_attribute): For constant
17811 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
17812 instead of DW_AT_data_member_location, DW_AT_bit_offset and
17813 DW_AT_byte_size attributes.
17815 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17817 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
17818 after forcing to constant memory when the code model is medium.
17820 2017-01-17 Julia Koval <julia.koval@intel.com>
17823 * config/i386/avx512fintrin.h
17824 (_mm512_i32gather_ps): Change __addr type to void const*.
17825 (_mm512_mask_i32gather_ps): Ditto.
17826 (_mm512_i32gather_pd): Ditto.
17827 (_mm512_mask_i32gather_pd): Ditto.
17828 (_mm512_i64gather_ps): Ditto.
17829 (_mm512_mask_i64gather_ps): Ditto.
17830 (_mm512_i64gather_pd): Ditto.
17831 (_mm512_mask_i64gather_pd): Ditto.
17832 (_mm512_i32gather_epi32): Ditto.
17833 (_mm512_mask_i32gather_epi32): Ditto.
17834 (_mm512_i32gather_epi64): Ditto.
17835 (_mm512_mask_i32gather_epi64): Ditto.
17836 (_mm512_i64gather_epi32): Ditto.
17837 (_mm512_mask_i64gather_epi32): Ditto.
17838 (_mm512_i64gather_epi64): Ditto.
17839 (_mm512_mask_i64gather_epi64): Ditto.
17840 (_mm512_i32scatter_ps): Change __addr type to void*.
17841 (_mm512_mask_i32scatter_ps): Ditto.
17842 (_mm512_i32scatter_pd): Ditto.
17843 (_mm512_mask_i32scatter_pd): Ditto.
17844 (_mm512_i64scatter_ps): Ditto.
17845 (_mm512_mask_i64scatter_ps): Ditto.
17846 (_mm512_i64scatter_pd): Ditto.
17847 (_mm512_mask_i64scatter_pd): Ditto.
17848 (_mm512_i32scatter_epi32): Ditto.
17849 (_mm512_mask_i32scatter_epi32): Ditto.
17850 (_mm512_i32scatter_epi64): Ditto.
17851 (_mm512_mask_i32scatter_epi64): Ditto.
17852 (_mm512_i64scatter_epi32): Ditto.
17853 (_mm512_mask_i64scatter_epi32): Ditto.
17854 (_mm512_i64scatter_epi64): Ditto.
17855 (_mm512_mask_i64scatter_epi64): Ditto.
17856 * config/i386/avx512pfintrin.h
17857 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
17858 (_mm512_mask_prefetch_i32gather_ps): Ditto.
17859 (_mm512_mask_prefetch_i64gather_pd): Ditto.
17860 (_mm512_mask_prefetch_i64gather_ps): Ditto.
17861 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
17862 (_mm512_prefetch_i32scatter_ps): Ditto.
17863 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
17864 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
17865 (_mm512_prefetch_i64scatter_pd): Ditto.
17866 (_mm512_prefetch_i64scatter_ps): Ditto.
17867 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
17868 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
17869 * config/i386/avx512vlintrin.h
17870 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
17871 (_mm_mmask_i32gather_ps): Ditto.
17872 (_mm256_mmask_i32gather_pd): Ditto.
17873 (_mm_mmask_i32gather_pd): Ditto.
17874 (_mm256_mmask_i64gather_ps): Ditto.
17875 (_mm_mmask_i64gather_ps): Ditto.
17876 (_mm256_mmask_i64gather_pd): Ditto.
17877 (_mm_mmask_i64gather_pd): Ditto.
17878 (_mm256_mmask_i32gather_epi32): Ditto.
17879 (_mm_mmask_i32gather_epi32): Ditto.
17880 (_mm256_mmask_i32gather_epi64): Ditto.
17881 (_mm_mmask_i32gather_epi64): Ditto.
17882 (_mm256_mmask_i64gather_epi32): Ditto.
17883 (_mm_mmask_i64gather_epi32): Ditto.
17884 (_mm256_mmask_i64gather_epi64): Ditto.
17885 (_mm_mmask_i64gather_epi64): Ditto.
17886 (_mm256_i32scatter_ps): Change __addr type to void*.
17887 (_mm256_mask_i32scatter_ps): Ditto.
17888 (_mm_i32scatter_ps): Ditto.
17889 (_mm_mask_i32scatter_ps): Ditto.
17890 (_mm256_i32scatter_pd): Ditto.
17891 (_mm256_mask_i32scatter_pd): Ditto.
17892 (_mm_i32scatter_pd): Ditto.
17893 (_mm_mask_i32scatter_pd): Ditto.
17894 (_mm256_i64scatter_ps): Ditto.
17895 (_mm256_mask_i64scatter_ps): Ditto.
17896 (_mm_i64scatter_ps): Ditto.
17897 (_mm_mask_i64scatter_ps): Ditto.
17898 (_mm256_i64scatter_pd): Ditto.
17899 (_mm256_mask_i64scatter_pd): Ditto.
17900 (_mm_i64scatter_pd): Ditto.
17901 (_mm_mask_i64scatter_pd): Ditto.
17902 (_mm256_i32scatter_epi32): Ditto.
17903 (_mm256_mask_i32scatter_epi32): Ditto.
17904 (_mm_i32scatter_epi32): Ditto.
17905 (_mm_mask_i32scatter_epi32): Ditto.
17906 (_mm256_i32scatter_epi64): Ditto.
17907 (_mm256_mask_i32scatter_epi64): Ditto.
17908 (_mm_i32scatter_epi64): Ditto.
17909 (_mm_mask_i32scatter_epi64): Ditto.
17910 (_mm256_i64scatter_epi32): Ditto.
17911 (_mm256_mask_i64scatter_epi32): Ditto.
17912 (_mm_i64scatter_epi32): Ditto.
17913 (_mm_mask_i64scatter_epi32): Ditto.
17914 (_mm256_i64scatter_epi64): Ditto.
17915 (_mm256_mask_i64scatter_epi64): Ditto.
17916 (_mm_i64scatter_epi64): Ditto.
17917 (_mm_mask_i64scatter_epi64): Ditto.
17918 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
17919 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
17920 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
17921 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
17922 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
17923 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
17924 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
17925 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
17926 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
17927 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
17928 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
17929 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
17930 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
17931 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
17932 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
17933 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
17934 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
17935 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
17936 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
17937 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
17938 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
17939 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
17940 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
17941 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
17942 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
17943 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
17944 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
17945 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
17946 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
17947 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
17948 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
17949 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
17950 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
17951 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
17952 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
17953 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
17954 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
17955 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
17956 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
17957 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
17958 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
17959 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
17960 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
17961 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
17962 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
17963 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
17964 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
17965 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
17966 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
17967 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
17968 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
17969 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
17970 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
17971 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
17972 definitions accordingly.
17974 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
17975 Kuan-Lin Chen <kuanlinchentw@gmail.com>
17978 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
17981 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
17984 * ira-conflicts.c (ira_build_conflicts): Update total conflict
17985 hard regs for inner regno.
17987 2017-01-17 Martin Liska <mliska@suse.cz>
17990 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
17991 assumption and add comment.
17993 2017-01-17 Nathan Sidwell <nathan@acm.org>
17995 * ipa-visibility.c (localize_node): New function, broken out of ...
17996 (function_and_variable_visibility): ... here. Call it.
17998 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
18000 PR middle-end/77445
18001 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
18002 correctly set frequency of oudgoing edge.
18003 (duplicate_thread_path): Fix profile updating.
18005 2017-01-17 Jakub Jelinek <jakub@redhat.com>
18008 * configure.ac: Add GCC_BASE_VER.
18009 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
18010 version from BASE-VER file.
18011 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
18012 (gcc.o): Depend on $(BASEVER).
18013 * common.opt (dumpfullversion): New option.
18014 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
18015 * doc/invoke.texi: Document -dumpfullversion.
18016 * doc/install.texi: Document --with-gcc-major-version-only.
18017 * configure: Regenerated.
18019 2017-01-17 Richard Biener <rguenther@suse.de>
18021 PR tree-optimization/71433
18022 * tree-vrp.c (register_new_assert_for): Merge same asserts
18023 on all incoming edges.
18024 (process_assert_insertions_for): Handle insertions at the
18027 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
18029 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
18030 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
18032 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
18035 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
18038 2017-01-17 Alan Modra <amodra@gmail.com>
18041 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
18042 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
18043 symbolic stack limit when pic.
18045 2017-01-16 Martin Sebor <msebor@redhat.com>
18047 PR tree-optimization/78608
18048 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
18050 2017-01-16 Jeff Law <law@redhat.com>
18053 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
18054 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
18055 for several include directories that may be relative to sysroot.
18056 * config/i386/x-mingw32 (gplus_includedir): Define.
18057 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
18058 (native_system_includedir): Likewise.
18059 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
18060 override if TARGET_SYSTEM_ROOT is defined.
18061 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
18063 PR tree-optimization/79090
18064 PR tree-optimization/33562
18065 PR tree-optimization/61912
18066 PR tree-optimization/77485
18067 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
18068 and computed trims into the dump file.
18070 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
18072 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
18074 2017-01-16 Jakub Jelinek <jakub@redhat.com>
18077 * gimplify.c (gimplify_init_constructor): If want_value and
18078 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
18082 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
18083 sequence. Formatting fixes.
18084 (doloop_optimize): Formatting fixes.
18087 * gcc.c (debug_level_greater_than_spec_func): New function.
18088 (static_spec_functions): Add debug-level-gt spec function.
18089 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
18091 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
18092 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
18093 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
18094 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
18095 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
18096 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
18098 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
18100 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
18101 QImode fixups to general and mask registers only.
18103 2017-01-16 Carl Love <cel@us.ibm.com>
18105 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
18106 for built-in functions
18107 vector signed char vec_nabs (vector signed char)
18108 vector signed short vec_nabs (vector signed short)
18109 vector signed int vec_nabs (vector signed int)
18110 vector signed long long vec_nabs (vector signed long long)
18111 vector float vec_nabs (vector float)
18112 vector double vec_nabs (vector double)
18113 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
18115 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
18116 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
18117 * doc/extend.texi: Update the documentation file for the new built-in
18120 2017-01-16 Martin Sebor <msebor@redhat.com>
18122 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
18125 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18127 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
18128 UNSPEC_VSX__XXSPLTD to require special splat handling.
18130 2017-01-16 David Malcolm <dmalcolm@redhat.com>
18133 * system.h: Poison strndup.
18135 2017-01-16 Alan Modra <amodra@gmail.com>
18138 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
18141 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
18143 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
18145 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
18147 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
18148 call recog here. Assert that INSN_CODE (insn) is non-negative.
18150 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
18153 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
18155 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
18156 in the currently scheduled RTL fragment.
18158 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
18160 PR rtl-optimization/78751
18161 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
18164 2017-01-14 Jeff Law <law@redhat.com>
18166 PR tree-optimization/79090
18167 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
18168 variable length stores.
18169 (compute_trims): Delete dead assignment to *trim_tail.
18170 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
18173 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
18175 PR rtl-optimization/78626
18176 PR rtl-optimization/78727
18177 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
18178 of a block, and split such blocks after everything else is finished.
18180 2017-01-14 Alan Modra <amodra@gmail.com>
18183 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
18184 target legitimate_combined_insn.
18185 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
18186 (rs6000_legitimate_combined_insn): New function.
18187 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
18189 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
18190 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
18191 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
18193 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
18195 * doc/frontends.texi (G++ and GCC): Remove references to Java.
18197 2017-01-13 Jeff Law <law@redhat.com>
18199 PR tree-optimization/33562
18200 PR tree-optimization/61912
18201 PR tree-optimization/77485
18202 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
18204 (delete_dead_assignment): Likewise.
18205 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
18206 statement to delete_dead_call and delete_dead_assignment.
18208 2017-01-13 David Malcolm <dmalcolm@redhat.com>
18211 * substring-locations.c (format_warning_va): Strengthen case 1 so
18212 that both endpoints of the substring must be within the format
18213 range for just the substring to be printed.
18215 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
18217 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
18218 * config/i386/i386.c (ix86_target_string): Add missing options
18219 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
18220 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
18221 flags_other and ix86_target_other to flags2_other. Display unknown
18223 (ix86_valid_target_attribute_inner_p): Add missing options and
18224 reorder options by implied ISAs, as in ix86_target_string.
18226 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
18228 * hash-table.h (hash_table::too_empty_p): New function.
18229 (hash_table::expand): Use it.
18230 (hash_table::traverse): Likewise.
18231 (hash_table::empty_slot): Use sizeof (value_type) instead of
18232 sizeof (PTR) to convert bytes to elements. Shrink the table
18233 if the current size is excessive for the current number of
18236 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
18238 * ira-costs.c (record_reg_classes): Break from the inner loop
18239 early once alt_fail is known to be true. Update outer loop
18240 handling accordingly.
18242 2017-01-13 Jeff Law <law@redhat.com>
18244 * tree-ssa-dse.c (decrement_count): New function.
18245 (increment_start_addr, maybe_trim_memstar_call): Likewise.
18246 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
18247 when we know the partially dead statement is a mem* function.
18249 PR tree-optimization/61912
18250 PR tree-optimization/77485
18251 * tree-ssa-dse.c: Include expr.h.
18252 (maybe_trim_constructor_store): New function.
18253 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
18255 PR tree-optimization/33562
18256 PR tree-optimization/61912
18257 PR tree-optimization/77485
18258 * doc/invoke.texi: Document new dse-max-object-size param.
18259 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
18260 * tree-ssa-dse.c: Include params.h.
18261 (dse_store_status): New enum.
18262 (initialize_ao_ref_for_dse): New, partially extracted from
18264 (valid_ao_ref_for_dse, normalize_ref): New.
18265 (setup_live_bytes_from_ref, compute_trims): Likewise.
18266 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
18267 (maybe_trim_partially_dead_store): Likewise.
18268 (maybe_trim_complex_store): Likewise.
18269 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
18270 Track what bytes live from the original store. Return tri-state
18271 for dead, partially dead or live.
18272 (dse_dom_walker): Add constructor, destructor and new private members.
18273 (delete_dead_call, delete_dead_assignment): New extracted from
18275 (dse_optimize_stmt): Make a member of dse_dom_walker.
18276 Use initialize_ao_ref_for_dse.
18278 PR tree-optimization/33562
18279 PR tree-optimization/61912
18280 PR tree-optimization/77485
18281 * sbitmap.h (bitmap_count_bits): Prototype.
18282 (bitmap_clear_range, bitmap_set_range): Likewise.
18283 * sbitmap.c (bitmap_clear_range): New function.
18284 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
18286 2017-01-13 Martin Liska <mliska@suse.cz>
18289 * function.c (set_cfun): Add new argument force.
18290 * function.h (set_cfun): Likewise.
18291 * ipa-inline-transform.c (inline_call): Use the function when
18292 strict alising from is dropped for function we inline to.
18294 2017-01-13 Richard Biener <rguenther@suse.de>
18296 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
18297 for dumping GIMPLE INTEGER_CSTs.
18299 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18301 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
18302 to 201112L since C++17.
18304 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
18307 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
18308 if -fsanitize=kernel-address is present.
18310 2017-01-13 Richard Biener <rguenther@suse.de>
18312 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
18313 as _Literal ( type ) number in case usual suffixes do not
18314 preserve all information.
18316 2017-01-13 Richard Biener <rguenther@suse.de>
18318 PR tree-optimization/77283
18319 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
18320 and ssa-iterators.h.
18321 (is_feasible_trace): Implement a cost model based on joiner
18324 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
18327 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
18328 char or short to __float128/_Float128 directly.
18330 2017-01-12 Martin Sebor <msebor@redhat.com>
18332 to -Wformat-overflow.
18333 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
18334 (min_bytes_remaining): Same.
18335 (get_string_length): Same.
18336 (format_string): Same.
18337 (format_directive): Same.
18339 (pass_sprintf_length::handle_gimple_call): Same.
18341 2017-01-12 Jakub Jelinek <jakub@redhat.com>
18343 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
18344 info.nowrite calls with no lhs that can't throw. Return bool
18345 whether gsi_remove has been called or not.
18346 (pass_sprintf_length::handle_gimple_call): Return bool whether
18347 try_substitute_return_value called gsi_remove. Formatting fix.
18348 (pass_sprintf_length::execute): Don't use gsi_remove if
18349 handle_gimple_call returned true.
18352 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
18353 be removed due to side-effects, don't remove following barrier nor
18354 turn the successor edge into fallthru edge.
18356 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18359 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
18360 element-reversing loads and stores as not swappable.
18362 2017-01-12 Nathan Sidwell <nathan@acm.org>
18363 Nicolai Stange <nicstange@gmail.com>
18365 * combine.c (try_combine): Don't ignore result of overlap checking
18366 loop. Combine overlap & asm check into single loop.
18368 2017-01-12 Richard Biener <rguenther@suse.de>
18370 * tree-pretty-print.c (dump_generic_node): Provide -gimple
18371 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
18373 2017-01-12 Richard Biener <rguenther@suse.de>
18375 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
18376 and TS_TARGET_OPTION directly derive from TS_BASE.
18377 * tree-core.h (tree_optimization_option): Derive from tree_base.
18378 (tree_target_option): Likewise.
18380 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
18382 * config/i386/i386.c (memory_address_length): Increase len
18383 only when rip_relative_addr_p returns false.
18385 2017-01-11 Julia Koval <julia.koval@intel.com>
18387 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
18388 (OPTION_MASK_ISA_SGX_SET): New.
18389 (ix86_handle_option): Handle OPT_msgx.
18390 * config.gcc: Added sgxintrin.h.
18391 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
18392 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
18393 * config/i386/i386.c (ix86_target_string): Add -msgx.
18395 (ix86_option_override_internal): Handle new options.
18396 (ix86_valid_target_attribute_inner_p): Add sgx.
18397 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
18398 * config/i386/i386.opt: Add msgx.
18399 * config/i386/sgxintrin.h: New file.
18400 * config/i386/x86intrin.h: Add sgxintrin.h.
18402 2017-01-11 Jakub Jelinek <jakub@redhat.com>
18405 * fold-const.c (maybe_nonzero_address): Return 1 for function
18407 (tree_single_nonzero_warnv_p): Don't handle function local objects
18411 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
18414 2017-01-11 David Malcolm <dmalcolm@redhat.com>
18417 * opts.c: Include "spellcheck.h"
18418 (struct string_fragment): New struct.
18419 (struct edit_distance_traits<const string_fragment &>): New
18421 (get_closest_sanitizer_option): New function.
18422 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
18424 2017-01-11 Jakub Jelinek <jakub@redhat.com>
18426 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
18428 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
18429 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
18430 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
18431 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
18432 for initial die_offset if dwarf_split_debug_info.
18433 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
18434 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
18436 (output_skeleton_debug_sections): Formatting fix. Use
18437 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
18438 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
18440 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
18442 * config/arm/cortex-a53.md: Add bypasses for
18443 cortex_a53_r2f_cvt.
18444 (cortex_a53_r2f): Only use for transfers.
18445 (cortex_a53_f2r): Likewise.
18446 (cortex_a53_r2f_cvt): Add reservation for conversions.
18447 (cortex_a53_f2r_cvt): Likewise.
18449 2017-01-11 Tamar Christina <tamar.christina@arm.com>
18451 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
18452 to all inlined functions, change static to extern.
18454 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
18457 * config/arm/arm.c (legitimize_pic_address): Handle reference to
18459 (arm_assemble_integer): Likewise.
18461 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
18463 * config.gcc: Use new awk script to check CPU, FPU and architecture
18464 parameters for --with-... options.
18465 * config/arm/parsecpu.awk: New file
18466 * config/arm/arm-cpus.in: New file.
18467 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
18469 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
18471 * config/arm/t-arm: Update dependency rules.
18472 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
18473 of processing .def files.
18474 * config/arm/genopt.sh: Deleted.
18475 * config/arm/gentune.sh: Deleted.
18476 * config/arm/arm-cores.def: Deleted.
18477 * config/arm/arm-arches.def: Deleted.
18478 * config/arm/arm-fpus.def: Deleted.
18479 * config/arm/arm-tune.md: Regenerated.
18480 * config/arm/arm-tables.opt: Regenerated.
18481 * config/arm/arm-cpu.h: New generated file.
18482 * config/arm/arm-cpu-data.h: New generated file.
18483 * config/arm/arm-cpu-cdata.h: New generated file.
18485 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
18488 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
18490 (input_varpool_node): Unpack dynamically_initialized bit.
18492 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
18494 PR rtl-optimization/79032
18495 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
18496 the alignment of the adjusted memory reference against that of MODE,
18497 instead of the alignment of the original memory reference.
18499 2017-01-11 Martin Jambor <mjambor@suse.cz>
18501 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
18503 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
18504 decorated functions.
18506 2017-01-11 Richard Biener <rguenther@suse.de>
18508 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
18509 set range/nonnull info for PHI results. Do not set it on
18510 stmts marked for removal.
18512 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
18514 * expr.c (store_field): In the bitfield case, fetch the return value
18515 from the registers before applying a single big-endian adjustment.
18516 Always do a final load for a BLKmode value not larger than a word.
18518 2017-01-10 David Malcolm <dmalcolm@redhat.com>
18521 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18522 that we correctly handle column numbers greater than
18523 LINE_MAP_MAX_COLUMN_NUMBER.
18525 2017-01-10 Martin Sebor <msebor@redhat.com>
18527 PR middle-end/78245
18528 * gimple-ssa-sprintf.c (get_destination_size): Call
18529 {init,fini}object_sizes.
18530 * tree-object-size.c (addr_object_size): Adjust.
18531 (pass_through_call): Adjust.
18532 (pass_object_sizes::execute): Adjust.
18533 * tree-object-size.h (fini_object_sizes): Declare.
18535 2017-01-10 Martin Sebor <msebor@redhat.com>
18537 PR tree-optimization/78775
18538 * builtins.c (get_size_range): Move...
18539 * calls.c: ...to here.
18540 (alloc_max_size): Accept zero argument.
18541 (operand_signed_p): Remove.
18542 (maybe_warn_alloc_args_overflow): Call get_size_range.
18543 * calls.h (get_size_range): Declare.
18545 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
18547 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
18548 from TI's devices.csv file as of September 2016.
18549 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
18551 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
18553 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
18554 * doc/invoke.texi: Likewise.
18555 * doc/md.texi: Likewise.
18556 * doc/objc.texi: Likewise.
18558 2017-01-10 Joshua Conner <joshconner@google.com>
18560 * config/arm/fuchsia-elf.h: New file.
18561 * config/fuchsia.h: New file.
18562 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
18563 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
18565 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
18567 2016-01-10 Richard Biener <rguenther@suse.de>
18569 PR tree-optimization/79034
18570 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18571 Propagate out degenerate PHIs in the joiner.
18573 2017-01-10 Martin Liska <mliska@suse.cz>
18575 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
18576 (sort_congruence_classes_by_decl_uid): Likewise.
18577 (sort_congruence_class_groups_by_decl_uid): Likewise.
18578 (sem_item_optimizer::merge_classes): Sort class, groups in these
18579 classes and members in the groups by DECL_UID of declarations.
18580 This would make merge operations stable.
18582 2017-01-10 Martin Liska <mliska@suse.cz>
18584 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
18585 usage of m_classes_vec.
18586 (sem_item_optimizer::~sem_item_optimizer): Likewise.
18587 (sem_item_optimizer::get_group_by_hash): Likewise.
18588 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
18589 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
18590 (sem_item_optimizer::verify_classes): Likewise.
18591 (sem_item_optimizer::process_cong_reduction): Likewise.
18592 (sem_item_optimizer::dump_cong_classes): Likewise.
18593 (sem_item_optimizer::merge_classes): Likewise.
18594 * ipa-icf.h (congruence_class_hash): Rename from
18595 congruence_class_group_hash. Remove declaration of m_classes_vec.
18597 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
18599 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
18600 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
18601 * config.gcc: Add avx512vpopcntdqintrin.h.
18602 * config/i386/avx512vpopcntdqintrin.h: New.
18603 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
18604 * config/i386/i386-builtin-types.def: Add new types.
18605 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
18606 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
18607 __builtin_ia32_vpopcountq_v8di_mask): New.
18608 * config/i386/i386-c.c (ix86_target_macros_internal): Define
18609 __AVX512VPOPCNTDQ__.
18610 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
18611 (PTA_AVX512VPOPCNTDQ): Define.
18612 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
18613 TARGET_AVX512VPOPCNTDQ_P): Define.
18614 * config/i386/i386.opt: Add mavx512vpopcntdq.
18615 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
18616 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
18618 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18620 PR middle-end/77484
18621 * predict.def (PRED_CALL): Set to 67.
18623 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
18625 * expr.c (store_field): In the bitfield case, if the value comes from
18626 a function call and is of an aggregate type returned in registers, do
18627 not modify the field mode; extract the value in all cases if the mode
18628 is BLKmode and the size is not larger than a word.
18630 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
18633 * config/i386/cpuid.h: Fix undefined behavior.
18635 2017-01-04 Jeff Law <law@redhat.com>
18637 PR tree-optimization/79007
18638 PR tree-optimization/67955
18639 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
18640 conservative for pt.null when flag_non_call_exceptions is on.
18642 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18644 PR translation/79019
18645 PR translation/79020
18646 * params.def (PARAM_INLINE_MIN_SPEEDUP,
18647 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
18648 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
18650 * config/avr/avr.opt (maccumulate-args): Likewise.
18651 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
18652 * common.opt (freport-bug): Likewise.
18653 * cif-code.def (CIF_FINAL_ERROR): Likewise.
18654 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
18655 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
18656 translatable string.
18657 * config/i386/i386.c (function_value_32): Likewise.
18658 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
18659 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
18661 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
18662 * common/config/msp430/msp430-common.c (msp430_handle_option):
18664 * symtab.c (symtab_node::verify_base): Likewise.
18665 * opts.c (set_debug_level): Likewise.
18666 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
18667 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
18668 missing whitespace to translatable strings.
18669 * config/avr/avr.md (bswapsi2): Fix typo in comment.
18670 * config/sh/superh.h: Likewise.
18671 * config/i386/xopintrin.h: Likewise.
18672 * config/i386/znver1.md: Likewise.
18673 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
18674 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
18675 * double-int.h (struct double_int): Likewise.
18676 * double-int.c (div_and_round_double): Likewise.
18677 * wide-int.cc: Likewise.
18678 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
18679 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
18680 * cfgcleanup.c (crossjumps_occured): Renamed to ...
18681 (crossjumps_occurred): ... this.
18682 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
18685 PR tree-optimization/78899
18686 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
18687 returning bool return struct loop *, NULL for failure and the new
18689 (versionable_outer_loop_p): Don't version outer loop if it has
18690 dont_vectorized bit set.
18691 (tree_if_conversion): When versioning outer loop, ensure
18692 tree_if_conversion is performed also on the inner loop of the
18693 non-vectorizable outer loop copy.
18694 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
18695 LOOP_VECTORIZED in inner loop of the scalar outer loop and
18696 prevent vectorization of it.
18697 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
18698 the outer loop vectorization of the non-scalar version is attempted
18699 before vectorization of the inner loop in scalar version. If
18700 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
18701 vectorization of its inner loop.
18702 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
18703 has 2 inner loops, rename also on edges from bb whose single pred
18704 is outer_loop->header. Fix typo in function comment.
18706 2017-01-09 Martin Sebor <msebor@redhat.com>
18709 * asan.c (asan_emit_stack_protection): Increase local buffer size
18710 to avoid snprintf truncation warning.
18712 2017-01-09 Andrew Pinski <apinski@cavium.com>
18714 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
18715 to reference thunderx2t99 for the tuning structure
18716 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
18718 (thunderx2t99_extra_costs): This.
18719 * config/aarch64/aarch64-tune.md: Regenerate.
18720 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
18721 (vulcan_addrcost_table): This.
18722 (vulcan_regmove_cost): Rename to ...
18723 (thunderx2t99_regmove_cost): This.
18724 (vulcan_vector_cost): Rename to ...
18725 (thunderx2t99_vector_cost): this.
18726 (vulcan_branch_cost): Rename to ...
18727 (thunderx2t99_branch_cost): This.
18728 (vulcan_tunings): Rename to ...
18729 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
18730 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
18732 2017-01-09 Martin Jambor <mjambor@suse.cz>
18736 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
18737 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
18738 (propagate_vr_accross_jump_function): Use the above function for all
18739 value range computations for pass-through jump functions and type
18740 converasion from explicit value range values.
18741 (ipcp_propagate_stage): Do not attempt to deduce types of formal
18742 parameters from TYPE_ARG_TYPES.
18743 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
18744 (ipa_write_node_info): Stream type of the actual argument.
18745 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
18747 2017-01-09 Martin Liska <mliska@suse.cz>
18750 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
18751 (lookup_compiler): Do not show error message with have_E.
18753 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18755 PR tree-optimization/78938
18756 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
18757 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
18758 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
18759 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
18762 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18764 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
18767 2017-01-09 Richard Biener <rguenther@suse.de>
18769 PR tree-optimization/78997
18770 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
18771 name condition properly.
18773 2017-01-09 Richard Biener <rguenther@suse.de>
18776 * dwarf2out.c (is_cxx): New overload with context.
18777 (is_naming_typedef_decl): Use it.
18779 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18781 * invoke.texi (Option Summary): Correct spacing in option lists
18782 and add line breaks to fix over-long lines.
18784 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18786 PR middle-end/17660
18788 * extend.texi (Common Variable Attributes): Add xref to GCC
18789 Internals manual to explain mode attribute keywords.
18791 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18794 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
18795 and Preprocessor Options.
18796 (Options for Linking): Document -pthread here....
18797 (RS/6000 and PowerPC Options): ...not here.
18798 (Solaris 2 Options): ...or here.
18799 * doc/cppopts.texi: Document -pthread.
18801 2017-01-08 Martin Sebor <msebor@redhat.com>
18803 PR middle-end/77708
18804 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
18805 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
18806 New member functions.
18807 (format_directive): Used them.
18809 (pass_sprintf_length::handle_gimple_call): Same.
18810 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
18811 to avoid truncation for any argument.
18812 (extract_affine_mul): Same.
18813 * tree.c (get_file_function_name): Same.
18815 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18817 PR middle-end/77484
18818 * predict.def (PRED_INDIR_CALL): Set to 86.
18820 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18822 PR preprocessor/54124
18823 * doc/cppopts.texi: Reformat -d subtable to list the full name
18824 of the options. Add cross-reference to the docs for the general
18825 compiler -d options.
18826 * doc/invoke.texi (Developer Options): Add cross-reference to the
18827 preprocessor-specific -d option documentation.
18829 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18831 PR preprocessor/13498
18832 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
18833 redudant material, and reflect new command-line options.
18834 (System Headers): Likewise.
18836 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18838 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
18839 -isystem, and -idirafter. Copy-edit.
18840 * doc/cppopts.texi: Copy-edit. Remove contradiction about
18841 default for -ftrack-macro-expansion. Delete obsolete and
18842 badly-formatted implementation details about -fdebug-cpp output.
18843 * doc/cppwarnopts.texi: Copy-edit.
18845 2017-01-07 David Malcolm <dmalcolm@redhat.com>
18848 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18849 that the transition from a max line width >= 1<<10 to narrower
18850 lines works correctly.
18852 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
18854 * doc/options.texi (PerFunction): New.
18855 * opt-functions.awk (switch_flags): Map both Optimization and
18856 PerFunction to CL_OPTIMIZATION.
18857 * opth-gen.awk: Test for PerFunction flag along with
18859 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
18860 it only when the latter is present. Skip those that don't in
18861 the hash function generator.
18862 * common.opt (fvar-tracking): Mark as PerFunction instead of
18864 (fvar-tracking-assignments): Likewise.
18865 (fvar-tracking-assignments-toggle): Likewise.
18866 (fvar-tracking-uninit): Likewise.
18868 2017-01-07 Jakub Jelinek <jakub@redhat.com>
18870 PR translation/79018
18871 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
18874 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
18877 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
18878 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
18879 TARGET_LONG_JUMP_TABLE_OFFSETS.
18880 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
18881 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
18882 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
18883 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
18884 * config/m68k/m68k.md (tablejump expander): Likewise.
18885 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
18886 TARGET_LONG_JUMP_TABLE_OFFSETS.
18887 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
18888 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
18890 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
18891 David Holsgrove <david.holsgrove@xilinx.com>
18893 * common/config/microblaze/microblaze-common.c
18894 (TARGET_EXCEPT_UNWIND_INFO): Remove.
18895 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
18897 * config/microblaze/microblaze.c (microblaze_must_save_register)
18898 (microblaze_expand_epilogue, microblaze_return_addr): Handle
18900 (microblaze_eh_return): New function.
18901 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
18902 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
18903 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
18904 * config/microblaze/microblaze.md (eh_return): New pattern.
18906 2017-01-06 Jakub Jelinek <jakub@redhat.com>
18908 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
18909 GCC_DIAGNOSTIC_STRINGIFY): Define.
18911 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
18913 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18915 * config/arm/arm.md (<mcrr>): New.
18917 * config/arm/arm.c (arm_arch5te): New.
18918 (arm_option_override): Set arm_arch5te.
18919 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
18921 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
18922 (arm_mcrr_qualifiers): ... this. New.
18923 (MRRC_QUALIFIERS): Define to...
18924 (arm_mrrc_qualifiers): ... this. New.
18925 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
18927 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
18928 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
18929 (MRRCI, mrrc, MRRC): New.
18930 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
18931 VUNSPEC_MRRC2): New.
18933 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18935 * config/arm/arm.md (<mcr>): New.
18937 * config/arm/arm.c (arm_coproc_builtin_available): Add
18938 support for mcr, mrc, mcr2 and mrc2.
18939 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
18940 (arm_mcr_qualifiers): ... this. New.
18941 (MRC_QUALIFIERS): Define to ...
18942 (arm_mrc_qualifiers): ... this. New.
18943 (MCR_QUALIFIERS): Define to ...
18944 (arm_mcr_qualifiers): ... this. New.
18945 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
18947 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
18948 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
18949 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
18950 VUNSPEC_MRC2): New.
18952 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18954 * config/arm/arm.md (*ldc): New.
18958 * config/arm/arm.c (arm_coproc_builtin_available): Add
18959 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
18960 (arm_coproc_ldc_stc_legitimate_address): New.
18961 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
18962 'qualifier_const_pointer'.
18963 (LDC_QUALIFIERS): Define to...
18964 (arm_ldc_qualifiers): ... this. New.
18965 (STC_QUALIFIERS): Define to...
18966 (arm_stc_qualifiers): ... this. New.
18967 * config/arm/arm-protos.h
18968 (arm_coproc_ldc_stc_legitimate_address): New.
18969 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
18970 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
18971 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
18972 stc2, stcl, stc2l): New.
18973 * config/arm/constraints.md (Uz): New.
18974 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
18975 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
18976 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
18977 VUNSPEC_STC2L): New.
18979 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18981 * config/arm/arm.md (<cdp>): New.
18982 * config/arm/arm.c (neon_const_bounds): Rename this ...
18983 (arm_const_bounds): ... this.
18984 (arm_coproc_builtin_available): New.
18985 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
18986 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
18987 (CDP_QUALIFIERS): Define to...
18988 (arm_cdp_qualifiers): ... this. New.
18990 (arm_expand_builtin_args): Add case for 6 arguments.
18991 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
18992 (arm_const_bounds): ... this.
18993 (arm_coproc_builtin_available): New.
18994 * config/arm/arm_acle.h (__arm_cdp): New.
18996 * config/arm/arm_acle_builtins.def (cdp): New.
18998 * config/arm/iterators.md (CDPI,CDP,cdp): New.
18999 * config/arm/neon.md: Rename all 'neon_const_bounds' to
19000 'arm_const_bounds'.
19001 * config/arm/types.md (coproc): New.
19002 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
19003 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
19004 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
19005 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
19007 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19009 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
19010 (UBINOP_QUALIFIERS): New.
19012 (acle_builtin_data): New. Change comment.
19013 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
19014 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
19015 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
19016 arm_acle_builtins.def.
19017 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
19018 (arm_init_acle_builtins): New.
19019 (CRC32_BUILTIN): Remove.
19020 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
19021 crc32cb, crc32ch and crc32cw.
19022 (arm_init_crc32_builtins): Remove.
19023 (arm_init_builtins): Use arm_init_acle_builtins rather
19024 than arm_init_crc32_builtins.
19025 (arm_expand_acle_builtin): New.
19026 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
19027 * config/arm/arm_acle_builtins.def: New.
19029 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19031 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
19032 (arm_builtin_datum): ... this.
19033 (arm_init_neon_builtin): Rename to ...
19034 (arm_init_builtin): ... this. Add a new parameters PREFIX
19035 and USE_SIG_IN_NAME.
19036 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
19037 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
19038 'arm_builtin_datum'.
19039 (arm_init_vfp_builtins): Likewise.
19040 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
19041 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
19042 (arm_expand_neon_args): Rename to ...
19043 (arm_expand_builtin_args): ... this. Rename builtin_arg
19044 enum values and differentiate between ARG_BUILTIN_MEMORY
19045 and ARG_BUILTIN_NEON_MEMORY.
19046 (arm_expand_neon_builtin_1): Rename to ...
19047 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
19048 values, arm_expand_builtin_args and add bool parameter NEON.
19049 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
19050 (arm_expand_vfp_builtin): Likewise.
19051 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
19053 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19055 PR middle-end/77484
19056 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
19057 * predict.c (tree_estimate_probability_bb): Reverse direction of
19058 polymorphic call predictor.
19060 2017-01-06 David Malcolm <dmalcolm@redhat.com>
19062 * passes.c (execute_one_pass): Split out pass-skipping logic into...
19063 (determine_pass_name_match): ...this new function and...
19064 (should_skip_pass_p): ...this new function.
19066 2017-01-06 Nathan Sidwell <nathan@acm.org>
19068 * ipa-visibility.c (function_and_variable_visibility): Reformat
19069 comments and long lines. Remove extrneous if.
19070 * symtab.c (symtab_node::make_decl_local): Fix code format.
19071 (symtab_node::set_section_for_node): Fix comment typo.
19073 2017-01-06 Martin Liska <mliska@suse.cz>
19076 * lra-constraints.c: Rename invariant to lra_invariant.
19077 * predict.c (set_even_probabilities): Initialize e to NULL.
19079 2017-01-05 Martin Sebor <msebor@redhat.com>
19081 PR tree-optimization/78910
19082 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
19083 (format_integer): Correct off-by-one error in the handling
19084 of precision with negative numbers in signed conversions..
19086 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
19088 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
19090 2017-01-05 Jakub Jelinek <jakub@redhat.com>
19092 PR tree-optimization/71016
19093 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
19094 factor_out_conditional_conversion. Formatting fix.
19095 (factor_out_conditional_conversion): Add cond_stmt argument.
19096 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
19097 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
19100 2017-01-05 David Malcolm <dmalcolm@redhat.com>
19102 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
19103 read-rtl-function.o, and selftest-rtl.o.
19104 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
19105 (selftest::aarch64_test_loading_full_dump): New function.
19106 (selftest::aarch64_run_selftests): New function.
19107 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
19108 selftest::aarch64_run_selftests.
19109 * config/i386/i386.c
19110 (selftest::ix86_test_loading_dump_fragment_1): New function.
19111 (selftest::ix86_test_loading_call_insn): New function.
19112 (selftest::ix86_test_loading_full_dump): New function.
19113 (selftest::ix86_test_loading_unspec): New function.
19114 (selftest::ix86_run_selftests): Call the new functions.
19115 * emit-rtl.c (maybe_set_max_label_num): New function.
19116 * emit-rtl.h (maybe_set_max_label_num): New decl.
19117 * function.c (instantiate_decls): Guard call to
19118 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
19119 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
19121 * gensupport.c (gen_reader::gen_reader): Pass "false"
19122 for new "compact" param of rtx_reader.
19123 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
19124 rather than an empty string for NULL strings.
19125 * read-md.c: Potentially include config.h rather than bconfig.h.
19126 Wrap include of errors.h with #ifdef GENERATOR_FILE.
19127 (have_error): New global, copied from errors.c.
19128 (md_reader::read_name): Rename to...
19129 (md_reader::read_name_1): ...this, adding "out_loc" param,
19130 and converting "missing name or number" to returning false, rather
19132 (md_reader::read_name): Reimplement in terms of read_name_1.
19133 (md_reader::read_name_or_nil): New function.
19134 (md_reader::read_string): Handle "(nil)" by returning NULL.
19135 (md_reader::md_reader): Add new param "compact".
19136 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
19137 (md_reader::read_file): New method.
19138 * read-md.h (md_reader::md_reader): Add new param "compact".
19139 (md_reader::read_file): New method.
19140 (md_reader::is_compact): New accessor.
19141 (md_reader::read_name): Convert return type from void to file_location.
19142 (md_reader::read_name_or_nil): New decl.
19143 (md_reader::read_name_1): New decl.
19144 (md_reader::m_compact): New field.
19145 (noop_reader::noop_reader): Pass "false" for new "compact" param
19147 (rtx_reader::rtx_reader): Add new "compact" param.
19148 (rtx_reader::read_rtx_operand): Make virtual and convert return
19149 type from void to rtx.
19150 (rtx_reader::read_until): New decl.
19151 (rtx_reader::handle_any_trailing_information): New virtual function.
19152 (rtx_reader::postprocess): New virtual function.
19153 (rtx_reader::finalize_string): New virtual function.
19154 (rtx_reader::m_in_call_function_usage): New field.
19155 (rtx_reader::m_reuse_rtx_by_id): New field.
19156 * read-rtl-function.c: New file.
19157 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
19158 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
19159 (selftest::verify_three_block_rtl_cfg): New decl.
19160 * read-rtl-function.h: New file.
19161 * read-rtl.c: Potentially include config.h rather than bconfig.h.
19162 For host, include function.h, memmodel.h, and emit-rtl.h.
19163 (one_time_initialization): New function.
19164 (struct compact_insn_name): New struct.
19165 (compact_insn_names): New array.
19166 (find_code): Handle insn codes in compact dumps.
19167 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
19168 (bind_subst_iter_and_attr): Likewise.
19169 (add_condition_to_string): Likewise.
19170 (add_condition_to_rtx): Likewise.
19171 (apply_attribute_uses): Likewise.
19172 (add_current_iterators): Likewise.
19173 (apply_iterators): Likewise.
19174 (initialize_iterators): Guard usage of apply_subst_iterator with
19175 #ifdef GENERATOR_FILE.
19176 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
19177 (md_reader::read_mapping): Likewise.
19178 (add_define_attr_for_define_subst): Likewise.
19179 (add_define_subst_attr): Likewise.
19180 (read_subst_mapping): Likewise.
19181 (check_code_iterator): Likewise.
19182 (rtx_reader::read_rtx): Likewise. Move one-time initialization
19184 (one_time_initialization): New function.
19185 (rtx_reader::read_until): New method.
19186 (read_flags): New function.
19187 (parse_reg_note_name): New function.
19188 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
19189 Handle reuse_rtx ids.
19190 Wrap iterator lookup within #ifdef GENERATOR_FILE.
19191 Add parsing support for RTL dumps, mirroring the special-cases in
19192 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
19193 values, and calling handle_any_trailing_information.
19194 (rtx_reader::read_rtx_operand): Convert return type from void
19195 to rtx, returning return_rtx. Handle case 'e'. Call
19196 finalize_string on XSTR and XTMPL fields.
19197 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
19198 "(nil)" values were omitted. Call the postprocess vfunc on the
19200 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
19201 class ctor. Initialize m_in_call_function_usage. Call
19202 one_time_initialization.
19203 * rtl-tests.c (selftest::test_uncond_jump): Call
19204 set_new_first_and_last_insn.
19205 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
19206 * selftest-rtl.c: New file.
19207 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
19208 (selftest::get_insn_by_uid): New decl.
19209 * selftest-run-tests.c (selftest::run_tests): Call
19210 read_rtl_function_c_tests.
19211 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
19212 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
19215 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
19217 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
19218 operands in a special way. Assert that pos+len <= mode precision.
19220 2017-01-05 Jakub Jelinek <jakub@redhat.com>
19222 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
19223 3 argument Alias with unlimited for the negative form.
19224 (fno-vect-cost-model): Removed.
19226 2017-01-05 Martin Liska <mliska@suse.cz>
19228 * hsa-gen.c (gen_hsa_divmod): New function.
19229 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
19231 2017-01-05 Martin Liska <mliska@suse.cz>
19234 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
19237 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19239 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
19240 small constant length operands.
19242 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19244 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
19245 between loop iterations.
19247 2017-01-05 Martin Liska <mliska@suse.cz>
19250 * gimplify.c (gimplify_decl_expr): Compare to
19251 asan_poisoned_variables instread of checking flags.
19252 (gimplify_target_expr): Likewise.
19253 (gimplify_expr): Likewise.
19254 (gimplify_function_tree): Conditionally initialize
19255 asan_poisoned_variables.
19257 2017-01-04 Jeff Law <law@redhat.com>
19259 PR tree-optimizatin/78812
19260 * rtl.h (contains_mem_rtx_p): Prototype.
19261 * ifcvt.c (containts_mem_rtx_p): Move from here to...
19262 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
19263 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
19264 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
19265 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
19267 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19269 * input.c (assert_char_at_range): Default-initialize actual_range.
19271 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19273 * df-scan.c (df_ref_create_structure): Make regno unsigned,
19274 to match the caller.
19276 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19278 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
19279 insns after final jump in test to emit dummy move.
19281 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19283 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
19284 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
19286 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19288 * multiple_target.c (create_dispatcher_calls): Init e_next.
19289 * tree-ssa-loop-split.c (split_loop): Init border.
19290 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
19293 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
19298 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
19299 (altivec_register_operand): Do not return true if the operand
19300 contains a SUBREG mixing SImode and SFmode.
19301 (vsx_register_operand): Likewise.
19302 (vsx_reg_sfsubreg_ok): New predicate.
19303 (vfloat_operand): Do not return true if the operand contains a
19304 SUBREG mixing SImode and SFmode.
19305 (vint_operand): Likewise.
19306 (vlogical_operand): Likewise.
19307 (gpc_reg_operand): Likewise.
19308 (int_reg_operand): Likewise.
19309 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
19310 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
19311 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
19313 (rs6000_emit_move_si_sf_subreg): New helper function.
19314 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
19315 fixup SUBREGs involving SImode and SFmode.
19316 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
19317 numbers for the new peephole2 optimization.
19318 (peephole2 for SFmode unions): New peephole2 to optimize cases in
19319 the GLIBC math library that do AND/IOR/XOR operations on single
19320 precision floating point.
19321 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
19322 target macros to say whether we need to avoid SUBREGs mixing
19324 (TARGET_ALLOW_SF_SUBREG): Likewise.
19325 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
19326 (UNSPEC_SI_FROM_SF): Likewise.
19327 (iorxor): Change spacing.
19328 (and_ior_xor): New iterator for AND, IOR, and XOR.
19329 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
19330 (movdi_from_sf_zero_ext): Likewise.
19331 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
19332 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
19333 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
19334 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
19335 (fms<mode>4): Likewise.
19336 (fnma<mode>4): Likewise.
19337 (fnms<mode>4): Likewise.
19338 (nfma<mode>4): Likewise.
19339 (nfms<mode>4): Likewise.
19341 2017-01-04 Marek Polacek <polacek@redhat.com>
19344 * doc/invoke.texi: Document -Wpointer-compare.
19346 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19348 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
19351 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
19352 descriptions for -gdwarf-5 and emit them as uleb128 instead of
19355 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
19358 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
19359 documentation of the powerpc_popcntb_ok attribute.
19360 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19361 code to issue warning messages if a requested CPU configuration is
19362 not supported by the binary (assembler and loader) toolchain.
19363 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
19364 made to define a built-in function that has been disabled.
19365 (paired_init_builtins): Add assertion to prevent ICE if attempt is
19366 made to define a built-in function that has been disabled.
19367 (altivec_init_builtins): Add comment explaining why definition
19368 of the DST built-in functions is not preceded by an assertion
19369 check. Add assertions to prevent ICE if attempts are made to
19370 define an altivec predicate or an abs* built-in function that has
19372 (htm_init_builtins): Add comment explaining why definition of the
19373 htm built-in functions is not preceded by an assertion check.
19375 2017-01-04 Jeff Law <law@redhat.com>
19377 PR tree-optimizatin/67955
19378 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
19379 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
19380 the points-to solution does not include pt_null. Use DECL_PT_UID
19383 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
19385 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
19386 Use gen_int_mode instead of gen_lopwart for const_int operands.
19388 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19390 PR tree-optimization/71563
19391 * match.pd: Simplify X << Y into X if Y is known to be 0 or
19392 out of range value - has low bits known to be zero.
19394 2017-01-04 Alan Modra <amodra@gmail.com>
19396 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
19397 * configure: Regenerate.
19398 * config.in: Regenerate.
19400 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19403 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
19404 a substring of the message, but strcmp with the whole message. Ifdef
19405 ENABLE_NLS, translate the message first using dgettext.
19407 2017-01-03 Jeff Law <law@redhat.com>
19409 PR tree-optimizatin/78856
19410 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
19411 (mark_threaded_blocks): Remove code to truncate thread paths that
19412 cross multiple loop headers. Instead invalidate the cached loop
19413 iteration information and handle case of a thread path walking
19414 into an irreducible region.
19416 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
19419 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
19420 assertions. Add support for doing the signbit if the IEEE 128-bit
19421 floating point value is in a GPR.
19422 * config/rs6000/rs6000.md (Fsignbit): Delete.
19423 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
19424 Update the length attribute if the value is in a GPR.
19425 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
19426 the sign or zero extension instruction, since the value is always 0/1.
19427 (signbit<mode>2_dm2): Delete using <Fsignbit>.
19430 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
19431 extracting SImode to a GPR register so that we can generate a
19432 store, limit the vector to be in a traditional Altivec register
19433 for the vextuwrx instruction.
19435 2017-01-03 Ian Lance Taylor <iant@google.com>
19437 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
19439 2017-01-03 Martin Sebor <msebor@redhat.com>
19441 PR tree-optimization/78696
19442 * gimple-ssa-sprintf.c (format_floating): Correct handling of
19443 precision. Use MPFR for %f for greater fidelity. Correct handling
19445 (pass_sprintf_length::compute_format_length): Set width and precision
19446 specified by asrerisk to void_node for vararg functions.
19447 (try_substitute_return_value): Adjust dump output.
19449 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
19451 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
19453 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
19455 * doc/invoke.texi (SPARC options): Document -mlra as the default.
19456 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
19457 -mlra/-mno-lra was passed to the compiler.
19459 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
19461 PR rtl-optimization/65618
19462 * emit-rtl.c (try_split): Move initialization of "before" and
19463 "after" to just before the call to emit_insn_after_setloc.
19465 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
19467 * doc/md.texi (Standard Names): Remove reference to Java frontend.
19469 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
19471 * dwarf2out.c (gen_enumeration_type_die): When
19472 -gno-strict-dwarf, add a DW_AT_encoding attribute.
19474 2017-01-03 Jakub Jelinek <jakub@redhat.com>
19476 PR tree-optimization/78965
19477 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
19478 Change first argument from const call_info & to call_info &. For %n
19479 set info.nowrite to false.
19481 PR middle-end/78901
19482 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
19483 possibly throwing calls.
19485 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
19486 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
19487 and fns handling, rather than in a separate case SSA_NAME.
19489 2017-01-02 Jeff Law <law@redhat.com>
19491 * config/darwin-driver.c (darwin_driver_init): Const-correctness
19492 fixes for first_period and second_period variables.
19494 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
19497 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
19498 (*insvqi_1): New insn pattern.
19499 (*insvqi_1_mem_rex64): Ditto.
19500 (*insvqi_2): Ditto.
19501 (*insvqi_3): Rename from *insvqi.
19503 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
19505 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
19507 * doc/cfg.texi (Edges): Remove reference to Java.
19508 (Maintaining the CFG): Ditto.
19510 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19512 PR middle-end/77674
19513 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
19514 transparent aliases.
19516 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19518 PR middle-end/77484
19519 * predict.def (PRED_CALL): Update hitrate.
19520 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
19521 * predict.c (tree_estimate_probability_bb): Split CALL predictor
19522 into direct/indirect/polymorphic variants.
19524 2017-01-01 Jakub Jelinek <jakub@redhat.com>
19526 Update copyright years.
19528 * gcc.c (process_command): Update copyright notice dates.
19529 * gcov-dump.c (print_version): Ditto.
19530 * gcov.c (print_version): Ditto.
19531 * gcov-tool.c (print_version): Ditto.
19532 * gengtype.c (create_file): Ditto.
19533 * doc/cpp.texi: Bump @copying's copyright year.
19534 * doc/cppinternals.texi: Ditto.
19535 * doc/gcc.texi: Ditto.
19536 * doc/gccint.texi: Ditto.
19537 * doc/gcov.texi: Ditto.
19538 * doc/install.texi: Ditto.
19539 * doc/invoke.texi: Ditto.
19541 Copyright (C) 2017 Free Software Foundation, Inc.
19543 Copying and distribution of this file, with or without modification,
19544 are permitted in any medium without royalty provided the copyright
19545 notice and this notice are preserved.