1 2017-07-28 Bin Cheng <bin.cheng@arm.com>
3 * tree-predcom.c: Revise general description of the pass.
4 (enum chain_type): New enum type for store elimination.
5 (struct chain): New field supporting store elimination.
6 (struct component): Ditto.
7 (dump_chain): Dump store-stores chain.
8 (release_chain): Release resources.
9 (split_data_refs_to_components): Compute and create component
10 contains only stores for elimination.
11 (get_chain_last_ref_at): New function.
12 (make_invariant_chain): Initialization.
13 (make_rooted_chain): Specify chain type in parameter and record it.
14 (add_looparound_copies): Skip for store-stores chain.
15 (determine_roots_comp): Compute type of chain and pass it to
17 (initialize_root_vars_store_elim_2): New function.
18 (finalize_eliminated_stores): New function.
19 (remove_stmt): Handle store for elimination.
20 (execute_pred_commoning_chain): Execute predictive commoning on
22 (determine_unroll_factor): Skip unroll for store-stores chain.
23 (prepare_initializers_chain_store_elim): New function.
24 (prepare_initializers_chain): Hanlde store-store chain.
25 (prepare_finalizers_chain, prepare_finalizers): New function.
26 (tree_predictive_commoning_loop): Return integer value indicating
27 if loop is unrolled or lcssa form is corrupted.
28 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
30 2017-07-28 Bin Cheng <bin.cheng@arm.com>
32 * tree-predcom.c (initialize_root): Delete.
33 (execute_pred_commoning_chain): Initialize root vars and replace
34 reference of non-combined chain directly, rather than call above
37 2017-07-28 Bin Cheng <bin.cheng@arm.com>
39 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
40 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
42 2017-07-28 Bin Cheng <bin.cheng@arm.com>
44 * tree-predcom.c (struct chain): New field init_seq.
45 (release_chain): Release init_seq.
46 (prepare_initializers_chain): Record intialization stmts in above
48 (insert_init_seqs): New function.
49 (tree_predictive_commoning_loop): Call insert_init_seqs.
51 2017-07-28 Bin Cheng <bin.cheng@arm.com>
53 * tree-predcom.c (determine_roots_comp): Skip trivial components.
55 2017-07-28 Richard Biener <rguenther@suse.de>
57 * match.pd: Remove superfluous :c.
58 * genmatch.c (simplify::id): Add member.
59 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
61 (current_id): New global.
62 (dt_node::parent): Move from ...
63 (dt_operand::parent): ... here. Add for_id member.
64 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
65 (decision_tree::find_node): Relax order requirement when
66 merging DT_TRUE nodes to ones inbetween the current simplify
67 and the one we try to merge with. Add diagnostic whenever
68 we need to enforce pattern order by not merging.
69 (decision_tree::insert): Set current_id.
70 (decision_tree::print_node): Dump parent node and for_id.
71 (parser::last_id): Add member.
72 (parser::push_simplify): Assign unique id.
73 (parser::parser): Initialize last_id.
75 2017-07-28 Martin Liska <mliska@suse.cz>
78 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
79 gimple_build_debug_bind.
81 2017-07-28 Richard Biener <rguenther@suse.de>
83 PR tree-optimization/81502
84 * match.pd: Add pattern combining BIT_INSERT_EXPR with
86 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
88 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
89 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
90 for BIT_FIELD_REF args.
91 * fold-const.c (make_bit_field_ref): Likewise.
92 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
94 2017-07-28 Jakub Jelinek <jakub@redhat.com>
97 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
98 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
99 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
100 Or it into SANITIZER_UNDEFINED.
101 * ubsan.c: Include gimple-fold.h and varasm.h.
102 (ubsan_expand_ptr_ifn): New function.
103 (instrument_pointer_overflow): New function.
104 (maybe_instrument_pointer_overflow): New function.
105 (instrument_object_size): Formatting fix.
106 (pass_ubsan::execute): Call instrument_pointer_overflow
107 and maybe_instrument_pointer_overflow.
108 * internal-fn.c (expand_UBSAN_PTR): New function.
109 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
110 * sanitizer.def (__ubsan_handle_pointer_overflow,
111 __ubsan_handle_pointer_overflow_abort): New builtins.
112 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
113 * internal-fn.def (UBSAN_PTR): New internal function.
114 * opts.c (sanitizer_opts): Add pointer-overflow.
115 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
116 * fold-const.c (build_range_check): Compute pointer range check in
117 integral type if pointer arithmetics would be needed. Formatting
120 2017-07-28 Martin Liska <mliska@suse.cz>
123 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
124 parameters that are of a variable-length.
126 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
128 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
130 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
131 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
132 (CRT_CALL_STATIC_FUNCTION): Likewise.
133 (ASM_DEFAULT_SPEC): New define.
134 (ASM_SPEC32): Likewise.
135 (ASM_SPEC64): Likewise.
136 (ASM_SPEC_COMMON): Likewise.
137 (ASM_SPEC): Likewise.
138 (INVALID_64BIT): Likewise.
139 (LINK_OS_DEFAULT_SPEC): Likewise.
140 (LINK_OS_SPEC32): Likewise.
141 (LINK_OS_SPEC64): Likewise.
142 (POWERPC_LINUX): Likewise.
143 (PTRDIFF_TYPE): Likewise.
144 (RESTORE_FP_PREFIX): Likewise.
145 (RESTORE_FP_SUFFIX): Likewise.
146 (SAVE_FP_PREFIX): Likewise.
147 (SAVE_FP_SUFFIX): Likewise.
148 (SIZE_TYPE): Likewise.
149 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
150 (TARGET_64BIT): Likewise.
151 (TARGET_64BIT): Likewise.
152 (TARGET_AIX): Likewise.
153 (WCHAR_TYPE_SIZE): Likewise.
154 (WCHAR_TYPE): Undefine.
155 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
156 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
157 (CPP_OS_RTEMS_SPEC): Delete.
158 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
159 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
161 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
163 2017-07-28 Jakub Jelinek <jakub@redhat.com>
165 PR tree-optimization/81578
166 * tree-parloops.c (build_new_reduction): Bail out if
167 reduction_code isn't one of the standard OpenMP reductions.
168 Move the details printing after that decision.
170 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
172 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
173 related to reload_in_progress.
174 (splat_input_operand): Likewise.
175 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
177 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
179 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
180 (TARGET_INSTANTIATE_DECLS): Likewise.
181 (legitimate_indexed_address_p): Delete reload_in_progress code.
182 (rs6000_debug_legitimate_address_p): Likewise.
183 (rs6000_eliminate_indexed_memrefs): Likewise.
184 (rs6000_emit_le_vsx_store): Likewise.
185 (rs6000_emit_move_si_sf_subreg): Likewise.
186 (rs6000_emit_move): Likewise.
187 (register_to_reg_type): Likewise.
188 (rs6000_pre_atomic_barrier): Likewise.
189 (rs6000_machopic_legitimize_pic_address): Likewise.
190 (rs6000_allocate_stack_temp): Likewise.
191 (rs6000_address_for_fpconvert): Likewise.
192 (rs6000_address_for_altivec): Likewise.
193 (rs6000_secondary_memory_needed_rtx): Delete function.
194 (rs6000_check_sdmode): Likewise.
195 (rs6000_alloc_sdmode_stack_slot): Likewise.
196 (rs6000_instantiate_decls): Likewise.
197 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
198 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
199 Delete reload_in_progress.
200 (*vec_reload_and_plus_<mptrsize>): Likewise.
201 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
202 (vsx_div_v2di): Likewise.
203 (vsx_udiv_v2di): Likewise.
205 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
207 * config/rs6000/rs6000.opt (mlra): Replace with stub.
208 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
209 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
210 (rs6000_debug_reg_global): Delete print of LRA status.
211 (rs6000_option_override_internal): Delete dead LRA related code.
212 (rs6000_lra_p): Delete function.
213 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
215 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
217 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
218 * config/riscv/rtems.h: New file.
220 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
221 Sudakshina Das <sudi.das@arm.com>
223 * config/aarch64/aarch64.md
224 (define_split for and<mode>3nr_compare): Move
225 non aarch64_logical_operand to a register.
226 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
227 register immediate operand to a register.
228 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
230 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
233 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
235 2017-07-27 Richard Biener <rguenther@suse.de>
237 PR tree-optimization/81573
238 PR tree-optimization/81494
239 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
240 multi defuse cycle case.
242 2017-07-27 Richard Biener <rguenther@suse.de>
244 PR tree-optimization/81571
245 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
248 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
250 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
251 earlier and only if MASK_FPU is set. Adjust formatting.
253 2017-07-27 Martin Liska <mliska@suse.cz>
255 * opt-functions.awk: Add validation of value of Init.
256 * optc-gen.awk: Pass new argument.
258 2017-07-27 Martin Liska <mliska@suse.cz>
260 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
263 2017-07-27 Martin Liska <mliska@suse.cz>
265 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
266 BBs and edges seen by autoFDO.
268 2017-07-27 Richard Biener <rguenther@suse.de>
270 PR tree-optimization/81502
271 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
272 with incompatible but same sized type.
273 (execute_update_addresses_taken): Likewise.
275 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
277 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
278 flag_tree_loop_vectorize rather than flag_tree_vectorize.
280 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
283 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
284 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
285 Change s_operand to memory_operand.
287 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
289 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
290 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
291 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
292 Emit instructions rather than returning an expression. Handle TFmode
293 and KFmode by casting to TImode.
294 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
295 (rs6000_emit_le_vsx_store): Likewise.
296 * config/rs6000/vsx.md (VSX_TI): New iterator.
297 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
298 (*vsx_le_undo_permute_<mode>): Likewise.
299 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
300 emit the split sequence.
301 (*vsx_le_perm_store_<mode>): Likewise.
303 2017-07-27 Jakub Jelinek <jakub@redhat.com>
305 PR tree-optimization/81555
306 PR tree-optimization/81556
307 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
308 if true, force CHANGED for the recursive invocation.
309 (reassociate_bb): Remember original length of ops array, pass
310 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
312 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
313 attributes for noipa attribute. For naked attribute use
314 lookup_attribute first before lookup_attribute_spec.
315 * final.c (rest_of_handle_final): Disable IPA RA for functions with
317 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
318 for functions with noipa attribute.
319 (cgraph_externally_visible_p): Return true for functions with noipa
321 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
322 for functions with noipa attribute.
323 * doc/extend.texi: Document noipa function attribute.
324 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
325 also for functions with noipa attribute.
326 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
328 2017-07-26 Andrew Pinski <apinski@cavium.com>
330 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
331 vec_unalign_load_cost and vec_unalign_store_cost.
333 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
335 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
336 -mvsx-small-integer option.
337 (ISA_3_0_MASKS_IEEE): Likewise.
338 (OTHER_VSX_VECTOR_MASKS): Likewise.
339 (POWERPC_MASKS): Likewise.
340 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
341 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
342 code, only testing for DImode being allowed in non-VSX floating
344 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
345 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
347 (rs6000_option_override_internal): Delete -mvsx-small-integer.
348 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
349 TARGET_P8_VECTOR test.
350 (rs6000_secondary_reload_simple_move): Likewise.
351 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
352 since TARGET_P9_VECTOR was already tested.
353 (rs6000_opt_masks): Remove -mvsx-small-integer.
354 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
355 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
357 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
358 test for TARGET_VEXTRACTUB was used, and that uses
360 (p9 extract splitter): Likewise.
361 (vsx_extract_<mode>_di_p9): Likewise.
362 (vsx_extract_<mode>_store_p9): Likewise.
363 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
364 for TARGET_P9_VECTOR was used. Delete code that is now dead with
365 the elimination of TARGET_VSX_SMALL_INTEGER.
366 (vsx_extract_<mode>_p8): Likewise.
367 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
368 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
369 (vsx_set_<mode>_p9): Likewise.
370 (vsx_set_v4sf_p9): Likewise.
371 (vsx_set_v4sf_p9_zero): Likewise.
372 (vsx_insert_extract_v4sf_p9): Likewise.
373 (vsx_insert_extract_v4sf_p9_2): Likewise.
374 * config/rs6000/rs6000.md (sign extend splitter): Change
375 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
376 (floatsi<mode>2_lfiwax_mem): Likewise.
377 (floatunssi<mode>2_lfiwzx_mem): Likewise.
378 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
379 since a test for TARGET_P9_VECTOR was used.
380 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
381 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
382 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
383 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
384 TARGET_P8_VECTOR test.
385 (fix_trunc<mode>si2_stfiwx): Likewise.
386 (fix_trunc<mode>si2_internal): Likewise.
387 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
388 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
390 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
391 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
392 TARGET_P8_VECTOR test.
393 (fixuns_trunc<mode>si2_stfiwx): Likewise.
394 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
395 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
397 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
398 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
399 since a test for TARGET_P9_VECTOR was used.
400 (splitter for loading small constants): Likewise.
402 2017-07-26 Andrew Pinski <apinski@cavium.com>
404 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
407 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
410 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
411 (fp_valid_at): Likewise.
413 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
415 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
416 (qdf24xx_addrcost_table): Likewise.
417 (cortexa57_tunings): Update to use generic_branch_cost.
418 (cortexa72_tunings): Likewise.
419 (cortexa73_tunings): Likewise.
420 (qdf24xx_tunings): Likewise.
422 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
424 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
425 (thunderx2t99_branch_cost): Likewise.
426 (cortexa35_tunings): Update to use generic_branch_cost.
427 (cortexa53_tunings): Likewise.
428 (cortexa57_tunings): Likewise.
429 (cortexa72_tunings): Likewise.
430 (cortexa73_tunings): Likewise.
431 (thunderx2t99_tunings): Likewise.
433 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
435 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
436 (sparc_option_override): Honour MASK_FSMULD.
437 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
438 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
439 * config/sparc/sparc.opt (mfsmuld): New option.
440 * doc/invoke.texi (mfsmuld): Document option.
442 2017-07-26 Marek Polacek <polacek@redhat.com>
445 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
447 2017-07-26 Richard Biener <rguenther@suse.de>
449 * gimple-match-head.c (do_valueize): Return OP if valueize
451 (get_def): New helper to get at the def stmt of a SSA name
453 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
454 do_valueize to get at the def stmt.
455 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
457 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
460 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
462 2017-07-26 Martin Liska <mliska@suse.cz>
465 * function.c (expand_function_start): Make expansion of
466 nonlocal_goto_save_area after parm_birth_insn.
468 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
470 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
471 from all CPU target flags enable members.
473 2017-07-26 Richard Biener <rguenther@suse.de>
475 * genmatch.c (dt_simplify::gen): Make iterator vars const.
476 (decision_tree::gen): Make 'type' const.
477 (write_predicate): Likewise.
479 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
481 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
482 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
483 (rs6000_option_override_internal): Likewise.
484 (rs6000_expand_vector_set): Likewise.
485 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
486 (TARGET_UPPER_REGS_SF): Likewise.
487 (TARGET_UPPER_REGS_DI): Likewise.
488 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
489 (TARGET_DIRECT_MOVE_64BIT): Likewise.
490 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
491 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
492 (Splitters for DI constants in Altivec registers): Likewise.
493 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
494 (vsx_set_v4sf_p9): Likewise.
495 (vsx_set_v4sf_p9_zero): Likewise.
496 (vsx_insert_extract_v4sf_p9): Likewise.
497 (vsx_insert_extract_v4sf_p9_2): Likewise.
499 2017-07-25 Carl Love <cel@us.ibm.com>
501 * doc/extend.texi: Update the built-in documentation file for the
502 existing built-in functions
503 vector signed char vec_cnttz (vector signed char);
504 vector unsigned char vec_cnttz (vector unsigned char);
505 vector signed short vec_cnttz (vector signed short);
506 vector unsigned short vec_cnttz (vector unsigned short);
507 vector signed int vec_cnttz (vector signed int);
508 vector unsigned int vec_cnttz (vector unsigned int);
509 vector signed long long vec_cnttz (vector signed long long);
510 vector unsigned long long vec_cnttz (vector unsigned long long);
512 2017-07-25 Andrew Pinski <apinski@cavium.com>
514 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
515 accesses where the use is for the first operand of a BIT_INSERT.
517 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
520 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
521 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
523 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
525 * config/i386/gstabs.h: Delete.
526 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
528 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
530 * config/i386/i386.c (ix86_decompose_address): Do not check for
531 register RTX when looking at index_reg or base_reg.
532 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
534 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
536 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
537 to update EH info here.
539 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
541 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
543 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
545 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
547 2017-07-25 Torsten Duwe <duwe@suse.de>
549 * common.opt: Introduce -fpatchable-function-entry
550 command line option, and its variables function_entry_patch_area_size
551 and function_entry_patch_area_start.
552 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
553 including a two-value parser.
554 * target.def (print_patchable_function_entry): New target hook.
555 * targhooks.h (default_print_patchable_function_entry): New function.
556 * targhooks.c (default_print_patchable_function_entry): Likewise.
557 * toplev.c (process_options): Switch off IPA-RA if
558 patchable function entries are being generated.
559 * varasm.c (assemble_start_function): Look at the
560 patchable-function-entry command line switch and current
561 function attributes and maybe generate NOP instructions by
562 calling the print_patchable_function_entry hook.
563 * doc/extend.texi: Document patchable_function_entry attribute.
564 * doc/invoke.texi: Document -fpatchable_function_entry
566 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
568 * doc/tm.texi: Re-generate.
570 2017-07-25 Jakub Jelinek <jakub@redhat.com>
573 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
574 TARGET_AVX512DQ rather than TARGET_AVX512BW.
576 2017-07-25 Tamar Christina <tamar.christina@arm.com>
578 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
580 2017-07-25 Richard Biener <rguenther@suse.de>
582 PR tree-optimization/81455
583 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
584 not walk in cycles when looking for guards.
586 2017-07-25 Richard Biener <rguenther@suse.de>
588 PR tree-optimization/81529
589 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
590 when optimizing backedge uses.
592 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
594 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
596 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
597 to dl_section_ref. On AIX, append an expression to subtract
598 the size of the section length to dl_section_ref.
600 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
602 * configure.ac: If any of the config.* scripts fail, exit 1.
603 * configure: Regenerate.
605 2017-07-25 Richard Biener <rguenther@suse.de>
608 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
609 of immediate uses, be more verbose on errors.
611 2017-07-25 Richard Biener <rguenther@suse.de>
613 PR tree-optimization/81510
614 * tree-vect-loop.c (vect_is_simple_reduction): When the
615 reduction stmt is not inside the loop bail out.
617 2017-07-25 Richard Biener <rguenther@suse.de>
619 PR tree-optimization/81303
620 * tree-vect-loop-manip.c (vect_loop_versioning): Build
621 profitability check against LOOP_VINFO_NITERSM1.
623 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
625 * domwalk.c (cmp_bb_postorder): Simplify.
626 (sort_bbs_postorder): New function. Use it...
627 (dom_walker::walk): ...here to optimize common cases.
629 2017-07-25 Martin Liska <mliska@suse.cz>
632 * ipa-visibility.c (function_and_variable_visibility): Make the
633 redirection just on target that supports aliasing.
634 Fix GNU coding style.
636 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
639 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
641 * config/arm/rtems.h: Likewise.
642 * config/bfin/rtems.h: Likewise.
643 * config/i386/rtemself.h: Likewise.
644 * config/lm32/rtems.h: Likewise.
645 * config/m32c/rtems.h: Likewise.
646 * config/m68k/rtemself.h: Likewise.
647 * config/microblaze/rtems.h: Likewise.
648 * config/mips/rtems.h: Likewise.
649 * config/moxie/rtems.h: Likewise.
650 * config/nios2/rtems.h: Likewise.
651 * config/powerpcspe/rtems.h: Likewise.
652 * config/rs6000/rtems.h: Likewise.
653 * config/rtems.h: Likewise.
654 * config/sh/rtems.h: Likewise.
655 * config/sh/rtemself.h: Likewise.
656 * config/sparc/rtemself.h: Likewise.
658 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
661 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
662 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
663 * tree-ssa-structalias.c (alias_get_name): Same.
665 2017-07-25 Bin Cheng <bin.cheng@arm.com>
668 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
669 instructions if no du chain is found.
671 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
673 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
675 2017-07-25 Richard Biener <rguenther@suse.de>
678 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
681 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
683 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
685 (ISA_2_7_MASKS_SERVER): Likewise.
686 (ISA_3_0_MASKS_IEEE): Likewise.
687 (OTHER_P8_VECTOR_MASKS): Likewise.
688 (OTHER_VSX_VECTOR_MASKS): Likewise.
689 (POWERPC_MASKS): Likewise.
690 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
691 duplicate list of options.
692 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
693 explicit -mupper-regs options.
694 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
695 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
696 alias for -mupper-regs-df.
697 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
698 (rs6000_init_hard_regno_mode_ok): Likewise.
699 (rs6000_option_override_internal): Likewise.
700 (rs6000_opt_masks): Likewise.
701 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
702 options in terms of whether -mvsx or -mpower8-vector was used.
703 (TARGET_UPPER_REGS_DI): Likewise.
704 (TARGET_UPPER_REGS_SF): Likewise.
705 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
706 -mupper-regs-* options.
708 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
710 * passes.c (emergency_dump_function): Print some empty lines and a
711 header before the RTL dump.
713 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
715 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
717 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
720 * config/aarch64/aarch64.c (aarch64_classify_symbol):
721 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
723 2017-07-24 Carl Love <cel@us.ibm.com>
725 * config/rs6000/rs6000-c.c: Add support for built-in functions
726 vector float vec_extract_fp32_from_shorth (vector unsigned short);
727 vector float vec_extract_fp32_from_shortl (vector unsigned short);
728 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
729 vec_extract_fp_from_shortl): Add defines for the two builtins.
730 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
731 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
733 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
734 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
735 * doc/extend.texi: Update the built-in documentation file for the
736 new built-in function.
738 2017-07-24 Jakub Jelinek <jakub@redhat.com>
741 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
743 * doc/generic.texi: Likewise.
744 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
745 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
747 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
749 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
750 (aarch64_mls_elt_merge<mode>): Likewise.
752 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
754 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
757 2017-07-23 Michael Collison <michael.collison@arm.com>
759 * config/arm/arm.c (arm_option_override): Deprecate
760 use of -mstructure-size-boundary.
761 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
762 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
764 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
767 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
768 Reduce cost estimate for direct moves.
770 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
773 * config/i386/i386.c (ix86_option_override_internal): Disable
774 BMI, BMI2 and TBM instructions for -m16.
776 2017-07-21 Carl Love <cel@us.ibm.com>
778 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
779 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
780 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
781 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
782 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
783 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
784 VMULOSW): New enum "unspec" values.
785 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
786 altivec_vmulosw): New patterns.
787 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
788 VMULOSW): Add definitions.
790 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
792 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
794 * config/aarch64/aarch64-options-extensions.def (rdma); New.
795 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
796 (AARCH64_FL_V8_1): Renumber.
797 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
798 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
799 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
800 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
801 rdma to feature modifiers list.
803 2017-07-21 Yury Gribov <tetra2005@gmail.com>
806 * ipa-visibility (function_and_variable_visibility): Convert
807 recursive PLT call to direct call if appropriate.
809 2017-07-21 Andrew Pinski <apinski@cavium.com>
811 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
812 operand 1 to see if the types precision matches.
813 * fold-const.c (operand_equal_p): Likewise.
815 2017-07-21 Richard Biener <rguenther@suse.de>
817 PR tree-optimization/81303
818 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
819 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
820 (vect_peeling_hash_get_lowest_cost): Adjust.
821 (vect_enhance_data_refs_alignment): Likewise. Use
822 vect_get_peeling_costs_all_drs to compute the penalty for no
823 peeling to match up costs.
825 2017-07-21 Richard Biener <rguenther@suse.de>
827 PR tree-optimization/81500
828 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
829 we didn't identify a reduction path.
831 2017-07-21 Tom de Vries <tom@codesourcery.com>
832 Cesar Philippidis <cesar@codesourcery.com>
834 PR gcov-profile/81442
835 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
838 2017-07-21 Tom de Vries <tom@codesourcery.com>
841 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
843 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
844 nvptx_override_options_after_change.
846 2017-07-21 Ulrich Drepper <drepper@redhat.com>
848 * dwarf2out.c (output_file_names): Avoid double testing for
851 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
853 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
855 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
857 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
859 (try_crossjump_to_edge): Do not punt on partitioned functions.
861 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
863 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
864 Put all BBs reachable only via paths crossing cold region to cold
866 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
868 2016-07-21 Richard Biener <rguenther@suse.de>
870 PR tree-optimization/81303
871 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
872 into account prologue and epilogue iterations when raising
873 min_profitable_iters to sth at least covering one vector iteration.
875 2017-07-21 Tamar Christina <tamar.christina@arm.com>
877 * config/arm/arm.c (arm_test_cpu_arch_dat):
880 2017-07-20 Nathan Sidwell <nathan@acm.org>
883 * tree.h (TYPE_METHODS): Delete.
884 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
885 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
886 (dbxout_type_methods): Scan TYPE_FIELDS.
887 (dbxout_type): Don't check TYPE_METHODS here.
888 * function.c (use_register_for_decl): Always ignore register for
889 class types when not optimizing.
890 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
891 * tree.c (free_lang_data_in_type): Stitch out member functions and
892 templates from TYPE_FIELDS.
893 (build_distinct_type_copy, verify_type_variant,
894 verify_type): Member fns are on TYPE_FIELDS.
895 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
896 * tree-pretty-print.c (dump_generic_node): Likewise.
898 2017-07-20 Jakub Jelinek <jakub@redhat.com>
901 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
902 V2TImode and V4TImode.
903 (ix86_expand_vector_extract): Likewise.
904 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
905 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
906 (ssescalarmode): Handle V4TImode and V2TImode.
907 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
908 (*vec_extractv2ti, *vec_extractv4ti): New insns.
909 (VEXTRACTI128_MODE): New mode iterator.
910 (splitter for *vec_extractv?ti first element): New.
911 (VEC_INIT_MODE): New mode iterator.
912 (vec_init<mode>): Consolidate 3 expanders into one using
913 VEC_INIT_MODE mode iterator.
915 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
917 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
918 non_spilled_static_chain_regno_p.
920 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
922 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
924 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
926 * bb-reorder.c (connect_traces): Allow copying of blocks within
929 2017-07-20 Richard Biener <rguenther@suse.de>
931 * gimple.h (gimple_phi_result): Add gphi * overload.
932 (gimple_phi_result_ptr): Likewise.
933 (gimple_phi_arg): Likewise. Adjust index assert to only
934 allow actual argument accesses rather than all slots available
936 (gimple_phi_arg_def): Add gphi * overload.
937 * tree-phinodes.c (make_phi_node): Initialize only actual
939 (resize_phi_node): Clear memory not covered by old node,
940 do not initialize excess argument slots.
941 (reserve_phi_args_for_new_edge): Initialize new argument slot
944 2017-07-20 Bin Cheng <bin.cheng@arm.com>
946 PR tree-optimization/81388
948 2016-07-21 Bin Cheng <bin.cheng@arm.com>
950 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
951 by removing computation of may_be_zero.
953 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
954 Tom de Vries <tom@codesourcery.com>
957 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
958 when gimple level profile disagrees with what RTL expander did.
960 2017-07-20 Richard Biener <rguenther@suse.de>
962 PR tree-optimization/61171
963 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
964 (vect_analyze_stmt): Add slp instance parameter.
965 (vectorizable_reduction): Likewise.
966 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
967 (vect_is_simple_reduction): Deal with chains not detected
968 as SLP reduction chain, specifically not properly associated
969 chains containing a mix of plus/minus.
970 (get_reduction_op): Remove.
971 (get_initial_defs_for_reduction): Simplify, pass in whether
972 this is a reduction chain, pass in the SLP node for the PHIs.
973 (vect_create_epilog_for_reduction): Get the SLP instance as
975 (vectorizable_reduction): Get the SLP instance as arg.
976 During analysis remember the SLP node with the PHIs in the
977 instance. Simplify getting at the vectorized reduction PHIs.
978 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
979 through SLP instance.
980 (vect_slp_analyze_operations): Likewise.
981 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
982 (vect_transform_stmt): Likewise.
984 2017-07-20 Tom de Vries <tom@codesourcery.com>
986 PR tree-optimization/81489
987 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
988 read of phi arg location to before loop that modifies phi.
990 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
992 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
995 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
998 * except.c (execute): Fix ordering issue.
1000 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
1002 PR rtl-optimization/81423
1003 * combine.c (make_compound_operation_int): Don't try to optimize
1004 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
1006 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
1008 PR rtl-optimization/81423
1009 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
1010 with a constant that is -1 in the truncated to mode.
1012 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1014 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
1015 (determine_unlikely_bbs): ... here.
1016 * predict.h (propagate_unlikely_bbs_forward): Declare.
1017 * cfgexpand.c (pass_expand::execute): Use it.
1018 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
1020 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
1021 propagate_unlikely_bbs_forward.
1023 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1026 * except.c (maybe_add_nop_after_section_switch): New function.
1029 2017-07-19 Tom de Vries <tom@codesourcery.com>
1031 * gimple.h (gimple_phi_set_arg): Make assert more strict.
1033 2017-07-19 Tom de Vries <tom@codesourcery.com>
1035 * gimple.h (gimple_phi_arg): Make assert more strict.
1037 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
1039 * config.gcc (powerpc*-*-*): Add mmintrin.h.
1040 * config/rs6000/mmintrin.h: New file.
1041 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
1043 2017-07-19 Jakub Jelinek <jakub@redhat.com>
1045 PR tree-optimization/81346
1046 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
1048 2017-07-19 Tom de Vries <tom@codesourcery.com>
1050 * config/nvptx/nvptx.md (VECIM): Add V2DI.
1052 2017-07-19 Tom de Vries <tom@codesourcery.com>
1054 * config/nvptx/nvptx-modes.def: Add V2DImode.
1055 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
1056 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
1057 (nvptx_output_mov_insn): Handle lack of mov.b128.
1058 (nvptx_print_operand): Handle 'H' and 'L' codes.
1059 (nvptx_vector_mode_supported): Allow V2DImode.
1060 (nvptx_preferred_simd_mode): New function.
1061 (nvptx_data_alignment): New function.
1062 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
1063 nvptx_preferred_simd_mode.
1064 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
1066 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
1068 2017-07-19 Tom de Vries <tom@codesourcery.com>
1070 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
1071 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
1072 (nvptx_vector_mode_supported): New function. Allow V2SImode.
1073 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
1074 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
1075 (mov<VECIM>_insn): New define_insn.
1076 (define_expand "mov<VECIM>): New define_expand.
1078 2017-07-19 Tom de Vries <tom@codesourcery.com>
1080 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
1082 2017-07-19 Jakub Jelinek <jakub@redhat.com>
1084 PR tree-optimization/81346
1085 * fold-const.h (fold_div_compare, range_check_type): Declare.
1086 * fold-const.c (range_check_type): New function.
1087 (build_range_check): Use range_check_type.
1088 (fold_div_compare): No longer static, rewritten into
1089 a match.pd helper function.
1090 (fold_comparison): Don't call fold_div_compare here.
1091 * match.pd (X / C1 op C2): New optimization using fold_div_compare
1094 2017-07-19 Nathan Sidwell <nathan@acm.org>
1096 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
1097 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
1098 * tree.c (find_decls_types_r, verify_type): Use
1099 TYPE_{MIN,MAX}_VALUE_RAW.
1100 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
1101 (hash_tree): Likewise.
1102 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
1104 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
1107 2017-07-18 Tom de Vries <tom@codesourcery.com>
1110 * omp-expand.c (expand_omp_for_static_chunk): Handle
1111 equal-argument loop exit phi.
1113 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
1116 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
1117 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
1118 operand 2 predicate.
1119 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
1120 operand 2 predicate.
1121 (ror,rol -> rorx splitters): Use const_int_operand as
1122 operand 2 predicate.
1124 2017-06-18 Richard Biener <rguenther@suse.de>
1126 PR tree-optimization/81410
1127 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
1128 the gap in the ! slp_perm SLP case after each group.
1130 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1133 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
1136 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1139 * predict.c (set_even_probabilities): Cleanup; do not affect
1140 probabilities that are already known.
1141 (combine_predictions_for_bb): Call even when count is set.
1143 2017-07-18 Nathan Sidwell <nathan@acm.org>
1145 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
1148 2017-07-18 Bin Cheng <bin.cheng@arm.com>
1151 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
1152 optimization for loop niter analysis.
1154 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
1157 * config/avr/avr.c (avr_optimize_casesi): Don't use
1158 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
1160 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1162 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
1163 body_cost_vec from _vect_peel_extended_info.
1164 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
1165 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
1168 2017-07-18 Bin Cheng <bin.cheng@arm.com>
1170 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
1172 2017-07-18 Richard Biener <rguenther@suse.de>
1174 PR tree-optimization/80620
1175 PR tree-optimization/81403
1176 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
1177 info when re-using a VN table entry.
1179 2017-07-18 Richard Biener <rguenther@suse.de>
1181 PR tree-optimization/81418
1182 * tree-vect-loop.c (vectorizable_reduction): Properly compute
1183 vectype_in. Verify that with lane-reducing reduction operations
1184 we have a single def-use cycle.
1186 2017-07-17 Carl Love <cel@us.ibm.com>
1188 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
1190 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1191 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
1192 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
1193 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
1194 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
1195 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
1196 VMULOSW): New enum "unspec" values.
1197 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
1198 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
1199 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
1200 altivec_vmulosw): New patterns.
1201 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
1202 VMULOSW): Add definitions.
1203 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
1205 * config/alpha/alpha.c: Include predict.h.
1207 2017-07-17 Yury Gribov <tetra2005@gmail.com>
1209 * tree-vrp.c (compare_assert_loc): Fix comparison function
1210 to return predictable results.
1212 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1214 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
1217 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
1218 * doc/invoke.texi (mexpand-adddi): Update text.
1220 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1222 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
1223 that also clobbers the CC register. The old expand code is moved
1225 (*arc_clzsi2): ... here.
1226 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
1227 the CC register. The old expand code is moved to ...
1228 (arc_ctzsi2): ... here.
1230 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1232 * config/arc/arc.opt (mindexed-loads): Use initial value
1233 TARGET_INDEXED_LOADS_DEFAULT.
1234 (mauto-modify-reg): Use initial value
1235 TARGET_AUTO_MODIFY_REG_DEFAULT.
1236 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
1237 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
1238 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
1239 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
1241 2017-07-17 Martin Liska <mliska@suse.cz>
1244 * opts.c (finish_options): Do not allow -fgnu-tm
1245 w/ -fsanitize={kernel-,}address. Say sorry.
1247 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1250 * tree-loop-distribution.c (classify_partition): Only assert on
1251 numer of iterations.
1252 (merge_dep_scc_partitions): Delete prameter. Update function call.
1253 (distribute_loop): Remove code handling loop with unknown niters.
1254 (pass_loop_distribution::execute): Skip loop with unknown niters.
1256 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1259 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
1260 function sort_partitions_by_post_order.
1262 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1264 PR tree-optimization/81374
1265 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
1266 the max index of basic blocks, rather than number of basic blocks.
1268 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1270 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
1272 (arc_legitimate_pic_operand_p): Likewise.
1273 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
1275 (arc_needs_pcl_p): Likewise.
1276 (arc_legitimate_pc_offset_p): Likewise.
1277 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
1278 function is also used in constrains.md.
1279 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
1280 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
1281 PLUS. Only return true/false in known cases, otherwise assert.
1282 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
1283 is already called in arc_legitimate_constant_p.
1284 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
1286 (LEGITIMATE_PIC_OPERAND_P): Use
1287 arc_raw_symbolic_reference_mentioned_p function.
1288 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
1293 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1294 Andrew Burgess <andrew.burgess@embecosm.com>
1296 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
1297 (arc_return_address_register): New function.
1298 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
1299 (arc_handle_fndecl_attribute): Add naked attribute.
1300 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
1301 (TARGET_WARN_FUNC_RETURN): Likewise.
1302 (arc_allocate_stack_slots_for_args): New function.
1303 (arc_warn_func_return): Likewise.
1304 (machine_function): Change type fn_type.
1305 (arc_compute_function_type): Consider new naked function type,
1306 change function return type.
1307 (arc_must_save_register): Adapt to handle new
1308 arc_compute_function_type's return type.
1309 (arc_expand_prologue): Likewise.
1310 (arc_expand_epilogue): Likewise.
1311 (arc_return_address_regs): Delete.
1312 (arc_return_address_register): New function.
1313 (arc_epilogue_uses): Use above function.
1314 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
1315 (arc_function_type): Change encoding, add naked type.
1316 (ARC_INTERRUPT_P): Change to handle the new encoding.
1317 (ARC_FAST_INTERRUPT_P): Likewise.
1318 (ARC_NORMAL_P): Define.
1319 (ARC_NAKED_P): Likewise.
1320 (arc_compute_function_type): Delete prototype.
1321 * config/arc/arc.md (in_ret_delay_slot): Use
1322 arc_return_address_register function.
1323 (simple_return): Likewise.
1324 (p_return_i): Likewise.
1326 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1328 PR tree-optimization/81428
1329 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
1330 can't be built for those types.
1332 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
1334 Remove stuff dead since r239246.
1336 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
1337 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
1338 (avr_inform_devices): Remove dead stuff.
1340 2017-07-17 Tamar Christina <tamar.christina@arm.com>
1342 * config/arm/arm_neon.h: Fix softp typo.
1344 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1346 PR tree-optimization/81365
1347 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
1348 aggregate moves onto bb predecessor edges, make sure there are no
1349 loads that could alias the lhs in between the start of bb and the
1352 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
1355 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
1356 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
1357 [LSHIFTRT, outer_code = TRUNCATE]: Same.
1359 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1361 PR tree-optimization/81396
1362 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
1363 (init_symbolic_number): Initialize it to 1.
1364 (perform_symbolic_merge): Add n_ops from both operands into the new
1366 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
1367 without base_addr as useless if they need more than one operation.
1368 (bswap_replace): Handle !bswap case for NULL base_addr.
1370 2017-07-17 Tom de Vries <tom@codesourcery.com>
1373 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
1376 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
1378 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
1379 conditional builtin define __FIX_LEON3FT_B2BST.
1381 2017-07-17 Daniel Cederman <cederman@gaisler.com>
1383 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
1384 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
1387 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1389 PR rtl-optimization/81424
1390 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
1391 to remove potential trapping from operands if -fnon-call-exceptions.
1393 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1395 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
1396 profile_proability for scalling.
1397 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
1399 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1401 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
1403 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1405 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
1406 fixpoint arithmetics.
1408 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1410 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
1411 fixpoint arithmetics.
1413 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1415 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
1416 fixpoint arithmetics.
1418 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1420 * profile-count.h (profile_probability::from_reg_br_prob_note,
1421 profile_probability::to_reg_br_prob_note): New functions.
1422 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
1423 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
1424 * predict.c (probability_reliable_p): Update.
1425 (edge_probability_reliable_p): Update.
1426 (br_prob_note_reliable_p): Update.
1427 (invert_br_probabilities): Update.
1428 (add_reg_br_prob_note): New function.
1429 (combine_predictions_for_insn): Update.
1430 * asan.c (asan_clear_shadow): Update.
1431 * cfgbuild.c (compute_outgoing_frequencies): Update.
1432 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
1433 (update_br_prob_note): Update.
1434 (rtl_verify_edges): Update.
1435 (purge_dead_edges): Update.
1436 (fixup_reorder_chain): Update.
1437 * emit-rtl.c (try_split): Update.
1438 * ifcvt.c (cond_exec_process_insns): Update.
1439 (cond_exec_process_if_block): Update.
1440 (dead_or_predicable): Update.
1441 * internal-fn.c (expand_addsub_overflow): Update.
1442 (expand_neg_overflow): Update.
1443 (expand_mul_overflow): Update.
1444 * loop-doloop.c (doloop_modify): Update.
1445 * loop-unroll.c (compare_and_jump_seq): Update.
1446 * optabs.c (emit_cmp_and_jump_insn_1): Update.
1447 * predict.h: Update.
1448 * reorg.c (mostly_true_jump): Update.
1450 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
1451 * config/alpha/alpha.c (emit_unlikely_jump): Update.
1452 * config/arc/arc.c: (emit_unlikely_jump): Update.
1453 * config/arm/arm.c: (emit_unlikely_jump): Update.
1454 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
1455 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
1456 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
1457 (ix86_print_operand): Update.
1458 (ix86_split_fp_branch): Update.
1459 (predict_jump): Update.
1460 * config/ia64/ia64.c (ia64_print_operand): Update.
1461 * config/mmix/mmix.c (mmix_print_operand): Update.
1462 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
1463 (rs6000_expand_split_stack_prologue): Update.
1464 * config/rs6000/rs6000.c: Update.
1465 * config/s390/s390.c (s390_expand_vec_strlen): Update.
1466 (s390_expand_vec_movstr): Update.
1467 (s390_expand_cs_tdsi): Update.
1468 (s390_expand_split_stack_prologue): Update.
1469 * config/sh/sh.c (sh_print_operand): Update.
1470 (expand_cbranchsi4): Update.
1471 (expand_cbranchdi4): Update.
1472 * config/sparc/sparc.c (output_v9branch): Update.
1473 * config/spu/spu.c (get_branch_target): Update.
1474 (ea_load_store_inline): Update.
1475 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
1476 * config/tilepro/tilepro.c: Update.
1478 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1480 * gimplify.c (mostly_copy_tree_r): Revert latest change.
1481 (gimplify_save_expr): Likewise.
1483 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1485 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
1487 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1489 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
1491 * timevar.def (TV_IPA_FNSUMMARY): Define.
1493 2017-07-16 Daniel Cederman <cederman@gaisler.com>
1495 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
1496 to back store errata sensitive sequence from being generated.
1497 (sqrtdf2_fix): Likewise.
1499 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1501 * tree-ssa-threadupdate.c (compute_path_counts,
1502 update_joiner_offpath_counts): Use profile_probability.
1504 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
1507 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1509 * config/arm/arm-c.c (arm_cpu_builtins): Define
1510 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1512 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
1514 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1515 array entries to represent __ieee128 versions of the
1516 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
1517 scalar_extract_sig, and scalar_insert_exp built-in functions.
1518 (altivec_resolve_overloaded_builtin): Add special case handling
1519 for the __builtin_scalar_insert_exp function, as represented by
1520 the P9V_BUILTIN_VEC_VSIEDP constant.
1521 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
1522 exponent support for __ieee128 argument.
1523 (VSESQP): Add scalar extract signature support for __ieee128
1525 (VSTDCNQP): Add scalar test negative support for __ieee128
1527 (VSIEQP): Add scalar insert exponent support for __int128 argument
1528 with __ieee128 result.
1529 (VSIEQPF): Add scalar insert exponent support for __ieee128
1530 argument with __ieee128 result.
1531 (VSTDCQP): Add scalar test data class support for __ieee128
1533 (VSTDCNQP): Add overload support for scalar test negative with
1535 (VSTDCQP): Add overload support for scalar test data class
1537 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
1539 (UNSPEC_VSX_SIEXPQP): New constant.
1540 (xsxexpqp): New insn for VSX scalar extract exponent quad
1542 (xsxsigqp): New insn for VSX scalar extract significand quad
1544 (xsiexpqpf): New insn for VSX scalar insert exponent quad
1545 precision with floating point argument.
1546 (xststdcqp): New expand for VSX scalar test data class quad
1548 (xststdcnegqp): New expand for VSX scalar test negative quad
1550 (xststdcqp): New insn to match expansions for VSX scalar test data
1551 class quad precision and VSX scalar test negative quad precision.
1552 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
1553 special case operand checking to enforce that second operand of
1554 VSX scalar test data class with quad precision argument is a 7-bit
1556 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
1557 prototypes and descriptions of __ieee128 versions of
1558 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
1559 scalar_test_data_class, and scalar_test_neg built-in functions.
1561 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1563 PR tree-optimization/81162
1564 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
1565 replace a negate with an add.
1567 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
1569 * doc/invoke.texi (arm/-mcpu): Document +crypto.
1571 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1573 * config/arm/arm-c.c (arm_cpu_builtins): Define
1574 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1576 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1578 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
1579 (armv8-r): Set ARM Cortex-R52 as default CPU.
1580 * config/arm/arm-tables.opt: Regenerate.
1581 * config/arm/arm-tune.md: Regenerate.
1582 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
1584 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
1585 extension for -mcpu=cortex-r52.
1587 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1589 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
1590 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
1591 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
1592 (fp-armv8): Define it as FP_ARMv8 only.
1593 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
1594 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
1596 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
1597 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
1598 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
1599 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
1600 than TARGET_FPU_ARMV8.
1601 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
1602 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
1603 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
1605 * config/arm/neon.md (neon_vrint): Likewise.
1606 (neon_vcvt): Likewise.
1607 (neon_<fmaxmin_op><mode>): Likewise.
1608 (<fmaxmin><mode>3): Likewise.
1609 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
1610 * config/arm/predicates.md (arm_cond_move_operator): Check against
1611 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
1613 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
1615 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
1618 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1620 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
1621 loop in comment with memset.
1623 2017-07-14 Martin Liska <mliska@suse.cz>
1625 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
1626 * dwarf2out.c (is_java): Remove the function.
1627 (output_pubname): Remove usage of the function.
1628 (lower_bound_default): Remove usage of DW_LANG_Java.
1629 (gen_compile_unit_die): Likewise.
1630 * gcc.c: Remove compiler defaults for .java and .zip files.
1631 * gimple-expr.c (remove_suffix): Change as there's no longer
1632 extension than 4-letter one.
1633 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
1634 (gimplify_save_expr): Likewise.
1635 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
1636 as it's possible even for other languages than Java.
1637 * langhooks.h (struct lang_hooks): Remove Java from a comment.
1638 * lto-opts.c (lto_write_options): Remove reference to Java.
1639 * opts.c (strip_off_ending): Update file extension handling.
1640 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
1641 * tree-eh.c (lower_resx): Likewise.
1642 * tree.c (free_lang_data_in_type): Remove dead code.
1643 (find_decls_types_r): Likewise.
1644 (build_common_builtin_nodes): Remove Java from a comment.
1645 (verify_type): Remove dead code.
1646 * varasm.c (assemble_external): Remove Java from a comment.
1648 2017-07-14 Martin Liska <mliska@suse.cz>
1650 * opts.c (finish_options): Add quotes.
1651 (common_handle_option): Likewise.
1653 2017-07-14 Martin Liska <mliska@suse.cz>
1655 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
1656 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
1658 * dwarf2out.c (lower_bound_default): Do not handle
1660 (gen_compile_unit_die): Likewise.
1661 * gcc.c: Remove default extension binding for GNU Pascal.
1662 * stmt.c: Remove Pascal language from a comment.
1663 * xcoffout.c: Likewise.
1665 2017-07-13 David Malcolm <dmalcolm@redhat.com>
1668 * diagnostic-show-locus.c (fixit_cmp): New function.
1669 (layout::layout): Sort m_fixit_hints.
1670 (column_range::column_range): Assert that the values are valid.
1671 (struct char_span): New struct.
1672 (correction::overwrite): New method.
1673 (struct source_line): New struct.
1674 (line_corrections::add_hint): Add assertions. Reimplement memcpy
1675 calls in terms of classes source_line and char_span, and
1676 correction::overwrite.
1677 (selftest::test_overlapped_fixit_printing_2): New function.
1678 (selftest::diagnostic_show_locus_c_tests): Call it.
1680 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
1682 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
1683 early if there is no lhs.
1685 2017-07-13 Martin Liska <mliska@suse.cz>
1687 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
1688 (gen_reference_type_die): Likewise.
1689 * stor-layout.c: Remove Pascal-related comment.
1691 2017-07-13 Martin Liska <mliska@suse.cz>
1693 * opts.c (finish_options): Add quotes to error messages.
1694 (parse_sanitizer_options): Likewise.
1696 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
1698 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
1700 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
1702 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
1704 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
1706 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
1708 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
1710 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1713 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
1714 provides the hardware capability bits, define the macro
1715 __BUILTIN_CPU_SUPPORTS__.
1716 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
1717 if GLIBC does not provide the hardware capability bits. Add a
1718 gcc_unreachable call if the built-in cpu function is neither
1719 __builtin_cpu_is nor __builtin_cpu_supports.
1720 (rs6000_get_function_versions_dispatcher): Change the warning
1721 that an old GLIBC is used which does not export the capability
1722 bits to be an error.
1723 * doc/extend.texi (target_clones attribute): Document the
1724 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
1725 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
1726 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
1727 the macros defined by GCC if the newer GLIBC is available.
1729 2017-07-12 Jeff Law <law@redhat.com>
1731 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
1732 remaining includes slightly.
1733 * config/riscv/riscv-builtins.c: Include profile-count.h.
1735 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1738 * config/avr/avr.c (avr_set_current_function): In diagnostic
1739 messages: Quote keywords and (parts of) identifiers.
1740 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
1743 2017-07-12 Carl Love <cel@us.ibm.com>
1745 * config/rs6000/rs6000-c.c: Add support for built-in functions
1746 vector bool char vec_revb (vector bool char);
1747 vector bool short vec_revb (vector short char);
1748 vector bool int vec_revb (vector bool int);
1749 vector bool long long vec_revb (vector bool long long);
1750 * doc/extend.texi: Update the built-in documentation file for the
1751 new built-in functions.
1753 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1755 * config/s390/s390.md: Remove movcc splitter.
1757 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1759 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
1760 load/store on condition.
1762 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
1765 * config/avr/avr.c (avr_encode_section_info)
1766 [progmem && !TREE_READONLY]: Error if progmem object needs
1769 2017-07-11 Michael Collison <michael.collison@arm.com>
1771 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
1774 2017-07-11 Carl Love <cel@us.ibm.com>
1776 * config/rs6000/rs6000-c.c: Add support for builtins
1777 vector unsigned int vec_parity_lsbb (vector signed int);
1778 vector unsigned int vec_parity_lsbb (vector unsigned int);
1779 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
1780 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
1781 vector unsigned long long vec_parity_lsbb (vector signed long long);
1782 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
1783 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
1784 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
1785 * doc/extend.texi: Update the built-in documentation file for the
1786 new built-in functions.
1788 2017-07-11 David Malcolm <dmalcolm@redhat.com>
1790 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
1791 (layout::m_primary_loc): New field.
1792 (layout::layout): Initialize new field. Move location filtering
1793 logic from here to...
1794 (layout::maybe_add_location_range): ...this new method. Add
1795 support for filtering to just the lines already specified by other
1797 (layout::will_show_line_p): New method.
1798 (gcc_rich_location::add_location_if_nearby): New method.
1799 (selftest::test_add_location_if_nearby): New test function.
1800 (selftest::diagnostic_show_locus_c_tests): Call it.
1801 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
1804 2017-07-11 Tom de Vries <tom@codesourcery.com>
1806 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
1807 (bb_first_real_insn): New function.
1808 (nvptx_single): Add extra initialization of broadcasted condition
1811 2017-07-11 Nathan Sidwell <nathan@acm.org>
1813 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
1815 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
1817 * doc/extend.texi (AVR Function Attributes): Remove weblink to
1818 Binutils doc as TEXI will mess them up.
1819 * doc/invoke.texi (AVR Options): Same here.
1821 2017-07-11 Daniel Cederman <cederman@gaisler.com>
1823 * config/sparc/sparc.opt (mfix-ut700): New option.
1824 (mfix-gr712rc): Likewise.
1825 (sparc_fix_b2bst): New variable.
1826 * doc/invoke.texi (SPARC options): Document them.
1827 (ARM options): Fix warnings.
1828 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
1829 instructions to prevent sequences that can trigger the store-store
1830 errata for certain LEON3FT processors.
1831 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
1832 (sparc_option_override): Set sparc_fix_b2bst appropriately.
1833 * config/sparc/sparc.md (fix_b2bst): New attribute.
1834 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
1836 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
1839 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
1841 (*rsqrtsf2_sse): Ditto.
1843 (div<mode>3): Macroize insn from divdf3 and divsf3
1844 using MODEF mode iterator.
1846 2017-07-10 Martin Sebor <msebor@redhat.com>
1848 PR tree-optimization/80397
1849 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
1850 instead of testing for equality to INTEGER_TYPE.
1852 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
1854 * config.gcc: Remove uclibc from arc target spec.
1856 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
1858 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
1860 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1863 * lto-wrapper.c (remove_option): New function.
1864 (merge_and_complain): Merge PIC/PIE options more realistically.
1866 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
1868 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
1872 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
1873 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
1874 * config.in: Regenerate.
1875 * configure: Regenerate.
1876 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
1877 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
1878 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
1879 (TARGET_GASISR_PROLOGUES): ...target mask.
1880 * common/config/avr/avr-common.c
1881 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
1882 Set -mgas-isr-prologues.
1883 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
1884 INSERT_PASS_BEFORE for it.
1885 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
1886 * config/avr/avr.c (avr_option_override)
1887 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
1888 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
1889 (avr_attribute_table) <no_gccisr>: Add new function attribute.
1890 (avr_set_current_function) <is_no_gccisr>: Init machine field.
1891 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
1893 (make_avr_pass_pre_proep): New function.
1894 (emit_push_sfr) <treg>: Add argument to function and use it
1896 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
1897 and set machine->gasisr.yes.
1898 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
1899 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
1900 __gcc_isr.n_pushed to .L__stack_usage.
1901 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
1902 (avr_asm_final_postscan_insn): ...this new static function.
1903 * config/avr/avr.h (machine_function)
1904 <is_no_gccisr, use_L__stack_usage>: New fields.
1905 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
1906 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
1907 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
1908 (gasisr, *gasisr): New expander and insn.
1909 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
1910 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
1911 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
1913 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
1915 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
1918 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
1920 Move jump-tables out of .text again.
1923 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
1924 (ASM_OUTPUT_ADDR_VEC): New function.
1925 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
1926 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
1927 INSN_ADDRESSes as asm comment.
1928 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
1929 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
1930 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
1931 * config/avr/avr.md (*tablejump): Adjust comment.
1932 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
1933 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
1935 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
1936 (avr_output_addr_vec): New proto.
1937 (avr_log_t) <insn_addresses>: New field.
1939 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
1942 * config/i386/i386.c (ix86_function_arg_advance): Set
1943 outgoing_args_on_stack to true if there are outgoing arguments
1945 (ix86_function_arg): Likewise.
1946 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
1947 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
1948 * config/i386/i386.h (machine_function): Add
1949 outgoing_args_on_stack.
1951 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
1953 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
1955 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
1957 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
1959 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
1960 (REAL_H): Remove $(MACHMODE_H).
1961 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
1963 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
1964 $(MACHMODE_H) and double-int.h.
1965 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
1967 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
1970 2017-07-07 Andrew Pinski <apinski@cavium.com>
1972 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
1973 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
1975 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
1977 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
1978 Add warning if GCC was not configured to link against a GLIBC that
1979 exports the hardware capability bits.
1980 (make_resolver_func): Make resolver function private and not a
1981 COMDAT function. Create the name with clone_function_name instead
1982 of make_unique_name.
1985 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
1986 correct operand in doing the split.
1988 2017-07-07 Carl Love <cel@us.ibm.com>
1990 * config/rs6000/rs6000-c: Add support for built-in function
1991 vector unsigned short vec_pack_to_short_fp32 (vector float,
1993 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
1994 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
1995 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
1996 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
1997 (convert_4f32_8i16): Add define_expand.
1998 * doc/extend.texi: Update the built-in documentation file for the
1999 new built-in function.
2001 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2003 * config/sparc/m8.md: New file.
2004 * config/sparc/sparc.md: Include m8.md.
2006 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2008 * config/sparc/sparc.opt: New option -mvis4b.
2009 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
2010 (sparc_option_override): Handle VIS4B.
2011 (enum sparc_builtins): Define
2012 SPARC_BUILTIN_DICTUNPACK{8,16,32},
2013 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
2014 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
2015 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
2016 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
2017 (check_constant_argument): New function.
2018 (sparc_vis_init_builtins): Define builtins
2019 __builtin_vis_dictunpack{8,16,32},
2020 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
2021 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
2022 __builtin_vis_fpcmpde{8,16,32}shl and
2023 __builtin_vis_fpcmpur{8,16,32}shl.
2024 (sparc_expand_builtin): Check that the constant operands to
2025 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
2026 constant and in range.
2027 * config/sparc/sparc-c.c (sparc_target_macros): Handle
2029 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
2030 (SPARC_IMM5_P): Likewise.
2031 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
2032 (enabled): Handle vis4b.
2033 (UNSPEC_DICTUNPACK): New unspec.
2034 (UNSPEC_FPCMPSHL): Likewise.
2035 (UNSPEC_FPUCMPSHL): Likewise.
2036 (UNSPEC_FPCMPDESHL): Likewise.
2037 (UNSPEC_FPCMPURSHL): Likewise.
2038 (cpu_feature): New CPU feature `vis4b'.
2039 (dictunpack{8,16,32}): New insns.
2040 (FPCSMODE): New mode iterator.
2041 (fpcscond): New code iterator.
2042 (fpcsucond): Likewise.
2043 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
2044 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
2045 (fpcmpde{8,16,32}{si,di}shl): Likewise.
2046 (fpcmpur{8,16,32}{si,di}shl): Likewise.
2047 * config/sparc/constraints.md: Define constraints `q' for unsigned
2048 2-bit integer constants and `t' for unsigned 5-bit integer
2050 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
2052 (imm5_operand_dictunpack16): Likewise.
2053 (imm5_operand_dictunpack32): Likewise.
2054 (imm2_operand): Likewise.
2055 * doc/invoke.texi (SPARC Options): Document -mvis4b.
2056 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
2057 ditunpack* and fpcmp*shl builtins.
2059 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2061 * config.gcc: Handle m8 in --with-{cpu,tune} options.
2062 * config.in: Add HAVE_AS_SPARC6 define.
2063 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
2065 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
2067 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
2068 (CPP_CPU_SPEC): Handle m8.
2069 (ASM_CPU_SPEC): Likewise.
2070 * config/sparc/sparc-opts.h (enum processor_type): Add
2072 * config/sparc/sparc.c (m8_costs): New struct.
2073 (sparc_option_override): Handle TARGET_CPU_m8.
2074 (sparc32_initialize_trampoline): Likewise.
2075 (sparc64_initialize_trampoline): Likewise.
2076 (sparc_issue_rate): Likewise.
2077 (sparc_register_move_cost): Likewise.
2078 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
2079 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
2080 (ASM_CPU64_DEFAULT_SPEC): Likewise.
2081 (CPP_CPU_SPEC): Handle M8.
2082 (ASM_CPU_SPEC): Likewise.
2083 (AS_M8_FLAG): Define.
2084 * config/sparc/sparc.md: Add m8 to the cpu attribute.
2085 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
2086 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
2088 * configure: Regenerate.
2089 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
2092 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2094 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
2096 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
2097 ("*movdi_insn_sp32"): Do not set v3pipe.
2098 ("*movsi_insn"): Likewise.
2099 ("*movdi_insn_sp64"): Likewise.
2100 ("*movsf_insn"): Likewise.
2101 ("*movdf_insn_sp32"): Likewise.
2102 ("*movdf_insn_sp64"): Likewise.
2103 ("*zero_extendsidi2_insn_sp64"): Likewise.
2104 ("*sign_extendsidi2_insn"): Likewise.
2105 ("*mov<VM32:mode>_insn"): Likewise.
2106 ("*mov<VM64:mode>_insn_sp64"): Likewise.
2107 ("*mov<VM64:mode>_insn_sp32"): Likewise.
2108 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
2109 ("<vlop:code><VL:mode>3"): Likewise.
2110 ("*not_<vlop:code><VL:mode>3"): Likewise.
2111 ("*nand<VL:mode>_vis"): Likewise.
2112 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
2113 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
2114 ("one_cmpl<VL:mode>2"): Likewise.
2115 ("faligndata<VM64:mode>_vis"): Likewise.
2116 ("alignaddrsi_vis"): Likewise.
2117 ("alignaddrdi_vis"): Likweise.
2118 ("alignaddrlsi_vis"): Likewise.
2119 ("alignaddrldi_vis"): Likewise.
2120 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
2121 ("bmaskdi_vis"): Likewise.
2122 ("bmasksi_vis"): Likewise.
2123 ("bshuffle<VM64:mode>_vis"): Likewise.
2124 ("cmask8<P:mode>_vis"): Likewise.
2125 ("cmask16<P:mode>_vis"): Likewise.
2126 ("cmask32<P:mode>_vis"): Likewise.
2127 ("pdistn<P:mode>_vis"): Likewise.
2128 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
2130 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2132 * config/sparc/sparc.md ("subtype"): New insn attribute.
2133 ("*wrgsr_sp64"): Set insn subtype.
2134 ("*rdgsr_sp64"): Likewise.
2135 ("alignaddrsi_vis"): Likewise.
2136 ("alignaddrdi_vis"): Likewise.
2137 ("alignaddrlsi_vis"): Likewise.
2138 ("alignaddrldi_vis"): Likewise.
2139 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
2140 ("fexpand_vis"): Likewise.
2141 ("fpmerge_vis"): Likewise.
2142 ("faligndata<VM64:mode>_vis"): Likewise.
2143 ("bshuffle<VM64:mode>_vis"): Likewise.
2144 ("cmask8<P:mode>_vis"): Likewise.
2145 ("cmask16<P:mode>_vis"): Likewise.
2146 ("cmask32<P:mode>_vis"): Likewise.
2147 ("fchksm16_vis"): Likewise.
2148 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
2149 ("fmean16_vis"): Likewise.
2150 ("fp<plusminus_insn>64_vis"): Likewise.
2151 ("<plusminus_insn>v8qi3"): Likewise.
2152 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
2153 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
2154 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
2155 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
2156 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
2157 ("*movqi_insn"): Likewise.
2158 ("*movhi_insn"): Likewise.
2159 ("*movsi_insn"): Likewise.
2160 ("movsi_pic_gotdata_op"): Likewise.
2161 ("*movdi_insn_sp32"): Likewise.
2162 ("*movdi_insn_sp64"): Likewise.
2163 ("movdi_pic_gotdata_op"): Likewise.
2164 ("*movsf_insn"): Likewise.
2165 ("*movdf_insn_sp32"): Likewise.
2166 ("*movdf_insn_sp64"): Likewise.
2167 ("*zero_extendhisi2_insn"): Likewise.
2168 ("*zero_extendqihi2_insn"): Likewise.
2169 ("*zero_extendqisi2_insn"): Likewise.
2170 ("*zero_extendqidi2_insn"): Likewise.
2171 ("*zero_extendhidi2_insn"): Likewise.
2172 ("*zero_extendsidi2_insn_sp64"): Likewise.
2173 ("ldfsr"): Likewise.
2174 ("prefetch_64"): Likewise.
2175 ("prefetch_32"): Likewise.
2176 ("tie_ld32"): Likewise.
2177 ("tie_ld64"): Likewise.
2178 ("*tldo_ldub_sp32"): Likewise.
2179 ("*tldo_ldub1_sp32"): Likewise.
2180 ("*tldo_ldub2_sp32"): Likewise.
2181 ("*tldo_ldub_sp64"): Likewise.
2182 ("*tldo_ldub1_sp64"): Likewise.
2183 ("*tldo_ldub2_sp64"): Likewise.
2184 ("*tldo_ldub3_sp64"): Likewise.
2185 ("*tldo_lduh_sp32"): Likewise.
2186 ("*tldo_lduh1_sp32"): Likewise.
2187 ("*tldo_lduh_sp64"): Likewise.
2188 ("*tldo_lduh1_sp64"): Likewise.
2189 ("*tldo_lduh2_sp64"): Likewise.
2190 ("*tldo_lduw_sp32"): Likewise.
2191 ("*tldo_lduw_sp64"): Likewise.
2192 ("*tldo_lduw1_sp64"): Likewise.
2193 ("*tldo_ldx_sp64"): Likewise.
2194 ("*mov<VM32:mode>_insn"): Likewise.
2195 ("*mov<VM64:mode>_insn_sp64"): Likewise.
2196 ("*mov<VM64:mode>_insn_sp32"): Likewise.
2198 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2200 * config/sparc/sparc.md ("type"): New insn type viscmp.
2201 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
2203 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
2204 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
2205 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
2206 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
2208 ("n7_vis_logical_11cycle"): Likewise.
2209 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
2210 * config/sparc/niagara2.md ("niag3_vis": Likewise.
2211 * config/sparc/niagara.md ("niag_vis"): Likewise.
2212 * config/sparc/ultra3.md ("us3_fga"): Likewise.
2213 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
2215 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2217 * config/sparc/sparc.md: New instruction type `bmask'.
2218 (bmaskdi_vis): Use the `bmask' type.
2219 (bmasksi_vis): Likewise.
2220 * config/sparc/ultra3.md (us3_array): Likewise.
2221 * config/sparc/niagara7.md (n7_array): Likewise.
2222 * config/sparc/niagara4.md (n4_array): Likewise.
2223 * config/sparc/niagara2.md (niag2_vis): Likewise.
2224 (niag3_vis): Likewise.
2225 * config/sparc/niagara.md (niag_vis): Likewise.
2227 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2229 * ipa-comdats.c: Remove optimize check from gate.
2230 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
2231 for functions not optimized.
2232 (ipa_fn_summary_read): Skip optimize check.
2233 (ipa_fn_summary_write): Likewise.
2234 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
2236 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
2238 (can_inline_edge_p): Check flag_pcc_struct_return for match.
2239 (check_callers): Give up on caller which is not optimized.
2240 (inline_small_functions): Likewise.
2241 (ipa_inline): Do not give up when not optimizing.
2242 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
2243 away unoptimizes cdtors.
2244 (whole_program_function_and_variable_visibility): Do
2245 ipa_discover_readonly_nonaddressable_vars in LTO mode.
2246 * ipa.c (process_references): Do not check optimize.
2247 (symbol_table::remove_unreachable_nodes): Update optimize check.
2248 (set_writeonly_bit): Update optimize check.
2249 (pass_ipa_cdtor_merge::gate): Do not check optimize.
2250 (pass_ipa_single_use::gate): Remove.
2252 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2254 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
2255 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
2256 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
2257 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
2258 permute_load, permute_store, adjust_extract, adjust_splat,
2259 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
2260 replace_swap_with_copy, dump_swap_insn_table,
2261 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
2262 recombine_lvx_pattern, recombine_stvx_pattern,
2263 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
2264 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
2265 to file rs6000-p8swap.c.
2266 * config/rs6000/rs6000-p8swap.c: New file.
2267 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
2268 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
2269 and rs6000*-*-* targets.
2271 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2273 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
2275 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2277 * lto-wrapper.c (merge_and_complain): Do not merge
2278 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
2279 fsigned_zeros, ftrapping_math, fwrapv.
2280 (append_compiler_options): Do not track these options.
2281 (append_linker_options): Likewie
2283 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2285 * cgraphunit.c (cgraph_node::finalize_function): When
2286 !flag_toplevel_reorde set no_reorder flag.
2287 (varpool_node::finalize_decl): Likewise.
2288 (symbol_table::compile): Drop no toplevel reorder path.
2290 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2292 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
2293 edges; zero probability is not better than uninitialized.
2295 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
2297 * asan.h (asan_sanitize_allocas_p): Declare.
2298 * asan.c (asan_sanitize_allocas_p): New function.
2299 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
2300 (handle_builtin_alloca): Likewise.
2301 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
2302 if !asan_sanitize_allocas_p.
2303 * params.def (asan-instrument-allocas): Add new option.
2304 * params.h (ASAN_PROTECT_ALLOCAS): Define.
2305 * opts.c (common_handle_option): Disable allocas sanitization for
2308 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
2310 * asan.c: Include gimple-fold.h.
2311 (get_last_alloca_addr): New function.
2312 (handle_builtin_stackrestore): Likewise.
2313 (handle_builtin_alloca): Likewise.
2314 (asan_emit_allocas_unpoison): Likewise.
2315 (get_mem_refs_of_builtin_call): Add new parameter, remove const
2316 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
2317 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
2318 (instrument_builtin_call): Pass gimple iterator to
2319 get_mem_refs_of_builtin_call.
2320 (last_alloca_addr): New global.
2321 * asan.h (asan_emit_allocas_unpoison): Declare.
2322 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
2323 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
2324 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
2325 if function calls alloca.
2326 * gimple-fold.c (replace_call_with_value): Remove static keyword.
2327 * gimple-fold.h (replace_call_with_value): Declare.
2328 * internal-fn.c: Include asan.h.
2329 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
2330 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
2332 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2334 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
2335 (C_SELFTEST_FLAGS): New.
2336 (CPP_SELFTEST_FLAGS): New.
2337 (SELFTEST_DEPS): New, from deps of s-selftest.
2338 (C_SELFTEST_DEPS): New, from deps of s-selftest.
2339 (CPP_SELFTEST_DEPS): New.
2340 (selftest): Add dependency on s-selftest-c++.
2341 (s-selftest): Rename to...
2342 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
2343 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
2344 than SELFTEST_FLAGS.
2345 (selftest-gdb): Rename to...
2346 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
2348 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
2349 (selftest-valgrind): Rename to...
2350 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
2352 (selftest-valgrind): Reintroduce as an alias for
2353 selftest-c-valgrind.
2354 (s-selftest-c++): New.
2355 (selftest-c++-gdb): New.
2356 (selftest-c++-valgrind): New.
2358 2017-07-06 Olivier Hainque <hainque@adacore.com>
2360 * gcc.c (process_command): When deciding if undefined variables
2361 should be ignored when processing specs, accept "gcc -v" as well.
2363 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2365 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
2366 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
2368 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
2370 * config/arm/arm-cpus.in (armv8-r): Add new entry.
2371 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
2372 * config/arm/arm-tables.opt: Regenerate.
2373 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
2375 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
2377 2017-07-06 Carl Love <cel@us.ibm.com>
2379 * ChangeLog: Clean up from mid air collision
2381 2017-07-06 Carl Love <cel@us.ibm.com>
2383 * config/rs6000/rs6000-c.c: Add support for built-in functions
2384 vector signed int vec_subc (vector signed int, vector signed int);
2385 vector signed __int128 vec_subc (vector signed __int128,
2386 vector signed __int128);
2387 vector unsigned __int128 vec_subc (vector unsigned __int128,
2388 vector unsigned __int128);
2389 vector signed int vec_sube (vector signed int, vector signed int,
2391 vector unsigned int vec_sube (vector unsigned int,
2392 vector unsigned int,
2393 vector unsigned int);
2394 vector signed __int128 vec_sube (vector signed __int128,
2395 vector signed __int128,
2396 vector signed__int128);
2397 vector unsigned __int128 vec_sube (vector unsigned __int128,
2398 vector unsigned __int128,
2399 vector unsigned __int128);
2400 vector signed int vec_subec (vector signed int, vector signed int,
2402 vector unsigned int vec_subec (vector unsigned int,
2403 vector unsigned int,
2404 vector unsigned int);
2405 vector signed __int128 vec_subec (vector signed __int128,
2406 vector signed __int128,
2407 vector signed__int128);
2408 vector unsigned __int128 vec_subec (vector unsigned __int128,
2409 vector unsigned __int128,
2410 vector unsigned __int128);
2411 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
2412 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
2413 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
2414 BU_ALTIVEC_OVERLOAD_X definitions.
2415 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
2416 * doc/extend.texi: Update the built-in documentation file for the new
2419 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2422 * diagnostic-show-locus.c (layout::layout): Use start and finish
2423 spelling location for the start and finish of each range.
2424 * genmatch.c (linemap_client_expand_location_to_spelling_point):
2425 Add unused aspect param.
2426 * input.c (expand_location_1): Add "aspect" param, and use it
2427 to access the correct part of the location.
2428 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
2430 (expand_location_to_spelling_point): Likewise.
2431 (linemap_client_expand_location_to_spelling_point): Add "aspect"
2432 param, and pass it to expand_location_1.
2434 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
2436 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
2437 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
2438 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
2439 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
2440 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
2441 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
2442 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
2443 _mm_maskz_getmant_ss): New intrinsics.
2444 (__builtin_ia32_getexpss128_mask): Changed to ...
2445 __builtin_ia32_getexpss128_round ... this.
2446 (__builtin_ia32_getexpsd128_mask): Changed to ...
2447 __builtin_ia32_getexpsd128_round ... this.
2448 * config/i386/i386-builtin-types.def
2449 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
2450 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
2451 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
2452 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
2453 __builtin_ia32_getmantss_mask_round): New builtins.
2454 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
2455 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
2456 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
2457 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
2458 * config/i386/sse.md
2459 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
2460 avx512f_sgetexp<mode><mask_scalar_name>
2461 <round_saeonly_scalar_name> ... this.
2462 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
2463 %0, %1, %2<round_saeonly_op3>}): Changed to ...
2464 vgetexp<ssescalarmodesuffix>
2465 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2466 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
2467 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
2468 avx512f_vgetmant<mode><mask_scalar_name>
2469 <round_saeonly_scalar_name> ... this.
2470 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
2471 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
2472 vgetmant<ssescalarmodesuffix>
2473 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
2474 %0<mask_scalar_operand4>, %1, %2
2475 <round_saeonly_scalar_mask_op4>, %3} ... this.
2476 * config/i386/subst.md (mask_scalar_operand4,
2477 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
2478 round_saeonly_scalar_nimm_predicate): New subst attributes.
2480 2017-07-06 Julia Koval <julia.koval@intel.com>
2482 * config/i386/i386.c (ix86_erase_embedded_rounding):
2483 Remove code for old rounding pattern.
2485 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
2487 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
2489 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
2491 * doc/sourcebuild.texi (Test Directives, Variants of
2492 dg-require-support): Add documentation for dg-require-stack-check.
2494 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
2496 * config/i386/subst.md (mask_scalar, round_scalar,
2497 round_saeonly_scalar): New meta-templates.
2498 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
2499 round_scalar_mask_operand3, round_scalar_mask_op3,
2500 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
2501 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
2502 round_saeonly_scalar_constraint,
2503 round_saeonly_scalar_prefix): New subst attribute.
2504 * config/i386/sse.md
2505 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
2506 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
2507 <round_scalar_name> ... this.
2508 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
2509 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
2510 <round_scalar_name> ... this.
2511 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
2512 <sse>_vm<code><mode>3<mask_scalar_name>
2513 <round_saeonly_scalar_name> ... this.
2514 (v<plusminus_mnemonic><ssescalarmodesuffix>
2515 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2516 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2517 v<plusminus_mnemonic><ssescalarmodesuffix>
2518 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2519 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2520 (v<multdiv_mnemonic><ssescalarmodesuffix>
2521 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2522 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2523 v<multdiv_mnemonic><ssescalarmodesuffix>
2524 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2525 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
2526 (v<maxmin_float><ssescalarmodesuffix>
2527 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
2528 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
2529 v<maxmin_float><ssescalarmodesuffix>
2530 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2531 %0<mask_scalar_operand3>, %1, %<iptr>2
2532 <round_saeonly_scalar_mask_op3>} ... this.
2534 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
2536 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
2537 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
2539 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
2540 Alan Hayward <alan.hayward@arm.com>
2541 David Sherwood <david.sherwood@arm.com>
2543 * combine.c (simplify_if_then_else): Remove "enum" before
2545 * compare-elim.c (can_eliminate_compare): Likewise.
2546 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
2548 (aarch64_lookup_simd_builtin_type): Likewise.
2549 (aarch64_simd_builtin_type): Likewise.
2550 (aarch64_init_simd_builtin_types): Likewise.
2551 (aarch64_simd_expand_args): Likewise.
2552 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
2554 (aarch64_reverse_mask): Likewise.
2555 (aarch64_simd_emit_reg_reg_move): Likewise.
2556 (aarch64_gen_adjusted_ldpstp): Likewise.
2557 (aarch64_ccmp_mode_to_code): Likewise.
2558 (aarch64_operands_ok_for_ldpstp): Likewise.
2559 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2560 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
2562 (aarch64_min_divisions_for_recip_mul): Likewise.
2563 (aarch64_reassociation_width): Likewise.
2564 (aarch64_get_condition_code_1): Likewise.
2565 (aarch64_simd_emit_reg_reg_move): Likewise.
2566 (aarch64_simd_attr_length_rglist): Likewise.
2567 (aarch64_reverse_mask): Likewise.
2568 (aarch64_operands_ok_for_ldpstp): Likewise.
2569 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
2570 (aarch64_gen_adjusted_ldpstp): Likewise.
2571 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
2573 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
2574 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
2575 (arm_lookup_simd_builtin_type): Likewise.
2576 (arm_simd_builtin_type): Likewise.
2577 (arm_init_simd_builtin_types): Likewise.
2578 (arm_expand_builtin_args): Likewise.
2579 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
2580 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
2581 (ft32_setup_incoming_varargs): Likewise.
2582 (ft32_function_arg): Likewise.
2583 (ft32_function_arg_advance): Likewise.
2584 (ft32_pass_by_reference): Likewise.
2585 (ft32_arg_partial_bytes): Likewise.
2586 (ft32_valid_pointer_mode): Likewise.
2587 (ft32_addr_space_pointer_mode): Likewise.
2588 (ft32_addr_space_legitimate_address_p): Likewise.
2589 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
2591 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
2592 (ix86_emit_outlined_ms2sysv_restore): Likewise.
2593 (iamcu_alignment): Likewise.
2594 (canonicalize_vector_int_perm): Likewise.
2595 (ix86_noce_conversion_profitable_p): Likewise.
2596 (ix86_mpx_bound_mode): Likewise.
2597 (ix86_operands_ok_for_move_multiple): Likewise.
2598 * config/microblaze/microblaze-protos.h
2599 (microblaze_expand_conditional_branch_reg): Likewise.
2600 * config/microblaze/microblaze.c
2601 (microblaze_expand_conditional_branch_reg): Likewise.
2602 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
2604 (rs6000_reassociation_width): Likewise.
2605 (rs6000_invalid_binary_op): Likewise.
2606 (fusion_p9_p): Likewise.
2607 (emit_fusion_p9_load): Likewise.
2608 (emit_fusion_p9_store): Likewise.
2609 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
2611 (riscv_hard_regno_mode_ok_p): Likewise.
2612 (riscv_address_insns): Likewise.
2613 (riscv_split_symbol): Likewise.
2614 (riscv_legitimize_move): Likewise.
2615 (riscv_function_value): Likewise.
2616 (riscv_hard_regno_nregs): Likewise.
2617 (riscv_expand_builtin): Likewise.
2618 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
2619 (riscv_build_integer): Likewise.
2620 (riscv_split_integer): Likewise.
2621 (riscv_legitimate_constant_p): Likewise.
2622 (riscv_cannot_force_const_mem): Likewise.
2623 (riscv_regno_mode_ok_for_base_p): Likewise.
2624 (riscv_valid_base_register_p): Likewise.
2625 (riscv_valid_offset_p): Likewise.
2626 (riscv_valid_lo_sum_p): Likewise.
2627 (riscv_classify_address): Likewise.
2628 (riscv_legitimate_address_p): Likewise.
2629 (riscv_address_insns): Likewise.
2630 (riscv_load_store_insns): Likewise.
2631 (riscv_force_binary): Likewise.
2632 (riscv_split_symbol): Likewise.
2633 (riscv_force_address): Likewise.
2634 (riscv_legitimize_address): Likewise.
2635 (riscv_move_integer): Likewise.
2636 (riscv_legitimize_const_move): Likewise.
2637 (riscv_legitimize_move): Likewise.
2638 (riscv_address_cost): Likewise.
2639 (riscv_subword): Likewise.
2640 (riscv_output_move): Likewise.
2641 (riscv_canonicalize_int_order_test): Likewise.
2642 (riscv_emit_int_order_test): Likewise.
2643 (riscv_function_arg_boundary): Likewise.
2644 (riscv_pass_mode_in_fpr_p): Likewise.
2645 (riscv_pass_fpr_single): Likewise.
2646 (riscv_pass_fpr_pair): Likewise.
2647 (riscv_get_arg_info): Likewise.
2648 (riscv_function_arg): Likewise.
2649 (riscv_function_arg_advance): Likewise.
2650 (riscv_arg_partial_bytes): Likewise.
2651 (riscv_function_value): Likewise.
2652 (riscv_pass_by_reference): Likewise.
2653 (riscv_setup_incoming_varargs): Likewise.
2654 (riscv_print_operand): Likewise.
2655 (riscv_elf_select_rtx_section): Likewise.
2656 (riscv_save_restore_reg): Likewise.
2657 (riscv_for_each_saved_reg): Likewise.
2658 (riscv_register_move_cost): Likewise.
2659 (riscv_hard_regno_mode_ok_p): Likewise.
2660 (riscv_hard_regno_nregs): Likewise.
2661 (riscv_class_max_nregs): Likewise.
2662 (riscv_memory_move_cost): Likewise.
2663 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
2664 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
2665 (rl78_addr_space_address_mode): Likewise.
2666 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2668 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2669 (rs6000_reassociation_width): Likewise.
2670 (rs6000_invalid_binary_op): Likewise.
2671 (fusion_p9_p): Likewise.
2672 (emit_fusion_p9_load): Likewise.
2673 (emit_fusion_p9_store): Likewise.
2674 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
2675 (ok_for_simple_move_operands): Likewise.
2676 (ok_for_simple_move_strict_operands): Likewise.
2677 (ok_for_simple_arith_logic_operands): Likewise.
2678 (visium_legitimize_reload_address): Likewise.
2679 (visium_select_cc_mode): Likewise.
2680 (output_cbranch): Likewise.
2681 (visium_split_double_move): Likewise.
2682 (visium_expand_copysign): Likewise.
2683 (visium_expand_int_cstore): Likewise.
2684 (visium_expand_fp_cstore): Likewise.
2685 * config/visium/visium.c (visium_pass_by_reference): Likewise.
2686 (visium_function_arg): Likewise.
2687 (visium_function_arg_advance): Likewise.
2688 (visium_libcall_value): Likewise.
2689 (visium_setup_incoming_varargs): Likewise.
2690 (visium_legitimate_constant_p): Likewise.
2691 (visium_legitimate_address_p): Likewise.
2692 (visium_legitimize_address): Likewise.
2693 (visium_secondary_reload): Likewise.
2694 (visium_register_move_cost): Likewise.
2695 (visium_memory_move_cost): Likewise.
2696 (prepare_move_operands): Likewise.
2697 (ok_for_simple_move_operands): Likewise.
2698 (ok_for_simple_move_strict_operands): Likewise.
2699 (ok_for_simple_arith_logic_operands): Likewise.
2700 (visium_function_value_1): Likewise.
2701 (rtx_ok_for_offset_p): Likewise.
2702 (visium_legitimize_reload_address): Likewise.
2703 (visium_split_double_move): Likewise.
2704 (visium_expand_copysign): Likewise.
2705 (visium_expand_int_cstore): Likewise.
2706 (visium_expand_fp_cstore): Likewise.
2707 (visium_split_cstore): Likewise.
2708 (visium_select_cc_mode): Likewise.
2709 (visium_split_cbranch): Likewise.
2710 (output_cbranch): Likewise.
2711 (visium_print_operand_address): Likewise.
2712 * expmed.c (flip_storage_order): Likewise.
2713 * expmed.h (emit_cstore): Likewise.
2714 (flip_storage_order): Likewise.
2715 * genrecog.c (validate_pattern): Likewise.
2716 * hsa-gen.c (gen_hsa_addr): Likewise.
2717 * internal-fn.c (expand_arith_overflow): Likewise.
2718 * ira-color.c (allocno_copy_cost_saving): Likewise.
2719 * lra-assigns.c (find_hard_regno_for_1): Likewise.
2720 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
2721 (process_invariant_for_inheritance): Likewise.
2722 * lra-eliminations.c (move_plus_up): Likewise.
2723 * omp-low.c (lower_oacc_reductions): Likewise.
2724 * simplify-rtx.c (simplify_subreg): Likewise.
2725 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
2726 (TARGET_CHKP_BOUND_MODE): Likewise..
2727 * targhooks.c (default_chkp_bound_mode): Likewise.
2728 (default_setup_incoming_vararg_bounds): Likewise.
2729 * targhooks.h (default_chkp_bound_mode): Likewise.
2730 (default_setup_incoming_vararg_bounds): Likewise.
2731 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
2732 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
2733 (have_whole_vector_shift): Likewise.
2734 * tree-vect-stmts.c (vectorizable_load): Likewise.
2735 * doc/tm.texi: Regenerate.
2737 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2739 Graceful degrade if Binutils PR21472 is not available.
2742 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
2743 .rodata in flash test fails.
2744 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
2745 * confgure: Regenerate.
2746 * config.in: Regenerate.
2747 * config/avr/avr.c (avr_asm_named_section)
2748 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
2749 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
2750 (avr_asm_init_sections): Same.
2752 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2754 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
2755 (fma<VH:mode>4_intrinsic): Likewise.
2756 (*fmsub<VCVTF:mode>4): Likewise.
2757 (*fmsub<VH:mode>4_intrinsic): Likewise.
2759 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
2762 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
2763 Don't depend on "optimize > 0".
2764 (out_movhi_r_mr, out_movqi_mr_r): Same.
2765 (out_movhi_mr_r, out_movqi_r_mr): Same.
2766 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
2767 io_address_operand on "optimize > 0".
2769 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2771 * tree-loop-distribution.c: Add general explanantion on the pass.
2772 (generate_loops_for_partition): Mark distributed loop.
2773 (pg_add_dependence_edges): New parameter. Handle alias data
2774 dependence specially and record it in the parameter if asked.
2775 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
2776 (init_partition_graph_vertices, add_partition_graph_edge): New.
2777 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
2778 (free_partition_graph_vdata, build_partition_graph): New.
2779 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
2780 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
2781 (data_ref_segment_size, latch_dominated_by_data_ref): New.
2782 (compute_alias_check_pairs, version_loop_by_alias_check): New.
2783 (version_for_distribution_p, finalize_partitions): New.
2784 (distribute_loop): Handle alias data dependence specially. Factor
2785 out loop fusion code as functions and call these functions.
2787 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2789 * tree-loop-distribution.c (classify_partition): New parameter and
2790 better handle reduction statement.
2791 (rdg_build_partitions): Revise comment.
2792 (distribute_loop): Compute statements in all partitions and pass it
2793 to classify_partition.
2795 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2797 * tree-loop-distribution.c (enum partition_type): New.
2798 (struct partition): New field type.
2799 (partition_merge_into): Add parameter. Update partition type.
2800 (data_dep_in_cycle_p, update_type_for_merge): New functions.
2801 (build_rdg_partition_for_vertex): Compute partition type.
2802 (rdg_build_partitions): Dump partition type.
2803 (distribute_loop): Update calls to partition_merge_into.
2805 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2807 * tree-loop-distribution.c (struct ddr_hasher): New.
2808 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
2810 (classify_partition): Call get_data_dependence.
2811 (pg_add_dependence_edges): Ditto.
2812 (distribute_loop): Release data dependence hash table.
2814 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2816 * tree-loop-distribution.c (ref_base_address): Delete.
2817 (similar_memory_accesses): Rename ...
2818 (share_memory_accesses): ... to this. Check if partitions access
2819 the same memory reference.
2820 (distribute_loop): Call share_memory_accesses.
2822 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2824 * tree-loop-distribution.c (struct partition): New field recording
2826 (partition_alloc, partition_free): Init and release data refs.
2827 (partition_merge_into): Merge data refs.
2828 (build_rdg_partition_for_vertex): Collect data refs for partition.
2829 (pg_add_dependence_edges): Change parameters from vector to bitmap.
2831 (distribute_loop): Remve data refs from vertice data of partition
2834 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2836 * tree-loop-distribution.c (params.h): Include header file.
2837 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
2838 (datarefs_vec): New global var.
2839 (create_rdg_vertices): Use datarefs_vec directly.
2840 (free_rdg): Don't free data references.
2841 (build_rdg): Update use. Don't free data references.
2842 (distribute_loop): Compute global variable for data references.
2843 Bail out if there are too many data references.
2845 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2847 * tree-loop-distribution.c (loop_nest): New global var.
2848 (build_rdg): Use loop directly, rather than loop nest.
2849 (pg_add_dependence_edges): Remove loop nest parameter. Use global
2851 (distribute_loop): Compute global variable loop nest. Update use.
2853 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2855 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
2856 (partition_merge_into): New parameter. Dump reason for fusion.
2857 (distribute_loop): Update use of partition_merge_into.
2859 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2861 * tree-loop-distribution.c (bb_top_order_index): New.
2862 (bb_top_order_index_size, bb_top_order_cmp): New.
2863 (stmts_from_loop): Use topological order.
2864 (pass_loop_distribution::execute): Compute and release topological
2865 order for basic blocks.
2867 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2869 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
2872 2017-07-05 Bin Cheng <bin.cheng@arm.com>
2874 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
2875 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
2876 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
2877 * internal-fn.def (LOOP_DIST_ALIAS): New.
2878 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
2879 (fold_loop_internal_call): ... this.
2880 (vect_loop_dist_alias_call): New function.
2881 (set_uid_loop_bbs): Call fold_loop_internal_call.
2882 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
2885 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
2888 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
2889 Require dead FLAGS_REG at the beginning of a peephole.
2891 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
2894 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
2895 arguments in the call to __builtin_ia32_sbb_u32.
2896 (_subborrow_u64): Swap _X and _Y arguments in the call to
2897 __builtin_ia32_sbb_u64.
2899 2017-07-04 Jakub Jelinek <jakub@redhat.com>
2902 * tree-vrp.c (compare_assert_loc): Turn into a function template
2903 with stable template parameter. Only test if a->e is NULL,
2904 !a->e == !b->e has been verified already. Use e == NULL or
2905 e != NULL instead of e or ! e tests. If stable is true, don't use
2906 iterative_hash_expr, on the other side allow a or b or both NULL
2907 and sort the NULLs last.
2908 (process_assert_insertions): Sort using compare_assert_loc<false>
2909 instead of compare_assert_loc, later sort using
2910 compare_assert_loc<true> before calling process_assert_insertions_for
2911 in a loop. Use break instead of continue once seen NULL pointer.
2913 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
2915 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
2916 Cortex-R7 and Cortex-R8 processors.
2918 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2920 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
2921 uninitialized while src is not.
2923 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
2925 * common/config/arm/arm-common.c: Adjust include path for
2927 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
2928 (arm-cpu.h): Create in build directory. Adjust dependency rules.
2929 (arm-cpu-data.h): Likewise.
2930 (arm-cpu-cdata.h): Likewise.
2931 * config/arm/arm-cpu.h: Delete.
2932 * config/arm/arm-cpu-cdata.h: Delete.
2933 * config/arm/arm-cpu-data.h: Delete.
2935 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
2937 * config/arm/arm-cpus.in (cortex-a55): New.
2938 (cortex-a75): Likewise.
2939 (cortex-a75.cortex-a55): Likewise.
2940 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
2942 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
2943 * config/arm/arm-cpu-cdata.h: Regenerate.
2944 * config/arm/arm-cpu-data.h: Regenerate.
2945 * config/arm/arm-cpu.h: Regenerate.
2946 * config/arm/arm-tables.opt: Regenerate.
2947 * config/arm/arm-tune.md: Regenerate.
2949 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2951 * haifa-sched.c (sched_create_recovery_edges): Update profile.
2953 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
2955 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
2958 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
2960 PR tree-optimization/81292
2961 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
2962 full_string_p, also call adjust_related_strinfos if the adjustment
2963 is simple, otherwise invalidate related strinfos.
2965 2017-07-04 Martin Liska <mliska@suse.cz>
2968 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
2969 newly created variable as DECL_IGNORED_P.
2971 2017-07-04 Martin Liska <mliska@suse.cz>
2974 * ipa-inline.c (inline_small_functions):
2975 Use xstrdup_for_dump.
2977 2017-07-04 Tom de Vries <tom@codesourcery.com>
2979 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
2981 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
2984 * config/darwin.c (darwin_function_switched_text_sections):
2987 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
2989 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
2991 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
2993 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
2995 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
2997 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
2998 min_profitable_iters, and th as inclusive lower bounds.
2999 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
3000 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
3001 for min_profitable_iters and min_profitable_estimate.
3002 (vect_transform_loop): Treat th as an inclusive lower bound.
3003 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
3005 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
3008 * config/darwin.c (darwin_function_switched_text_sections):
3009 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
3010 in two pieces, and suppress the use of buf.
3012 2017-07-03 Nathan Sidwell <nathan@acm.org>
3014 * hash-table.h (hash_table_mod1): Fix indentation.
3016 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3019 * predict.c (force_edge_cold): Be more careful about propagation
3021 * profile-count.h (profile_probability::guessed,
3022 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
3024 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
3026 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
3028 * doc/invoke.texi (rcpc architecture extension): Document it.
3030 2017-07-03 Richard Biener <rguenther@suse.de>
3032 PR tree-optimization/60510
3033 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
3034 the scalar reduction PHI and use it.
3035 (vectorizable_reduction): Properly guard the single_defuse_cycle
3036 path for non-SLP reduction chains where we cannot use it.
3037 Rework reduc_def/index and vector type deduction. Rework
3038 vector operand gathering during reduction op code-gen.
3039 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
3040 chains dissolve the chain and leave it to non-SLP reduction
3043 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3045 * tree-data-ref.h (dr_alignment): Declare.
3046 * tree-data-ref.c (dr_alignment): New function.
3047 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
3048 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
3050 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
3052 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3054 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
3055 and base_misalignment fields.
3056 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
3057 * tree-data-ref.c: Include builtins.h.
3058 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
3059 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
3060 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
3061 * tree-vect-data-refs.c: Include tree-cfg.h.
3062 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
3063 fields instead of calculating an alignment here.
3064 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
3065 innermost_loop_behavior fields.
3067 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3069 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
3071 (DR_STEP_ALIGNMENT): New macro.
3072 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
3073 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
3074 (create_data_ref): Print it.
3075 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
3076 to tell whether the step preserves vector (mis)alignment.
3077 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
3078 Move the check for an integer step and generalise to all INTEGER_CST.
3079 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
3080 Print the outer step alignment.
3082 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3084 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
3085 with offset_alignment.
3086 (DR_ALIGNED_TO): Delete.
3087 (DR_OFFSET_ALIGNMENT): New macro.
3088 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
3089 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
3090 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
3091 (create_data_ref): Likewise.
3092 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
3093 (vect_analyze_data_refs): Likewise.
3094 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
3095 creating dummy innermost behavior.
3097 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3099 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
3100 with a "innermost_loop_behavior *" and refeence tree.
3101 * tree-data-ref.c (dr_analyze_innermost): Likewise.
3102 (create_data_ref): Update call accordingly.
3103 * tree-predcom.c (find_looparound_phi): Likewise.
3105 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3107 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
3108 fields with dr_wrt_vec_loop.
3109 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
3110 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
3111 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
3112 (vect_dr_behavior): New function.
3113 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
3114 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
3115 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
3116 track whether the step preserves the misalignment.
3117 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
3118 Use vect_dr_behavior.
3119 (vect_setup_realignment): Update call accordingly.
3120 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
3121 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
3122 call to vect_create_addr_base_for_vector_ref.
3123 (vect_create_cond_for_align_checks): Likewise.
3124 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
3125 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
3126 (vect_recog_mask_conversion_pattern): Likewise.
3127 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
3128 (new_stmt_vec_info): Remove redundant zeroing.
3130 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
3132 * common/config/arm/arm-common.c (arm_be8_option): New function.
3133 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
3134 (ISA_ARMv6): Add isa_bit_be8.
3135 * config/arm/arm.h (arm_be8_option): Add prototype.
3136 (BE8_SPEC_FUNCTION): New define.
3137 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
3138 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
3139 (mlittle-endian): Similarly.
3140 (mbe8, mbe32): New options.
3141 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
3142 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
3144 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3146 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
3148 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3150 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
3151 (cleanup_tree_cfg_bb): Use it.
3152 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
3154 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
3156 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3159 * loop-doloop.c (add_test): Update profile.
3161 2017-07-03 Martin Liska <mliska@suse.cz>
3164 * sanopt.c (rewrite_usage_of_param): New function.
3165 (sanitize_rewrite_addressable_params): Likewise.
3166 (pass_sanopt::execute): Call rewrite_usage_of_param.
3168 2017-07-03 Richard Biener <rguenther@suse.de>
3170 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
3171 back to using VIEW_CONVERT_EXPR.
3173 2017-07-03 Martin Liska <mliska@suse.cz>
3176 * doc/extend.texi: Document when a resolver function is
3177 generated for target_clones.
3179 2017-07-03 Martin Liska <mliska@suse.cz>
3181 * asan.c (asan_emit_stack_protection): Unpoison just red zones
3182 and shadow memory of auto variables which are subject of
3183 use-after-scope sanitization.
3184 (asan_expand_mark_ifn): Add do set only when is_poison.
3186 2016-07-03 Richard Biener <rguenther@suse.de>
3188 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
3190 (vect_force_simple_reduction): Record reduction def -> phi mapping.
3191 (vectorizable_reduction): Perform reduction PHI creation when
3192 visiting a reduction PHI and adjust and simplify code generation
3193 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
3194 (vect_transform_loop): Visit reduction PHIs.
3195 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
3196 defs into the SLP tree.
3197 (vect_build_slp_tree): Reduction defs terminate the recursion.
3198 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
3200 (vect_get_vec_defs_for_stmt_copy): Export.
3201 (vect_get_vec_defs): Likewise.
3202 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
3204 (vect_get_vec_defs_for_stmt_copy): Declare.
3205 (vect_get_vec_defs): Likewise.
3207 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3209 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
3210 parameter with a "loop" parameter and use it instead of the
3211 loop containing DR_STMT. Don't check simple_iv when doing
3212 BB analysis. Describe the two analysis modes in the comment.
3214 2017-07-03 Tom de Vries <tom@codesourcery.com>
3216 PR tree-optimization/69468
3217 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
3218 (find_same_succ_bb): Handle ignore_edge_flags.
3220 2017-07-03 Tom de Vries <tom@codesourcery.com>
3222 PR tree-optimization/81192
3223 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
3225 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
3227 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
3229 2017-07-03 Tom de Vries <tom@codesourcery.com>
3231 PR tree-optimization/81192
3232 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
3233 BB_SAME_SUCC (bb) == NULL.
3235 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3237 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
3240 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3242 * dumpfile.c: Include profile-count.h
3243 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
3245 (insert_cond_bb): Update profile.
3246 * tree-cfg.h (insert_cond_bb): Update prototype.
3247 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
3248 * tree-dump.c: Do not include tree-cfg.
3250 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3252 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
3254 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3256 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
3259 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3261 * tree-complex.c (expand_complex_div_wide): update profile.
3263 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3264 Alan Hayward <alan.hayward@arm.com>
3265 David Sherwood <david.sherwood@arm.com>
3267 * Makefile.in (MACHMODE_H): Remove insn-modes.h
3268 (CORETYPES_H): New define.
3269 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
3270 (insn-modes-inline.h, s-modes-inline-h): New rules.
3271 (generated_files): Add insn-modes-inline.h.
3272 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
3273 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
3274 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
3275 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
3276 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
3277 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
3278 (build/gencodes.o, build/genconditions.o): Likewise.
3279 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
3280 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
3281 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
3282 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
3283 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
3284 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
3285 * coretypes.h: Include everything up to real.h for generators.
3286 Include insn-modes.h first. Include wide-int-print.h after
3287 wide-int.h. Include insn-modes-inline.h and then machmode.h.
3288 * machmode.h: Don't include insn-modes.h here.
3289 * function-tests.c: Remove includes of signop.h, machmode.h,
3290 double-int.h and wide-int.h.
3292 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
3294 * optc-save-gen.awk: Likewise.
3295 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
3296 * godump.c: Remove include of wide-int-print.h.
3297 * pretty-print.h: Likewise.
3298 * wide-int-print.cc: Likewise.
3299 * wide-int.cc: Likewise.
3300 * hash-map-tests.c: Remove include of signop.h.
3301 * hash-set-tests.c: Likewise.
3302 * rtl-tests.c: Likewise.
3303 * mkconfig.sh: Remove include of machmode.h.
3304 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
3306 (emit_insn_modes_inline_h): ...this new function. Emit the code
3307 into an insn-modes-inline.h header file, adding appropriate
3308 include guards and end comments.
3309 (emit_insn_modes_c_header): Remove include of machmode.h.
3310 (emit_min_insn_modes_c_header): Include coretypes.h rather than
3312 (main): Handle -i flag and call emit_insn_modes_inline_h when
3315 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3317 * tree-ssa-strlen.c (strinfo): Rename the length field to
3318 nonzero_chars. Add a full_string_p field.
3319 (compare_nonzero_chars, zero_length_string_p): New functions.
3320 (get_addr_stridx): Add an offset_out parameter.
3321 Use compare_nonzero_chars.
3322 (get_stridx): Update accordingly. Use compare_nonzero_chars.
3323 (new_strinfo): Update after above changes to strinfo.
3324 (set_endptr_and_length): Set full_string_p.
3325 (get_string_length): Update after above changes to strinfo.
3326 (unshare_strinfo): Update call to new_strinfo.
3327 (maybe_invalidate): Likewise.
3328 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
3329 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
3330 as a uhwi instead of an shwi. Update after above changes to
3331 strinfo and new_strinfo.
3332 (zero_length_string): Assert that chainsi contains full strings.
3333 Use zero_length_string_p. Update call to new_strinfo.
3334 (adjust_related_strinfos): Update after above changes to strinfo.
3335 Copy full_string_p from origsi.
3336 (adjust_last_stmt): Use zero_length_string_p.
3337 (handle_builtin_strlen): Update after above changes to strinfo and
3338 new_strinfo. Install the lhs as the string length if the previous
3339 entry didn't describe a full string.
3340 (handle_builtin_strchr): Update after above changes to strinfo
3342 (handle_builtin_strcpy): Likewise.
3343 (handle_builtin_strcat): Likewise.
3344 (handle_builtin_malloc): Likewise.
3345 (handle_pointer_plus): Likewise.
3346 (handle_builtin_memcpy): Likewise. Track nonzero characters
3347 that aren't necessarily followed by a nul terminator.
3348 (handle_char_store): Likewise.
3350 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3352 PR tree-optimization/80769
3353 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
3354 for malloc and calloc. Document the new invariant that all related
3355 strinfos have delayed lengths or none do.
3356 (verify_related_strinfos): Move earlier in file.
3357 (set_endptr_and_length): New function, split out from...
3358 (get_string_length): ...here. Also set the lengths of related
3360 (zero_length_string): Assert that chainsi has known (rather than
3362 (adjust_related_strinfos): Likewise.
3364 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3366 PR tree-optimization/81136
3367 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
3368 assert that two references with the same misalignment have the same
3369 compile-time misalignment if those compile-time misalignments
3372 2017-07-01 Andi Kleen <ak@linux.intel.com>
3374 * print-tree.c (print_node): Print all attributes.
3376 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3378 * cfg.c (scale_bbs_frequencies): New function.
3379 * cfg.h (scale_bbs_frequencies): Declare it.
3380 * cfgloopanal.c (single_likely_exit): Cleanup.
3381 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
3383 (scale_loop_profile): Likewise.
3384 (loop_version): Likewise.
3385 (create_empty_loop_on_edge): Update.
3386 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
3387 scale_loop_frequencies, scale_loop_profile, loopify,
3388 loop_version): Update prototypes.
3389 * modulo-sched.c (sms_schedule): Update.
3390 * predict.c (unlikely_executed_edge_p): Also check probability.
3391 (probably_never_executed_edge_p): Fix typo.
3392 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3393 * tree-parloops.c (gen_parallel_loop): Update.
3394 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
3395 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3396 * tree-ssa-loop-split.c (split_loop): Update.
3397 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3398 * tree-vect-loop-manip.c (vect_do_peeling): Update.
3399 (vect_loop_versioning): Update.
3400 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3402 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3404 * trans-mem.c (split_bb_make_tm_edge): Update profile.
3406 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3408 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
3409 to keep profile consistent.
3411 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3413 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
3414 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
3415 * profile-count.h (max_safe_multiplier): Make unsigned.
3416 (profile_count::guessed_zero): New.
3418 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3420 * bb-reorder.c (fix_up_crossing_landing_pad,
3421 fix_crossing_conditional_branches): Use make_single_succ_edge
3422 to keep profile consistent.
3424 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3426 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
3429 2017-07-01 Jakub Jelinek <jakub@redhat.com>
3432 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
3433 the right scopes, make sure cond_jump isn't preserved between multiple
3434 iterations. Search for fallthru edge whenever there are 3+ edges and
3435 use find_fallthru_edge for it.
3437 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3439 Patch by Alexander Monakov <amonakov@ispras.ru>
3440 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
3441 probabilities consistently.
3443 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3445 * pa.c (pa_expand_compare_and_swap_loop): Update call of
3446 emit_cmp_and_jump_insns.
3448 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3451 * tree-inline.c (expand_call_inline): Combine profile statuses.
3453 2017-06-30 Andrew Pinski <apinski@cavium.com>
3455 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
3456 fold_stmt returned true.
3458 2017-06-30 Nathan Sidwell <nathan@acm.org>
3460 * ggc.h (empty_string): Delete.
3461 * cfgexpand.c (expand_asm_stmt): Use plain "".
3462 * optabs.c (expand_asm_memory_barrier): Likewise.
3463 * stringpool.c (empty_string): Delete.
3464 (digit_vector, digit_string): Delete.
3465 (ggc_alloc_string): Use plain "", don't optimize single digit
3466 strings. Use ggc_alloc_atomic.
3468 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
3470 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
3471 comparison set and one other set, use the cost of the non-comparison
3474 2017-06-30 Nathan Sidwell <nathan@acm.org>
3476 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
3479 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
3481 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
3482 loops. Remove now unneeded calls to gimple_switch_set_label() that
3483 just set removed labels to NULL_TREE.
3485 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
3487 * tree-ssanames.c (set_range_info_raw): Abstract from ...
3488 (set_range_info): ...here. Only call set_range_info_raw if domain
3490 (set_nonzero_bits): Call set_range_info_raw.
3491 * tree-ssanames.h (set_range_info_raw): New.
3493 2017-06-30 Jakub Jelinek <jakub@redhat.com>
3496 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
3497 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
3498 of nonimmediate_operand and <store_mask_constraint> instead of m
3499 for the input operand. For V8FI iterator, always split if input
3500 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
3501 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
3502 <store_mask_predicate> instead of register_operand and
3503 <store_mask_constraint> instead of v for the input operand. Make
3504 sure both operands aren't MEMs for if not <mask_applied>.
3506 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
3508 * lto-wrapper.c (copy_file) Close both file descriptors before
3511 2017-06-30 Martin Liska <mliska@suse.cz>
3514 * multiple_target.c (create_dispatcher_calls): Make ifunc
3515 also for function that don't have calls or are not referenced.
3517 2017-06-30 Richard Biener <rguenther@suse.de>
3519 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
3520 analyze the first scalar stmt. Move vector type computation
3521 for the BB case here from ...
3522 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
3523 live operation processing in the SLP case properly.
3525 2017-06-30 Richard Biener <rguenther@suse.de>
3527 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
3529 2017-06-30 Martin Liska <mliska@suse.cz>
3532 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
3533 before BUILT_IN_UNWIND_RESUME when ASAN is used.
3535 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
3537 * doc/invoke.texi (AArch64): Add missing options and remove redundant
3540 2017-06-30 Richard Biener <rguenther@suse.de>
3542 PR tree-optimization/81249
3543 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
3544 condition reduction result to original scalar type.
3546 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3548 * profile-count.h (enum profile_quality): Fix typos and whitespace
3551 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3553 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
3554 type for branch probabilities.
3556 2017-06-29 Julian Brown <julian@codesourcery.com>
3557 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3559 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
3560 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
3561 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
3562 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
3564 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3566 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
3567 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
3568 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
3569 CC usage from generic code to here.
3570 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
3571 CC usage into the target macros.
3573 2017-06-29 Maya Rashish <coypu@sdf.org>
3575 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
3578 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3580 * arm/arm-builtins.c: Include profile-count.h
3581 * except.c (sjlj_emit_function_enter): Use
3582 profile_probability::unlikely.
3584 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3586 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
3587 and tocrel_offset be pointer args rather than implicitly using
3589 (legitimate_constant_pool_address_p, rs6000_emit_move,
3590 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
3591 tocrel_offset and use in toc_relative_expr_p call.
3592 (print_operand, print_operand_address): Use static tocrel_base_oac
3593 and tocrel_offset_oac.
3594 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
3597 2017-06-29 Maya Rashish <coypu@sdf.org>
3599 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
3601 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
3603 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
3604 objects, take into account only the alignment of 'op0' and 'mode1' if
3607 2017-06-29 Steve Ellcey <sellcey@cavium.com>
3609 * ccmp.c (ccmp_tree_comparison_p): New function.
3610 (ccmp_candidate_p): Update to use above function.
3611 (get_compare_parts): New function.
3612 (expand_ccmp_next): Update to use new functions.
3613 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
3615 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
3616 take mode as argument.
3617 * ccmp.h (expand_ccmp_expr): Add mode as argument.
3618 * expr.c (expand_expr_real_1): Pass mode as argument.
3620 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
3622 * combine.c (combine_instructions): Print insns to dump_file, together
3625 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3627 * asan.c (asan_emit_stack_protection): Update.
3628 (create_cond_insert_point): Update.
3629 * auto-profile.c (afdo_propagate_circuit): Update.
3630 * basic-block.h (struct edge_def): Turn probability to
3631 profile_probability.
3632 (EDGE_FREQUENCY): Update.
3633 * bb-reorder.c (find_traces_1_round): Update.
3634 (better_edge_p): Update.
3635 (sanitize_hot_paths): Update.
3636 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
3637 (make_single_succ_edge): Update.
3638 (check_bb_profile): Update.
3639 (dump_edge_info): Update.
3640 (update_bb_profile_for_threading): Update.
3641 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
3642 probabilitycount to 0.
3643 * cfgbuild.c (compute_outgoing_frequencies): Update.
3644 * cfgcleanup.c (try_forward_edges): Update.
3645 (outgoing_edges_match): Update.
3646 (try_crossjump_to_edge): Update.
3647 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
3648 (expand_gimple_tailcall): Update.
3649 (construct_init_block): Use make_single_succ_edge.
3650 (construct_exit_block): Use make_single_succ_edge.
3651 * cfghooks.c (verify_flow_info): Update.
3652 (redirect_edge_succ_nodup): Update.
3653 (split_edge): Update.
3654 (account_profile_record): Update.
3655 * cfgloopanal.c (single_likely_exit): Update.
3656 * cfgloopmanip.c (scale_loop_profile): Update.
3657 (set_zero_probability): Remove.
3658 (duplicate_loop_to_header_edge): Update.
3659 * cfgloopmanip.h (loop_version): Update prototype.
3660 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3661 (force_nonfallthru_and_redirect): Update.
3662 (update_br_prob_note): Update.
3663 (rtl_verify_edges): Update.
3664 (purge_dead_edges): Update.
3665 (rtl_lv_add_condition_to_bb): Update.
3666 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
3667 * cgraphunit.c (init_lowered_empty_function): Update.
3668 (cgraph_node::expand_thunk): Update.
3669 * cilk-common.c: Include profile-count.h
3670 * dojump.c (inv): Remove.
3671 (jumpifnot): Update.
3672 (jumpifnot_1): Update.
3673 (do_jump_1): Update.
3675 (do_jump_by_parts_greater_rtx): Update.
3676 (do_compare_rtx_and_jump): Update.
3677 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
3678 do_jump_1. do_compare_rtx_and_jump): Update prototype.
3679 * dwarf2cfi.c: Include profile-count.h
3680 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
3681 (sjlj_emit_dispatch_table): Likewise.
3682 * explow.c: Include profile-count.h
3683 * expmed.c (emit_store_flag_force): Update.
3684 (do_cmp_and_jump): Update.
3685 * expr.c (compare_by_pieces_d::generate): Update.
3686 (compare_by_pieces_d::finish_mode): Update.
3687 (emit_block_move_via_loop): Update.
3688 (store_expr_with_bounds): Update.
3689 (store_constructor): Update.
3690 (expand_expr_real_2): Update.
3691 (expand_expr_real_1): Update.
3692 * expr.h (try_casesi, try_tablejump): Update prototypes.
3693 * gimple-pretty-print.c (dump_probability): Update.
3694 (dump_profile): New.
3695 (dump_gimple_label): Update.
3696 (dump_gimple_bb_header): Update.
3697 * graph.c (draw_cfg_node_succ_edges): Update.
3698 * hsa-gen.c (convert_switch_statements): Update.
3699 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
3700 (find_if_case_1): Update.
3701 (find_if_case_2): Update.
3702 * internal-fn.c (expand_arith_overflow_result_store): Update.
3703 (expand_addsub_overflow): Update.
3704 (expand_neg_overflow): Update.
3705 (expand_mul_overflow): Update.
3706 (expand_vector_ubsan_overflow): Update.
3707 * ipa-cp.c (good_cloning_opportunity_p): Update.
3708 * ipa-split.c (split_function): Use make_single_succ_edge.
3709 * ipa-utils.c (ipa_merge_profiles): Update.
3710 * loop-doloop.c (add_test): Update.
3711 (doloop_modify): Update.
3712 * loop-unroll.c (compare_and_jump_seq): Update.
3713 (unroll_loop_runtime_iterations): Update.
3714 * lra-constraints.c (lra_inheritance): Update.
3715 * lto-streamer-in.c (input_cfg): Update.
3716 * lto-streamer-out.c (output_cfg): Update.
3717 * mcf.c (adjust_cfg_counts): Update.
3718 * modulo-sched.c (sms_schedule): Update.
3719 * omp-expand.c (expand_omp_for_init_counts): Update.
3720 (extract_omp_for_update_vars): Update.
3721 (expand_omp_ordered_sink): Update.
3722 (expand_omp_for_ordered_loops): Update.
3723 (expand_omp_for_generic): Update.
3724 (expand_omp_for_static_nochunk): Update.
3725 (expand_omp_for_static_chunk): Update.
3726 (expand_cilk_for): Update.
3727 (expand_omp_simd): Update.
3728 (expand_omp_taskloop_for_outer): Update.
3729 (expand_omp_taskloop_for_inner): Update.
3730 * omp-simd-clone.c (simd_clone_adjust): Update.
3731 * optabs.c (expand_doubleword_shift): Update.
3732 (expand_abs): Update.
3733 (emit_cmp_and_jump_insn_1): Update.
3734 (expand_compare_and_swap_loop): Update.
3735 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
3736 * predict.c (predictable_edge_p): Update.
3737 (edge_probability_reliable_p): Update.
3738 (set_even_probabilities): Update.
3739 (combine_predictions_for_insn): Update.
3740 (combine_predictions_for_bb): Update.
3741 (propagate_freq): Update.
3742 (estimate_bb_frequencies): Update.
3743 (force_edge_cold): Update.
3744 * profile-count.c (profile_count::dump): Add missing space into dump.
3745 (profile_count::debug): Add newline.
3746 (profile_count::differs_from_p): Explicitly convert to unsigned.
3747 (profile_count::stream_in): Update.
3748 (profile_probability::dump): New member function.
3749 (profile_probability::debug): New member function.
3750 (profile_probability::differs_from_p): New member function.
3751 (profile_probability::differs_lot_from_p): New member function.
3752 (profile_probability::stream_in): New member function.
3753 (profile_probability::stream_out): New member function.
3754 * profile-count.h (profile_count_quality): Rename to ...
3755 (profile_quality): ... this one.
3756 (profile_probability): New.
3757 (profile_count): Update.
3758 * profile.c (compute_branch_probabilities): Update.
3759 * recog.c (peep2_attempt): Update.
3760 * sched-ebb.c (schedule_ebbs): Update.
3761 * sched-rgn.c (find_single_block_region): Update.
3762 (compute_dom_prob_ps): Update.
3763 (schedule_region): Update.
3764 * sel-sched-ir.c (compute_succs_info): Update.
3765 * stmt.c (struct case_node): Update.
3766 (do_jump_if_equal): Update.
3767 (get_outgoing_edge_probs): Update.
3768 (conditional_probability): Update.
3769 (emit_case_dispatch_table): Update.
3770 (expand_case): Update.
3771 (expand_sjlj_dispatch_table): Update.
3772 (emit_case_nodes): Update.
3773 * targhooks.c: Update.
3774 * tracer.c (better_p): Update.
3775 (find_best_successor): Update.
3776 * trans-mem.c (expand_transaction): Update.
3777 * tree-call-cdce.c: Update.
3778 * tree-cfg.c (gimple_split_edge): Upate.
3779 (move_sese_region_to_fn): Upate.
3780 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
3781 * tree-eh.c (lower_resx): Upate.
3782 (cleanup_empty_eh_move_lp): Upate.
3783 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3784 * tree-inline.c (copy_edges_for_bb): Update.
3785 (copy_cfg_body): Update.
3786 * tree-parloops.c (gen_parallel_loop): Update.
3787 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
3788 (gimple_gen_time_profiler): Update.
3789 * tree-ssa-dce.c (remove_dead_stmt): Update.
3790 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3791 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3792 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3793 (unloop_loops): Update.
3794 (try_peel_loop): Update.
3795 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3796 * tree-ssa-loop-split.c (connect_loops): Update.
3797 (split_loop): Update.
3798 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3799 (hoist_guard): Update.
3800 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3801 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3802 (value_replacement): Update.
3803 * tree-ssa-reassoc.c (branch_fixup): Update.
3804 * tree-ssa-tail-merge.c (replace_block_by): Update.
3805 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3806 (create_edge_and_update_destination_phis): Update.
3807 (compute_path_counts): Update.
3808 (recompute_probabilities): Update.
3809 (update_joiner_offpath_counts): Update.
3810 (freqs_to_counts_path): Update.
3811 (duplicate_thread_path): Update.
3812 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3813 (struct switch_conv_info): Update.
3814 (gen_inbound_check): Update.
3815 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3816 (vect_do_peeling): Update.
3817 (vect_loop_versioning): Update.
3818 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3819 (optimize_mask_stores): Update.
3820 * ubsan.c (ubsan_expand_null_ifn): Update.
3821 * value-prof.c (gimple_divmod_fixed_value): Update.
3822 (gimple_divmod_fixed_value_transform): Update.
3823 (gimple_mod_pow2): Update.
3824 (gimple_mod_pow2_value_transform): Update.
3825 (gimple_mod_subtract): Update.
3826 (gimple_mod_subtract_transform): Update.
3827 (gimple_ic): Update.
3828 (gimple_stringop_fixed_value): Update.
3829 (gimple_stringops_transform): Update.
3830 * value-prof.h: Update.
3832 2017-06-29 Carl Love <cel@us.ibm.com>
3834 * config/rs6000/rs6000-c.c: Add support for built-in functions
3835 vector signed int vec_signed (vector float);
3836 vector signed long long vec_signed (vector double);
3837 vector signed int vec_signed2 (vector double, vector double);
3838 vector signed int vec_signede (vector double);
3839 vector signed int vec_signedo (vector double);
3840 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
3841 instruction generator.
3842 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3843 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
3844 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
3846 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
3847 vunsignede_v2df): Add define_expands.
3848 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
3849 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
3850 VEC_UNSIGNEDO): Add definitions.
3851 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
3852 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
3853 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
3854 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
3855 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
3856 * config/rs6000/altivec.h (vec_signed, vec_signed2,
3857 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
3858 vec_unsignede, vec_unsignedo): Add builtin defines.
3859 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
3861 * doc/extend.texi: Update the built-in documentation file for the
3862 new built-in functions.
3864 2017-06-29 Richard Biener <rguenther@suse.de>
3866 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
3867 reduction chains to LOOP_VINFO_REDUCTIONS.
3868 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
3869 SLP reductions after processing reduction chains.
3871 2017-06-29 Nathan Sidwell <nathan@acm.org>
3873 * builtins.c (fold_builtin_FUNCTION): Use
3874 lang_hooks.decl_printable_name.
3876 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
3879 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
3880 with only one label.
3881 * stmt.c (expand_case): Assert NCASES is greater than one.
3883 2017-06-29 Richard Biener <rguenther@suse.de>
3885 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
3887 (group_case_labels): Likewise.
3888 (find_taken_edge): Push sanity checking on val to workers...
3889 (find_taken_edge_cond_expr): ... here
3890 (find_taken_edge_switch_expr): ... and here, handle cases
3891 with just a default label.
3892 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
3893 (group_case_labels): Likewise.
3894 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
3895 group_case_labels does anything cleanup the CFG again.
3897 2017-06-29 Bin Cheng <bin.cheng@arm.com>
3899 PR tree-optimization/81196
3900 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
3901 exit condition comparing two IVs.
3903 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
3905 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
3906 profile to the dummy entry at the end of the list of architectures.
3907 * config/arm/arm-cpu-cdata.h: Regenerated.
3909 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3910 Michael Collison <michael.collison@arm.com>
3913 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
3915 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
3916 (*aarch64_reg_<mode>3_minus_mask): New pattern.
3917 (*aarch64_<optab>_reg_di3_mask2): New pattern.
3918 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
3919 of shift when the shift amount is masked with constant equal to
3920 the size of the mode.
3921 * config/aarch64/predicates.md (subreg_lowpart_operator): New
3924 2017-06-29 Martin Liska <mliska@suse.cz>
3926 * config/i386/i386.opt: Change range from [1,5] to [0,5].
3928 2017-06-29 Yury Gribov <tetra2005@gmail.com>
3931 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
3934 (edge_growth_cache_entry::edge_growth_cache_entry): New
3936 (reset_edge_growth_cache): Update to use constructor.
3938 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3940 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
3941 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
3942 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
3944 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
3946 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
3947 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
3949 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
3951 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
3952 (*-linux-uclibc*): Add t-uclibc tmake_file.
3953 * config/t-musl: New.
3954 * config/t-uclibc: New.
3956 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
3958 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
3960 (gen_comm_data): Emit architectural setting of arch_prof.
3961 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
3963 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
3964 (armv8-m.base, armv8-m.main): Likewise.
3965 * arm-protos.h (arm_build_target): Add profile field.
3966 (arch_option): Likewise.
3967 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
3969 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
3970 arm_active_target.profile.
3972 2017-06-28 Richard Biener <rguenther@suse.de>
3975 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
3976 TYPE_OVERFLOW_WRAPS.
3977 * match.pd (negate_expr_p): Likewise.
3978 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
3979 fold_build2, not fold_binary.
3981 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
3983 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
3984 Convert memory address to Pmode.
3985 (aarch64_print_operand): Assert MEM operands are always Pmode.
3987 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
3990 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
3991 Remove redundant if.
3992 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
3993 * config/arm/aarch-common-protos.h
3994 (aarch_forward_to_shift_is_not_shifted_re): Remove.
3995 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
3997 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
4000 * multiple_target.c (create_dispatcher_calls): Set the default
4001 clone to be static, not public.
4003 2017-06-28 Richard Biener <rguenther@suse.de>
4005 * tree-vect-loop.c (vectorizable_reduction): Move special
4006 cond reduction IV var creation ...
4007 (vect_create_epilog_for_reduction): ... here. Remove induction_index
4008 parameter. Use STMT_VINFO_VECTYPE.
4009 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
4012 2017-06-28 Martin Liska <mliska@suse.cz>
4015 * ipa-visibility.c (non_local_p): Handle visibility.
4017 2017-06-28 Martin Liska <mliska@suse.cz>
4020 * common.opt: Add IntegerRange to various options.
4021 * opt-functions.awk (integer_range_info): New function.
4022 * optc-gen.awk: Add integer_range_info to cl_options struct.
4023 * opts-common.c (decode_cmdline_option): Handle
4024 CL_ERR_INT_RANGE_ARG.
4025 (cmdline_handle_error): Likewise.
4026 * opts.c (print_filtered_help): Show valid interval in
4027 when --help is provided.
4028 * opts.h (struct cl_option): Add range_min and range_max fields.
4029 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
4031 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
4033 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
4034 (x * C EQ/NE y * C): New transformation.
4036 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
4038 * genmultilib (combination_space): Accept '+' in option names.
4040 2017-06-28 Martin Liska <mliska@suse.cz>
4043 * asan.c (instrument_derefs): Bail out inner references
4044 that are hard register variables.
4046 2017-06-28 Jakub Jelinek <jakub@redhat.com>
4049 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
4050 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
4052 2017-06-28 Richard Biener <rguenther@suse.de>
4054 * tree-vectorizer.h (vect_get_vec_defs): Remove.
4055 (vect_get_slp_defs): Adjust.
4056 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
4058 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
4060 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4061 get_initial_defs_for_reduction instead of vect_get_vec_defs.
4062 (vectorizable_reduction): Adjust.
4063 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
4065 (vect_get_slp_defs): Likewise.
4066 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
4067 (vectorizable_bswap): Adjust.
4068 (vectorizable_call): Likewise.
4069 (vectorizable_conversion): Likewise.
4070 (vectorizable_assignment): Likewise.
4071 (vectorizable_shift): Likewise.
4072 (vectorizable_operation): Likewise.
4073 (vectorizable_store): Likewise.
4074 (vectorizable_condition): Likewise.
4075 (vectorizable_comparison): Likewise.
4077 2017-06-28 Michael Collison <michael.collison@arm.com>
4080 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
4082 (arm_gen_movmemqi): Removed unused variable 'i'.
4083 Convert 'for' loop into 'while' loop.
4084 (arm_expand_prologue): Remove last unnecessary set of insn.
4085 (thumb_pop): Remove unused variable 'pushed_words'.
4086 (thumb_exit): Remove last unnecessary set of regs_to_pop.
4088 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4090 * config/s390/predicates.md: Use s390_rel_address_ok_p.
4091 * config/s390/s390-protos.h: Add prototype of
4092 s390_rel_address_ok_p.
4093 * config/s390/s390.c (s390_got_symbol): New function.
4094 (s390_rel_address_ok_p): New function.
4095 (legitimize_pic_address): Use s390_rel_address_ok_p.
4096 (s390_load_got): Use s390_got_symbol.
4097 (s390_option_override): Issue error if
4098 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
4099 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
4101 * config/s390/s390.opt: New option mpic-data-is-text-relative.
4103 2017-06-27 Andrew Pinski <apinski@cavium.com>
4105 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
4106 (X * copysign (1.0, X)): New pattern.
4107 (X * copysign (1.0, -X)): New pattern.
4108 (copysign (-1.0, CST)): New pattern.
4110 2017-06-27 Joseph Myers <joseph@codesourcery.com>
4112 * genmultilib (combination_space): Remove variable.
4113 Validate reuse rules against regular expression for any sequence
4114 of multilib options in any order.
4116 2017-06-27 Michael Collison <michael.collison@arm.com>
4118 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
4119 call aarch64_split_simd_combine.
4120 * (aarch64_combine_internal<mode>): Delete pattern.
4121 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
4122 Allow register and subreg operands.
4124 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4126 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
4127 specific need, just fallback on defaults.
4128 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
4130 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4131 Olivier Hainque <hainque@adacore.com>
4133 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
4135 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
4136 targets. Pick a default if no particular attempt applied.
4137 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
4140 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4142 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
4143 (x86_64-wrs-vxworks7): Likewise.
4145 2017-06-27 Marek Polacek <polacek@redhat.com>
4148 * ubsan.c (instrument_null): Check get_base_address's result for null.
4150 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
4152 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
4154 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
4156 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
4157 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
4158 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
4159 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
4161 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
4162 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
4163 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
4164 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
4165 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
4166 BUILT_IN_FEUPDATEENV): New builtins.
4167 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
4168 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
4169 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
4170 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
4172 (builtin_structptr_types): Adjust size.
4173 * tree.c (builtin_structptr_types): Add four entries.
4175 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4176 Olivier Hainque <hainque@adacore.com>
4178 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
4179 (TLS_SYM): New local macro, forcing reference to __tls__ on
4180 link command lines for VxWorks 7 RTPs, triggering initialization
4182 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
4183 OS features TLS support, true for RTPs on VxWorks 7.
4184 * config/vxworks.c (vxworks_override_options): Setup emutls
4187 2017-06-27 Jakub Jelinek <jakub@redhat.com>
4189 * predict.c (test_prediction_value_range): Use -1U instead of -1
4190 to avoid narrowing conversion warning.
4191 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
4192 to avoid narrowing conversion warning.
4193 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
4195 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
4197 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4199 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
4200 64bit configurations.
4201 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
4202 (SIZE_TYPE): Likewise.
4203 * config/vxworks.c (vxworks_emutls_var_fields): Use
4204 long_unsigned_type_node instead of unsigned_type_node as the offset
4205 field type, which is "pointer" mode in emutls.c.
4207 2017-06-27 Jakub Jelinek <jakub@redhat.com>
4210 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
4213 * gimple-fold.c (replace_call_with_call_and_fold): Handle
4214 gimple_vuse copying separately from gimple_vdef copying.
4216 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4218 * value-prof.c (free_hist): Remove call to memset and the enclosing if
4221 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
4222 Olivier Hainque <hainque@adacore.com>
4224 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
4225 for all vxworks7 targets.
4226 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
4227 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
4228 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
4229 variations for VX6/VX7 and 32/64bits later on in ...
4230 (VXWORKS_LIB_SPEC): Leverage new macros.
4231 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
4232 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
4234 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
4236 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
4237 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
4239 2017-06-26 Carl Love <cel@us.ibm.com>
4241 * config/rs6000/rs6000-c.c: Add support for built-in functions
4242 vector bool char vec_reve (vector bool char);
4243 vector signed char vec_reve (vector signed char);
4244 vector unsigned char vec_reve (vector unsigned char);
4245 vector bool int vec_reve (vector bool int);
4246 vector signed int vec_reve (vector signed int);
4247 vector unsigned int vec_reve (vector unsigned int);
4248 vector bool long long vec_reve (vector bool long long);
4249 vector signed long long vec_reve (vector signed long long);
4250 vector unsigned long long vec_reve (vector unsigned long long);
4251 vector bool short vec_reve (vector bool short);
4252 vector signed short vec_reve (vector signed short);
4253 vector double vec_reve (vector double);
4254 vector float vec_reve (vector float);
4255 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
4256 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
4257 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
4258 (altivec_vreve): New pattern.
4259 * config/rs6000/altivec.h (vec_reve): New define.
4260 * doc/extend.texi (vec_rev): Update the built-in documentation file
4261 for the new built-in functions.
4263 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4265 PR tree-optimization/71815
4266 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
4268 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
4270 (slsr_process_phi): Likewise.
4271 (replace_uncond_cands_and_profitable_phis): Don't replace a
4272 multiply candidate with a stride of 1 (copy or cast).
4273 (phi_incr_cost): Call uses_consumed_by_stmt rather than
4275 (lowest_cost_path): Likewise.
4276 (total_savings): Likewise.
4278 2017-06-26 Richard Biener <rguenther@suse.de>
4281 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
4282 Use def_builtin_pure for all gather builtins.
4284 2017-06-26 Richard Biener <rguenther@suse.de>
4286 PR tree-optimization/81203
4287 * tree-tailcall.c (find_tail_calls): Do not move stmts into
4290 2017-06-26 Marek Polacek <polacek@redhat.com>
4293 * doc/invoke.texi: Document -Wmultistatement-macros.
4295 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
4297 * doc/sourcebuild.texi (ARM-specific attributes): Document new
4298 arm_neon_ok_no_float_abi effective target.
4300 2017-06-26 Richard Biener <rguenther@suse.de>
4302 PR tree-optimization/80928
4303 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
4304 (copy_bbs): Set BB_DUPLICATED flag early.
4305 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
4307 (execute_on_shrinking_pred): Likewise.
4308 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
4309 BB_DUPLICATED blocks.
4310 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
4311 iterate over all PHIs considering removal of *gsi.
4313 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
4315 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
4318 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4320 * config/rs6000/rs6000-string.c: (expand_block_clear,
4321 do_load_for_compare, select_block_compare_mode,
4322 compute_current_alignment, expand_block_compare,
4323 expand_strncmp_align_check, expand_strn_compare,
4324 expand_block_move, rs6000_output_load_multiple)
4325 Move functions related to string/block move/compare
4327 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
4328 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
4329 for this function which is now used in two files.
4330 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
4331 * config.gcc: Add rs6000-string.o to extra_objs for
4332 targets powerpc*-*-* and rs6000*-*-*.
4334 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4337 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
4338 32-bit, since indexed is not valid for DImode.
4339 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
4340 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
4341 (define_peephole2 for Altivec d-form load): Add 32-bit support.
4342 (define_peephole2 for Altivec d-form store): Likewise.
4345 * multiple_target.c (create_dispatcher_calls): Only create the
4346 dispatcher call if the function is the default clone of a
4349 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
4352 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
4353 a call, force the call to not be a tail call.
4355 2017-06-23 Jeff Law <law@redhat.com>
4357 * doc/contrib.texi: Add entry for Steven Pemberton's work on
4360 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
4362 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
4363 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
4364 handling for early expansion of vector shifts (sl,sr,sra,rl).
4365 (builtin_function_type): Add vector shift right instructions
4366 to the unsigned argument list.
4368 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
4370 rtl-optimizatoin/79286
4371 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
4372 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
4373 trap. PIC register plus a const unspec without offset can never trap.
4375 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
4377 * tree.h (builtin_structptr_type): New type.
4378 (builtin_structptr_types): Declare new array.
4379 * tree.c (builtin_structptr_types): New array.
4380 (free_lang_data, build_common_tree_nodes): Use it.
4382 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
4385 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
4388 2017-06-22 Matt Turner <mattst88@gmail.com>
4390 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
4391 Lake models to skylake case. Assume skylake for unknown
4392 models with clflushopt.
4394 2017-06-22 Jeff Law <law@redhat.com>
4396 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
4397 frame sizes that do not satisfy aarch64_uimm12_shift.
4399 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
4401 * profile-count.h (apply_probability,
4402 apply_scale, probability_in): Fix checks for zero.
4404 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4406 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
4407 * doc/cppdiropts.texi (-I @var{dir}): Document it.
4409 2016-06-22 Richard Biener <rguenther@suse.de>
4411 * tree-vect-loop.c (vect_model_reduction_cost): Handle
4412 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
4413 REDUC_MAX_EXPR support.
4414 (vectorizable_reduction): Likewise.
4415 (vect_create_epilog_for_reduction): Likewise.
4417 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
4419 * match.pd (A / (1 << B) -> A >> B): New.
4420 * generic-match-head.c: Include optabs-tree.h.
4421 * gimple-match-head.c: Likewise.
4422 * optabs-tree.h (target_supports_op_p): New.
4423 * optabs-tree.c (target_supports_op_p): New.
4425 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4427 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
4428 $gcc_cv_ld --help output.
4429 (gcc_cv_ld_demangle): Likewise.
4430 (gcc_cv_ld_eh_frame_hdr): Likewise.
4431 (gcc_cv_ld_pie): Likewise.
4432 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
4433 (gcc_cv_ld_buildid): Likewise.
4434 (gcc_cv_ld_sysroot): Likewise.
4435 (ld_bndplt_support): Likewise.
4436 (ld_pushpopstate_support): Likewise.
4437 * configure: Regenerate.
4438 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
4440 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4443 * config/i386/sse.md (round<mode>2): Renumber match_dup and
4444 operands indexes to avoid gap between operands and match_dups.
4446 2017-06-21 Andrew Pinski <apinski@cavium.com>
4448 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
4449 Increment Arith_shift and Arith_shift_reg by 1.
4450 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
4452 * config/aarch64/aarch64.c (thunderx_tunings): Enable
4453 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
4454 (aarch64_strip_extend): Add new argument and test for it.
4455 (aarch64_cheap_mult_shift_p): New function.
4456 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
4457 add a cost if it is true.
4458 Update calls to aarch64_strip_extend.
4459 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
4461 2017-06-21 Andrew Pinski <apinski@cavium.com>
4463 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
4465 (thunderxt88): Likewise.
4466 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
4467 (thunderx_prefetch_tune): New variable.
4468 (thunderx2t99_prefetch_tune): Update for the correct values.
4469 (thunderxt88_tunings): New variable.
4470 (thunderx_tunings): Use thunderx_prefetch_tune instead of
4471 generic_prefetch_tune.
4472 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
4474 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4476 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
4477 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
4478 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
4479 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
4480 (aarch64_atomic_cas<mode>, GPI): Likewise.
4482 2017-06-21 Martin Liska <mliska@suse.cz>
4484 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
4485 statements on cold and hot labels.
4486 * predict.c (tree_estimate_probability_bb): Remove the
4487 prediction from this place.
4489 2017-06-21 Martin Liska <mliska@suse.cz>
4491 PR tree-optimization/79489
4492 * gimplify.c (maybe_add_early_return_predict_stmt): New
4494 (gimplify_return_expr): Call the function.
4495 * predict.c (tree_estimate_probability_bb): Remove handling
4497 * predict.def: Update comment about early return predictor.
4498 * gimple-predict.h (is_gimple_predict): New function.
4499 * predict.def: Change default value of early return to 66.
4500 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
4502 * passes.def: Put pass_strip_predict_hints to the beginning of
4505 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
4507 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
4508 FUNCTION_DECL declarations.
4509 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
4511 (dwaf2out_decl): Likewise.
4512 * godump.c (go_early_global_decl): Skip call to the real debug hook
4513 for FUNCTION_DECL declarations.
4514 * passes.c (rest_of_decl_compilation): Skip call to the
4515 early_global_decl debug hook for FUNCTION_DECL declarations, unless
4516 -fdump-go-spec is passed.
4518 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4520 * config/i386/i386.c (struct builtin_isa): New field pure_p.
4521 Reorder for compactness.
4522 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
4523 (def_builtin_pure, def_builtin_pure2): New functions.
4524 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
4526 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
4528 * match.pd (nop_convert): New predicate.
4529 ((A +- CST1) +- CST2): Allow some NOP conversions.
4531 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4534 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
4535 with ctors/dtors if GOVD_SHARED is set.
4537 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4539 * config/aarch64/aarch64.md (movti_aarch64):
4540 Emit mov rather than orr.
4541 (movtf_aarch64): Likewise.
4542 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
4543 Emit mov rather than orr.
4545 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4547 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
4548 Swap alternatives, make integer dup more expensive.
4550 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
4552 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
4553 Return true for non-tls symbols.
4555 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
4557 * config/aarch64/aarch64-cores.def (cortex-a55): New.
4558 (cortex-a75): Likewise.
4559 (cortex-a75.cortex-a55): Likewise.
4560 * config/aarch64/aarch64-tune.md: Regenerate.
4561 * doc/invoke.texi (-mtune): Document new values for -mtune.
4563 2017-06-21 Tom de Vries <tom@codesourcery.com>
4565 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
4567 (Effective-Target Keywords, Other attributes): Suggest using
4568 dg-add-options stack_size feature to get stack limit in stack_size
4569 effective target documentation.
4571 2017-06-21 Julian Brown <julian@codesourcery.com>
4572 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4574 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
4575 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
4576 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
4578 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
4579 attribute type list for neon_multiply.
4580 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
4581 attribute type list for neon_multiply.
4582 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
4583 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
4584 attribute type list for neon_multiply.
4585 * config/arm/types.md (crypto_pmull): Add.
4586 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
4587 attribute type list.
4589 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
4591 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
4594 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4596 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
4597 to make sure not to dereference a NULL cost_classes_ptr pointer.
4599 2017-06-20 Carl Love <cel@us.ibm.com>
4601 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4602 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4603 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4604 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4605 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4606 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4607 VMULOSW): New enum "unspec" values.
4608 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
4609 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
4610 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4611 altivec_vmulosw): New patterns.
4612 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4613 VMULOSW): Add definitions.
4615 2017-06-20 Julia Koval <julia.koval@intel.com>
4617 * config/i386/i386.c: Fix rounding expand for new pattern.
4618 * config/i386/subst.md: Fix pattern (parallel -> unspec).
4620 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4622 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
4623 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
4625 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4627 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
4630 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
4632 * config/aarch64/aarch64-cores.def: Rearrange to sort by
4633 architecture, then by implementer ID.
4634 * config/aarch64/aarch64-tune.md: Regenerate.
4636 2017-06-20 Richard Biener <rguenther@suse.de>
4639 * fold-const.c (split_tree): Fold to type before negating.
4641 2017-06-20 David Malcolm <dmalcolm@redhat.com>
4643 * diagnostic-show-locus.c
4644 (selftest::test_fixit_deletion_affecting_newline): New function.
4645 (selftest::diagnostic_show_locus_c_tests): Call it.
4647 2017-06-20 Andreas Schwab <schwab@suse.de>
4650 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
4653 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
4655 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
4656 __ARM_FEATURE_COPROC according to support.
4658 2017-06-20 Jakub Jelinek <jakub@redhat.com>
4660 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
4661 Rewritten to avoid overflow for > 32-bit pointers.
4664 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
4665 by removing enum keyword.
4666 (ubsan_type_descriptor): Likewise. Formatting fix.
4669 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
4670 splitter): Require TARGET_SSE2 in the condition.
4672 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
4675 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
4676 for doing vector set of SFmode on ISA 3.0.
4677 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
4678 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
4680 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
4681 SFmode value into a V4SF variable that was extracted from another
4682 V4SF variable without converting the element to double precision
4683 and back to single precision vector format.
4684 (vsx_insert_extract_v4sf_p9_2): Likewise.
4686 2017-06-19 Jakub Jelinek <jakub@redhat.com>
4688 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
4689 in UWHI to avoid undefined overflow.
4692 * ubsan.h (enum ubsan_encode_value_phase): New.
4693 (ubsan_encode_value): Change second argument to
4694 enum ubsan_encode_value_phase with default value of
4695 UBSAN_ENCODE_VALUE_GENERIC.
4696 * ubsan.c (ubsan_encode_value): Change second argument to
4697 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
4698 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
4699 create_tmp_var_raw instead of create_tmp_var and use a
4701 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
4702 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
4703 ubsan_encode_value callers.
4706 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
4707 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
4708 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
4710 2017-06-19 Richard Biener <rguenther@suse.de>
4713 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
4714 estimates if we changed anything.
4716 2017-06-19 Richard Biener <rguenther@suse.de>
4718 PR tree-optimization/80887
4719 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
4720 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
4721 simplified lookups, then reset mprts_hook.
4722 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
4724 (try_to_simplify): Likewise.
4726 2017-06-19 Martin Liska <mliska@suse.cz>
4729 * gimplify.c (gimplify_switch_expr):
4730 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
4732 2017-06-19 Martin Liska <mliska@suse.cz>
4734 * doc/install.texi: Document that PGO runs in 4 stages.
4736 2017-06-19 Martin Liska <mliska@suse.cz>
4739 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
4740 to dispatcher function name.
4741 * multiple_target.c (replace_function_decl): New function.
4742 (create_dispatcher_calls): Redirect both edges and references.
4744 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
4746 * profile-count.c (profile_count::dump): Dump quality.
4747 (profile_count::differs_from_p): Update for unsigned val.
4748 * profile-count.h (profile_count_quality): New enum.
4749 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
4751 2017-06-19 Richard Biener <rguenther@suse.de>
4753 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
4754 struct function as arg.
4755 (estimate_numbers_of_iterations): Export overload with loop arg.
4756 (free_numbers_of_iterations_estimates_loop): Use an overload of
4757 free_numbers_of_iterations_estimates instead.
4758 * tree-cfg.c (remove_bb): Adjust.
4759 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
4760 * tree-parloops.c (gen_parallel_loop): Likewise.
4761 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
4763 (tree_unroll_loops_completely): Likewise.
4764 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
4765 Use an overload instead and export.
4766 (estimated_loop_iterations): Adjust.
4767 (max_loop_iterations): Likewise.
4768 (likely_max_loop_iterations): Likewise.
4769 (estimate_numbers_of_iterations): Take struct function as arg
4771 (loop_exits_before_overflow): Adjust.
4772 (free_numbers_of_iterations_estimates_loop): Use an overload.
4773 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
4774 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
4776 2017-06-19 Richard Biener <rguenther@suse.de>
4779 * ipa-prop.c (find_constructor_constant_at_offset): Handle
4780 RANGE_EXPR conservatively.
4782 2017-06-16 Carl Love <cel@us.ibm.com>
4784 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4785 definitions for vec_float, vec_float2, vec_floato,
4786 vec_floate built-ins.
4787 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
4788 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
4790 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
4791 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
4792 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
4793 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
4794 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
4795 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
4796 vec_floato): Add builtin defines.
4797 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
4798 Update the built-in documentation file for the new built-in
4801 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4803 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
4804 (mthumb): Mark as the negative of -marm.
4806 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4808 * doc/invoke.texi (ARM Options, -mcpu): Document supported
4810 (ARM Options, -mtune): Document that this accepts the same
4811 extension options as -mcpu.
4812 (ARM Options, -mfpu): Document addition of -mfpu=auto.
4814 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4816 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
4817 permitted extensions.
4819 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4821 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
4822 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
4823 (armv8-m.main): Add option +nodsp.
4824 * config/arm/arm-cpu-cdata.h: Regenerated.
4826 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4828 * config/arm/t-fuchsia: New file.
4829 * config.gcc (arm*-*-fuchsia*): Use it.
4831 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4833 * config/arm/t-symbian: Rewrite for new option infrastructure.
4835 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4837 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
4838 (MULTILIB_REQUIRED): Likewise.
4840 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4842 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
4843 (MULTILIB_RESUE): Likewise.
4844 (MULTILIB_MATCHES): Likewise.
4845 (MULTLIB_REQUIRED): Likewise.
4847 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4849 * config/arm/t-rtems: Rewrite for new option framework.
4851 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4853 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
4854 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
4855 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
4856 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
4857 * config/arm/t-multilib: ... here.
4858 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
4859 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
4860 armv7-a and armv8*-a when A-profile libraries have not been built.
4861 * config/arm/t-rmprofile: Rewrite.
4863 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4865 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
4866 with a backslash. Remove the backslash after substituting unescaped
4868 * doc/fragments.texi (MULTILIB_REUSE): Document it.
4870 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4872 * config.gcc: (arm*-*-*): When building a-profile libraries, force
4873 the driver to pass through the default setting of -mfloat-abi.
4874 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
4876 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
4877 (all_feat_combs): New rule.
4878 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
4880 * config/arm/t-aprofile: Rewrite.
4882 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4884 * config/arm/arm.h (FPUTYPE_AUTO): Define.
4885 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
4886 fpu is not specified by the user/command-line.
4887 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
4888 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
4889 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
4890 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
4891 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
4892 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
4894 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4896 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
4897 * config/arm/t-arm-elf: Rewritten.
4899 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4901 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
4902 have some floating-point instructions.
4903 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
4904 (TARGET_MAYBE_HARD_FLOAT): New macro.
4905 * config/arm/arm-builtins.c (arm_init_builtins): Use
4906 TARGET_MAYBE_HARD_FLOAT.
4907 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
4909 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4911 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
4912 (configargs.h): Include it.
4913 (arm_print_hint_for_fpu_option): New function.
4914 (arm_parse_fpu_option): New function.
4915 (candidate_extension): New class.
4916 (arm_canon_for_multilib): New function.
4917 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
4918 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
4919 (ARCH_CANONICAL_SPECS): New macro.
4920 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
4922 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4924 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
4925 are set after handling multilib fragments. Set target_cpu_default2
4928 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4930 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
4932 (arm*-*-*): Set target_cpu_default2 to a quoted string.
4933 * config/arm/parsecpu.awk (check_cpu): Validate any extension
4935 (check_arch): Likewise.
4936 * config/arm/arm.c (arm_configure_build_target): Handle
4937 TARGET_CPU_DEFAULT being a string constant. Scan any feature
4938 options in the default.
4940 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4942 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
4943 when an option is an alias of another.
4944 * config/arm/parsecpu.awk (optalias): New parser token.
4945 (gen_comm_data): Mark non-alias options as such. Emit entries
4946 for extension aliases.
4947 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
4948 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
4949 (armv6kz, armv6zk, armv6t2): Likewise.
4950 (armv7): Make vfpv3-d16 an alias.
4951 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
4953 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
4954 Sort in canonical order.
4955 (armv8-a): Sort in canonical order.
4956 (armv8.1-a, armv8.2-a): Likewise.
4957 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
4959 (cortex-a9): Sort in canonical order.
4960 * config/arm/arm.c (selftests.h): Include it.
4961 (arm_test_cpu_arch_data): New function.
4962 (arm_run_self_tests): New function.
4963 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
4964 (targetm): Move declaration to the end of the file.
4965 * arm-cpu-cdata.h: Regenerated.
4967 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4969 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
4970 call to target_mode_check describing the type of option passed.
4971 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
4972 (arm_target_thumb_only): Use arm_parse_arch_option_name or
4973 arm_parse_cpu_option_name to match parameters against list of
4975 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
4976 arm_arch_core_flags data structure.
4977 * config/arm/arm-cpu_cdata.h: Regenerated.
4979 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
4981 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
4983 (arm_print_hint_for_cpu_option): Likewise.
4984 (arm_print_hint_for_arch_option): Likewise.
4985 (arm_parse_cpu_option_name): Likewise.
4986 (arm_parse_arch_option_name): Likewise.
4987 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
4988 of entries in the all_fpus list.
4989 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
4990 (arm_parse_cpu_option_name): Declare.
4991 (arm_parse_arch_option_name): Declare.
4992 (arm_parse_option_features): Declare.
4993 (arm_intialize_isa): Declare.
4994 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
4996 (gen_comm_data): ... here. Make definitions non-static.
4997 * config/arm/arm-cpu-data.h: Regenerated.
4998 * config/arm/arm-cpu-cdata.h: Regenerated.
5000 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5002 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
5003 (cpu_arch_extension): New structure.
5004 (cpu_arch_option, arch_option, cpu_option): New structures.
5005 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
5007 (gen_data): Generate new format data tables.
5008 * config/arm/arm.c (cpu_tune): New structure.
5009 (cpu_option, processors): Delete.
5010 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
5011 (arm_print_hint_for_cpu_option): ... this and ...
5012 (arm_print_hint_for_arch_option): ... this.
5013 (arm_parse_arch_cpu_name): Delete. Replace with ...
5014 (arm_parse_cpu_option_name): ... this and ...
5015 (arm_parse_arch_option_name): ... this.
5016 (arm_unrecognized_feature): Change type of target parameter to
5018 (arm_parse_arch_cpu_features): Delete. Replace with ...
5019 (arm_parse_option_features): ... this.
5020 (arm_configure_build_target): Rework to use new configuration data
5022 (arm_print_tune_info): Rework for new configuration data tables.
5023 * config/arm/arm-cpu-data.h: Regenerated.
5024 * config/arm/arm-cpu.h: Regenerated.
5026 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5028 * Makefile.in (OBJS): Move sbitmap.o from here ...
5029 (OBJS-libcommon): ... to here.
5031 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5033 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
5034 (ISA_ALL_CRYPTO): New macro.
5035 (ISA_ALL_SIMD): New macro
5036 (ISA_ALL_FP): New macro.
5037 * config/arm/arm.c (fpu_bitlist): Update initializer.
5038 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
5040 (arm9e): Add fpu. Add option for nofp
5041 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
5042 (arm926ej-s, arm1026ej-s): Likewise.
5043 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
5044 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
5045 neon-fp16, neon-vfpv4, nofp and nosimd.
5046 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
5047 (cortex-a8): Add fpu. Add option for nofp.
5048 (cortex-a9): Add fpu. Add options for nosimd and nofp.
5049 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
5050 (cortex-r4f): Add fpu.
5051 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
5052 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
5054 (cortex-r8): Likewise.
5055 (cortex-m4): Add fpu. Add option for nofp.
5056 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
5057 (cortex-a17.cortex-a7): Likewise.
5058 (cortex-a32): Add fpu. Add options for crypto and nofp.
5059 (cortex-a35, cortex-a53): Likewise.
5060 (cortex-a57): Add fpu. Add option for crypto.
5061 (cortex-a72, cortex-a73): Likewise.
5062 (exynos-m1): Likewise.
5063 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
5064 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
5065 (cortex-m33): Add fpu. Add option for nofp.
5066 * config/arm/arm-cpu-cdata.h: Regenerated
5067 * config/arm/arm-cpu-data.h: Regenerated.
5069 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5071 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
5072 (armv5te, armv5tej): Likewise.
5073 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
5074 (armv7): Add options fp and vfpv3-d16.
5075 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
5076 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
5078 (armv7ve): Likewise.
5079 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
5080 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
5081 (armv8-a): Add nocrypto option.
5082 (armv8.1-a, armv8.2-a): Likewise.
5083 (armv8-m.main): add options fp, fp.dp and nofp.
5085 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5087 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
5089 (armv8-a+crc): Delete.
5090 (armv8.1-a): Add options simd, crypto and nofp.
5091 (armv8.2-a): Add options fp16, simd, crypto and nofp.
5092 (armv8.2-a+fp16): Delete.
5093 (armv8-m.main): Add option dsp.
5094 (armv8-m.main+dsp): Delete.
5095 (cortex-a8): Add fpu. Add nofp option.
5096 (cortex-a9): Add fpu. Add nofp and nosimd options.
5097 * config/arm/parsecpu.awk (gen_data): Generate option tables and
5098 link to main cpu and architecture data structures.
5099 (gen_comm_data): Only put isa attributes from the main architecture
5101 (option): New statement for architecture and CPU entries.
5102 * arm.c (struct cpu_option): New structure.
5103 (struct processors): Add entry for options.
5104 (arm_unrecognized_feature): New function.
5105 (arm_parse_arch_cpu_name): Ignore any characters after the first
5107 (arm_parse_arch_cpu_feature): New function.
5108 (arm_configure_build_target): Separate out any CPU and architecture
5109 features and parse separately. Don't error out if -mfpu=auto is
5110 used with only an architecture string.
5111 (arm_print_asm_arch_directives): New function.
5112 (arm_file_start): Call it.
5113 * config/arm/arm-cpu-cdata.h: Regenerated.
5114 * config/arm/arm-cpu-data.h: Likewise.
5115 * config/arm/arm-tables.opt: Likewise.
5117 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5119 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
5120 assembler when it is not -mfpu=auto.
5122 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5124 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
5125 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
5126 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
5127 (ASM_CPU_SPEC): Rewrite.
5128 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
5129 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
5130 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
5131 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
5132 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
5133 copied string is NUL-terminated. Also strip any characters prefixed
5135 (arm_rewrite_selected_arch): New function.
5136 (arm_rewrite_march): New function.
5138 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5140 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
5141 (x_arm_cpu_string, x_arm_tune_string): Likewise.
5142 (march, mcpu, mtune): Convert to string-based options.
5143 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
5144 (arm_parse_arch_cpu_name): New function.
5145 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
5146 identify selected architecture or CPU.
5147 (arm_option_save): New function.
5148 (TARGET_OPTION_SAVE): Redefine.
5149 (arm_option_restore): Restore string options.
5150 (arm_option_print): Print string options.
5152 2017-06-16 Martin Sebor <msebor@redhat.com>
5154 PR tree-optimization/80933
5155 PR tree-optimization/80934
5156 * builtins.c (fold_builtin_3): Do not handle bcmp here.
5157 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
5158 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
5159 (gimple_fold_builtin): Call them.
5161 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5163 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
5164 as unlikely; update profile.
5166 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5168 * predict.c (force_edge_cold): Handle declaring edges impossible
5171 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5173 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
5175 (try_unroll_loop_completely): Fix reporting.
5177 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5179 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
5181 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
5184 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
5185 if given a non-constant argument for an intrinsic which requires a
5188 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5190 * profile.c (compare_freqs): New function.
5191 (branch_prob): Sort edge list.
5192 (find_spanning_tree): Assume that the list is priority sorted.
5194 2017-06-16 Richard Biener <rguenther@suse.de>
5196 PR tree-optimization/81090
5197 * passes.def (pass_record_bounds): Remove.
5198 * tree-pass.h (make_pass_record_bounds): Likewise.
5199 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
5200 make_pass_record_bounds): Likewise.
5201 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
5202 not free niter estimates at the beginning but at the end.
5203 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
5205 2017-06-16 Richard Biener <rguenther@suse.de>
5207 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
5208 initializer to workaround ICE in host GCC 4.8.
5210 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5212 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
5214 (clone_inlined_nodes): Update.
5216 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5218 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
5219 prefetch settings, and enable prefetching by default at -O3.
5221 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5223 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5224 Set flag_prefetch_loop_arrays according to tuning data.
5226 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5228 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
5230 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
5231 [Unrelated to main purpose of the patch] Place the pointer field last
5232 to enable type checking errors when tune structure are wrongly merged.
5233 * config/aarch64/aarch64.c (generic_prefetch_tune,)
5234 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
5235 (thunderx2t99_prefetch_tune): New tune constants.
5236 (tune_params *_tunings): Update all tunings (no functional change).
5237 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
5238 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
5239 from tunings structures.
5241 2017-06-16 Jakub Jelinek <jakub@redhat.com>
5244 * ubsan.c (instrument_null): Add T argument, use it instead
5245 of computing it based on IS_LHS.
5246 (instrument_object_size): Likewise.
5247 (pass_ubsan::execute): Adjust instrument_null and
5248 instrument_object_size callers to pass gimple_get_lhs or
5249 gimple_assign_rhs1 result to it. Use instrument_null instead of
5250 calling get_base_address and instrument_mem_ref. Handle
5251 aggregate call arguments for object-size sanitization.
5253 2017-06-16 Yury Gribov <tetra2005@gmail.com>
5255 PR tree-optimization/81089
5256 * tree-vrp.c (is_masked_range_test): Validate operands of
5259 2017-06-15 Martin Sebor <msebor@redhat.com>
5262 * dumpfile.c (dump_register): Avoid calling memset to initialize
5263 a class with a default ctor.
5264 * gcc.c (struct compiler): Remove const qualification.
5265 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
5266 * hash-table.h: Ditto.
5267 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
5269 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
5270 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
5272 * params.h (struct param_info): Make struct members non-const.
5273 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
5274 with default initialization.
5275 * vec.h (vec_copy_construct, vec_default_construct): New helper
5277 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
5278 with vec_copy_construct.
5279 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
5280 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
5281 * doc/invoke.texi (-Wclass-memaccess): Document.
5283 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5285 * emit-rtl.h (is_leaf): Update comment about local
5288 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5291 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
5292 for a variable to have a section before checking if the section has a
5294 Set section to.persistent if persistent attribute is set.
5295 Warn if .persistent attribute is used on an automatic variable.
5297 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
5299 PR rtl-optimization/80474
5300 * reorg.c (update_block): Do not ignore instructions in a delay slot.
5302 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
5304 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
5307 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
5309 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
5310 (casesi): Emit bounds checking as RTL.
5311 (casesi_internal_mips16_<mode>): Remove bounds checking.
5313 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
5315 * config/xtensa/xtensa.c (xtensa_option_override): Append
5316 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
5317 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
5318 xtensa_doloop_hooks): Define unconditionally.
5319 (xtensa_reorg_loops): Only call reorg_loops in the presence of
5321 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
5322 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
5323 for it in xtensa_option_override.
5324 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
5325 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
5327 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
5329 * doc/cppopts.texi: Document '-' special value to -MF.
5331 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
5333 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
5334 (cortex_a53_fconst): Likewise.
5335 (cortex_a53_fpmul): Likewise.
5336 (cortex_a53_f_load_64): Likewise.
5337 (cortex_a53_f_load_many): Likewise.
5338 (cortex_a53_advsimd_alu): Likewise.
5339 (cortex_a53_advsimd_alu_q): Likewise.
5340 (cortex_a53_advsimd_mul): Likewise.
5341 (cortex_a53_advsimd_mul_q): Likewise.
5342 (fpmac bypass): Add new bypass for fpmac-fpmac case.
5343 Add missing fmul, r2f_cvt and fconst cases.
5345 2017-06-14 Richard Biener <rguenther@suse.de>
5348 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
5350 (fold_binary_loc): When associating do not treat pre-existing
5351 TREE_OVERFLOW on literal constants as a reason to allow
5352 TREE_OVERFLOW on associated literal constants.
5354 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5356 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
5357 (MASK_FEATURES): New macro.
5358 * config/sparc/sparc.c (sparc_option_override): Remove the special
5359 handling of -mfpu and generalize it to all MASK_FEATURES switches.
5361 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5363 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
5364 a division of 0 if non-call exceptions are enabled.
5366 2017-06-14 Andrew Pinski <apinski@cavium.com>
5367 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5370 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
5371 Improve vector initialization code gen for only variable case.
5373 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5375 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
5377 2017-06-14 Richard Biener <rguenther@suse.de>
5379 PR tree-optimization/81083
5380 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
5383 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5385 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
5386 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
5387 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
5388 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
5389 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
5390 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
5392 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5394 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
5395 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
5397 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5399 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
5401 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5403 * config/rs6000/t-rtems: Don't handle SPE.
5405 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5407 * config/rs6000/t-linux: Don't handle SPE.
5409 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5411 * config/rs6000/eabispe.h: Delete file.
5413 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5415 * config/rs6000/t-spe: Delete file.
5417 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5419 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
5420 (rs6000_legitimate_offset_address_p): Return false for anything in
5421 V2SImode or V2SFmode.
5423 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5425 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
5426 except V2SF and V2SI. Rearrange the vector modes, and add comments.
5427 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
5429 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
5430 (rs6000_legitimate_offset_address_p): Ditto.
5431 (rs6000_emit_move): Ditto.
5432 (rs6000_init_builtins): Remove V4HI_type_node.
5434 2017-06-13 Martin Liska <mliska@suse.cz>
5437 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
5438 (gate_asan): Likewise.
5439 * asan.h (asan_no_sanitize_address_p): Remove the function.
5440 (sanitize_flags_p): New function.
5441 * builtins.def: Fix coding style.
5442 * common.opt: Use renamed enum value.
5443 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
5444 * doc/extend.texi: Document no_sanitize attribute.
5445 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
5446 to SANITIZE_UNDEFINED_NONDEFAULT.
5447 * gcc.c (sanitize_spec_function): Use the renamed enum value.
5448 * gimple-fold.c (optimize_atomic_compare_exchange_p):
5449 Use sanitize_flags_p.
5450 * gimplify.c (gimplify_function_tree): Likewise.
5451 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
5452 * opts.c (parse_no_sanitize_attribute): New function.
5453 (common_handle_option): Use renamed enum value.
5454 * opts.h (parse_no_sanitize_attribute): Declare.
5455 * tree.c (sanitize_flags_p): New function.
5456 * tree.h: Declared here.
5457 * tsan.c: Use sanitize_flags_p.
5458 * ubsan.c (ubsan_expand_null_ifn): Likewise.
5459 (instrument_mem_ref): Likewise.
5460 (instrument_bool_enum_load): Likewise.
5461 (do_ubsan_in_current_function): Remove the function.
5462 (pass_ubsan::execute): Use sanitize_flags_p.
5463 * ubsan.h: Remove do_ubsan_in_current_function
5464 * tree-cfg.c (print_no_sanitize_attr_value): New function.
5465 (dump_function_to_file): Use it here.
5467 2017-06-13 Martin Jambor <mjambor@suse.cz>
5469 PR tree-optimization/80803
5470 PR tree-optimization/81063
5471 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
5472 (propagate_subaccesses_across_link): Enqueue subtree whenever
5473 necessary instead of relying on the caller.
5475 2017-06-13 Martin Jambor <mjambor@suse.cz>
5477 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
5478 that have a first_link.
5479 (sort_and_splice_var_accesses): Do not check first_link before
5481 (subtree_mark_written_and_enqueue): Likewise.
5482 (propagate_all_subaccesses): Likewise and do not stop at first
5483 parent with a first_link.
5485 2017-06-13 Martin Jambor <mjambor@suse.cz>
5487 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
5490 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5492 * match.pd: New pattern.
5494 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5496 * tree-vrp.c (is_masked_range_test): New function.
5497 (register_edge_assert_for): Determine ranges for
5500 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5502 PR tree-optimization/67328
5503 * fold-const.c (maskable_range_p): New function.
5504 (build_range_check): Generate bittests if possible.
5506 2017-06-13 Martin Liska <mliska@suse.cz>
5508 * gimple-pretty-print.c (dump_probability): Add new argument.
5509 (dump_edge_probability): Dump both probability and count.
5510 (dump_gimple_label): Likewise.
5511 (dump_gimple_bb_header): Likewise.
5513 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
5516 * config/avr/avr-devices.c: Fix indentation.
5517 * config/avr/gen-avr-mmcu-specs.c: Dito.
5519 2017-06-13 Richard Biener <rguenther@suse.de>
5521 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
5522 instead get vector type from stmt_info.
5523 (vectorizable_reduction): Adjust. Remove dead code.
5525 2017-06-13 Richard Biener <rguenther@suse.de>
5528 * fold-const.c (extract_muldiv_1): Remove bogus distribution
5529 case of C * (x * C2 + C3).
5530 (fold_addr_of_array_ref_difference): Properly fold index difference.
5532 2017-06-12 David S. Miller <davem@davemloft.net>
5535 * config/sparc/sparc.md (return expander): Emit frame blockage if
5536 function uses alloca.
5538 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
5540 * combine.c (make_field_assignment): Check len rather than the mode
5541 precision when calling force_to_mode.
5543 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
5545 Support multilibs and devices that see flash in RAM address range.
5548 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
5549 (avr_mcu_t) <flash_pm_offset>: New field.
5550 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
5551 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
5552 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
5553 (AVR_TINY_PM_OFFSET): Remove macro.
5554 * config/avr/avr.opt (-mshort-calls): New option.
5555 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
5556 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
5557 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
5558 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
5559 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
5560 instead of avr_arch->have_jmp_call.
5561 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
5562 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
5563 avr_arch->flash_pm_offset to define.
5564 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
5565 new field flash_pm_offset. Add entry for avrxmega3.
5566 (avr_texinfo): Add entry for avrxmega3.
5567 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
5568 attiny212, attiny214,
5569 attiny412, attiny414, attiny416, attiny417,
5570 attiny814, attiny816, attiny817,
5571 attiny1614, attiny1616, attiny1617,
5572 attiny3214, attiny3216, attiny3217.
5573 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
5574 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
5575 (avr_print_operand_address) [AVR_TINY]: Same.
5576 (avr_asm_init_sections) <readonly_data_section>: Only patch
5577 callback if avr_arch->flash_pm_offset = 0.
5578 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
5579 for rodata if avr_arch->flash_pm_offset != 0.
5580 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
5581 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
5582 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
5583 (m_options): Append opt_rcall.
5584 (m_dirnames): Append dir_rcall.
5585 * config/avr/t-multilib: Regenerate.
5587 * configure.ac [target=avr]: Check whether avrxmega3 default
5588 linker description file works as needed.
5589 * configure: Regenerate.
5590 * doc/avr-mmcu.texi: Regenerate.
5591 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
5592 <__AVR_ARCH__>: Document avrxmega3 and 103.
5593 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
5594 <__AVR_SHORT_CALLS__>: Document it.
5595 <__AVR_PM_BASE_ADDRESS__>: Document it.
5596 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
5597 (AVR Variable Attributes) <progmem>: Document this is
5598 not needed for avrxmega3.
5599 (AVR Named Address Spaces) <__flash>: Dito.
5601 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
5603 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
5605 2017-06-12 Doug Rupp <rupp@adacore.com>
5607 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
5608 Append vxworks-stdint.h to the tm_file list.
5609 * config/vxworks-stdint.h: New file.
5611 2017-06-12 Martin Liska <mliska@suse.cz>
5613 PR tree-optimization/81041
5614 * tree-profile.c (gimple_gen_ic_func_profiler):
5615 Create an extra BB in profile-generate
5616 (gimple_gen_time_profiler): Likewise.
5618 2017-06-12 Jakub Jelinek <jakub@redhat.com>
5620 PR tree-optimization/81003
5621 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
5622 (update_range_test): Use it instead of force_gimple_operand_gsi.
5624 2017-06-12 Richard Biener <rguenther@suse.de>
5626 PR tree-optimization/81053
5627 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
5628 with backedge value not defined in loop. Simplify def stmt
5631 2017-06-11 Tom de Vries <tom@codesourcery.com>
5634 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
5636 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
5637 nvptx_cannot_force_const_mem.
5639 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5641 * opts.c (finish_options): Move test for flag_split_stack after
5642 it has been initialized.
5644 2017-06-11 Jason Merrill <jason@redhat.com>
5646 * tree.h (id_equal): New.
5647 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
5648 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
5649 instead of strcmp of IDENTIFIER_POINTER.
5651 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5653 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
5654 (mark_all_inlined_calls_cdtor): Fix formating.
5655 (inline_transform): Rescale profile before inlining.
5657 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5659 * cgraph.h (cgraph_edge::clone): Update prototype.
5660 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
5661 (cgraph_node::create_clone): Update.
5662 (cgraph_node::create_version_clone): Update.
5663 * tree-inline.c (copy_bb): Update.
5664 (expand_call_inline): Update.
5666 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5668 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
5669 factored out from ...
5670 (rs6000_emit_prologue): ... here.
5672 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
5674 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
5675 factored out from ...
5676 (rs6000_emit_prologue): ... here.
5678 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5680 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
5682 (handle_missing_profiles): Fix computation of tp_first_run.
5683 (counts_to_freqs): Do not touch freqs when count is 0.
5685 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
5687 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
5690 2017-06-10 Tom de Vries <tom@codesourcery.com>
5692 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
5693 attributes): Document signal effective target.
5695 2017-06-10 Tom de Vries <tom@codesourcery.com>
5697 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
5698 Document effective target stack_size.
5700 2017-06-09 David Malcolm <dmalcolm@redhat.com>
5702 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
5703 to the edit_context if they can be auto-applied.
5705 2017-06-9 Ian Lance Taylor <iant@golang.org>
5707 * opts.c (finish_options): If -fsplit-stack, disable implicit
5708 -forder-blocks-and-partition.
5709 * doc/invoke.texi (Optimize Options): Document that when using
5710 -fsplit-stack -forder-blocks-and-partition is not implicitly
5713 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5715 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
5716 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
5717 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
5718 * builtins.def (abort, trap, unreachable): Declare cold.
5719 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
5720 * tree-core.h (ECF_COLD): New.
5721 * tree.c (set_call_expr_flags): Handle ECF_COLD.
5722 (build_common_builtin_nodes): Mark unreachable and abort as cold.
5724 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5726 * predict.c (unlikely_executed_stmt_p): Cleanup.
5728 2017-06-09 Richard Biener <rguenther@suse.de>
5730 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
5731 model if the ref is always written to.
5733 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5735 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
5737 2017-06-09 Tamar Christina <tamar.christina@arm.com>
5739 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
5742 2017-06-09 Tom de Vries <tom@codesourcery.com>
5745 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
5746 "target cannot support label values" when encountering LABEL_REF.
5748 2017-06-09 Martin Liska <mliska@suse.cz>
5750 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
5751 (gimple_gen_ic_func_profiler): Emit direct comparison
5752 of __gcov_indirect_call_callee with NULL.
5753 (gimple_gen_time_profiler): Change probability from
5754 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
5756 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5758 * profile.c (edge_gcov_counts): Turn to pointer.
5759 (compute_branch_probabilities, compute_branch_probabilities): Update.
5760 (branch_prob): Do not clear edge_gcov_count.
5761 * profile.h (edge_gcov_counts): Turn to pointer.
5762 (edge_gcov_count): Update.
5764 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
5766 * gimple.h (gimple_check_failed): Mark cold.
5768 2017-06-09 Richard Biener <rguenther@suse.de>
5770 PR tree-optimization/66623
5771 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
5772 refactor check_reduction into two parts, properly computing
5773 whether we have to check reduction validity for outer loop
5776 2017-06-09 Richard Biener <rguenther@suse.de>
5778 PR tree-optimization/79483
5779 * graphite-scop-detection.c (order): New global.
5780 (get_order): Compute bb to order mapping that satisfies code
5781 generation constraints.
5782 (cmp_pbbs): New helper.
5783 (build_scops): Start domwalk at entry block, sort generated
5786 2017-06-09 Richard Biener <rguenther@suse.de>
5789 * ipa-polymorphic-call.c
5790 (ipa_polymorphic_call_context::restrict_to_inner_class):
5791 Skip FIELD_DECLs with error_mark_node type.
5792 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
5795 2017-06-09 Martin Liska <mliska@suse.cz>
5797 * predict.c (struct branch_predictor): New struct.
5798 (test_prediction_value_range): New test.
5799 (predict_c_tests): New function.
5800 * selftest-run-tests.c (selftest::run_tests): Run the function.
5801 * selftest.h: Declare new tests.
5803 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
5806 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
5807 gen_add3_insn did not fail.
5808 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
5809 r0, construct that number in a temporary reg and add that reg to r0.
5810 If asked to put the result in r0 as well, fail.
5812 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
5814 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5815 for early expansion of vec_eqv.
5817 2017-06-08 Jakub Jelinek <jakub@redhat.com>
5820 * ubsan.c (instrument_null): Avoid pointless code temporary.
5821 (pass_ubsan::execute): Instrument aggregate arguments of calls.
5823 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
5827 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
5830 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
5833 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5835 * predict.c (unlikely_executed_edge_p): Move ahead.
5836 (probably_never_executed_edge_p): Use it.
5838 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5841 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
5842 gimple_call_builtin_p call.
5844 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5846 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
5847 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
5848 rtl_check_failed_type2, rtl_check_failed_code1,
5849 rtl_check_failed_code2, rtl_check_failed_code_mode,
5850 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
5851 rtvec_check_failed_bounds, rtl_check_failed_flag,
5852 _fatal_insn_not_found, _fatal_insn): Likewise.
5853 * tree.h (tree_contains_struct_check_failed,
5854 tree_check_failed, tree_not_check_failed,
5855 tree_class_check_failed, tree_range_check_failed,
5856 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
5857 tree_vec_elt_check_failed, phi_node_elt_check_failed,
5858 tree_operand_check_failed, omp_clause_check_failed,
5859 omp_clause_operand_check_failed, omp_clause_range_check_failed):
5862 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5864 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
5865 flag_branch_probabilities.
5866 * ipa-inline.c (edge_badness): Likewise.
5867 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
5868 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
5869 * predict.c (maybe_hot_frequency_p): Likewise.
5870 (probably_never_executed): Likewise.
5871 * sched-ebb.c (schedule_ebbs): Likewise.
5872 * sched-rgn.c (find_single_block_region): Likewise.
5873 * tracer.c (tail_duplicate): Likewise.
5875 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5877 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
5878 longer requires x_flag_profile_use.
5880 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5882 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
5883 instead of flag_reorder_blocks_and_partition.
5884 * dbxout.c (dbxout_function_end): Likewise.
5885 * dwarf2out.c (gen_subprogram_die): Likewise.
5886 * haifa-sched.c (sched_create_recovery_edges): Likewise.
5887 * hw-doloop.c (reorg_loops): Likewise.
5888 * varasm.c (assemble_start_function,
5889 assemble_end_function): Likewise.
5890 (decide_function_section): Do not check for
5891 flag_reorder_blocks_and_partition.
5893 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
5895 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
5897 (chkp_get_hard_register_fake_addr_expr): Ditto.
5898 (chkp_build_addr_expr): Add check for hard reg case.
5899 (chkp_parse_array_and_component_ref): Ditto.
5900 (chkp_find_bounds_1): Ditto.
5901 (chkp_process_stmt): Don't generate bounds store for
5904 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5906 * predict.c (maybe_hot_bb_p): Do not check profile status.
5907 (maybe_hot_edge_p): Likewise.
5908 (probably_never_executed): Check for zero counts even if profile
5910 (unlikely_executed_edge_p): New function.
5911 (unlikely_executed_stmt_p): New function.
5912 (unlikely_executed_bb_p): New function.
5913 (set_even_probabilities): Use unlikely predicates.
5914 (combine_predictions_for_bb): Likewise.
5915 (predict_paths_for_bb): Likewise.
5916 (predict_paths_leading_to_edge): Likewise.
5917 (determine_unlikely_bbs): New function.
5918 (estimate_bb_frequencies): Use it.
5919 (compute_function_frequency): Use zero counts even if profile is
5921 * profile-count.h: Fix typo.
5923 2017-08-08 Julia Koval <julia.koval@intel.com>
5925 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
5926 _mm512_mask_cvtsepi16_storeu_epi8,
5927 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
5928 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
5929 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
5930 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
5931 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
5932 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
5933 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
5934 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
5935 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
5936 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
5937 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
5938 __builtin_ia32_pmovuswb256mem_mask,
5939 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
5940 __builtin_ia32_pmovwb512mem_mask): New builtins.
5942 2017-08-08 Julia Koval <julia.koval@intel.com>
5944 PR target/73350,80862
5945 * config/i386/subst.md (round): Fix round pattern.
5946 * config/i386/i386.c (ix86_erase_embedded_rounding):
5947 Fix erasing rounding for the fixed pattern.
5949 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
5951 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
5953 2017-06-08 Martin Liska <mliska@suse.cz>
5955 PR gcov-profile/80911
5956 * gcov.c (block_info::block_info): New constructor.
5958 2017-06-07 Carl Love <cel@us.ibm.com>
5960 * config/rs6000/rs6000-c: The return type of the following
5961 built-in functions was implemented as int not long long. Fix sign
5962 of return value for the unsigned version of vec_mulo and vec_mule.
5963 vector unsigned long long vec_bperm (vector unsigned long long,
5964 vector unsigned char)
5965 vector signed long long vec_mule (vector signed int,
5967 vector unsigned long long vec_mule (vector unsigned int,
5968 vector unsigned int)
5969 vector signed long long vec_mulo (vector signed int,
5971 vector unsigned long long vec_mulo (vector unsigned int,
5972 vector unsigned int)
5973 * doc/extend.texi: Fix the documentation for the built-in
5976 2017-06-07 Carl Love <cel@us.ibm.com>
5979 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
5982 2017-06-07 Carl Love <cel@us.ibm.com>
5984 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
5985 support, Generate doublehv for signed int/float for BE case only.
5987 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
5989 * doc/invoke.texi (mcx16): Rewrite.
5991 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
5993 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
5994 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
5995 *mov<mode>_softfloat, and an anonymous splitter): Use
5996 nonimmediate_operand instead of rs6000_nonimmediate_operand.
5998 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6000 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
6002 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
6003 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
6004 (rs6000_debug_reg_global): Adjust.
6005 (rs6000_init_hard_regno_mode_ok): Adjust.
6006 (rs6000_dbx_register_number): Adjust.
6007 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
6008 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
6009 Remove SPE_ACC and SPEFSCR.
6010 (REG_ALLOC_ORDER): Ditto.
6011 (FRAME_POINTER_REGNUM): Change to 111.
6012 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
6013 (REG_CLASS_NAMES): Ditto.
6014 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
6015 (REGISTER_NAMES): Ditto.
6016 (ADDITIONAL_REG_NAMES): Ditto.
6017 (rs6000_reg_names): Ditto.
6018 * config/rs6000/rs6000.md: Renumber some register number
6021 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6023 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
6025 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
6026 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
6028 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
6029 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
6030 Delete the SPE high registers.
6031 (REG_ALLOC_ORDER): Ditto.
6032 (enum reg_class): Remove SPE_HIGH_REGS.
6033 (REG_CLASS_NAMES): Ditto.
6034 (REG_CLASS_CONTENTS): Delete the SPE high registers.
6035 (REGISTER_NAMES): Ditto.
6036 (rs6000_reg_names): Ditto.
6037 * doc/tm.texi.in: Remove SPE as example.
6038 * doc/tm.texi: Regenerate.
6040 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6042 * config/rs6000/8540.md (ppc8540_brinc): Delete.
6043 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
6044 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
6045 * config/rs6000/rs6000.md (type): Remove "brinc".
6047 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6049 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
6050 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
6051 * config/rs6000/linuxspe.h: Delete file.
6052 * config/rs6000/rs6000.md: Don't include spe.md.
6053 * config/rs6000/spe.h: Delete file.
6054 * config/rs6000/spe.md: Delete file.
6055 * config/rs6000/t-rs6000: Remove spe.md.
6057 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6059 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
6060 (reg_or_none500mem_operand): Delete.
6061 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
6062 instead of reg_or_none500mem_operand.
6064 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6066 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
6067 handling of SPE flags.
6068 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
6070 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6072 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
6074 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
6075 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
6076 paired_divv2sf3): Similar.
6077 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
6078 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
6079 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
6081 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
6082 Rename the paired_* instruction patterns.
6083 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
6085 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
6086 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
6087 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
6088 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
6089 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
6091 (struct machine_function): Delete field spe_insn_chain_scanned_p.
6092 (spe_func_has_64bit_regs_p): Delete.
6093 (spe_expand_predicate_builtin): Delete.
6094 (spe_expand_evsel_builtin): Delete.
6095 (TARGET_DWARF_REGISTER_SPAN): Do not define.
6096 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
6097 (invalid_e500_subreg): Delete.
6098 (rs6000_legitimize_address): Always force_reg op2 as well, for
6099 paired single memory accesses.
6100 (rs6000_member_type_forces_blk): Delete.
6101 (rs6000_spe_function_arg): Delete.
6102 (rs6000_expand_unop_builtin): Delete SPE handling.
6103 (rs6000_expand_binop_builtin): Ditto.
6104 (spe_expand_stv_builtin): Delete.
6105 (bdesc_2arg_spe): Delete.
6106 (spe_expand_builtin): Delete.
6107 (spe_expand_predicate_builtin): Delete.
6108 (spe_expand_evsel_builtin): Delete.
6109 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
6110 (spe_init_builtins): Delete.
6111 (spe_func_has_64bit_regs_p): Delete.
6112 (savres_routine_name): Delete "info" parameter. Adjust callers.
6113 (rs6000_emit_stack_reset): Ditto.
6114 (rs6000_dwarf_register_span): Delete.
6115 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
6116 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
6117 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
6119 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
6121 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
6122 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
6123 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
6124 mulv2sf3, divv2sf3): Delete expanders.
6126 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6128 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
6130 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6132 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
6133 * config/rs6000/rs6000.c: Ditto.
6135 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6137 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
6138 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
6139 comparison_operator.
6141 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6143 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
6144 * config/rs6000/rs6000.opt: Ditto.
6145 * config/rs6000/t-rtems: Ditto.
6147 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6149 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
6150 TARGET_E500_SINGLE by 0, simplify.
6151 * config/rs6000/rs6000.c: Ditto.
6152 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
6153 (spe_build_register_parallel): Delete.
6154 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
6155 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
6156 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
6157 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
6158 (E500_CONVERT): Delete.
6159 * config/rs6000/spe.md: Remove many patterns and all define_constants.
6161 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6163 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
6164 * config/rs6000/dfp.md: Ditto.
6165 (negdd2, *negdd2_fpr): Merge.
6166 (absdd2, *absdd2_fpr): Merge.
6167 (negtd2, *negtd2_fpr): Merge.
6168 (abstd2, *abstd2_fpr): Merge.
6169 * config/rs6000/e500.h: Delete file.
6170 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
6171 TARGET_FPRS by 1 and simplify.
6172 * config/rs6000/rs6000-c.c: Ditto.
6173 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
6175 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
6177 * config/rs6000/rs6000.md: Ditto.
6178 (floatdidf2, *floatdidf2_fpr): Merge.
6179 (move_from_CR_gt_bit): Delete.
6180 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
6181 (E500_CR_IOR_COMPARE): Delete.
6182 (All patterns that require !TARGET_FPRS): Delete.
6183 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
6185 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6187 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
6189 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6191 * graphds.c (add_edge): Intitialize edge's attached data.
6192 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
6193 pointer parameter. Call pointed function on each edge during
6194 graph traversing. Skip traversing the edge when the function
6196 (graphds_dfs, graphds_scc): Ditto.
6197 (for_each_edge): New parameter. Pass the new parameter to callback
6199 * graphds.h (skip_edge_callback): New function pointer type.
6200 (graphds_dfs, graphds_scc): New function pointer parameter.
6201 (graphds_edge_callback, for_each_edge): New parameter.
6203 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6205 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
6206 out code checking if runtime alias check is possible to below ...
6207 Call the new function.
6208 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
6209 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
6211 2017-06-07 Marek Polacek <polacek@redhat.com>
6214 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
6215 TYPE_OVERFLOW_WRAPS check.
6217 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6219 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
6220 if versioning is required.
6221 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
6222 peeling with the check for versioning.
6224 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6226 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
6227 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
6228 Set true to new parameter if new ssa variable is defined.
6229 (vect_gen_vector_loop_niters): Refactor. Set range information
6230 for the new vector loop bound variable.
6231 (vect_do_peeling): Ditto.
6233 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6235 * tree-affine.c (ssa.h): Include header file.
6236 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
6237 has wrapping overflow behavior.
6239 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6241 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
6243 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6245 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
6246 (tree_to_aff_combination): ... here.
6248 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6250 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
6251 reg_pressure model function.
6252 (ivopts_global_cost_for_size): Delete.
6253 (determine_set_costs, iv_ca_recount_cost): Call new model function
6254 ivopts_estimate_reg_pressure.
6256 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6258 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
6259 expensive than udiv. Remove floating point cases from mod.
6261 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6263 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
6266 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6268 * config/aarch64/aarch64.md
6269 (copysignsf3): Fix mask generation.
6271 2017-06-07 Jakub Jelinek <jakub@redhat.com>
6273 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
6275 (class dump_manager): Add register_dumps method.
6276 * dumpfile.c: Include langhooks.h.
6277 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
6278 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
6279 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
6280 (dump_manager::dump_register): Start with 512 entries instead of 32.
6281 (dump_manager::register_dumps): New method.
6282 * toplev.c (general_init): Instead of invoking register_dumps
6283 langhook, invoke register_dumps method on the dump manager.
6284 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
6287 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
6289 * doc/md.texi: Clarify the restrictions on a define_insn condition.
6290 Say that # requires an associated define_split to exist, and that
6291 the define_split must be suitable for use after register allocation.
6293 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6295 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
6296 (compute_outgoing_frequencies): Also initialize zero counts.
6297 (find_many_sub_basic_blocks): Do not produce uninitialized profile
6298 around loops; preserve more of profile when nothing changes.
6300 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
6302 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
6304 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
6305 * config/arm/arm-cpu-cdata.h: Regenerate.
6306 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
6307 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
6308 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
6309 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
6310 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
6312 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
6314 * config/arm/t-rmprofile: Likewise.
6315 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
6317 2017-06-06 David S. Miller <davem@davemloft.net>
6320 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
6321 blockage if function uses alloca.
6323 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
6325 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
6326 New "uid" fields to hold pretty-print IDs of group and ref.
6327 Memory references are now identified as <group_id>:<ref_id>
6328 instead of using [random] addresses.
6329 (dump_mem_details): Simplify, no functional change.
6330 (dump_mem_ref): Simplify and make output more concise.
6331 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
6332 (find_or_create_group): Initialize group uid.
6333 (record_ref): Initialize ref uid. Improve debug output.
6334 (prune_group_by_reuse, should_issue_prefetch_p,)
6335 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
6336 (mark_nontemporal_store, determine_loop_nest_reuse):
6337 Improve debug output.
6339 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
6341 * dbgcnt.def (prefetch): New debug counter.
6342 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
6343 (schedule_prefetches): Stop issueing prefetches if debug counter
6346 2017-06-06 Tom de Vries <tom@codesourcery.com>
6348 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
6349 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
6351 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6353 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
6354 Use aarch64_reg_or_zero predicate for operand 4.
6355 (aarch64_compare_and_swap<mode> define_insn_and_split):
6356 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
6357 (aarch64_store_exclusive<mode>): Likewise for operand 2.
6359 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
6361 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
6362 (arm_compute_save_core_reg_mask): This.
6363 (thumb1_compute_save_reg_mask): Rename into ...
6364 (thumb1_compute_save_core_reg_mask): This.
6365 (arm_compute_save_reg0_reg12_mask): Adapt comment.
6366 (arm_compute_frame_layout): Likewise.
6368 2017-06-06 Richard Biener <rguenther@suse.de>
6370 PR tree-optimization/80974
6371 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
6372 keep or clear leaders SSA info.
6374 2017-06-06 Tom de Vries <tom@codesourcery.com>
6376 * config/nvptx/nvptx.c (split_mode_p): New function.
6377 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
6379 2017-06-06 Tom de Vries <tom@codesourcery.com>
6381 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
6383 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6386 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
6389 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6391 * shrink-wrap.c (handle_simple_exit): Update profile.
6392 (try_shrink_wrapping): Upate profile.
6394 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6396 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
6397 (tree_guess_outgoing_edge_probabilities): New.
6398 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
6399 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
6401 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6403 * ipa-split.c (split_function): Initialize return bb profile.
6405 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6407 * profile.c (compute_branch_probabilities): Also initialize
6410 2017-06-06 Richard Biener <rguenther@suse.de>
6412 PR tree-optimization/80928
6413 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
6414 (vect_analyze_loop_operations): Properly guard analysis for
6416 (vect_transform_loop): Likewise.
6417 (vect_analyze_loop_2): Also reset SLP type on PHIs.
6418 (vect_model_induction_cost): Do not cost for pure SLP.
6419 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
6420 of induction in inner loop vectorization.
6421 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
6422 (vect_get_and_check_slp_defs): Handle vect_induction_def.
6423 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
6425 (vect_analyze_slp_cost_1): Cost induction.
6426 (vect_detect_hybrid_slp_stmts): Handle PHIs.
6427 (vect_get_slp_vect_defs): Likewise.
6428 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
6429 (vect_transform_stmt): Handle SLP reductions.
6430 * tree-vectorizer.h (vectorizable_induction): Adjust.
6432 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6434 * config/rs6000/rs6000.c (make_resolver_func): Update
6435 init_lowered_empty_function call.
6437 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
6439 * doc/invoke.texi: Document the -fprofile-abs-path option.
6440 * common.opt (fprofile-abs-path): New option.
6441 * gcov-io.h (gcov_write_filename): Declare.
6442 * gcov-io.c (gcov_write_filename): New function.
6443 * coverage.c (coverage_begin_function): Use gcov_write_filename.
6444 * profile.c (output_location): Likewise.
6446 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6448 * shring-wrap.c: Revert accidental commit.
6450 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
6452 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
6454 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6456 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
6458 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
6459 profile in callgraph edge.
6460 * profile-count.h (apply_probability): If THIS is 0, then result is 0
6461 (apply_scale): Likewise.
6462 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
6463 Also scale profile when inlining function with zero profile.
6464 (initialize_cfun): Update exit block profile even when it is zero.
6465 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
6466 when profile is read.
6468 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6470 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
6471 (CLONE_*): New constants to define the processors we can generate
6472 code for with the target_clone attribute.
6473 (rs6000_clone_map): New array to identify which clone processors
6474 the current program is running on.
6475 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
6476 target_clone attribute.
6477 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
6478 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
6479 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
6480 (cpu_expand_builtin): Add support for target_clone attribute.
6481 (rs6000_valid_attribute_p): Allow "default" attribute.
6482 (get_decl_name): New debug function to simplify printing the
6483 current function name in debugging statements.
6484 (rs6000_clone_priority): New functions to support the target_clone
6485 attribute, and be able to generate code to switch between ISA 2.05
6486 through ISA 3.0 (power6 through power9).
6487 (rs6000_compare_version_priority): Likewise.
6488 (rs6000_get_function_versions_dispatcher): Likewise.
6489 (make_resolver_func): Likewise.
6490 (add_condition_to_bb): Likewise.
6491 (dispatch_function_versions): Likewise.
6492 (rs6000_generate_version_dispatcher_body): Likewise.
6493 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
6494 (fusion_gpr_load_p): Fix a spacing issue.
6495 * doc/extend.texi (Common Function Attributes): Document that the
6496 PowerPC supports the target_clone attribute.
6498 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
6500 * config/arm/arm.h: explain F symbol found in description of ARM
6501 register allocation in its legend.
6503 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6505 * config/mips/frame-header-opt.c: Include profile-count.h.
6506 * config/riscv/riscv.c: Include profile-count.h
6508 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6510 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
6512 (sm_set_flag_if_changed): Add bbs field.
6513 (execute_sm_if_changed_flag_set): Pass BBS.
6514 (execute_sm): Update.
6516 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6518 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6521 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6523 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
6524 (peephole2): New peephole2 to emit the above.
6525 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
6527 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6529 * config/aarch64/aarch64.c (define_peephole2 above
6530 *sub_<shift>_<mode>): New peephole.
6532 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
6534 * config/i386/i386.c (make_resolver_func): Update.
6535 * Makefile.in: Add profile-count.h and profile-count.o
6536 * auto-profile.c (afdo_indirect_call): Update to new API.
6537 (afdo_set_bb_count): Update.
6538 (afdo_propagate_edge): Update.
6539 (afdo_propagate_circuit): Update.
6540 (afdo_calculate_branch_prob): Update.
6541 (afdo_annotate_cfg): Update.
6542 * basic-block.h: Include profile-count.h
6543 (struct edge_def): Turn count to profile_count.
6544 (struct basic_block_def): Likewie.
6545 (REG_BR_PROB_BASE): Move to profile-count.h
6546 (RDIV): Move to profile-count.h
6547 * bb-reorder.c (max_entry_count): Turn to profile_count.
6548 (find_traces): Update.
6549 (rotate_loop):Update.
6550 (connect_traces):Update.
6551 (sanitize_hot_paths):Update.
6552 * bt-load.c (migrate_btr_defs): Update.
6553 * cfg.c (RDIV): Remove.
6554 (init_flow): Use alloc_block.
6555 (alloc_block): Uninitialize count.
6556 (unchecked_make_edge): Uninitialize count.
6557 (check_bb_profile): Update.
6558 (dump_edge_info): Update.
6559 (dump_bb_info): Update.
6560 (update_bb_profile_for_threading): Update.
6561 (scale_bbs_frequencies_int): Update.
6562 (scale_bbs_frequencies_gcov_type): Update.
6563 (scale_bbs_frequencies_profile_count): New.
6564 * cfg.h (update_bb_profile_for_threading): Update.
6565 (scale_bbs_frequencies_profile_count): Declare.
6566 * cfgbuild.c (compute_outgoing_frequencies): Update.
6567 (find_many_sub_basic_blocks): Update.
6568 * cfgcleanup.c (try_forward_edges): Update.
6569 (try_crossjump_to_edge): Update.
6570 * cfgexpand.c (expand_gimple_tailcall): Update.
6571 (construct_exit_block): Update.
6572 * cfghooks.c (verify_flow_info): Update.
6573 (dump_bb_for_graph): Update.
6574 (split_edge): Update.
6575 (make_forwarder_block): Update.
6576 (duplicate_block): Update.
6577 (account_profile_record): Update.
6578 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
6579 (get_estimated_loop_iterations): Update.
6580 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
6581 (single_likely_exit): Update.
6582 * cfgloopmanip.c (scale_loop_profile): Update.
6584 (set_zero_probability): Update.
6585 (lv_adjust_loop_entry_edge): Update.
6586 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
6587 (purge_dead_edges): Update.
6588 (rtl_account_profile_record): Update.
6589 * cgraph.c (cgraph_node::create): Uninitialize count.
6590 (symbol_table::create_edge): Uninitialize count.
6591 (cgraph_update_edges_for_call_stmt_node): Update.
6592 (cgraph_edge::dump_edge_flags): Update.
6593 (cgraph_node::dump): Update.
6594 (cgraph_edge::maybe_hot_p): Update.
6595 * cgraph.h: Include profile-count.h
6596 (create_clone), create_edge, create_indirect_edge): Update.
6597 (cgraph_node): Turn count to profile_count.
6598 (cgraph_edge0: Likewise.
6599 (make_speculative, clone): Update.
6600 (create_edge): Update.
6601 (init_lowered_empty_function): Update.
6602 * cgraphclones.c (cgraph_edge::clone): Update.
6603 (duplicate_thunk_for_node): Update.
6604 (cgraph_node::create_clone): Update.
6605 * cgraphunit.c (cgraph_node::analyze): Update.
6606 (cgraph_node::expand_thunk): Update.
6607 * final.c (dump_basic_block_info): Update.
6608 * gimple-streamer-in.c (input_bb): Update.
6609 * gimple-streamer-out.c (output_bb): Update.
6610 * graphite.c (print_global_statistics): Update.
6611 (print_graphite_scop_statistics): Update.
6612 * hsa-brig.c: Include basic-block.h.
6613 * hsa-dump.c: Include basic-block.h.
6614 * hsa-gen.c (T sum_slice): Update.
6615 (convert_switch_statements):Update.
6616 * hsa-regalloc.c: Include basic-block.h.
6617 * ipa-chkp.c (chkp_produce_thunks): Update.
6618 * ipa-cp.c (struct caller_statistics): Update.
6619 (init_caller_stats): Update.
6620 (gather_caller_stats): Update.
6621 (ipcp_cloning_candidate_p): Update.
6622 (good_cloning_opportunity_p): Update.
6623 (get_info_about_necessary_edges): Update.
6624 (dump_profile_updates): Update.
6625 (update_profiling_info): Update.
6626 (update_specialized_profile): Update.
6627 (perhaps_add_new_callers): Update.
6628 (decide_about_value): Update.
6629 (ipa_cp_c_finalize): Update.
6630 * ipa-devirt.c (struct odr_type_warn_count): Update.
6631 (struct decl_warn_count): Update.
6632 (struct final_warning_record): Update.
6633 (possible_polymorphic_call_targets): Update.
6634 (ipa_devirt): Update.
6635 * ipa-fnsummary.c (redirect_to_unreachable): Update.
6636 * ipa-icf.c (sem_function::merge): Update.
6637 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
6638 * ipa-inline.c (compute_uninlined_call_time): Update.
6639 (compute_inlined_call_time): Update.
6640 (want_inline_small_function_p): Update.
6641 (want_inline_self_recursive_call_p): Update.
6642 (edge_badness): Update.
6643 (lookup_recursive_calls): Update.
6644 (recursive_inlining): Update.
6645 (inline_small_functions): Update.
6646 (dump_overall_stats): Update.
6647 (dump_inline_stats): Update.
6648 * ipa-profile.c (ipa_profile_generate_summary): Update.
6649 (ipa_propagate_frequency): Update.
6650 (ipa_profile): Update.
6651 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
6652 * ipa-utils.c (ipa_merge_profiles): Update.
6653 * loop-doloop.c (doloop_modify): Update.
6654 * loop-unroll.c (report_unroll): Update.
6655 (unroll_loop_runtime_iterations): Update.
6656 * lto-cgraph.c (lto_output_edge): Update.
6657 (lto_output_node): Update.
6658 (input_node): Update.
6659 (input_edge): Update.
6660 (merge_profile_summaries): Update.
6661 * lto-streamer-in.c (input_cfg): Update.
6662 * lto-streamer-out.c (output_cfg): Update.
6663 * mcf.c (create_fixup_graph): Update.
6664 (adjust_cfg_counts): Update.
6665 (sum_edge_counts): Update.
6666 * modulo-sched.c (sms_schedule): Update.
6667 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
6668 * predict.c (maybe_hot_count_p): Update.
6669 (probably_never_executed): Update.
6670 (dump_prediction): Update.
6671 (combine_predictions_for_bb): Update.
6672 (propagate_freq): Update.
6673 (handle_missing_profiles): Update.
6674 (counts_to_freqs): Update.
6675 (rebuild_frequencies): Update.
6676 (force_edge_cold): Update.
6677 * predict.h: Include profile-count.h
6678 (maybe_hot_count_p, counts_to_freqs): UPdate.
6679 * print-rtl-function.c: Do not include cfg.h
6680 * print-rtl.c: Include basic-block.h
6681 * profile-count.c: New file.
6682 * profile-count.h: New file.
6683 * profile.c (is_edge_inconsistent): Update.
6684 (correct_negative_edge_counts): Update.
6685 (is_inconsistent): Update.
6686 (set_bb_counts): Update.
6687 (read_profile_edge_counts): Update.
6688 (compute_frequency_overlap): Update.
6689 (compute_branch_probabilities): Update; Initialize and deinitialize
6691 (branch_prob): Update.
6692 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
6693 (edge_gcov_count): New.
6694 (bb_gcov_count): New.
6695 * shrink-wrap.c (try_shrink_wrapping): Update.
6696 * tracer.c (better_p): Update.
6697 * trans-mem.c (expand_transaction): Update.
6698 (ipa_tm_insert_irr_call): Update.
6699 (ipa_tm_insert_gettmclone_call): Update.
6700 * tree-call-cdce.c: Update.
6701 * tree-cfg.c (gimple_duplicate_sese_region): Update.
6702 (gimple_duplicate_sese_tail): Update.
6703 (gimple_account_profile_record): Update.
6704 (execute_fixup_cfg): Update.
6705 * tree-inline.c (copy_bb): Update.
6706 (copy_edges_for_bb): Update.
6707 (initialize_cfun): Update.
6708 (freqs_to_counts): Update.
6709 (copy_cfg_body): Update.
6710 (expand_call_inline): Update.
6711 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
6712 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
6713 (try_unroll_loop_completely): Update.
6714 (try_peel_loop): Update.
6715 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
6716 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
6717 * tree-ssa-loop-split.c (connect_loops): Update.
6718 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
6719 * tree-ssa-reassoc.c (branch_fixup): Update.
6720 * tree-ssa-tail-merge.c (replace_block_by): Update.
6721 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
6722 (compute_path_counts): Update.
6723 (update_profile): Update.
6724 (recompute_probabilities): Update.
6725 (update_joiner_offpath_counts): Update.
6726 (estimated_freqs_path): Update.
6727 (freqs_to_counts_path): Update.
6728 (clear_counts_path): Update.
6729 (ssa_fix_duplicate_block_edges): Update.
6730 (duplicate_thread_path): Update.
6731 * tree-switch-conversion.c (case_bit_test_cmp): Update.
6732 (struct switch_conv_info): Update.
6733 * tree-tailcall.c (decrease_profile): Update.
6734 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
6735 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
6736 * value-prof.c (check_counter): Update.
6737 (gimple_divmod_fixed_value): Update.
6738 (gimple_mod_pow2): Update.
6739 (gimple_mod_subtract): Update.
6740 (gimple_ic_transform): Update.
6741 (gimple_stringop_fixed_value): Update.
6742 * value-prof.h (gimple_ic): Update.
6744 2017-06-02 Carl Love <cel@us.ibm.com>
6746 * config/rs6000/rs6000-c: Add support for built-in functions
6747 vector double vec_doublee (vector signed int);
6748 vector double vec_doublee (vector unsigned int);
6749 vector double vec_doublee (vector float);
6750 vector double vec_doubleh (vector signed int);
6751 vector double vec_doubleh (vector unsigned int);
6752 vector double vec_doubleh (vector float);
6753 vector double vec_doublel (vector signed int);
6754 vector double vec_doublel (vector unsigned int);
6755 vector double vec_doublel (vector float);
6756 vector double vec_doubleo (vector signed int);
6757 vector double vec_doubleo (vector unsigned int);
6758 vector double vec_doubleo (vector float);.
6759 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
6760 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
6762 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
6763 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
6764 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
6766 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
6767 vec_doublel, vec_doubleh.
6768 * doc/extend.texi: Update the built-in documentation file for the
6769 new built-in functions.
6771 2017-06-02 David Malcolm <dmalcolm@redhat.com>
6774 * ipa-inline-analysis.c (free_growth_caches): Set
6775 edge_removal_hook_holder to NULL after removing it.
6777 2017-06-02 Sudakshina Das <sudi.das@arm.com>
6779 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
6780 comparision with zero.
6782 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6783 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6784 for early expansion of vec_min and vec_max builtins.
6785 (builtin_function_type): Add min/max unsigned variants to those
6786 identified as having unsigned arguments.
6788 2017-06-02 Olivier Hainque <hainque@adacore.com>
6790 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
6792 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6794 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
6795 Use VALL_F16 iterator rather than VALL.
6797 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6799 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
6800 Emit CBNZ inside loop when doing a strong exchange and comparing
6801 against zero. Generate the CC flags after the loop.
6803 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
6805 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
6806 (dl_section_ref): New.
6807 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
6808 On AIX, append an expression to subtract the size of the
6809 section length to dl_section_ref.
6811 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
6813 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6814 for early expansion of vector absolute builtins.
6816 2017-06-02 Richard Biener <rguenther@suse.de>
6818 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
6819 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
6821 2017-06-02 Richard Biener <rguenther@suse.de>
6823 PR tree-optimization/80948
6824 * tree-tailcall.c (find_tail_calls): Track stmts to move in
6827 2017-06-02 Richard Biener <rguenther@suse.de>
6829 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
6831 * tree-vect-stmts.c (process_use): Do not mark backedge defs
6832 for inductions as relevant.
6834 2017-06-02 Richard Biener <rguenther@suse.de>
6836 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
6837 (vectorizable_induction): ... this. Remove dead code.
6839 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
6841 * builtins. (expand_builtin_alloca): Remove second parameter and
6842 infer its value from the first parameter instead.
6843 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
6845 2017-06-02 Jakub Jelinek <jakub@redhat.com>
6847 PR rtl-optimization/80903
6848 * loop-doloop.c (add_test): Unshare sequence.
6850 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
6852 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
6854 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
6856 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
6858 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
6859 xlogue_layout::get_instance, logue_layout::xlogue_layout,
6860 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
6861 (xlogue_layout::get_stub_rtx): Make static.
6862 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
6863 (xlogue_layout::compute_stub_managed_regs): Rename to...
6864 (xlogue_layout::count_stub_managed_regs): ...this.
6865 (xlogue_layout::is_stub_managed_reg): New function.
6866 (xlogue_layout::m_stub_names): Rename to...
6867 (xlogue_layout::s_stub_names): ...this, make static.
6868 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
6869 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
6870 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
6871 xlogue_layout::s_stub_names): Instantiate statics.
6872 (stub_managed_regs): Remove.
6873 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
6874 (disable_call_ms2sysv_xlogues): Rename to...
6875 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
6876 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
6878 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
6879 change after reload_completed.
6880 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
6882 (ix86_expand_prologue): Likewise.
6883 (ix86_expand_epilogue): Likewise.
6884 (ix86_expand_split_stack_prologue): Likewise.
6885 (ix86_compute_frame_layout): Remove frame parameter ...
6886 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
6887 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
6889 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
6890 (ix86_frame): Move from here ...
6891 * config/i386/i386.h (ix86_frame): ... to here.
6892 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
6893 complete ix86_frame data structure instead. Remove some_ld_name.
6895 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
6897 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
6898 symbols that hold a DECL_VALUE_EXPR.
6900 2017-06-01 Martin Jambor <mjambor@suse.cz>
6902 PR tree-optimization/80898
6903 * tree-sra.c (process_subtree_disqualification): Removed.
6904 (disqualify_candidate): Do not acll
6905 process_subtree_disqualification.
6906 (subtree_mark_written_and_enqueue): New function.
6907 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
6908 RHS has been disqualified and re-queue LHS if necessary. Apart
6909 from that, ignore disqualified RHS.
6911 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6913 * config/s390/s390.c (s390_emit_epilogue): Disable early return
6914 address fetch for z10 or later.
6916 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6918 * config/arc/arc.md (tst_movb): Add guard when splitting.
6920 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6922 * config/arc/arc.c (arc_can_eliminate): Test against
6923 arc_frame_pointer_needed.
6925 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6927 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
6928 to prevent store reordering.
6929 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
6930 (type): Add block type.
6931 (stack_tie): Define special instruction to be used in
6934 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6936 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
6937 constraint. It is not valid for the pattern.
6938 (noncommutative_binary_comparison): Likewise.
6940 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6942 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
6945 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6947 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
6948 be used by the reg-alloc.
6950 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6952 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
6953 reg-alloc when having mul64 or mul32x16 instructions.
6954 (mulsidi3): Likewise.
6955 (umulsidi3): Likewise.
6956 (mulsi32x16): New pattern.
6957 (mulsi64): Likewise.
6958 (mulsidi64): Likewise.
6959 (umulsidi64): Likewise.
6960 (MUL32x16_REG): Define.
6961 (mul64_600): Use MUL32x16_REG.
6962 (mac64_600): Likewise.
6963 (umul64_600): Likewise.
6964 (umac64_600): Likewise.
6966 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
6968 * config/arc/arc.md (mulsi3_700): Make it commutative.
6970 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
6972 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
6974 (*sign_extendsidi2_insn): Likewise for movstosw.
6976 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
6978 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
6979 the type of the input discriminant value. Convert the
6980 discriminant value of signedness vary.
6982 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
6984 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
6985 Add to -Wall section.
6987 2017-06-01 Richard Biener <rguenther@suse.de>
6990 * fold-const.c (fold_plusminus_mult_expr): If the factored
6991 factor may be zero use a wrapping type for the inner operation.
6992 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
6993 and handle moved defs.
6994 (process_assignment): Properly guard the unary op case. Return a
6995 tri-state indicating that moving the stmt before the call may allow
6996 to continue. Pass through to_move.
6997 (find_tail_calls): Handle moving unrelated defs before
7000 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
7003 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
7004 splitter result in the canonical way.
7006 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
7008 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
7009 also for 32bit target. Update insn attributes.
7010 (zero-extendsidi2 splitter): Allow all registers for operand 1.
7012 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
7014 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
7015 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
7016 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
7017 (_mm_maskz_min_ss): New intrinsics.
7019 2017-05-31 Martin Liska <mliska@suse.cz>
7021 * tree-vect-loop.c (vect_create_epilog_for_reduction):
7022 Change comment style to one we normally use.
7023 (vectorizable_reduction): Likewise.
7024 (vectorizable_induction): Likewise.
7025 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7026 (vectorizable_call): Likewise.
7027 (vectorizable_simd_clone_call): Likewise.
7028 (vectorizable_conversion): Likewise.
7029 (vectorizable_assignment): Likewise.
7030 (vectorizable_shift): Likewise.
7031 (vectorizable_operation): Likewise.
7032 (vectorizable_store): Likewise.
7033 (vectorizable_load): Likewise.
7034 * tree-vectorizer.h: Likewise.
7036 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
7038 * passes.c (emergency_dump_function): New.
7039 * tree-pass.h (emergency_dump_function): Declare.
7040 * plugin.c (plugins_internal_error_function): Remove.
7041 * plugin.h (plugins_internal_error_function): Remove declaration.
7042 * toplev.c (internal_error_function): New static function. Use it...
7043 (general_init): ...here.
7045 2017-05-31 Graham Markall <graham.markall@embecosm.com>
7047 * config/arc/arc.c (arc_print_operand): Handle constant operands.
7048 (arc_rtx_costs): Add costs for new patterns.
7049 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
7050 * config/arc/predicates.md: Add _1_2_3_operand predicate.
7052 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
7054 * tree-ssa-strlen.c (get_next_strinfo): New function.
7055 (get_stridx_plus_constant): Use it.
7056 (zero_length_string): Likewise.
7057 (adjust_related_strinfos): Likewise.
7058 (adjust_last_stmt): Likewise.
7060 2017-05-31 Richard Biener <rguenther@suse.de>
7063 * config/i386/i386.c (ix86_expand_builtin): Remove assert
7064 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
7066 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
7068 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
7069 loop_vinfo argument and use of dependence distance vectors.
7070 Check instead whether the two references differ only in their
7071 initial value and assume that they have the same alignment if the
7072 difference is a multiple of the vector alignment.
7073 (vect_analyze_data_refs_alignment): Update call accordingly.
7075 2017-05-31 Martin Liska <mliska@suse.cz>
7078 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
7080 2017-05-31 Bin Cheng <bin.cheng@arm.com>
7082 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
7083 (create_intersect_range_checks): Move from ...
7084 * tree-data-ref.c (create_intersect_range_checks_index)
7085 (create_intersect_range_checks): ... to here.
7086 (create_runtime_alias_checks): New function factored from ...
7087 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
7088 here. Call above function.
7089 * tree-data-ref.h (create_runtime_alias_checks): New function.
7091 2017-05-31 Bin Cheng <bin.cheng@arm.com>
7093 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
7094 segment length for dr_b and compute it in wide_int.
7096 2017-05-31 Richard Biener <rguenther@suse.de>
7098 PR tree-optimization/80906
7099 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
7100 and pass through iv_map.
7101 (copy_bb_and_scalar_dependences): Adjust.
7102 (translate_pending_phi_nodes): Likewise.
7103 (copy_loop_close_phi_args): Handle code-generating IVs instead
7106 2017-05-30 David Malcolm <dmalcolm@redhat.com>
7108 * diagnostic-color.c (color_dict): Add "type-diff".
7109 (parse_gcc_colors): Update comment.
7110 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
7111 -fdiagnostics-show-template-tree and -fno-elide-type.
7112 (GCC_COLORS): Add type-diff to example.
7114 (-fdiagnostics-show-template-tree): New.
7115 (-fno-elide-type): New.
7116 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
7117 the pp_format_decoder callback. Call any m_format_postprocessor's
7119 (pretty_printer::pretty_printer): Initialize
7120 m_format_postprocessor.
7121 (pretty_printer::~pretty_printer): Delete any
7122 m_format_postprocessor.
7123 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
7124 (class format_postprocessor): New class.
7125 (struct pretty_printer::format_decoder): Document the new parameters.
7126 (struct pretty_printer::m_format_postprocessor): New field.
7127 * tree-diagnostic.c (default_tree_printer): Update for new
7128 bool and const char ** params.
7129 * tree-diagnostic.h (default_tree_printer): Likewise.
7131 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
7133 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
7134 (lwa_operand): Delete rs6000_gen_cell_microcode test.
7135 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
7136 rs6000_gen_cell_microcode code.
7137 (rs6000_final_prescan_insn): Delete.
7138 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
7139 "warn-cell-microcode" entries.
7140 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
7141 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
7142 throughout. Change cc_reg_not_micro_cr0_operand to
7143 cc_reg_not_cr0_operand throughout.
7144 (*extendhi<mode>2_noload): Delete.
7145 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
7146 (mwarn-cell-microcode): Delete.
7147 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
7148 -mgen-cell-microcode and -mwarn-cell-microcode.
7150 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
7153 * config/i386/constraints.md (Yd): New constraint.
7155 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
7156 and (?Yd, r) alternatives. Update insn attributes.
7157 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
7158 and (?*Yd, r) alternatives. Update insn attributes.
7159 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
7161 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
7163 * gimplify.c (gimplify_modify_expr): Don't create a
7164 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
7167 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
7169 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
7171 2017-05-30 Richard Biener <rguenther@suse.de>
7173 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
7174 and reduc_def fields.
7175 (STMT_VINFO_REDUC_TYPE): New define.
7176 (STMT_VINFO_REDUC_DEF): Likewise.
7177 (vect_force_simple_reduction): Adjust prototype.
7178 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
7179 (vect_is_simple_reduction): Remove check_reduction argument.
7180 (vect_force_simple_reduction): Adjust and set
7181 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
7182 (vectorizable_reduction): Do not re-do reduction analysis
7183 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
7184 * tree-parloops.c (gather_scalar_reductions): Adjust.
7186 2017-05-30 Richard Biener <rguenther@suse.de>
7189 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
7192 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
7194 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
7195 Introduce unknown_misalignment parameter and remove vf.
7196 (vect_peeling_hash_get_lowest_cost):
7197 Pass unknown_misalignment parameter.
7198 (vect_enhance_data_refs_alignment):
7199 Fix unsupportable data ref treatment.
7201 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7203 * tree-vect-data-refs.c (vect_get_data_access_cost):
7204 Workaround for SLP handling.
7205 (vect_enhance_data_refs_alignment):
7206 Compute costs for doing no peeling at all, compare to the best
7207 peeling costs so far and avoid peeling if cheaper.
7209 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7211 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
7212 Return peeling info and set costs to zero for unlimited cost
7214 (vect_enhance_data_refs_alignment): Also inspect all datarefs
7215 with unknown misalignment. Compute and costs for unknown
7216 misalignment, compare them to the costs for known misalignment
7217 and choose the cheapest for peeling.
7219 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7221 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
7222 (vect_get_peeling_costs_all_drs): Create function.
7223 (vect_peeling_hash_get_lowest_cost):
7224 Use vect_get_peeling_costs_all_drs.
7225 (vect_peeling_supportable): Create function.
7226 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
7228 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7230 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
7231 DR_HAS_NEGATIVE_STEP.
7232 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
7233 (vect_enhance_data_refs_alignment): Use.
7234 (vect_duplicate_ssa_name_ptr_info): Use.
7235 * tree-vectorizer.h (dr_misalignment): Use.
7236 (known_alignment_for_access_p): Use.
7238 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7241 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
7243 (has_section_name): New function.
7245 2017-05-30 Martin Liska <mliska@suse.cz>
7248 * auto-profile.c (get_function_decl_from_block): Fix
7251 2017-05-30 Richard Biener <rguenther@suse.de>
7254 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
7256 2017-05-30 Martin Liska <mliska@suse.cz>
7258 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
7259 * dumpfile.h (struct dump_file_info): Remove ctors.
7261 2017-05-30 Martin Liska <mliska@suse.cz>
7263 * predict.def: Fix GNU coding style.
7265 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
7267 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
7268 Mark 'to' argument with ATTRIBUTE_UNUSED.
7270 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
7272 * config/xtensa/xtensa.c (xtensa_emit_call): Use
7273 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
7274 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
7277 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
7279 * doc/install.texi (Options specification): Restore entry of
7280 --enable-sjlj-exceptions.
7282 2017-05-27 Michael Eager <eager@eagercon.com>
7285 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
7287 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
7289 * config/microblaze/microblaze.h
7290 (FIXED_REGISTERS): Update in macro.
7291 (CALL_USED_REGISTERS): Update in macro.
7293 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
7295 * doc/install.texi: Add links to macOS binary distributions.
7297 2017-05-27 Jakub Jelinek <jakub@redhat.com>
7301 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7303 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
7305 2017-05-26 Martin Liska <mliska@suse.cz>
7307 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
7309 2017-05-26 Martin Liska <mliska@suse.cz>
7311 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
7312 always leading ';; '.
7313 (dump_bb_info): Likewise.
7314 (brief_dump_cfg): Likewise.
7315 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
7316 * dumpfile.c: Remove usage of TDF_VERBOSE.
7317 * dumpfile.h (enum dump_kind): Likewise.
7318 (dump_gimple_bb_header): Do not use TDF_COMMENT.
7319 * print-tree.c (debug_verbose): Remove.
7320 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
7321 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
7322 * tree-diagnostic.c (default_tree_printer): Replace
7323 TDF_DIAGNOSTIC with TDF_SLIM.
7325 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7327 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
7328 in parameter loop, rather than loop_vinfo.
7329 (create_intersect_range_checks): Ditto.
7330 (vect_create_cond_for_alias_checks): Update call to above functions.
7332 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7334 PR tree-optimization/80815
7335 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
7336 for merging runtime alias checks. Handle negative DR_STEPs.
7338 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7340 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
7342 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
7343 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
7344 out code pruning runtime alias checks.
7345 * tree-data-ref.c (prune_runtime_alias_test_list): New function
7346 factored out from above.
7347 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
7349 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
7351 (prune_runtime_alias_test_list): New decalaration.
7353 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7355 * tree-vect-data-refs.c (compare_tree): Rename and move ...
7356 * tree-data-ref.c (data_ref_compare_tree): ... to here.
7357 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
7358 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
7359 (operator==, comp_dr_with_seg_len_pair): Ditto.
7360 (vect_prune_runtime_alias_test_list): Ditto.
7362 2017-05-26 Martin Liska <mliska@suse.cz>
7365 * params.def: Bound partial-inlining-entry-probability param.
7367 2017-05-26 Marek Polacek <polacek@redhat.com>
7370 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
7373 2017-05-26 Richard Biener <rguenther@suse.de>
7375 PR tree-optimization/80842
7376 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
7379 2017-05-26 Richard Biener <rguenther@suse.de>
7381 PR tree-optimization/80844
7382 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
7384 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
7386 * doc/md.texi (Machine Constraints): Update x86 family
7387 machine constraints section to match 'config/i386/constraints.md'.
7389 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
7391 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
7393 2017-05-25 Nathan Sidwell <nathan@acm.org>
7395 * doc/invoke.texi (--enable-languages): Update documentation.
7397 2017-05-25 Martin Liska <mliska@suse.cz>
7399 * dumpfile.c: Add TDF_FOLDING.
7400 * dumpfile.h (enum dump_kind): Likewise.
7401 * genmatch.c (dt_simplify::gen_1): Use it.
7403 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7405 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
7407 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7409 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
7410 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
7412 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7414 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
7415 * match.pd (X == C): Rewrite it here.
7416 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
7417 with_certain_nonzero_bits2): New predicates.
7418 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
7420 2017-05-24 Nathan Sidwell <nathan@acm.org>
7422 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
7425 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
7428 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
7430 * config/powerpcspe: New port. Files are copied from the rs6000
7431 port, with "rs6000" in filenames replaced by "powerpcspe".
7433 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
7435 PR rtl-optimization/80754
7436 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
7438 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
7440 * config/sparc/sparc.md (length): Return the correct value for -mflat
7441 sibcalls to match output_sibcall.
7443 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
7447 * config/rs6000/rs6000.c (struct machine_function): Add new field
7449 (rs6000_get_separate_components): Init that field, use it.
7450 (rs6000_components_for_bb): Use the field.
7452 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
7454 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
7456 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
7459 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
7461 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7464 * config/s390/s390.c (s390_check_qrst_address): Check incoming
7465 address against address_operand predicate.
7466 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
7468 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
7470 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
7471 parameters passed indirectly.
7473 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
7475 * config/i386/i386.md (*movdi_internal): Remove SSE4
7476 alternative 18 (?r, *v). Update insn attributes.
7477 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
7478 Update insn attributes.
7479 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
7480 Update insn attributes.
7481 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
7482 alternative 1 (r, v). Remove isa attribute.
7483 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
7484 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
7485 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
7487 2017-05-23 Tom de Vries <tom@codesourcery.com>
7489 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
7492 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7494 * cgraphunit.c (symbol_table::process_new_functions): Update.
7495 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
7496 (inline_generate_summary): Rename to ...
7497 (ipa_fn_summary_generate): ... this one.
7498 (inline_read_summary): Rename to ...
7499 (ipa_fn_summary_read): ... this one.
7500 (inline_write_summary): Rename to ...
7501 (ipa_fn_summary_write): ... this one.
7502 (inline_free_summary): Rename to ...
7503 (ipa_free_fn_summary): ... this one.
7504 (pass_data_local_fn_summary, pass_local_fn_summary,
7505 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
7506 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
7507 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
7508 make_pass_ipa_fn_summary): New.
7509 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
7510 inline_write_summary, inline_free_summary): Remove.
7511 (ipa_free_fn_summary) : New.
7512 * ipa-inline.c (ipa_inline): Update.
7513 (pass_ipa_inline): Do not generate summaries.
7514 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
7516 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
7517 and add pass_ipa_fn_summary.
7518 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
7520 (make_pass_inline_parameters): Remove.
7522 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
7524 * omp-low.c (struct omp_context): Remove "default_kind" member.
7527 * omp-offload.c (execute_oacc_device_lower): Remove the
7528 parallelism dimensions function attributes for unparallelized
7529 OpenACC kernels constructs.
7531 2017-05-23 Martin Liska <mliska@suse.cz>
7533 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
7535 (cgraph_edge::make_speculative): Likewise.
7536 (cgraph_edge::resolve_speculation): Likewise.
7537 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
7538 (cgraph_node::dump): Likewise.
7539 * cgraph.h: Likewise.
7540 * cgraphunit.c (analyze_functions): Likewise.
7541 (symbol_table::compile): Likewise.
7542 * ipa-cp.c (print_all_lattices): Likewise.
7543 (determine_versionability): Likewise.
7544 (initialize_node_lattices): Likewise.
7545 (ipcp_verify_propagated_values): Likewise.
7546 (estimate_local_effects): Likewise.
7547 (update_profiling_info): Likewise.
7548 (create_specialized_node): Likewise.
7549 (perhaps_add_new_callers): Likewise.
7550 (decide_about_value): Likewise.
7551 (decide_whether_version_node): Likewise.
7552 (identify_dead_nodes): Likewise.
7553 (ipcp_store_bits_results): Likewise.
7554 * ipa-devirt.c (dump_targets): Likewise.
7555 (ipa_devirt): Likewise.
7556 * ipa-icf.c (sem_item::dump): Likewise.
7557 (sem_function::equals): Likewise.
7558 (sem_variable::equals): Likewise.
7559 (sem_item_optimizer::read_section): Likewise.
7560 (sem_item_optimizer::execute): Likewise.
7561 (congruence_class::dump): Likewise.
7562 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
7563 (dump_inline_summary): Likewise.
7564 (estimate_node_size_and_time): Likewise.
7565 (inline_analyze_function): Likewise.
7566 * ipa-inline-transform.c (inline_call): Likewise.
7567 * ipa-inline.c (report_inline_failed_reason): Likewise.
7568 (want_early_inline_function_p): Likewise.
7569 (edge_badness): Likewise.
7570 (update_edge_key): Likewise.
7571 (inline_small_functions): Likewise.
7572 * ipa-profile.c (ipa_profile): Likewise.
7573 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
7574 (ipa_make_edge_direct_to_target): Likewise.
7575 (remove_described_reference): Likewise.
7576 (ipa_impossible_devirt_target): Likewise.
7577 (propagate_controlled_uses): Likewise.
7578 (ipa_print_node_params): Likewise.
7579 (ipcp_transform_function): Likewise.
7580 * ipa-pure-const.c (pure_const_read_summary): Likewise.
7581 (propagate_pure_const): Likewise.
7582 * ipa-reference.c (generate_summary): Likewise.
7583 (read_write_all_from_decl): Likewise.
7584 (propagate): Likewise.
7585 (ipa_reference_read_optimization_summary): Likewise.
7586 * ipa-utils.c (ipa_merge_profiles): Likewise.
7587 * ipa.c (walk_polymorphic_call_targets): Likewise.
7588 (symbol_table::remove_unreachable_nodes): Likewise.
7589 (ipa_single_use): Likewise.
7590 * passes.c (execute_todo): Likewise.
7591 * predict.c (drop_profile): Likewise.
7592 * symtab.c (symtab_node::get_dump_name): New function.
7593 (symtab_node::dump_name): Likewise.
7594 (symtab_node::dump_asm_name): Likewise.
7595 (symtab_node::dump_references): Likewise.
7596 (symtab_node::dump_referring): Likewise.
7597 (symtab_node::dump_base): Likewise.
7598 (symtab_node::debug_symtab): Likewise.
7599 * tree-sra.c (convert_callers_for_node): Likewise.
7600 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
7601 * value-prof.c (init_node_map): Likewise.
7603 2017-05-23 Martin Liska <mliska@suse.cz>
7605 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
7606 and symtab_node::debug_symtab to symbol_table::debug.
7607 * cgraphunit.c (analyze_functions): Use the renamed function.
7608 (symbol_table::compile): Likewise.
7609 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
7610 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
7611 * passes.c (execute_todo): Likewise.
7612 * symtab.c (symbol_table::dump): New function.
7613 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
7615 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7617 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
7618 that nonconst implies exec.
7620 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7622 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7623 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7624 (inline_edge_summary_vec): Turn into ...
7625 (ipa_call_summaries): ... this one.
7626 (redirect_to_unreachable, edge_set_predicate,
7627 evaluate_properties_for_edge, inline_summary_alloc,
7628 reset_ipa_call_summary, reset_inline_summary,
7629 inline_summary_t::duplicate): Update.
7630 (inline_edge_duplication_hook): Turn to ...
7631 (ipa_call_summary_t::duplicate): ... this one.
7632 (inline_edge_removal_hook): Turn to ...
7633 (ipa_call_summary_t::remove): ... this one.
7634 (dump_inline_edge_summary): Turn to ...
7635 (dump_ipa_call_summary): ... this one.
7636 (estimate_function_body_sizes): Update.
7637 (inline_update_callee_summaries): Update.
7638 (remap_edge_change_prob): Update.
7639 (remap_edge_summaries): Update.
7640 (inline_merge_summary): Update.
7641 (do_estimate_edge_time): Update.
7642 (inline_generate_summary): Update.
7643 (inline_read_section): Update.
7644 (inline_read_summary): Update.
7645 (inline_free_summary): Update.
7646 * ipa-inline.c (can_inline_edge_p): Update.
7647 (compute_inlined_call_time): Update.
7648 (want_inline_small_function_p): Update.
7649 (edge_badness): Update.
7650 (early_inliner): Update.
7651 * ipa-inline.h (inline_edge_summary): Turn to ...
7652 (ipa_call_summary): ... this one.
7653 (ipa_call_summary_t): New class.
7654 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
7655 (ipa_call_summaries): New.
7656 (inline_edge_summary): Remove.
7657 (estimate_edge_growth): Update.
7658 * ipa-profile.c (ipa_propagate_frequency_1): Update.
7659 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7660 * ipa-split.c (execute_split_functions): Update.
7661 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
7663 2017-05-23 Tom de Vries <tom@codesourcery.com>
7665 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7666 attributes): Document rdrand effective target.
7668 2017-05-23 Tom de Vries <tom@codesourcery.com>
7670 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
7671 attributes): Sort alphabetically.
7673 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
7675 * config/avr/genmultilib.awk: Use gsub instead of gensub.
7677 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
7680 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
7681 V2DF/V2DI splat into two separate patterns, one that handles
7682 registers, and the other that only handles memory. Drop support
7683 for splatting from a GPR on ISA 2.07 and then splitting the
7684 splat into direct move and splat.
7685 (vsx_splat_<mode>_reg): Likewise.
7686 (vsx_splat_<mode>_mem): Likewise.
7688 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
7690 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
7692 2017-05-22 Jakub Jelinek <jakub@redhat.com>
7695 * omp-low.c (finish_taskreg_remap): New function.
7696 (finish_taskreg_scan): If unit size of ctx->record_type
7697 is non-constant, unshare the size expression and replace
7698 decls in it with possible outer var refs.
7701 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
7702 GOVD_SHARED rather than GOVD_PRIVATE with it.
7703 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
7704 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
7707 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
7708 as last argument to build_outer_var_ref for pointer bases of array
7711 2017-05-19 Martin Sebor <msebor@redhat.com>
7713 * print-tree.c (print_node): Print DECL_READ_P flag.
7715 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7717 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
7718 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
7719 * cgraph.c: Likewise.
7720 * cgraphunit.c: Likewise.
7721 * gengtype.c: Likewise.
7722 * ipa-cp.c: Likewise.
7723 * ipa-devirt.c: Likewise.
7724 * ipa-icf.c: Likewise.
7725 * ipa-predicate.c: Likewise.
7726 * ipa-profile.c: Likewise.
7727 * ipa-prop.c: Likewise.
7728 * ipa-split.c: Likewise.
7730 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
7731 edge_predicate_pool, dump_inline_hints,
7732 inline_summary::account_size_time, redirect_to_unreachable,
7733 edge_set_predicate, set_hint_predicate,
7734 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
7735 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
7736 inline_summary_t::remove, remap_hint_predicate_after_duplication,
7737 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
7738 ipa_call_summary_t::remove, initialize_growth_caches,
7739 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
7740 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
7741 mark_modified, unmodified_parm_1, unmodified_parm,
7742 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
7743 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
7744 compute_bb_predicates, will_be_nonconstant_expr_predicate,
7745 will_be_nonconstant_predicate, record_modified_bb_info,
7746 get_minimal_bb, record_modified, param_change_prob,
7747 phi_result_unknown_predicate, predicate_for_phi_result,
7748 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
7749 estimate_function_body_sizes, compute_inline_parameters,
7750 compute_inline_parameters_for_curren, pass_data_inline_parameters,
7751 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
7752 inline_update_callee_summaries, remap_edge_change_prob,
7753 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
7754 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
7755 inline_analyze_function, inline_summary_t::insert,
7756 inline_generate_summary, read_ipa_call_summary, inline_read_section,
7757 inline_read_summary, write_ipa_call_summary, inline_write_summary,
7758 inline_free_summary): Move to ipa-fnsummary.h
7759 (predicate_t): Remove.
7760 * ipa-fnsummary.c: New file.
7761 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
7762 (enum inline_hints_vals, inline_hints, agg_position_info,
7763 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
7764 inline_summaries, ipa_call_summary, ipa_call_summary_t,
7765 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
7766 dump_inline_summary, dump_inline_hints, inline_generate_summary,
7767 inline_read_summary, inline_write_summary, inline_free_summary,
7768 inline_analyze_function, initialize_inline_failed,
7769 inline_merge_summary, inline_update_overall_summary,
7770 compute_inline_parameters): Move to ipa-fnsummary.h
7771 * ipa-fnsummary.h: New file.
7772 * ipa-inline-transform.h: Include ipa-inline.h.
7773 * ipa-inline.c: LIkewise.
7775 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7777 * ipa-inline.c (edge_badness): Use inlined_time instead of
7778 inline_summaries->get.
7780 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7782 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
7784 2017-05-22 Nathan Sidwell <nathan@acm.org>
7786 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
7787 (fdump-lang): Document 'raw' option.
7788 * dumpfile.h (TDI_tu): Delete.
7789 * dumpfile.c (dump_files): Remove translation-unit.
7790 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
7792 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
7794 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
7795 command option from $(AWK) call.
7796 * config/avr/genmultilib.awk: Simplify and rewrite so that it
7797 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
7798 [FORMAT]: Remove handling of variable.
7799 * config/avr/t-multilib: Regenerate.
7801 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7803 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
7805 (dump_inline_summary): Do not print self_time.
7806 (estimate_function_body_sizes): Do not set self_time.
7807 (compute_inline_parameters): Likewise.
7808 (inline_read_section, inline_write_summary): Do not stream self_time.
7809 * ipa-inline.h (inline_summary): Drop self_time.
7811 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
7813 * ipa-inline-analysis.c (account_size_time): Rename to ...
7814 (inline_summary::account_size_time): ... this one.
7815 (reset_ipa_call_summary): Turn to ...
7816 (ipa_call_summary::reset): ... this one.
7817 (reset_inline_summary): Turn to ...
7818 (inline_summary::reset): ... this one.
7819 (inline_summary_t::remove): Update.
7820 (inline_summary_t::duplicate): Update.
7821 (ipa_call_summary_t::remove): Update.
7822 (dump_inline_summary): Update.
7823 (estimate_function_body_sizes): Update.
7824 (compute_inline_parameters): Update.
7825 (estimate_node_size_and_time): Update.
7826 (inline_merge_summary): Update.
7827 (inline_update_overall_summary): Update.
7828 (inline_read_section): Update.
7829 (inline_write_summary): Update.
7830 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
7831 add account_size_time and reset member functions.
7832 (ipa_call_summary): Add reset function.
7833 * ipa-predicate.h (predicate::operator &): Constify.
7835 2017-05-22 Richard Biener <rguenther@suse.de>
7837 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
7839 2017-05-19 Jason Merrill <jason@redhat.com>
7841 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
7843 2017-05-19 Marek Polacek <polacek@redhat.com>
7846 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
7847 TYPE_OVERFLOW_WRAPS checks.
7849 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
7851 * tree-core.h (enum omp_clause_default_kind): Add
7852 "OMP_CLAUSE_DEFAULT_PRESENT".
7853 * tree-pretty-print.c (dump_omp_clause): Handle it.
7854 * gimplify.c (enum gimplify_omp_var_data): Add
7855 "GOVD_MAP_FORCE_PRESENT".
7856 (gimplify_adjust_omp_clauses_1): Map it to
7857 "GOMP_MAP_FORCE_PRESENT".
7858 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
7860 * gimplify.c (oacc_default_clause): Clarify.
7862 2017-05-19 Nathan Sidwell <nathan@acm.org>
7864 LANG_HOOK_REGISTER_DUMPS
7865 * toplev.c (general_init): Call register dump lang hook.
7866 * doc/invoke.texi: Document -fdump-lang option family.
7867 * dumpfile.c (dump_files): Remove class dump here.
7868 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
7869 * dumpfile.h (tree_dump_index): Remove TDI_class.
7870 * langhooks-def.h (lhd_register_dumps): Declare.
7871 (LANG_HOOKS_REGISTER_DUMPS): Define.
7872 (LANG_HOOKS_INITIALIZER): Add it.
7873 * langhooks.c (lhd_register_dumps): Define.
7874 * langhooks.h (struct lang_hooks): Add register_dumps.
7876 2017-05-19 Nathan Sidwell <nathan@acm.org>
7878 * context.h (context::set_passes): New.
7879 * context.c (context::context): Do not create pass manager.
7880 * toplev.c (general_init): Create pass manager here.
7882 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
7884 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
7885 use this splitter if two add or or instructions would also work for
7886 the constant we want to generate.
7888 2017-05-19 Richard Biener <rguenther@suse.de>
7891 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
7892 predicate evaluation.
7894 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
7896 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
7898 * ipa-inline.c (want_inline_small_function_p): Do not cast to
7901 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
7903 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
7904 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
7905 (inline_edge_summary_vec): Turn into ...
7906 (ipa_call_summaries): ... this one.
7907 (redirect_to_unreachable, edge_set_predicate,
7908 evaluate_properties_for_edge, inline_summary_alloc,
7909 reset_ipa_call_summary, reset_inline_summary,
7910 inline_summary_t::duplicate): Update.
7911 (inline_edge_duplication_hook): Turn to ...
7912 (ipa_call_summary_t::duplicate): ... this one.
7913 (inline_edge_removal_hook): Turn to ...
7914 (ipa_call_summary_t::remove): ... this one.
7915 (dump_inline_edge_summary): Turn to ...
7916 (dump_ipa_call_summary): ... this one.
7917 (estimate_function_body_sizes): Update.
7918 (inline_update_callee_summaries): Update.
7919 (remap_edge_change_prob): Update.
7920 (remap_edge_summaries): Update.
7921 (inline_merge_summary): Update.
7922 (do_estimate_edge_time): Update.
7923 (inline_generate_summary): Update.
7924 (inline_read_section): Update.
7925 (inline_read_summary): Update.
7926 (inline_free_summary): Update.
7927 * ipa-inline.c (can_inline_edge_p): Update.
7928 (compute_inlined_call_time): Update.
7929 (want_inline_small_function_p): Update.
7930 (edge_badness): Update.
7931 (early_inliner): Update.
7932 * ipa-inline.h (inline_edge_summary): Turn to ...
7933 (ipa_call_summary): ... this one.
7934 (ipa_call_summary_t): New class.
7935 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
7936 (ipa_call_summaries): New.
7937 (inline_edge_summary): Remove.
7938 (estimate_edge_growth): Update.
7939 * ipa-profile.c (ipa_propagate_frequency_1): Update.
7940 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7941 * ipa-split.c (execute_split_functions): Update.
7942 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
7944 2017-05-19 Richard Biener <rguenther@suse.de>
7947 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
7949 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
7951 * config/rs6000/rs6000.c (struct machine_function): Add field
7952 fpr_is_wrapped_separately.
7953 (rs6000_get_separate_components): Use 64 components. Handle the
7955 (rs6000_components_for_bb): Handle the FPR components.
7956 (rs6000_emit_prologue_components): Handle the FPR components.
7957 (rs6000_emit_epilogue_components): Handle the FPR components.
7958 (rs6000_set_handled_components): Handle the FPR components.
7959 (rs6000_emit_prologue): Don't output prologue code for those FPRs
7960 that are already separately shrink-wrapped.
7961 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
7962 that are already separately shrink-wrapped.
7964 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
7967 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
7970 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
7971 (define_peephole2 for Altivec d-form load): Add peepholes to catch
7972 cases where the register allocator uses a move and an offsettable
7973 memory operation to/from a FPR register on ISA 2.06/2.07.
7974 (define_peephole2 for Altivec d-form store): Likewise.
7976 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
7979 * config/i386/mmx.md (*mov<mode>_internal): Enable
7980 alternatives 11, 12, 13 and 14 also for 32bit targets.
7981 Remove alternatives 15, 16, 17 and 18.
7982 * config/i386/sse.md (vec_concatv2di): Change
7983 alternative (!x, *y) to (x, ?!*Yn).
7985 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
7987 * dumpfile.h (enum dump_kind): Remove stray comma.
7989 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
7991 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
7992 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
7993 predicate::num_conditions
7994 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
7995 (CHANGED): turn into predicate::changed.
7996 (agg_position_info): Move to ipa-predicate.h
7997 (add_condition, predicate::add_clause, predicate::operator &=,
7998 predicate::or_with, predicate::evaluate, predicate::probability,
7999 dump_condition, dump_clause, predicate::dump,
8000 predicate::remap_after_duplication, predicate::remap_after_inlining,
8001 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
8002 (evaluate_conditions_for_known_args): Update.
8003 (set_cond_stmt_execution_predicate): Update.
8004 * ipa-inline.h: Include ipa-predicate.h
8005 (condition, inline_param_summary, conditions, agg_position_info,
8006 predicate): Move to ipa-predicate.h
8007 * ipa-predicate.c: New file.
8008 * ipa-predicate.h: New file.
8010 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
8012 * final.c (leaf_function_p): Check we are not in a sequence.
8014 2017-05-18 Martin Liska <mliska@suse.cz>
8016 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
8017 * dumpfile.c (dump_register): Use new enum dump_kind.
8018 (get_dump_file_name): Likewise.
8019 (dump_enable_all): Likewise.
8020 (dump_switch_p_1): Likewise.
8021 (enable_rtl_dump_file): Remove usage of TDF_RTL.
8022 * dumpfile.h (enum dump_kind): New enum type.
8023 (struct dump_file_info): Create constructor and
8024 format fields and comments.
8025 * passes.c (pass_manager::register_one_dump_file):
8027 * statistics.c (statistics_early_init): Likewise.
8028 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
8029 TDF_TREE with TDF_SLIM.
8030 (gather_memory_references_ref): Likewise.
8032 2017-05-18 Martin Liska <mliska@suse.cz>
8034 * vec.h (struct vnull): Use it.
8036 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
8038 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
8039 (true_predicate, false_predicate, true_predicate_p,
8040 false_predicate_p): Remove.
8041 (single_cond_predicate, not_inlined_predicate): Turn to member function
8043 (add_condition): Update.
8044 (add_clause): Turn to...
8045 (predicate::add_clause): ... this one; update; allow passing NULL
8047 (and_predicates): Turn to ...
8048 (predicate::operator &=): ... this one.
8049 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
8050 (or_predicates): Turn to ...
8051 (predicate::or_with): ... this one.
8052 (evaluate_predicate): Turn to ...
8053 (predicate::evaluate): ... this one.
8054 (predicate_probability): Turn to ...
8055 (predicate::probability): ... this one.
8056 (dump_condition): Update.
8057 (dump_predicate): Turn to ...
8058 (predicate::dump): ... this one.
8059 (account_size_time): Update.
8060 (edge_set_predicate): Update.
8061 (set_hint_predicate): UPdate.
8062 (evaluate_conditions_for_known_args): Update.
8063 (evaluate_properties_for_edge): Update.
8064 (remap_predicate_after_duplication): Turn to...
8065 (predicate::remap_after_duplication): ... this one.
8066 (remap_hint_predicate_after_duplication): Update.
8067 (inline_summary_t::duplicate): UPdate.
8068 (dump_inline_edge_summary): Update.
8069 (dump_inline_summary): Update.
8070 (set_cond_stmt_execution_predicate): Update.
8071 (set_switch_stmt_execution_predicate): Update.
8072 (compute_bb_predicates): Update.
8073 (will_be_nonconstant_expr_predicate): Update.
8074 (will_be_nonconstant_predicate): Update.
8075 (phi_result_unknown_predicate): Update.
8076 (predicate_for_phi_result): Update.
8077 (array_index_predicate): Update.
8078 (estimate_function_body_sizes): Update.
8079 (estimate_node_size_and_time): Update.
8080 (estimate_ipcp_clone_size_and_time): Update.
8081 (remap_predicate): Rename to ...
8082 (predicate::remap_after_inlining): ... this one.
8083 (remap_hint_predicate): Update.
8084 (inline_merge_summary): Update.
8085 (inline_update_overall_summary): Update.
8086 (estimate_size_after_inlining): Update.
8087 (read_predicate): Rename to ...
8088 (predicate::stream_in): ... this one.
8089 (read_inline_edge_summary): Update.
8090 (write_predicate): Rename to ...
8091 (predicate::stream_out): ... this one.
8092 (write_inline_edge_summary): Update.
8093 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
8094 (clause_t): Turn to uint32_t
8095 (predicate): Turn to class; implement constructor and operators
8097 (size_time_entry): Update.
8098 (inline_summary): Update.
8099 (inline_edge_summary): Update.
8101 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
8103 * fold-const.c (fold_binary_loc): Move transformation...
8104 * match.pd (C - X CMP X): ... here.
8106 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
8108 * config/sparc/sparc.c (sparc_option_override): Set function
8109 alignment for -mcpu=niagara7 to 64 to match the I$ line.
8110 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
8112 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
8114 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
8116 2017-05-18 Marek Polacek <polacek@redhat.com>
8119 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
8120 (pass_ubsan::execute): Call gimple_assign_single_p instead of
8121 gimple_assign_load_p.
8123 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
8126 * real.c (do_compare): Give decimal_do_compare preference over
8127 comparing just the signs.
8129 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
8131 * doc/md.texi (Canonicalization of Instructions): Describe the
8132 canonical form of instructions that inherently set a condition
8135 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
8138 * tree-cfg.c: Move deletion of unreachable case statements to after
8139 the merging of consecutive case labels.
8141 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
8143 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
8144 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
8145 restoring of callee-saved registers.
8147 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
8149 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
8150 * config/visium/visium.c (single_set_and_flags): Likewise.
8151 * config/visium/visium.md (Substitutions): Likewise.
8153 2017-05-17 Martin Liska <mliska@suse.cz>
8155 * cfg.c: Introduce dump_flags_t type and
8156 use it instead of int type.
8158 * cfghooks.c: Likewise.
8159 * cfghooks.h (struct cfg_hooks): Likewise.
8160 * cfgrtl.c: Likewise.
8161 * cfgrtl.h: Likewise.
8162 * cgraph.c (cgraph_node::get_body): Likewise.
8163 * coretypes.h: Likewise.
8164 * domwalk.c: Likewise.
8165 * domwalk.h: Likewise.
8166 * dumpfile.c (struct dump_option_value_info): Likewise.
8167 (dump_enable_all): Likewise.
8168 (dump_switch_p_1): Likewise.
8169 (opt_info_switch_p): Likewise.
8170 * dumpfile.h (enum tree_dump_index): Likewise.
8171 (struct dump_file_info): Likewise.
8172 * genemit.c: Likewise.
8173 * generic-match-head.c: Likewise.
8174 * gengtype.c (open_base_files): Likewise.
8175 * gimple-pretty-print.c: Likewise.
8176 * gimple-pretty-print.h: Likewise.
8177 * graph.c (print_graph_cfg): Likewise.
8178 * graphite-scop-detection.c (dot_all_sese): Likewise.
8179 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
8180 * loop-unroll.c (report_unroll): Likewise.
8181 * passes.c (pass_manager::register_one_dump_file): Likewise.
8182 * print-tree.c: Likewise.
8183 * statistics.c: Likewise.
8184 * tree-cfg.c: Likewise.
8185 * tree-cfg.h: Likewise.
8186 * tree-dfa.c: Likewise.
8187 * tree-dfa.h: Likewise.
8188 * tree-dump.c (dump_function): Likewise.
8189 * tree-dump.h (struct dump_info): Likewise.
8190 * tree-pretty-print.c: Likewise.
8191 * tree-pretty-print.h: Likewise.
8192 * tree-ssa-live.c: Likewise.
8193 * tree-ssa-live.h: Likewise.
8194 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
8195 * tree-vect-loop.c: Likewise.
8196 * tree-vect-slp.c: Likewise.
8198 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
8199 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8201 PR tree-optimization/80457
8202 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
8203 of all arguments to a statement as scalar_to_vec operations.
8204 (vectorizable_call): Adjust call to vect_model_simple_cost for
8206 (vectorizable_conversion): Likewise.
8207 (vectorizable_assignment): Likewise.
8208 (vectorizable_shift): Likewise.
8209 (vectorizable_operation): Likewise.
8210 (vectorizable_comparison): Likewise.
8211 (vect_is_simple_cond): Record the def types for operands.
8212 (vectorizable_condition): Likewise, call vect_model_simple_cost.
8213 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
8214 for statement argument count.
8216 2017-05-16 Carl Love <cel@us.ibm.com>
8218 * config/rs6000/rs6000-c: Add support for built-in functions
8219 vector unsigned long long vec_bperm (vector unsigned long long,
8220 vector unsigned char)
8221 vector signed long long vec_mule (vector signed int,
8223 vector unsigned long long vec_mule (vector unsigned int,
8224 vector unsigned int)
8225 vector signed long long vec_mulo (vector signed int,
8227 vector unsigned long long vec_mulo (vector unsigned int,
8228 vector unsigned int)
8229 vector signed char vec_sldw (vector signed char,
8232 vector unsigned char vec_sldw (vector unsigned char,
8233 vector unsigned char,
8235 vector signed short vec_sldw (vector signed short,
8236 vector signed short,
8238 vector unsigned short vec_sldw (vector unsigned short,
8239 vector unsigned short,
8241 vector signed int vec_sldw (vector signed int,
8244 vector unsigned int vec_sldw (vector unsigned int,
8245 vector unsigned int,
8247 vector signed long long vec_sldw (vector signed long long,
8248 vector signed long long,
8250 vector unsigned long long vec_sldw (vector unsigned long long,
8251 vector unsigned long long,
8253 * config/rs6000/rs6000-c: Add support for built-in functions
8254 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
8255 * config/rs6000/altivec.h: Add defintion for vec_sldw.
8256 * doc/extend.texi: Update the built-in documentation for the
8257 new built-in functions.
8259 2017-05-16 Marek Polacek <polacek@redhat.com>
8263 * tree.c (save_expr): Don't fold the expression.
8265 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
8267 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
8268 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
8269 and (?*y,m). Update insn attributes.
8271 2017-05-16 Martin Liska <mliska@suse.cz>
8273 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
8274 flags argument of print_gimple_stmt, print_gimple_expr,
8275 print_generic_stmt and print_generic_expr.
8276 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
8277 * coretypes.h: Likewise.
8278 * except.c (dump_eh_tree): Likewise.
8279 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8280 * gimple-pretty-print.h: Likewise.
8281 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
8282 (backprop::push_to_worklist): Likewise.
8283 (backprop::pop_from_worklist): Likewise.
8284 (backprop::process_use): Likewise.
8285 (backprop::intersect_uses): Likewise.
8286 (note_replacement): Likewise.
8287 * gimple-ssa-store-merging.c
8288 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
8289 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
8290 (pass_store_merging::execute): Likewise.
8291 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
8292 (ssa_base_cand_dump_callback): Likewise.
8293 (dump_incr_vec): Likewise.
8294 (replace_refs): Likewise.
8295 (replace_mult_candidate): Likewise.
8296 (create_add_on_incoming_edge): Likewise.
8297 (create_phi_basis): Likewise.
8298 (insert_initializers): Likewise.
8299 (all_phi_incrs_profitable): Likewise.
8300 (introduce_cast_before_cand): Likewise.
8301 (replace_one_candidate): Likewise.
8302 * gimplify.c (gimplify_expr): Likewise.
8303 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
8304 (set_rename): Likewise.
8305 (rename_uses): Likewise.
8306 (copy_loop_phi_nodes): Likewise.
8307 (add_close_phis_to_merge_points): Likewise.
8308 (copy_loop_close_phi_args): Likewise.
8309 (copy_cond_phi_args): Likewise.
8310 (graphite_copy_stmts_from_block): Likewise.
8311 (translate_pending_phi_nodes): Likewise.
8312 * graphite-poly.c (print_pdr): Likewise.
8313 (dump_gbb_cases): Likewise.
8314 (dump_gbb_conditions): Likewise.
8315 (print_scop_params): Likewise.
8316 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
8317 (build_cross_bb_scalars_use): Likewise.
8318 (gather_bbs::before_dom_children): Likewise.
8319 * hsa-dump.c (dump_hsa_immed): Likewise.
8320 * ipa-cp.c (print_ipcp_constant_value): Likewise.
8321 (get_replacement_map): Likewise.
8322 * ipa-inline-analysis.c (dump_condition): Likewise.
8323 (estimate_function_body_sizes): Likewise.
8324 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
8325 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
8326 * ipa-prop.c (ipa_dump_param): Likewise.
8327 (ipa_print_node_jump_functions_for_edge): Likewise.
8328 (ipa_modify_call_arguments): Likewise.
8329 (ipa_modify_expr): Likewise.
8330 (ipa_dump_param_adjustments): Likewise.
8331 (ipa_dump_agg_replacement_values): Likewise.
8332 (ipcp_modif_dom_walker::before_dom_children): Likewise.
8333 * ipa-pure-const.c (check_stmt): Likewise.
8334 (pass_nothrow::execute): Likewise.
8335 * ipa-split.c (execute_split_functions): Likewise.
8336 * omp-offload.c (dump_oacc_loop_part): Likewise.
8337 (dump_oacc_loop): Likewise.
8338 * trans-mem.c (tm_log_emit): Likewise.
8339 (tm_memopt_accumulate_memops): Likewise.
8340 (dump_tm_memopt_set): Likewise.
8341 (dump_tm_memopt_transform): Likewise.
8342 * tree-cfg.c (gimple_verify_flow_info): Likewise.
8343 (print_loop): Likewise.
8344 * tree-chkp-opt.c (chkp_print_addr): Likewise.
8345 (chkp_gather_checks_info): Likewise.
8346 (chkp_get_check_result): Likewise.
8347 (chkp_remove_check_if_pass): Likewise.
8348 (chkp_use_outer_bounds_if_possible): Likewise.
8349 (chkp_reduce_bounds_lifetime): Likewise.
8350 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
8351 (chkp_mark_completed_bounds): Likewise.
8352 (chkp_register_incomplete_bounds): Likewise.
8353 (chkp_mark_invalid_bounds): Likewise.
8354 (chkp_maybe_copy_and_register_bounds): Likewise.
8355 (chkp_build_returned_bound): Likewise.
8356 (chkp_get_bound_for_parm): Likewise.
8357 (chkp_build_bndldx): Likewise.
8358 (chkp_get_bounds_by_definition): Likewise.
8359 (chkp_generate_extern_var_bounds): Likewise.
8360 (chkp_get_bounds_for_decl_addr): Likewise.
8361 * tree-chrec.c (chrec_apply): Likewise.
8362 * tree-data-ref.c (dump_data_reference): Likewise.
8363 (dump_subscript): Likewise.
8364 (dump_data_dependence_relation): Likewise.
8365 (analyze_overlapping_iterations): Likewise.
8366 * tree-inline.c (expand_call_inline): Likewise.
8367 (tree_function_versioning): Likewise.
8368 * tree-into-ssa.c (dump_defs_stack): Likewise.
8369 (dump_currdefs): Likewise.
8370 (dump_names_replaced_by): Likewise.
8371 (dump_update_ssa): Likewise.
8372 (update_ssa): Likewise.
8373 * tree-object-size.c (pass_object_sizes::execute): Likewise.
8374 * tree-parloops.c (build_new_reduction): Likewise.
8375 (try_create_reduction_list): Likewise.
8376 (ref_conflicts_with_region): Likewise.
8377 (oacc_entry_exit_ok_1): Likewise.
8378 (oacc_entry_exit_single_gang): Likewise.
8379 * tree-pretty-print.h: Likewise.
8380 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
8381 (get_scalar_evolution): Likewise.
8382 (add_to_evolution): Likewise.
8383 (get_loop_exit_condition): Likewise.
8384 (analyze_evolution_in_loop): Likewise.
8385 (analyze_initial_condition): Likewise.
8386 (analyze_scalar_evolution): Likewise.
8387 (instantiate_scev): Likewise.
8388 (number_of_latch_executions): Likewise.
8389 (gather_chrec_stats): Likewise.
8390 (final_value_replacement_loop): Likewise.
8391 (scev_const_prop): Likewise.
8392 * tree-sra.c (dump_access): Likewise.
8393 (disqualify_candidate): Likewise.
8394 (create_access): Likewise.
8396 (maybe_add_sra_candidate): Likewise.
8397 (create_access_replacement): Likewise.
8398 (analyze_access_subtree): Likewise.
8399 (analyze_all_variable_accesses): Likewise.
8400 (sra_modify_assign): Likewise.
8401 (initialize_constant_pool_replacements): Likewise.
8402 (find_param_candidates): Likewise.
8403 (decide_one_param_reduction): Likewise.
8404 (replace_removed_params_ssa_names): Likewise.
8405 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
8406 * tree-ssa-copy.c (dump_copy_of): Likewise.
8407 (copy_prop_visit_cond_stmt): Likewise.
8408 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
8409 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
8410 (record_equivalences_from_stmt): Likewise.
8411 * tree-ssa-dse.c (compute_trims): Likewise.
8412 (delete_dead_call): Likewise.
8413 (delete_dead_assignment): Likewise.
8414 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
8415 (forward_propagate_into_cond): Likewise.
8416 (pass_forwprop::execute): Likewise.
8417 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8418 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
8420 (move_computations_worker): Likewise.
8421 (execute_sm): Likewise.
8422 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
8423 (remove_exits_and_undefined_stmts): Likewise.
8424 (remove_redundant_iv_tests): Likewise.
8425 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
8426 (adjust_iv_update_pos): Likewise.
8427 * tree-ssa-math-opts.c (bswap_replace): Likewise.
8428 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
8429 (value_replacement): Likewise.
8430 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
8431 * tree-ssa-pre.c (print_pre_expr): Likewise.
8432 (get_representative_for): Likewise.
8433 (create_expression_by_pieces): Likewise.
8434 (insert_into_preds_of_block): Likewise.
8435 (eliminate_insert): Likewise.
8436 (eliminate_dom_walker::before_dom_children): Likewise.
8437 (eliminate): Likewise.
8438 (remove_dead_inserted_code): Likewise.
8439 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
8440 * tree-ssa-reassoc.c (get_rank): Likewise.
8441 (eliminate_duplicate_pair): Likewise.
8442 (eliminate_plus_minus_pair): Likewise.
8443 (eliminate_not_pairs): Likewise.
8444 (undistribute_ops_list): Likewise.
8445 (eliminate_redundant_comparison): Likewise.
8446 (update_range_test): Likewise.
8447 (optimize_range_tests_var_bound): Likewise.
8448 (optimize_vec_cond_expr): Likewise.
8449 (rewrite_expr_tree): Likewise.
8450 (rewrite_expr_tree_parallel): Likewise.
8451 (linearize_expr): Likewise.
8452 (break_up_subtract): Likewise.
8453 (linearize_expr_tree): Likewise.
8454 (attempt_builtin_powi): Likewise.
8455 (attempt_builtin_copysign): Likewise.
8456 (transform_stmt_to_copy): Likewise.
8457 (transform_stmt_to_multiply): Likewise.
8458 (dump_ops_vector): Likewise.
8459 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
8460 (print_scc): Likewise.
8461 (set_ssa_val_to): Likewise.
8462 (visit_reference_op_store): Likewise.
8463 (visit_use): Likewise.
8464 (sccvn_dom_walker::before_dom_children): Likewise.
8465 (run_scc_vn): Likewise.
8466 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
8468 (expr_hash_elt::print): Likewise.
8469 (const_and_copies::pop_to_marker): Likewise.
8470 (const_and_copies::record_const_or_copy_raw): Likewise.
8471 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
8472 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
8473 (dump_predicates): Likewise.
8474 (find_uninit_use): Likewise.
8475 (warn_uninitialized_phi): Likewise.
8476 (pass_late_warn_uninitialized::execute): Likewise.
8477 * tree-ssa.c (verify_vssa): Likewise.
8478 (verify_ssa): Likewise.
8479 (maybe_optimize_var): Likewise.
8480 * tree-vrp.c (dump_value_range): Likewise.
8481 (dump_all_value_ranges): Likewise.
8482 (dump_asserts_for): Likewise.
8483 (register_edge_assert_for_2): Likewise.
8484 (vrp_visit_cond_stmt): Likewise.
8485 (vrp_visit_switch_stmt): Likewise.
8486 (vrp_visit_stmt): Likewise.
8487 (vrp_visit_phi_node): Likewise.
8488 (simplify_cond_using_ranges_1): Likewise.
8489 (fold_predicate_in): Likewise.
8490 (evrp_dom_walker::before_dom_children): Likewise.
8491 (evrp_dom_walker::push_value_range): Likewise.
8492 (evrp_dom_walker::pop_value_range): Likewise.
8493 (execute_early_vrp): Likewise.
8495 2017-05-16 Richard Biener <rguenther@suse.de>
8497 * dwarf2out.c (loc_list_from_tree_1): Do not create
8498 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
8500 2017-05-16 Richard Biener <rguenther@suse.de>
8502 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
8504 (note_variable_value_in_expr): If we resolved the decl ref
8505 do not push to the stack.
8507 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
8509 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
8510 operations in fast-math mode.
8511 (vaddq_f16): Likewise.
8512 (vmul_f16): Likewise.
8513 (vmulq_f16): Likewise.
8514 (vsub_f16): Likewise.
8515 (vsubq_f16): Likewise.
8516 * config/arm/neon.md (add<mode>3): New.
8518 (fma:<VH:mode>3): New. Also remove outdated comment.
8521 2017-05-16 Martin Liska <mliska@suse.cz>
8525 * ipa-devirt.c (warn_types_mismatch): Fix typo.
8526 (odr_types_equivalent_p): Likewise.
8528 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
8530 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
8532 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
8535 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
8536 non-interunit SSE move alternatives with '?'.
8537 (zero-extendsidi peephole2): New peephole to skip intermediate
8538 general register in SSE zero-extend sequence.
8540 2017-05-15 Jeff Law <law@redhat.com>
8542 * reorg.c (relax_delay_slots): Create a new variable to hold
8543 the temporary target rather than clobbering TARGET_LABEL.
8545 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
8546 missing argument to extract_bit_field call.
8547 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
8549 2017-05-15 Martin Liska <mliska@suse.cz>
8552 * gcc.c (process_command): Do not allow empty argument of -o option.
8554 2017-05-15 Renlin Li <renlin.li@arm.com>
8556 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
8557 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
8558 * config/aarch64/constraints.md (Usf): Add long call check.
8559 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
8560 (call_value): Likewise.
8561 (sibcall): Likewise.
8562 (sibcall_value): Likewise.
8564 (call_value_insn): New.
8565 (sibcall_insn): Update rtx pattern.
8566 (sibcall_value_insn): Likewise.
8567 (call_internal): Remove.
8568 (call_value_internal): Likewise.
8569 (sibcall_internal): Likewise.
8570 (sibcall_value_internal): Likewise.
8571 (call_reg): Likewise.
8572 (call_symbol): Likewise.
8573 (call_value_reg): Likewise.
8574 (call_value_symbol): Likewise.
8576 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
8579 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
8581 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
8583 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
8584 compatible with CCGOCmode and with CCZmode.
8586 2017-05-14 Martin Sebor <msebor@redhat.com>
8589 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
8590 (gimple_fold_builtin_snprintf): Same.
8591 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
8592 (gimple_fold_builtin_snprintf): Same.
8593 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
8595 (is_call_safe): New function.
8596 (try_substitute_return_value): Call it.
8597 (try_simplify_call): New function.
8598 (pass_sprintf_length::handle_gimple_call): Call it.
8600 2017-05-14 Martin Sebor <msebor@redhat.com>
8603 * builtins.c (expand_builtin_stpncpy): Simplify.
8605 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
8607 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
8608 * config/i386/i386.h
8609 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
8610 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
8611 (struct machine_function): Add new members call_ms2sysv,
8612 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
8613 (struct machine_frame_state): New fields sp_realigned and
8614 sp_realigned_offset.
8615 * config/i386/i386.c
8616 (enum xlogue_stub): New enum.
8617 (enum xlogue_stub_sets): New enum.
8618 (class xlogue_layout): New class.
8619 (struct ix86_frame): New fields stack_realign_allocate_offset,
8620 stack_realign_offset and outlined_save_offset. Modify comments to
8621 detail stack layout when using out-of-line stubs.
8622 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
8623 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
8624 -mcall-ms2sysv-xlogues.
8625 (stub_managed_regs): New static variable.
8626 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
8627 registers managed by out-of-line stub.
8628 (disable_call_ms2sysv_xlogues): New function.
8629 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
8630 m->call_ms2sysv when appropriate and compute frame layout for
8632 (sp_valid_at, fp_valid_at): New inline functions.
8633 (choose_basereg): New function.
8634 (choose_baseaddr): Add align parameter, use choose_basereg and modify
8636 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
8637 Use align parameter of choose_baseaddr to generated aligned SSE movs
8639 (pro_epilogue_adjust_stack): Modify to track
8640 machine_frame_state::sp_realigned.
8641 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
8642 (ix86_nsaved_sseregs): Likewise.
8643 (ix86_emit_save_regs): Likewise.
8644 (ix86_emit_save_regs_using_mov): Likewise.
8645 (ix86_emit_save_sse_regs_using_mov): Likewise.
8646 (get_scratch_register_on_entry): Likewise.
8647 (gen_frame_set): New function.
8648 (gen_frame_load): Likewise.
8649 (gen_frame_store): Likewise.
8650 (emit_outlined_ms2sysv_save): Likewise.
8651 (emit_outlined_ms2sysv_restore): Likewise.
8652 (ix86_expand_prologue): Modify stack re-alignment code and call
8653 emit_outlined_ms2sysv_save when appropriate.
8654 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
8655 parameter rtx_insn *insn, which allows the function to be used to only
8657 (ix86_expand_epilogue): Modify validity checks of frame and stack
8658 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
8659 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
8660 * config/i386/predicates.md
8661 (save_multiple): New predicate.
8662 (restore_multiple): Likewise.
8663 * config/i386/sse.md
8664 (save_multiple<mode>): New pattern.
8665 (save_multiple_realign<mode>): Likewise.
8666 (restore_multiple<mode>): Likewise.
8667 (restore_multiple_and_return<mode>): Likewise.
8668 (restore_multiple_leave_return<mode>): Likewise.
8669 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
8671 2017-05-14 Julia Koval <julia.koval@intel.com>
8673 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
8674 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
8675 (__builtin_ia32_xsetbv): New builtins.
8676 * config/i386/i386.c (ix86_expand_special_args_builtin):
8678 (ix86_expand_builtin): Special expand for new intrinsics.
8679 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
8680 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
8681 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
8683 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8685 * cfganal.c (inverted_post_order_compute): Change argument type
8687 * cfganal.h (inverted_post_order_compute): Adjust prototype.
8688 * df-core.c (rest_of_handle_df_initialize): Adjust.
8689 (rest_of_handle_df_finish): Likewise.
8690 (df_analyze_1): Likewise.
8691 (df_analyze): Likewise.
8692 (loop_inverted_post_order_compute): Change argument to be a vec *.
8693 (df_analyze_loop): Adjust.
8694 (df_get_n_blocks): Likewise.
8695 (df_get_postorder): Likewise.
8696 * df.h (struct df_d): Change field to be a vec.
8697 * lcm.c (compute_laterin): Adjust.
8698 (compute_available): Likewise.
8699 * lra-lives.c (lra_create_live_ranges_1): Likewise.
8700 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
8701 * tree-ssa-pre.c (compute_antic): Likewise.
8703 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8705 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
8706 (depth_first_search::depth_first_search): Change structure init
8707 function to this constructor.
8708 (depth_first_search::add_bb): Rename function to this member.
8709 (depth_first_search::execute): Likewise.
8710 (flow_dfs_compute_reverse_finish): Adjust.
8712 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8714 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
8715 (longest_simple_path): Likewise.
8716 * shrink-wrap.c (spread_components): Likewise.
8717 (disqualify_problematic_components): Likewise.
8718 (emit_common_heads_for_components): Likewise.
8719 (emit_common_tails_for_components): Likewise.
8720 (insert_prologue_epilogue_for_components): Likewise.
8722 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8724 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
8727 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8729 * df-core.c (df_set_blocks): Start using auto_bitmap.
8730 (df_compact_blocks): Likewise.
8731 * df-problems.c (df_rd_confluence_n): Likewise.
8732 * df-scan.c (df_insn_rescan_all): Likewise.
8733 (df_process_deferred_rescans): Likewise.
8734 (df_update_entry_block_defs): Likewise.
8735 (df_update_exit_block_uses): Likewise.
8736 (df_entry_block_bitmap_verify): Likewise.
8737 (df_exit_block_bitmap_verify): Likewise.
8738 (df_scan_verify): Likewise.
8739 * lra-constraints.c (lra_constraints): Likewise.
8740 (undo_optional_reloads): Likewise.
8741 (lra_undo_inheritance): Likewise.
8742 * lra-remat.c (calculate_gen_cands): Likewise.
8743 (do_remat): Likewise.
8744 * lra-spills.c (assign_spill_hard_regs): Likewise.
8745 (spill_pseudos): Likewise.
8746 * tree-ssa-pre.c (bitmap_set_and): Likewise.
8747 (bitmap_set_subtract_values): Likewise.
8749 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8751 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
8752 management with auto_bitmap.
8753 (fix_inter_tick): Likewise.
8754 (fix_recovery_deps): Likewise.
8755 * ira.c (add_store_equivs): Likewise.
8756 (find_moveable_pseudos): Likewise.
8757 (split_live_ranges_for_shrink_wrap): Likewise.
8758 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
8759 (rtx_reuse_manager::seen_def_p): Likewise.
8760 (rtx_reuse_manager::set_seen_def): Likewise.
8761 * print-rtl.h (class rtx_reuse_manager): Likewise.
8763 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8765 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
8767 (migrate_btr_def): Likewise.
8768 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
8769 * df-core.c (loop_post_order_compute): Likewise.
8770 (loop_inverted_post_order_compute): Likewise.
8771 * hsa-common.h: Likewise.
8772 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
8773 * init-regs.c (initialize_uninitialized_regs): Likewise.
8774 * ipa-inline.c (resolve_noninline_speculation): Likewise.
8775 (inline_small_functions): Likewise.
8776 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
8777 * ira.c (combine_and_move_insns): Likewise.
8778 (build_insn_chain): Likewise.
8779 * loop-invariant.c (find_invariants): Likewise.
8780 * lower-subreg.c (propagate_pseudo_copies): Likewise.
8781 * predict.c (tree_predict_by_opcode): Likewise.
8782 (predict_paths_leading_to): Likewise.
8783 (predict_paths_leading_to_edge): Likewise.
8784 (estimate_loops_at_level): Likewise.
8785 (estimate_loops): Likewise.
8786 * shrink-wrap.c (try_shrink_wrapping): Likewise.
8787 (spread_components): Likewise.
8788 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
8789 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
8790 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
8791 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
8792 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
8793 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
8794 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
8795 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
8796 (mark_threaded_blocks): Likewise.
8797 (thread_through_all_blocks): Likewise.
8798 * tree-ssa.c (verify_ssa): Likewise.
8799 (execute_update_addresses_taken): Likewise.
8800 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
8802 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8804 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
8806 (post_order_compute): Likewise.
8807 (inverted_post_order_compute): Likewise.
8808 (pre_and_rev_post_order_compute_fn): Likewise.
8810 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8812 * genrecog.c (int_set::int_set): Explicitly construct our
8813 auto_vec base class.
8814 * vec.h (auto_vec::auto_vec): New constructor.
8816 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8818 * bitmap.h (class auto_bitmap): New constructor taking
8819 bitmap_obstack * argument.
8821 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
8823 * bitmap.h (class auto_bitmap): Change type of m_bits to
8824 bitmap_head, and adjust ctor / dtor and member operators.
8826 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
8828 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
8829 when returned register mode doesn't match original mode.
8831 2017-05-12 Jeff Law <law@redhat.com>
8832 Jakub Jelinek <jakub@redhat.com>
8834 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
8835 we look for cc setter after the compare-elim changes.
8836 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
8837 within the vector to match what compare-elim now expects.
8838 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
8839 (xorsi3_flags, one_cmplsi2_flags): Likewise.
8841 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
8842 after the compare-elim changes.
8843 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
8844 the vector to match what compare-elim now expects.
8845 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
8846 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
8847 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
8848 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
8849 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
8851 * config/visium/visium.c (single_set_and_flags): Fix where
8852 we look for cc setter after the compare-elim changes.
8853 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
8854 with the vector to match what compare-elim now expects.
8855 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
8856 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
8857 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
8858 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
8859 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
8860 (neg<mode>2_insn_set_overflow): Likewise.
8862 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
8865 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
8866 maybe_expand_insn call, set ops[0].target. If still set after call,
8867 set alt_rtl. Add extra arg to recursive calls.
8868 (extract_bit_field): Add alt_rtl argument. Pass to
8870 * expmed.h (extract_bit_field): Fix prototype.
8871 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
8872 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
8873 to extract_bit_field_calls.
8874 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
8875 Pass alt_rtl to extract_bit_field calls.
8876 * calls.c (store_unaligned_arguments_into_psuedos)
8877 load_register_parameters): Pass extra NULL to extract_bit_field calls.
8878 * optabs.c (maybe_legitimize_operand): Clear op->target when call
8880 * optabs.h (struct expand_operand): Add target bitfield.
8882 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
8884 * compare-elim.c (try_eliminate_compare): Canonicalize
8885 operation with embedded compare to
8886 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
8887 (set (reg) (operation)].
8889 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
8891 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
8894 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
8895 cost of adding a carry flag for ADC instruction.
8896 [case MINUS]: Ignore the cost of subtracting a carry flag
8897 for SBB instruction.
8899 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
8901 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
8903 * config/rs6000/bmiintrin.h: New file.
8904 * config/rs6000/bmi2intrin.h: New file.
8905 * config/rs6000/x86intrin.h: New file.
8907 2017-05-12 Jeff Law <law@redhat.com>
8909 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
8912 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
8915 * tree-cfg.c: Remove cfg edges of unreachable case statements.
8917 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8919 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
8920 early expansion of vector divide builtins.
8921 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
8922 builtins identified as having unsigned arguments.
8924 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8926 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
8927 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
8928 expansion of vector logical operations (and, andc, or, xor,
8931 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8933 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
8934 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
8936 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8938 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
8939 early GIMPLE expansion of vector multiplies.
8941 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
8943 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
8944 TARGET_HAVE_MOVT conditional.
8945 (movt splitter): Likewise.
8947 2017-05-12 Richard Biener <rguenther@suse.de>
8949 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
8950 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8951 Fold all stmts not inplace.
8953 2017-05-12 Richard Biener <rguenther@suse.de>
8955 PR tree-optimization/80713
8956 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
8957 inserted_exprs bit for not removed stmts.
8959 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
8962 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
8963 parallelized" attribute for parallelized OpenACC kernels.
8964 * omp-offload.c (execute_oacc_device_lower): Use it.
8966 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
8967 Set "oacc kernels" attribute.
8968 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
8969 parameter. Adjust all users.
8970 (oacc_fn_attrib_kernels_p): Remove function.
8971 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
8972 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
8973 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
8974 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
8975 assert "oacc kernels" attribute is set.
8977 2017-05-11 Carl Love <cel@us.ibm.com>
8979 * config/rs6000/rs6000-c: Add support for built-in functions
8980 vector unsigned char vec_popcnt (vector signed char)
8981 vector unsigned char vec_popcnt (vector unsigned char)
8982 vector unsigned short vec_popcnt (vector signed short)
8983 vector unsigned short vec_popcnt (vector unsigned short)
8984 vector unsigned int vec_popcnt (vector signed int)
8985 vector unsigned int vec_popcnt (vector unsigned int)
8986 vector unsigned long long vec_popcnt (vector signed long long)
8987 vector unsigned long long vec_popcnt (vector unsigned long long)
8988 vector signed long long vec_slo (vector signed long long,
8990 vector signed long long vec_slo (vector signed long long,
8991 vector unsigned char)
8992 vector unsigned long long vec_slo (vector unsigned long long,
8994 vector unsigned long long vec_slo (vector unsigned long long,
8995 vector unsigned char)
8996 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
8997 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
8998 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
8999 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
9000 * doc/extend.texi: Update the built-in documentation file for the
9001 new built-in functions.
9003 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
9005 * attribs.h (sorted_attr_string): Move machine independent
9006 functions for target clone support from the i386 port to common
9007 code. Rename ix86_function_versions to common_function_versions.
9008 Rename make_name to make_unique_name.
9009 (common_function_versions): Likewise.
9010 (make_unique_name): Likewise.
9011 (make_dispatcher_decl): Likewise.
9012 (is_function_default_version): Likewise.
9013 * attribs.c (attr_strcmp): Likewise.
9014 (sorted_attr_string): Likewise.
9015 (common_function_versions): Likewise.
9016 (make_unique_name): Likewise.
9017 (make_dispatcher_decl): Likewise.
9018 (is_function_default_version): Likewise.
9019 * config/i386/i386.c (attr_strcmp): Likewise.
9020 (sorted_attr_string): Likewise.
9021 (ix86_function_versions): Likewise.
9022 (make_name): Likewise.
9023 (make_dispatcher_decl): Likewise.
9024 (is_function_default_version): Likewise.
9025 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
9027 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9030 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
9031 Account for direct move costs for vec_construct of integer
9034 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
9037 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
9038 (UNSPEC_STX_ATOMIC): Ditto.
9039 (loaddi_via_sse): New insn.
9040 (storedi_via_sse): Ditto.
9041 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
9042 Update corresponding peephole2 patterns.
9043 (atomic_storedi_fpu): Ditto.
9045 2017-05-11 Julia Koval <julia.koval@intel.com>
9047 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
9048 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
9050 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
9051 (__builtin_ia32_rsqrt14ss_mask): New builtins.
9052 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
9054 2017-05-11 Nathan Sidwell <nathan@acm.org>
9056 * graphite-poly.c: Include dumpfile.h.
9058 * dumpfle.h (dump_function): Declare here ...
9059 * tree-dump.h (dump_function): ... not here.
9060 * dumpfile.c: #include tree-cfg.h.
9061 (dump_function): Move here from ...
9062 * tree-dump.c (dump_function): ... here.
9063 * gimplify.c: #include splay-tree.h, not tree-dump.h.
9064 * graphite-poly.c: Don't include tree-dump.h.
9065 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
9066 * print-tree.c: Likewise.
9067 * stor-layout.c: Likewise.
9068 * tree-nested.c: Likewise.
9070 * dumpfile.c (dump_start): Use TDF_FLAGS.
9071 (dump_enable_all): Fix TDF_KIND check thinko.
9073 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
9075 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9076 array entries to represent two legal parameterizations of the
9077 overloaded __builtin_cmpb function, as represented by the
9078 P6_OV_BUILTIN_CMPB constant.
9079 (altivec_resolve_overloaded_builtin): Add special case handling
9080 for the __builtin_cmpb function, as represented by the
9081 P6_OV_BUILTIN_CMPB constant.
9082 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
9083 (BU_P6_64BIT_2): New macro.
9084 (BU_P6_OVERLOAD_2): New macro
9085 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
9086 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
9087 (CMPB): Add overload support to represent both 32-bit and 64-bit
9088 compare-bytes function.
9089 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
9090 support for TARGET_CMPB.
9091 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
9092 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
9093 documentation of the __builtin_cmpb overloaded built-in function.
9095 2017-05-11 Richard Biener <rguenther@suse.de>
9097 PR tree-optimization/80705
9098 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
9099 bases are not vectorizable.
9101 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9103 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
9104 when counting register pressure.
9106 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9108 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
9109 (iv_ca_more_deps): Renamed to ...
9110 (iv_ca_compare_deps): ... this.
9111 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
9113 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9115 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
9117 (determine_group_iv_costs): ... here.
9118 (find_inv_vars_cb): Record inv var if it's not recorded before.
9120 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9122 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
9123 (get_shiftadd_cost): Ditto.
9125 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9127 * tree-ssa-address.c: Include header file.
9128 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
9130 (add_to_parts): Refactor.
9131 (addr_to_parts): New parameter. Update use of move_hint_to_base.
9132 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
9135 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9137 PR tree-optimization/53090
9138 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
9140 (extract_cond_operands): Detect condition with IV on both sides
9141 and return COMP_IV_EXPR_2.
9142 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
9143 (rewrite_use_compare): Simplify by removing call to function
9144 extract_cond_operands.
9146 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9148 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
9149 (extract_cond_operands): Detect condition comparing against non-
9150 invariant bound and return appropriate enum value.
9151 (find_interesting_uses_cond): Update use of extract_cond_operands.
9152 Handle its return value accordingly.
9153 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
9155 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9157 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
9158 nonlinear iv_use computation in loop invariant sensitive way.
9160 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9162 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
9163 (find_iv_candidates): Call relate_compare_use_with_all_cands.
9165 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9167 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
9168 (dump_cand): Support iv_cand.inv_exprs.
9169 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
9171 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
9174 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9176 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
9178 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
9179 as local function. Include necessary header files.
9180 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
9182 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9184 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
9186 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9188 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
9189 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
9190 RSHIFT_EXPR and BIT_NOT_EXPR.
9192 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9194 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
9195 (adjust_setup_cost): New parameter supporting round up adjustment.
9196 (struct address_cost_data): Delete.
9197 (force_expr_to_var_cost): Don't bound cost with spill_cost.
9198 (split_address_cost, ptr_difference_cost): Delete.
9199 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
9200 (struct ainc_cost_data): New struct.
9201 (get_address_cost_ainc): New function.
9202 (get_address_cost, get_computation_cost): Reimplement.
9203 (determine_group_iv_cost_address): Record inv_expr for all uses of
9205 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
9206 (iv_ca_has_deps): Reimplemented to ...
9207 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
9209 (iv_ca_extend): Call iv_ca_more_deps.
9211 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9213 * tree-ssa-address.c (struct mem_address): Move to header file.
9214 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
9215 * tree-ssa-address.h (struct mem_address): Move from C file.
9216 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
9218 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9220 * tree-affine.h (aff_combination_type): New interface.
9221 (aff_combination_zero_p): Remove static.
9222 (aff_combination_const_p): New interface.
9223 (aff_combination_singleton_var_p): New interfaces.
9225 2017-05-11 Richard Biener <rguenther@suse.de>
9227 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9228 Skip unreachable blocks and destinations.
9229 (eliminate): Move stmt removal and fixup ...
9230 (fini_eliminate): ... here. Skip inserted exprs.
9231 (pass_pre::execute): Move fini_pre after fini_eliminate.
9232 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
9233 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
9234 PRE to get rid of dead code that has invalid SSA form and
9235 split critical edges again.
9237 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9239 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
9241 2017-05-11 Richard Biener <rguenther@suse.de>
9243 * passes.c (execute_function_todo): Verify loops if they are
9244 said to be up-to-date.
9245 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
9246 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
9248 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
9251 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
9252 handle calling assemble_external ourself.
9255 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
9256 modes with zero size. Enhance comment.
9258 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9260 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
9261 built-ins for vec_xl and vec_xst with short and char pointer
9264 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
9266 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
9267 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
9268 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
9269 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
9270 (_mm_maskz_min_round_ss): New intrinsics.
9271 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
9272 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
9273 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
9274 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
9275 (__builtin_ia32_minss_mask_round): New builtins.
9276 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9277 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9278 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
9280 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
9281 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
9283 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
9286 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
9288 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
9289 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
9290 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
9291 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
9292 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
9293 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
9294 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
9295 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9296 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
9297 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
9298 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
9299 (__builtin_ia32_mulss_mask_round): New builtins.
9300 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9301 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9302 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
9304 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
9305 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
9307 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
9310 2017-05-10 Julia Koval <julia.koval@intel.com>
9312 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
9313 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
9314 (_mm256_setr_m128i): New intrinsics.
9316 2017-05-10 Julia Koval <julia.koval@intel.com>
9318 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
9319 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
9320 (_mm_maskz_rcp14_ss): New intrinsics.
9321 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
9322 (__builtin_ia32_rcp14ss_mask): New builtins.
9323 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
9325 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
9327 PR tree-optimization/51513
9328 * tree-cfg.c (gimple_seq_unreachable_p): New function.
9329 (assert_unreachable_fallthru_edge_p): Use it.
9330 (group_case_labels_stmt): Likewise.
9331 * tree-cfg.h: Prototype it.
9332 * stmt.c: Include cfghooks.h and tree-cfg.h.
9333 (emit_case_dispatch_table) <gap_label>: New local variable.
9334 Use it to fill dispatch table gaps.
9335 Test for default_label before updating probabilities.
9336 (expand_case) <default_label>: Remove unneeded initialization.
9337 Test for unreachable default case statement and remove its edge.
9338 Set default_label accordingly.
9339 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
9341 2017-05-10 Carl Love <cel@us.ibm.com>
9343 * config/rs6000/rs6000-c: Add support for built-in functions
9344 vector signed char vec_neg (vector signed char)
9345 vector signed short int vec_neg (vector short int)
9346 vector signed int vec_neg (vector signed int)
9347 vector signed long long vec_neg (vector signed long long)
9348 vector float vec_neg (vector float)
9349 vector double vec_neg (vector double)
9350 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
9352 * config/rs6000/altivec.h: Add define for vec_neg
9353 * doc/extend.texi: Update the built-in documentation for the
9354 new built-in functions.
9356 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9358 PR tree-optimization/77644
9359 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
9361 2017-05-10 Nathan Sidwell <nathan@acm.org>
9363 * dumpfile.h (TDI_lang_all): New.
9364 (TDF_KIND): New. Renumber others
9365 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
9367 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
9369 (get_dump_file_name): Adjust suffix generation.
9370 (dump_enable_all): Use TDF_KIND.
9371 * doc/invoke.texi (-fdump-lang-all): Document.
9373 * dumpfile.h: Tabify.
9375 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
9378 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
9379 Move member access before delete.
9381 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
9383 * tree-inline.c (expand_call_inline): Split block at stmt
9386 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9389 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
9390 are now unused after splitting mov{sf,sd}_hardfloat.
9391 (f32_lr2): Likewise.
9393 (f32_lm2): Likewise.
9395 (f32_li2): Likewise.
9398 (f32_sr2): Likewise.
9400 (f32_sm2): Likewise.
9402 (f32_si2): Likewise.
9405 (f32_vsx): Likewise.
9407 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
9408 For movsf, order stores so the VSX stores occur before the GPR
9409 store which encourages the register allocator to use a traditional
9410 FPR instead of a GPR. For movsd, order the stores so that the GPR
9411 store comes before the VSX stores to allow the power6 to work.
9412 This is due to the power6 not having a 32-bit integer store
9413 instruction from a FPR.
9414 (movsf_hardfloat): Likewise.
9415 (movsd_hardfloat): Likewise.
9417 2017-05-09 Martin Sebor <msebor@redhat.com>
9419 PR translation/80280
9420 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
9423 PR translation/80280
9424 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
9425 data member added in r247778.
9426 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
9428 2017-05-09 Nathan Sidwell <nathan@acm.org>
9430 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
9432 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
9435 2017-05-09 Marek Polacek <polacek@redhat.com>
9437 * doc/invoke.texi: Fix typo.
9439 2017-05-09 Richard Biener <rguenther@suse.de>
9441 * tree-vrp.c (vrp_val_is_max): Adjust comment.
9442 (vrp_val_is_min): Likewise.
9443 (set_value_range_to_value): Likewise.
9444 (set_value_range_to_nonnegative): Likewise.
9445 (gimple_assign_nonzero_p): Likewise.
9446 (gimple_stmt_nonzero_p): Likewise.
9447 (vrp_int_const_binop): Likewise. Remove unreachable case.
9448 (adjust_range_with_scev): Adjust comments.
9449 (compare_range_with_value): Likewise.
9450 (extract_range_from_phi_node): Likewise.
9451 (test_for_singularity): Likewise.
9453 2017-05-09 Richard Biener <rguenther@suse.de>
9455 * tree-vrp.c (get_single_symbol): Add assert that we don't
9456 get overflowed constants as invariant part.
9457 (compare_values_warnv): Add comment before the TREE_NO_WARNING
9458 checks. Use wi::cmp instead of recursing for integer constants.
9459 (compare_values): Just ignore whether we assumed undefined
9460 overflow instead of failing the compare.
9461 (extract_range_for_var_from_comparison_expr): Add comment before the
9462 TREE_NO_WARNING sets.
9463 (test_for_singularity): Likewise.
9464 (extract_range_from_comparison): Do not disable optimization
9465 when we assumed undefined overflow.
9466 (extract_range_basic): Remove init of unused var.
9468 2017-05-09 Richard Biener <rguenther@suse.de>
9470 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
9471 (extract_range_from_multiplicative_op_1): Adjust.
9472 (extract_range_from_binary_expr_1): Use int_const_binop.
9474 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
9477 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
9478 rs6000_store_data_bypass_p in seven define_bypass directives and
9479 in several comments.
9480 * config/rs6000/rs6000-protos.h: Add prototype for
9481 rs6000_store_data_bypass_p function.
9482 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
9483 function implements slightly different (rs6000-specific) semantics
9484 than store_data_bypass_p, returning false rather than aborting
9485 with assertion error when arguments do not satisfy the
9486 requirements of store data bypass.
9487 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
9488 rs6000_store_data_bypass_p.
9490 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
9492 * config/xtensa/xtensa-protos.h
9493 (xtensa_initial_elimination_offset): New declaration.
9494 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
9495 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
9496 macro definition, add case for FRAME_POINTER_REGNUM when
9497 FRAME_GROWS_DOWNWARD.
9498 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
9499 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
9500 xtensa_initial_elimination_offset.
9502 2017-05-08 Nathan Sidwell <nathan@acm.org>
9504 * doc/invoke.texi: Alphabetize -fdump options.
9506 2017-05-08 Martin Sebor <msebor@redhat.com>
9508 PR translation/80280
9509 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
9511 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
9513 * target.def (compute_frame_layout): New optional target hook.
9514 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
9515 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
9516 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
9518 * reload1.c (verify_initial_elim_offsets): Likewise.
9519 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
9520 (use_simple_return_p): Call arm_compute_frame_layout if needed.
9521 (arm_get_frame_offsets): Split up into this ...
9522 (arm_compute_frame_layout): ... and this function.
9524 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
9526 * config/aarch64/constraints.md (Usa): New constraint.
9527 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
9529 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
9531 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
9532 with_multilib_list after it has been checked.
9534 2017-05-08 Richard Biener <rguenther@suse.de>
9536 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
9537 (bitmap_set_subtract_values): Likewise.
9539 2017-05-08 Richard Biener <rguenther@suse.de>
9541 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
9542 (gimple_assign_nonzero): ... this and remove strict_overflow_p
9544 (gimple_stmt_nonzero_warnv_p): Rename to ...
9545 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
9547 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
9548 (extract_range_basic): Adjust, do not disable propagation on
9549 strict overflow sensitive simplification.
9550 (vrp_visit_cond_stmt): Likewise.
9552 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
9554 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
9555 body size unconditionally.
9557 2017-05-07 Jeff Law <law@redhat.com>
9560 2017-05-06 Jeff Law <law@redhat.com>
9561 PR tree-optimization/78496
9562 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9565 PR tree-optimization/78496
9566 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9567 (simplify_stmt_using_ranges): Call it.
9568 (vrp_dom_walker::before_dom_children): Extract equivalences
9569 from an ASSERT_EXPR with an equality comparison against a
9572 2017-05-06 Jeff Law <law@redhat.com>
9574 PR tree-optimization/78496
9575 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
9578 PR tree-optimization/78496
9579 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
9580 (simplify_stmt_using_ranges): Call it.
9581 (vrp_dom_walker::before_dom_children): Extract equivalences
9582 from an ASSERT_EXPR with an equality comparison against a
9585 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9587 * lra-constraints.c (lra_copy_reg_equiv): New function.
9588 (split_reg): Use it to copy equivalence information from the
9589 original register to the spill register.
9591 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
9593 PR rtl-optimization/75964
9594 * simplify-rtx.c (simplify_const_relational_operation): Remove
9595 invalid handling of comparisons of integer ABS.
9597 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
9599 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
9601 (init_regs): Remove declaration.
9602 (function_arg_advance_32): Initialize error_p as boolean variable.
9604 2017-05-05 Nathan Sidwell <nathan@acm.org>
9606 * store-motion.c (remove_reachable_equiv_notes): Reformat long
9607 lines. Use for (;;).
9609 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9611 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
9612 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
9613 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
9614 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
9615 VF=2 that require versioning.
9617 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9619 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
9622 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9624 * diagnostic.h (diagnostic_override_option_index): Convert from
9625 macro to inline function.
9627 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9629 * diagnostic.c (last_module_changed_p): New function.
9630 (set_last_module): New function.
9631 (diagnostic_report_current_module): Convert macro usage to
9632 the above functions.
9633 * diagnostic.h (diagnostic_context::last_module): Strengthen
9634 from const line_map * to const line_map_ordinary *.
9635 (diagnostic_last_module_changed): Delete macro.
9636 (diagnostic_set_last_module): Delete macro.
9638 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9640 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
9641 with diagnostic_report_diagnostic.
9642 (diagnostic_n_impl_richloc): Likewise.
9643 * diagnostic.h (report_diagnostic): Delete macro.
9644 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
9645 with diagnostic_report_diagnostic.
9646 * substring-locations.c (format_warning_va): Likewise.
9648 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9650 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
9651 save/restor of format_spec. Move option-printing code to...
9652 (print_option_information): ...this new function, and
9653 reimplement by simply printing to the pretty_printer,
9654 rather than appending to the format string.
9656 2017-05-05 David Malcolm <dmalcolm@redhat.com>
9658 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
9659 handling logic into...
9660 (update_effective_level_from_pragmas): ...this new function.
9662 2017-05-04 Andrew Waterman <andrew@sifive.com>
9664 * config/riscv/riscv.opt (mstrict-align): New option.
9665 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
9666 (SLOW_UNALIGNED_ACCESS): Define.
9667 (riscv_slow_unaligned_access): Declare.
9668 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
9670 (riscv_slow_unaligned_access): New variable.
9671 (rocket_tune_info): Set slow_unaligned_access to true.
9672 (optimize_size_tune_info): Set slow_unaligned_access to false.
9673 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
9674 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
9675 (riscv_option_override): Set riscv_slow_unaligned_access.
9676 * doc/invoke.texi: Add -mstrict-align to RISC-V.
9678 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
9680 * config/riscv/riscv.md: Unify indentation.
9682 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
9687 * config/rs6000/rs6000.md (u code attribute): Add FIX and
9689 (extendsi<mode>2): Add support for doing sign extension via
9690 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
9691 don't have ISA 3.0 instructions.
9692 (extendsi<mode>2 splitter): Likewise.
9693 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
9694 generate the normal insns since SImode can now go in vector
9695 registers. Disallow the special UNSPECs needed for previous
9696 machines to hide SImode being used. Add new insns
9697 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
9698 (fix_trunc<mode>si2_stfiwx): Likewise.
9699 (fix_trunc<mode>si2_internal): Likewise.
9700 (fixuns_trunc<mode>si2): Likewise.
9701 (fixuns_trunc<mode>si2_stfiwx): Likewise.
9702 (fctiw<u>z_<mode>_smallint): Likewise.
9703 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
9704 of floating point to 32-bit integer from doing a direct move to
9705 the GPR registers to do a store.
9706 (fctiwz_<mode>): Break long line.
9708 2017-05-05 Bin Cheng <bin.cheng@arm.com>
9710 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
9711 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
9712 (addr_list, addr_offset_valid_p): New.
9713 (split_address_groups): Check offset validity with above function.
9714 (gt-tree-ssa-loop-ivopts.h): Include header file.
9716 2017-05-05 Nathan Sidwell <nathan@acm.org>
9718 * config.gcc (arm*-*-*): Add missing 'fi'.
9720 2017-05-05 Steve Ellcey <sellcey@cavium.com>
9722 * doc/invoke.texi (-fopt-info): Explicitly say order of options
9723 included in -fopt-info does not matter.
9724 * doc/optinfo.texi (-fopt-info): Fix description of default
9725 behavour. Explicitly say order of options included in -fopt-info
9728 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
9730 * config.gcc: Allow combinations of aprofile and rmprofile values for
9731 --with-multilib-list.
9732 * config/arm/t-multilib: New file.
9733 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
9734 variables. Remove setting of ISA and floating-point ABI in
9735 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
9736 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
9737 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
9738 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
9740 * config/arm/t-rmprofile: Likewise except for the matches changes.
9741 * doc/install.texi (--with-multilib-list): Document the combination of
9742 aprofile and rmprofile values and warn about pitfalls in doing that.
9744 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9746 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
9747 (movdi_aarch64): Likewise.
9749 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9751 PR tree-optimization/80632
9752 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
9754 (build_arrays): Initialize it for virtual phis.
9755 (fix_phi_nodes): Use it for virtual phis.
9757 PR tree-optimization/80558
9758 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
9759 [x, y] op z into [x op, y op z] for op & or | if conditions
9762 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
9763 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9766 * config/arm/arm.md (use_literal_pool): Remove.
9767 (64-bit immediate split): No longer takes cost into consideration
9768 if arm_disable_literal_pool is enabled.
9769 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
9770 used when arm_disable_literal_pool is enabled.
9771 (arm_max_const_double_inline_cost): Remove use of
9772 arm_disable_literal_pool.
9773 (push_minipool_fix): Add assert.
9774 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
9775 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
9776 (no_literal_pool_sf_immediate): New.
9778 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9780 PR tree-optimization/80613
9781 * tree-ssa-dce.c (propagate_necessity): Remove cases for
9782 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
9784 2017-05-05 Richard Biener <rguenther@suse.de>
9786 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
9788 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
9790 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
9791 of this flag from insn conditions due to removal from r247495.
9793 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
9795 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
9797 (arm_early_store_addr_dep_ptr): Likewise.
9798 * config/arm/aarch-common-protos.h
9799 (arm_early_load_addr_dep_ptr): Add prototype.
9800 (arm_early_store_addr_dep_ptr): Likewise.
9801 * config/arm/cortex-a53.md: Add new bypasses.
9803 2017-05-05 Jakub Jelinek <jakub@redhat.com>
9805 * tree.c (next_type_uid): Change type to unsigned.
9806 (type_hash_canon): Decrement back next_type_uid if
9807 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
9808 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
9811 2017-05-04 Martin Sebor <msebor@redhat.com>
9813 * builtins.c: Fix a trivial typo in a comment.
9816 * builtins.c (check_sizes): Adjust to handle reading past the end.
9817 Avoid printing excessive upper bound of ranges. Use %E to print
9818 tree nodes instead of converting them to %wu.
9819 (expand_builtin_memchr): New function.
9820 (compute_dest_size): Rename...
9821 (compute_objsize): ...to this.
9822 (expand_builtin_memcpy): Adjust.
9823 (expand_builtin_mempcpy): Adjust.
9824 (expand_builtin_strcat): Adjust.
9825 (expand_builtin_strcpy): Adjust.
9826 (check_strncat_sizes): Adjust.
9827 (expand_builtin_strncat): Adjust.
9828 (expand_builtin_strncpy): Adjust and simplify.
9829 (expand_builtin_memset): Adjust.
9830 (expand_builtin_bzero): Adjust.
9831 (expand_builtin_memcmp): Adjust.
9832 (expand_builtin): Handle memcmp.
9833 (maybe_emit_chk_warning): Check strncat just once.
9835 2017-05-04 Martin Sebor <msebor@redhat.com>
9837 PR preprocessor/79214
9840 * builtins.c (check_sizes): Add inlining context and issue
9841 warnings even when -Wno-system-headers is set.
9842 (check_strncat_sizes): Same.
9843 (expand_builtin_strncat): Same.
9844 (expand_builtin_memmove): New function.
9845 (expand_builtin_stpncpy): Same.
9846 (expand_builtin): Handle memmove and stpncpy.
9848 2017-05-04 Bin Cheng <bin.cheng@arm.com>
9850 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
9851 which is not used any more.
9853 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9855 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
9857 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9859 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
9860 (cortexa53_tunings): Likewise.
9861 (cortexa57_tunings): Likewise.
9862 (cortexa72_tunings): Likewise.
9863 (cortexa73_tunings): Likewise.
9865 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9867 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
9868 Set loop alignment to 8.
9870 2017-05-04 Martin Sebor <msebor@redhat.com>
9872 PR translation/80280
9873 * builtins.c (expand_builtin_object_size): Add missing quoting to
9874 %D and like directives.
9875 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
9876 (hsa_type_for_tree_type): Same.
9877 (verify_function_arguments): Same.
9878 * symtab.c (symbol_table::change_decl_assembler_name): Same.
9879 * varasm.c (get_section): Same.
9882 2017-05-04 Martin Sebor <msebor@redhat.com>
9884 PR translation/80280
9885 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
9887 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
9889 * config/aarch64/aarch64.c (generic_addrcost_table):
9890 Change HI/TI mode setting.
9892 2017-05-04 Martin Jambor <mjambor@suse.cz>
9894 PR tree-optimization/80622
9895 * tree-sra.c (comes_initialized_p): New function.
9896 (build_accesses_from_assign): Only set write lazily when
9897 comes_initialized_p is false.
9898 (analyze_access_subtree): Use comes_initialized_p.
9899 (propagate_subaccesses_across_link): Assert !comes_initialized_p
9900 instead of testing for PARM_DECL.
9902 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9904 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
9905 constraint on operand 0 to allow more general addressing modes.
9906 Adjust output template.
9907 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
9909 * config/aarch64/aarch64-protos.h
9910 (aarch64_address_valid_for_prefetch_p): Declare prototype.
9911 * config/aarch64/constraints.md (Dp): New address constraint.
9912 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
9915 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
9917 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
9918 update use of estimate_ipcp_clone_size_and_time.
9919 (estimate_local_effects): Update use of
9920 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
9921 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
9922 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
9923 Return nonspecialized time.
9925 2017-05-04 Richard Biener <rguenther@suse.de>
9927 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
9928 for the last VUSE which def dominates the PHI. Directly call
9930 (get_continuation_for_phi_1): Remove.
9932 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
9934 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
9935 to explain the use of truncating division. Cap the number of
9936 iterations to the maximum given by nb_iterations_upper_bound,
9939 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
9941 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
9942 * configure: Regenerate.
9943 * config.in: Regenerate.
9944 * config/i386/driver-mingw32.c: new file.
9945 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
9946 * config.host: Link driver-mingw32.o on MinGW host.
9947 * doc/install.texi: Document new --enable-mingw-wildcard configure
9950 2017-05-04 Marek Polacek <polacek@redhat.com>
9952 PR tree-optimization/80612
9953 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
9955 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9956 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
9958 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
9959 (movt splitter): Likewise.
9960 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
9961 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
9962 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
9963 block for Thumb-1 with MOVT.
9964 (thumb2_legitimate_address_p): Move code block ...
9965 (can_avoid_literal_pool_for_label_p): ... into this new function.
9966 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
9968 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
9969 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
9970 "M-profile targets with the MOVT instruction".
9972 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9974 * config/arm/arm-builtins.c (arm_init_builtins): Rename
9975 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
9976 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
9978 2017-05-04 Martin Liska <mliska@suse.cz>
9980 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
9983 2017-05-04 Richard Biener <rguenther@suse.de>
9985 * tree.c (array_at_struct_end_p): Handle arrays at struct
9986 end with flexarrays more conservatively. Refactor and treat
9987 arrays of arrays or aggregates more strict. Fix
9988 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
9989 * tree.c (array_at_struct_end_p): Adjust prototype.
9990 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
9991 * gimple-fold.c (get_range_strlen): Likewise.
9992 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
9994 2017-05-04 Richard Biener <rguenther@suse.de>
9996 PR tree-optimization/31130
9997 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
9999 (supports_overflow_infinity): Likewise.
10000 (is_negative_overflow_infinity): Likewise.
10001 (is_positive_overflow_infinity): Likewise.
10002 (is_overflow_infinity): Likewise.
10003 (stmt_overflow_infinity): Likewise.
10004 (overflow_infinity_range_p): Likewise.
10005 (usable_range_p): Remove as always returning true.
10006 (make_overflow_infinity): Remove.
10007 (negative_overflow_infinity): Likewise.
10008 (positive_overflow_infinity): Likewise.
10009 (avoid_overflow_infinity): Likewise.
10010 (set_value_range): Adjust accordingly.
10011 (set_value_range_to_nonnegative): Likewise, remove now unused
10012 overflow_infinity arg.
10013 (vrp_operand_equal_p): Adjust.
10014 (update_value_range): Likewise.
10015 (range_int_cst_singleton_p): Likewise.
10016 (operand_less_p): Likewise.
10017 (compare_values_warnv): Likewise.
10018 (extract_range_for_var_from_comparison_expr): Likewise.
10019 (vrp_int_const_binop): Likewise.
10020 (zero_nonzero_bits_from_vr): Likewise.
10021 (extract_range_from_multiplicative_op_1): Likewise.
10022 (extract_range_from_binary_expr_1): Likewise.
10023 (extract_range_from_unary_expr): Likewise.
10024 (extract_range_from_comparison): Likewise.
10025 (extract_range_basic): Likewise.
10026 (adjust_range_with_scev): Likewise.
10027 (compare_ranges): Likewise.
10028 (compare_range_with_value): Likewise.
10029 (dump_value_range): Likewise.
10030 (test_for_singularity): Likewise, remove strict_overflow_p parameter
10032 (simplify_cond_using_ranges): Adjust.
10034 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
10036 * brig-builtins.def: Added a builtin for class_f64.
10037 * builtin-types.def: Added a builtin type needed by class_f64.
10039 2017-05-03 Jason Merrill <jason@redhat.com>
10041 * timevar.def: Add TV_CONSTEXPR.
10043 2017-05-03 David Malcolm <dmalcolm@redhat.com>
10045 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
10047 2017-05-03 Martin Jambor <mjambor@suse.cz>
10049 * ipa-prop.c (ipa_update_after_lto_read): Removed.
10050 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
10051 * ipa-cp.c (ipcp_propagate_stage): Do not call
10052 ipa_update_after_lto_read.
10053 * ipa-inline.c (ipa_inline): Likewise.
10055 2017-05-03 Martin Jambor <mjambor@suse.cz>
10057 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
10058 tag. Added a default constructor and a destructor.
10059 (ipa_edge_args_sum_t): New class;
10060 (ipa_edge_args_sum): Declare.
10061 (ipa_edge_args_vector): Remove declaration.
10062 (IPA_EDGE_REF): Use ipa_edge_args_sum.
10063 (ipa_free_edge_args_substructures): Remove declaration.
10064 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
10065 (ipa_edge_args_info_available_for_edge_p): Likewise.
10066 * ipa-prop.c (ipa_edge_args_vector): Removed.
10067 (edge_removal_hook_holder): Likewise.
10068 (edge_duplication_hook_holder): Likewise.
10069 (ipa_edge_args_sum): New variable.
10070 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
10071 ipa_edge_args_vector.
10072 (ipa_free_edge_args_substructures): Likewise.
10073 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
10074 ipa_edge_args_vector.
10075 (ipa_edge_removal_hook): Turned into method
10076 ipa_edge_args_sum_t::remove.
10077 (ipa_edge_duplication_hook): Turned into method
10078 ipa_edge_args_sum_t::duplicate.
10079 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
10080 registering edge hooks.
10081 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
10082 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
10083 ipa_edge_args_sum instead of ipa_edge_args_vector.
10084 * ipa-profile.c (ipa_profile): Likewise.
10086 2017-05-03 Martin Jambor <mjambor@suse.cz>
10088 * symbol-summary.h (function_summary): New method exists.
10089 (function_summary::symtab_removal): Deallocate through release.
10090 (call_summary): New class.
10091 (gt_ggc_mx): New overload.
10092 (gt_pch_nx): Likewise.
10093 (gt_pch_nx): Likewise.
10095 2017-05-03 Jeff Law <law@redhat.com>
10097 PR tree-optimization/78496
10098 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
10099 from simplify_cond_using_ranges. Split off code to walk
10100 backwards through casts into ...
10101 (simplify_cond_using_ranges_2): New function.
10102 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
10103 (execute_vrp): After identifying jump threads, call
10104 simplify_cond_using_ranges_2.
10106 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
10109 * ipa-inline.h (inline_summary): Add ctor.
10110 (create_ggc): Do not use ggc_cleared_alloc.
10112 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
10113 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10115 * gcc.c (handle_braces): Support escaping in switch matching
10117 * doc/invoke.texi (Spec Files): Document it.
10118 Remove superfluous @code markup in items.
10120 2017-05-03 David Malcolm <dmalcolm@redhat.com>
10122 * diagnostic-show-locus.c (struct column_range): New struct.
10123 (get_affected_columns): New function.
10124 (get_printed_columns): New function.
10125 (struct correction): New struct.
10126 (correction::ensure_capacity): New function.
10127 (correction::ensure_terminated): New function.
10128 (struct line_corrections): New struct.
10129 (line_corrections::~line_corrections): New dtor.
10130 (line_corrections::add_hint): New function.
10131 (layout::print_trailing_fixits): Reimplement in terms of the new
10133 (selftest::test_overlapped_fixit_printing): New function.
10134 (selftest::diagnostic_show_locus_c_tests): Call it.
10136 2017-05-03 Nathan Sidwell <nathan@acm.org>
10138 Canonicalize canonical type hashing
10139 * tree.h (type_hash_canon_hash): Declare.
10140 * tree.c (type_hash_list, attribute_hash_list): Move into
10141 type_hash_canon_hash.
10142 (build_type_attribute_qual_variant): Break out hash code calc into
10143 type_hash_canon_hash.
10144 (type_hash_canon_hash): New. Generic type hash computation.
10145 (build_range_type_1, build_array_type_1, build_function_type,
10146 build_method_type_directly, build_offset_type, build_complex_type,
10147 make_vector_type): Call it.
10149 2017-05-03 Richard Biener <rguenther@suse.de>
10151 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
10152 When all DRs have unknown misaligned do not always peel
10153 when there is a store but apply the same costing model as if
10154 there were only loads.
10156 2017-05-03 Richard Biener <rguenther@suse.de>
10159 PR tree-optimization/80492
10160 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
10161 compare_base_decls returning dont-know properly.
10163 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
10165 * config/arm/iterators.md (CCSI): New mode iterator.
10166 (arch): New mode attribute.
10167 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
10168 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
10169 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
10170 code iterator for success result mode.
10171 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
10172 the corresponding new insn generators.
10174 2017-05-03 Bin Cheng <bin.cheng@arm.com>
10177 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10178 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
10180 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
10182 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
10183 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
10184 (DDR_A): Wrap DDR argument in brackets.
10185 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
10186 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
10187 (DDR_REVERSED_P): Likewise.
10189 2017-05-03 Jakub Jelinek <jakub@redhat.com>
10191 PR tree-optimization/79472
10192 * tree-switch-conversion.c (struct switch_conv_info): Add
10193 contiguous_range and default_case_nonstandard fields.
10194 (collect_switch_conv_info): Compute contiguous_range and
10195 default_case_nonstandard fields, don't clear final_bb if
10196 contiguous_range and only the default case doesn't have the required
10198 (check_all_empty_except_final): Set default_case_nonstandard instead
10199 of failing if contiguous_range and the default case doesn't have empty
10201 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
10202 and only the default case doesn't have the required constants. Skip
10204 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
10205 if default_case_nonstandard.
10206 (build_constructors): Build constant 1 just once. Assert that default
10207 values aren't inserted in between cases if contiguous_range. Skip
10209 (build_arrays): Skip virtual phis.
10210 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
10211 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
10212 Handle virtual phis.
10213 (gen_inbound_check): Handle default_case_nonstandard case.
10214 (process_switch): Adjust check_final_bb caller. Call
10215 gather_default_values with the first non-default case instead of
10216 default case if default_case_nonstandard.
10218 2017-05-02 Nathan Sidwell <nathan@acm.org>
10220 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
10221 check. Fix formatting.
10223 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
10225 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
10226 errors when comparing specialized and unspecialized times.
10228 2017-05-02 David Malcolm <dmalcolm@redhat.com>
10230 * diagnostic-show-locus.c
10231 (layout::should_print_annotation_line_p): Make private.
10232 (layout::print_annotation_line): Make private.
10233 (layout::annotation_line_showed_range_p): Make private.
10234 (layout::show_ruler): Make private.
10235 (layout::print_source_line): Make private. Pass in line and
10236 line_width, rather than calling location_get_source_line. Drop
10238 (layout::print_leading_fixits): New method.
10239 (layout::print_any_fixits): Rename to...
10240 (layout::print_trailing_fixits): ...this, and make private.
10241 Don't print newline fixits.
10242 (diagnostic_show_locus): Move logic for printing one row into...
10243 (layout::print_line): ...this new function. Move the
10244 location_get_source_line call and error-handling from
10245 print_source_line to here. Call print_leading_fixits, and rename
10246 print_any_fixits to print_trailing_fixits.
10247 (selftest::test_fixit_insert_containing_newline): Update now that
10248 newlines are partially supported.
10249 (selftest::test_fixit_insert_containing_newline_2): New test.
10250 (selftest::test_fixit_replace_containing_newline): Update comments.
10251 (selftest::diagnostic_show_locus_c_tests): Call the new test.
10252 * edit-context.c (class added_line): New class.
10253 (class edited_line): Describe newline handling in comment.
10254 (edited_line::actually_edited_p): New method.
10255 (edited_line::print_content): Delete redundant decl.
10256 (edited_line::m_predecessors): New field.
10257 (edited_file::print_content): Call edited_line::print_content.
10258 (edited_file::print_diff): Update to support newlines.
10259 (edited_file::print_diff_hunk): Likewise.
10260 (edited_file::print_run_of_changed_lines): New function.
10261 (edited_file::print_diff_line): Convert to...
10262 (print_diff_line): ...this.
10263 (edited_file::get_effective_line_count): New function.
10264 (edited_line::edited_line): Initialize new field m_predecessors.
10265 (edited_line::~edited_line): Clean up m_predecessors.
10266 (edited_line::apply_fixit): Handle newlines.
10267 (edited_line::get_effective_line_count): New function.
10268 (edited_line::print_content): New function.
10269 (edited_line::print_diff_lines): New function.
10270 (selftest::test_applying_fixits_insert_containing_newline): New
10272 (selftest::test_applying_fixits_replace_containing_newline): New
10274 (selftest::insert_line): New function.
10275 (selftest::test_applying_fixits_multiple_lines): Add example of
10277 (selftest::edit_context_c_tests): Call the new tests.
10279 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10281 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
10282 parameter cand. Update dump information.
10283 (get_computation_cost): Update uses.
10285 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10287 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
10288 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
10289 (get_computation_at, rewrite_use_address): Update use of
10290 get_computation_aff.
10292 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10294 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
10295 (get_computation): Delete.
10296 (get_computation_cost): Implement like get_computation_cost_at.
10297 Use get_computation_at.
10298 (get_computation_cost_at): Delete.
10299 (rewrite_use_nonlinear_expr): Use get_computation_at.
10300 (rewrite_use_compare, remove_unused_ivs): Ditto.
10302 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10304 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
10306 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10308 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
10309 (ivopts_global_cost_for_size): Rename parameter and update uses.
10310 (iv_ca_recount_cost): Update uses.
10311 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
10312 candidates seperately in n_invs and n_cands.
10313 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
10315 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10317 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
10318 (find_inv_vars_cb): New.
10319 (find_depends): Renamed to ...
10320 (find_inv_vars): ... this.
10321 (add_candidate_1, force_var_cost): Call find_inv_vars.
10322 (split_address_cost, determine_group_iv_cost_cond): Ditto.
10324 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10326 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
10327 inv_vars. Add inv_exprs.
10328 (struct iv_cand): Rename depends_on to inv_vars.
10329 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
10330 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
10331 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
10332 (dump_cand): Dump inv_vars.
10333 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
10334 (record_invariant, find_depends, add_candidate_1): Ditto.
10335 (set_group_iv_cost, force_var_cost): Ditto.
10336 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
10337 (get_computation_cost_at, get_computation_cost): Ditto.
10338 (determine_group_iv_cost_generic): Ditto.
10339 (determine_group_iv_cost_address): Ditto.
10340 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
10341 (determine_group_iv_costs): Ditto.
10342 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
10343 (iv_ca_set_remove_invariants): Renamed to ...
10344 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
10345 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
10346 (iv_ca_set_add_invariants): Renamed to ...
10347 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
10348 (iv_ca_set_cp): Use iv_ca_set_add_invs.
10349 (iv_ca_has_deps): Support inv_vars and inv_exprs.
10350 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
10351 (create_new_ivs): Remove useless dump.
10353 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10355 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
10357 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
10358 (iv_ca_set_no_cp, create_new_iv): Ditto.
10360 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10362 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
10364 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10366 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
10367 function tree_check2.
10369 2017-05-02 Martin Liska <mliska@suse.cz>
10371 * doc/gcov.texi: Add missing preposition.
10372 * gcov.c (function_info::function_info): Properly fill up
10373 all member variables.
10375 2017-05-02 Tamar Christina <tamar.christina@arm.com>
10377 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
10379 2017-05-02 Tamar Christina <tamar.christina@arm.com>
10381 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
10383 2017-05-02 Martin Liska <mliska@suse.cz>
10386 * lto-streamer-in.c (lto_read_tree_1): Remove
10387 LTO_STREAMER_DEBUG.
10388 * lto-streamer.c (struct tree_hash_entry): Likewise.
10389 (struct tree_entry_hasher): Likewise.
10390 (tree_entry_hasher::hash): Likewise.
10391 (tree_entry_hasher::equal): Likewise.
10392 (lto_streamer_init): Likewise.
10393 (lto_orig_address_map): Likewise.
10394 (lto_orig_address_get): Likewise.
10395 (lto_orig_address_remove): Likewise.
10396 * lto-streamer.h: Likewise.
10397 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
10398 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
10400 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
10402 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
10403 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
10404 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
10405 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
10406 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
10407 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
10408 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
10409 (mm_maskz_sub_ss): New intrinsics.
10410 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
10411 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
10412 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
10413 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
10414 (__builtin_ia32_subss_mask_round): New builtins.
10415 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
10416 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
10417 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
10419 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
10420 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
10422 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
10425 2017-05-02 Martin Jambor <mjambor@suse.cz>
10427 PR tree-optimization/78687
10428 * tree-sra.c (access): New field parent.
10429 (process_subtree_disqualification): New function.
10430 (disqualify_candidate): Call it.
10431 (build_accesses_from_assign): Reset write flag if creating an
10433 (build_access_subtree): Fill in parent field and also prpagate
10434 down grp_write flag.
10435 (create_artificial_child_access): New parameter set_grp_write, set
10436 grp_write to its value.
10437 (propagate_subaccesses_across_link): Also propagate grp_write flag
10439 (propagate_all_subaccesses): Push the closest parent back to work
10440 queue if add_access_to_work_queue returned true.
10442 2017-05-02 Richard Biener <rguenther@suse.de>
10444 * common.opt (fstrict-overflow): Alias negative to fwrapv.
10445 * doc/invoke.texi (fstrict-overflow): Remove all traces of
10446 -fstrict-overflow documentation.
10447 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
10448 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
10449 flag_strict_overflow.
10450 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
10451 * lto-opts.c (lto_write_options): Do not stream it.
10452 * lto-wrapper.c (merge_and_complain): Do not handle it.
10453 * opts.c (default_options_table): Do not set -fstrict-overflow.
10454 (finish_options): Likewise do not clear it when sanitizing.
10455 * simplify-rtx.c (simplify_const_relational_operation): Do not
10456 test flag_strict_overflow.
10458 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10460 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
10461 using enabled attribute.
10462 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
10463 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
10464 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
10465 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
10466 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
10467 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
10468 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
10469 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
10470 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
10471 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
10473 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10475 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
10477 2017-05-02 Richard Biener <rguenther@suse.de>
10479 PR tree-optimization/80591
10481 2017-04-10 Richard Biener <rguenther@suse.de>
10483 * tree-ssa-structalias.c (find_func_aliases): Properly handle
10486 2017-05-02 Richard Biener <rguenther@suse.de>
10488 PR tree-optimization/80549
10489 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
10490 (cleanup_tree_cfg_noloop): Create forwarders to known loop
10491 headers if they do not have a preheader.
10493 2017-05-02 Martin Liska <mliska@suse.cz>
10496 * common.opt: Fix typo.
10497 * doc/invoke.texi: Likewise.
10499 2017-05-01 Jan Beulich <jbeulich@suse.com>
10501 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
10502 swapping, add (x,x,m,x,n) alternative.
10504 2017-05-01 Nathan Sidwell <nathan@acm.org>
10506 * calls.c (combine_pending_stack_adjustment_and_call): Remove
10507 unnecessary unadjusted_alignment check.
10509 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10512 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
10514 * gimplify.c (gimplify_cilk_detach): New function.
10515 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
10516 * tree-core.h: Document EXPR_CILK_SPAWN.
10517 * tree.h (EXPR_CILK_SPAWN): Define.
10519 2017-05-01 David Malcolm <dmalcolm@redhat.com>
10521 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
10522 to use new fixit_hint representation, using the "replace" logic.
10523 (get_line_span_for_fixit_hint): Likewise.
10524 (layout::print_any_fixits): Likewise.
10525 (selftest::test_one_liner_many_fixits): Rename to...
10526 (selftest::test_one_liner_many_fixits_1): ...this, and update
10527 comment and expected output to reflect that the multiple fix-it
10528 hints are now consolidated into one insertion.
10529 (selftest::test_one_liner_many_fixits_2): New test.
10530 (selftest::test_diagnostic_show_locus_one_liner): Update for
10532 (selftest::test_fixit_consolidation): Update for fix-it API
10534 * diagnostic.c (print_parseable_fixits): Likewise.
10535 * edit-context.c (edited_line::m_line_events): Convert from
10536 auto_vec <line_event *> to auto_vec <line_event>.
10537 (class line_event): Convert from abstract base class to a concrete
10538 class, taking over the role of replace_event.
10539 (class insert_event): Delete.
10540 (class replace_event): Rename to class line_event. Convert to
10542 (edit_context::add_fixits): Reimplement.
10543 (edit_context::apply_insert): Delete.
10544 (edit_context::apply_replace): Rename to...
10545 (edit_context::apply_fixit): ...this. Convert to half-open range.
10546 (edited_file::apply_insert): Delete.
10547 (edited_file::apply_replace): Rename to...
10548 (edited_file::apply_fixit): ...this.
10549 (edited_line::~edited_line): Drop deletion of events.
10550 (edited_line::apply_insert): Delete.
10551 (edited_line::apply_replace): Rename to...
10552 (edited_line::apply_fixit): ...this. Convert to half-open range.
10553 Update for change to type of m_line_events.
10554 * edit-context.h (edit_context::apply_insert): Delete.
10555 (edit_context::apply_replace): Rename to...
10556 (edit_context::apply_fixit): ...this.
10558 2017-05-01 Martin Sebor <msebor@redhat.com>
10560 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
10563 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
10566 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
10567 __get_cpuid_max returns 0.
10568 (__get_cpuid_count): Ditto.
10570 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
10572 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
10573 replacement expression is another instance of one of its arguments.
10575 2017-05-01 Jakub Jelinek <jakub@redhat.com>
10578 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
10579 check for stack push/pop autoinc.
10580 * config/i386/i386.c (ix86_agi_dependent): Return false
10581 if the only reason why modified_in_p returned true is that
10582 addr is SP based and set_insn is a push or pop.
10584 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10586 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
10589 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
10592 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
10593 (account_size_time): Use two predicates - exec_pred and
10595 (evaluate_conditions_for_known_args): Compute both clause and
10597 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
10598 (inline_summary_t::duplicate): Update.
10599 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
10601 (compute_inline_parameters): Likewise.
10602 (estimate_edge_size_and_time): Update caluclation of time.
10603 (estimate_node_size_and_time): Compute both time and nonspecialized
10605 (estimate_ipcp_clone_size_and_time): Update.
10606 (inline_merge_summary): Update.
10607 (do_estimate_edge_time): Update.
10608 (do_estimate_edge_size): Update.
10609 (do_estimate_edge_hints): Update.
10610 (inline_read_section, inline_write_summary): Stream both new predicates.
10611 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
10613 (compute_inlined_call_time): Cleanup.
10614 (big_speedup_p): Update.
10615 (edge_badness): Update.
10616 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
10617 (size_time_entry): Replace predicate by exec_predicate and
10618 nonconst_predicate.
10619 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
10620 (estimate_edge_time): Return also nonspec_time.
10621 (reset_edge_growth_cache): Update.
10623 2017-04-29 Jakub Jelinek <jakub@redhat.com>
10625 PR rtl-optimization/80491
10626 * ifcvt.c (noce_process_if_block): When looking for x setter
10627 with missing else_bb, don't check only the insn right before
10628 cond_earliest, but look for the last insn that x is modified in
10629 within the same bb.
10631 PR rtl-optimization/80491
10632 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
10634 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
10636 PR tree-optimization/80487
10637 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
10639 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10641 PR tree-optimization/79697
10642 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
10643 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
10644 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
10646 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
10647 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
10649 2017-04-28 Martin Sebor <msebor@redhat.com>
10651 PR tree-optimization/80523
10652 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
10653 (init_target_to_host_charmap, target_to_host, target_strtol10): New
10655 (maybe_warn, format_directive, parse_directive): Use new functions.
10656 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
10658 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
10660 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
10662 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
10664 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
10665 target_header_dir): Set correctly.
10666 * configure: Regenerated.
10667 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
10668 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
10669 instead of SYSTEM_HEADER_DIR.
10671 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
10673 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
10674 (estimate_local_effects): Likewise.
10675 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
10676 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
10677 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
10678 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
10679 do_estimate_edge_time, estimate_edge_time): Likewise.
10680 * ipa-inline-analysis.c (estimate_node_size_and_time,
10681 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
10682 (estimate_time_after_inlining): Remove.
10684 2017-04-28 Martin Liska <mliska@suse.cz>
10686 * doc/gcov.texi: Enhance documentation of gcov.
10688 2017-04-28 Martin Liska <mliska@suse.cz>
10690 * doc/gcov.texi: Sort options in alphabetic order.
10691 * doc/gcov-dump.texi: Likewise.
10692 * doc/gcov-tool.texi: Likewise.
10693 * gcov.c (print_usage): Likewise.
10694 * gcov-dump.c (print_usage): Likewise.
10695 * gcov-tool.c (print_merge_usage_message): Likewise.
10696 (print_rewrite_usage_message): Likewise.
10697 (print_overlap_usage_message): Likewise.
10699 2017-04-28 Martin Liska <mliska@suse.cz>
10701 PR gcov-profile/53915
10702 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
10704 2017-04-28 Martin Liska <mliska@suse.cz>
10706 PR gcov-profile/79891
10707 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
10708 is marked by compiler as living on a line.
10709 (get_cycles_count): Remove usage of the union.
10710 (output_intermediate_file): Likewise.
10711 (find_source): Fix GNU coding style.
10712 (accumulate_line_counts): Remove old non-all block mode.
10713 (output_lines): Remove usage of the union.
10714 * profile.c (output_location): Include all BBs, even if
10715 belonging to a same line (and file) as a previous BB.
10717 2017-04-28 Martin Liska <mliska@suse.cz>
10719 * gcov.c (process_args): Handle new argument 'w'.
10720 (read_graph_file): Assign ID to BBs.
10721 (output_branch_count): Display BB # if verbose flag is set.
10722 (output_lines): Likewise for arcs.
10723 (print_usage): Add '--verbose' option help.
10724 * doc/gcov.texi: Document --verbose (-w) option.
10726 2017-04-28 Martin Liska <mliska@suse.cz>
10728 * gcov.c (struct block_location_info): New struct.
10729 (process_file): Fill up the new structure.
10730 (read_graph_file): Replace usage of encoding by the newly added
10732 (add_line_counts): Likewise.
10733 (accumulate_line_counts): Remove usage of the union.
10734 (function_info::function_info): New function.
10735 (function_info::~function_info): Likewise.
10736 (process_file): Call delete instead of release_function.
10737 (release_function): Release the function.
10738 (release_structures): Call delete instead of release_function.
10739 (solve_flow_graph): Replace usage of num_blocks.
10740 (find_exception_blocks): Likewise.
10741 (output_lines): Fix GNU coding style.
10743 2017-04-28 Martin Liska <mliska@suse.cz>
10746 * coverage.c (coverage_remove_note_file): New function.
10747 * coverage.h: Declare the function.
10748 * toplev.c (finalize): Clean if an error has been seen.
10750 2017-04-28 Martin Liska <mliska@suse.cz>
10752 PR gcov-profile/80031
10753 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
10754 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
10755 * gcov.c (read_graph_file): Read just number of blocks.
10756 * profile.c (branch_prob): Do not stream 0 flags per a basic
10759 2017-04-28 Martin Liska <mliska@suse.cz>
10761 * gcov-dump.c (tag_*): Add new argument to declarations.
10762 (dump_gcov_file): Likewise.
10763 (tag_blocks): Add and use new argument depth.
10764 (tag_arcs): Likewise.
10765 (tag_lines): Likewise.
10766 (tag_counters): Likewise.
10767 (tag_summary): Likewise.
10768 (dump_working_sets): Use depth to do a proper indentation.
10770 2017-04-28 Jakub Jelinek <jakub@redhat.com>
10773 * cgraph.h (symtab_node::debug_symtab): No longer inline.
10774 * symtab.c (symtab_node::debug_symtab): Move definition here.
10776 2017-04-28 Richard Biener <rguenther@suse.de>
10778 * lto-streamer.h (LTO_major_version): Bump to 7.
10780 2017-04-28 Richard Biener <rguenther@suse.de>
10782 * tree-vrp.c (assert_info): New struct.
10783 (add_assert_info): New helper.
10784 (register_edge_assert_for_2): Refactor to add asserts to a vector
10786 (register_edge_assert_for_1): Likewise.
10787 (register_edge_assert_for): Likewise.
10788 (finish_register_edge_assert_for): New helper actually registering
10789 asserts where live on edge.
10790 (find_conditional_asserts): Adjust.
10791 (find_switch_asserts): Likewise.
10792 (evrp_dom_walker::try_find_new_range): Generalize.
10793 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
10795 2017-04-27 Marek Polacek <polacek@redhat.com>
10798 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
10799 arg10 and arg11 to itype.
10801 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
10803 * doc/extend.texi (Object Size Checking): Improve grammar.
10805 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
10808 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
10809 that the logic for permitting reciprocal estimates matches that
10812 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10815 * tree.c (type_cache_hasher::equal): Only compare
10816 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
10817 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
10818 non-aggregate element types.
10819 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
10820 about the flag on ARRAY_TYPEs in the comment, formatting fix.
10822 2017-04-27 Richard Biener <rguenther@suse.de>
10824 PR middle-end/80533
10825 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
10826 stripping ARRAY_REFs from MEM_EXPR make sure we're not
10827 keeping a reference to a trailing array.
10829 2017-04-27 Richard Biener <rguenther@suse.de>
10831 PR middle-end/80539
10832 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
10833 being in loop-closed SSA form conservatively.
10834 (chrec_fold_multiply_poly_poly): Likewise.
10836 2017-04-27 Tamar Christina <tamar.christina@arm.com>
10838 PR middle-end/79665
10839 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
10840 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
10842 2017-04-27 Jakub Jelinek <jakub@redhat.com>
10845 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
10846 (aarch64_function_arg_alignment): Return unsigned int again, but still
10847 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
10848 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
10849 Don't emit -Wpsabi note.
10850 (aarch64_function_arg_boundary): Likewise.
10851 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
10854 2017-04-26 Nathan Sidwell <nathan@acm.org>
10856 * tree.h (crc32_unsigned_n): Declare.
10857 (crc32_unsigned, crc32_unsigned): Make inline.
10858 * tree.c (crc32_unsigned_bits): Replace with ...
10859 (crc32_unsigned_n): ... this.
10860 (crc32_unsigned, crc32_byte): Remove.
10861 (crc32_string): Remove unnecessary braces.
10863 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
10865 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
10866 * ipa-inline-analysis.c (MAX_TIME): Remove.
10867 (account_size_time): Use sreal for time.
10868 (dump_inline_summary): Update.
10869 (estimate_function_body_sizes): Update.
10870 (estimate_edge_size_and_time): Update.
10871 (estimate_calls_size_and_time): Update.
10872 (estimate_node_size_and_time): Update.
10873 (inline_merge_summary): Update.
10874 (inline_update_overall_summary): Update.
10875 (estimate_time_after_inlining): Update.
10876 (inline_read_section): Update.
10877 (inline_write_summary): Update.
10878 * ipa-inline.c (compute_uninlined_call_time): Update.
10879 (compute_inlined_call_time): Update.
10880 (recursive_inlining): Update.
10881 (inline_small_functions): Update.
10882 (dump_overall_stats): Update.
10883 * ipa-inline.h: Include sreal.h.
10884 (size_time_entry): Turn time to sreal.
10885 (inline_summary): Turn self_time nad time to sreal.
10887 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
10889 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
10891 (sreal::stream_out, sreal::stream_in): New.
10892 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
10894 2017-04-25 Jakub Jelinek <jakub@redhat.com>
10896 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
10899 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
10902 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
10904 (dimode_scalar_chain::compute_convert_gain): Ditto.
10905 (dimode_scalar_chain::make_vector_copies): Ditto.
10906 (dimode_scalar_chain::convert_reg): Ditto.
10907 (dimode_scalar_chain::convert_insn): Ditto.
10908 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
10909 (VI248_AVX512BW_1): New mode iterator.
10910 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
10911 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
10914 2017-04-25 Martin Sebor <msebor@redhat.com>
10916 PR tree-optimization/80497
10917 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
10918 constants are representable in HOST_WIDE_INT.
10919 (parse_directive): Ditto.
10921 2017-04-25 Martin Sebor <msebor@redhat.com>
10924 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
10925 (new_zero_array): Adjust signature.
10926 (dom_info::dom_init): Used unsigned rather that size_t.
10927 (dom_info::dom_info): Same.
10929 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10930 Jakub Jelinek <jakub@redhat.com>
10933 * config/arm/arm.c: Include gimple.h.
10934 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
10935 returns negative, increment ncrn only if it returned positive.
10936 (arm_needs_doubleword_align): Return int instead of bool,
10937 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
10938 members, but if there is any such non-FIELD_DECL
10939 > PARM_BOUNDARY aligned decl, return -1 instead of false.
10940 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
10941 returns negative, increment nregs only if it returned positive.
10942 (arm_setup_incoming_varargs): Likewise.
10943 (arm_function_arg_boundary): Emit -Wpsabi note if
10944 arm_needs_doubleword_align returns negative, return
10945 DOUBLEWORD_ALIGNMENT only if it returned positive.
10947 2017-04-25 Marek Polacek <polacek@redhat.com>
10950 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
10951 first argument to type.
10953 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
10956 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
10957 type checks to test for compatibility instead of equality.
10959 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10960 Jakub Jelinek <jakub@redhat.com>
10963 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
10965 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
10966 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
10967 the alignment computation, but return their maximum in warn_alignment.
10968 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
10969 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
10971 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
10972 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
10975 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10977 * config/arc/simdext.md (dmpyh): Fix typo.
10979 2017-04-25 Richard Biener <rguenther@suse.de>
10981 PR tree-optimization/80492
10982 * alias.c (compare_base_decls): Handle registers with asm
10983 specification conservatively.
10984 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
10985 compare_base_decls returning dont-know properly.
10987 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10989 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
10990 (legitimate_offset_address_p): New function.
10991 (arc_legitimate_address_p): Use above function.
10993 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10995 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
10997 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
10999 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
11000 ACCH registers whenever they are available.
11002 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11004 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
11005 double regs fix when not used.
11007 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11009 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
11011 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
11012 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
11014 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11016 * config/arc/arc.c (arc_output_addsi): Check for h-register class
11017 when emitting short ADD instructions.
11019 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11021 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
11023 (cmpsi_cc_c_insn): Likewise.
11024 (cbranchsi4_scratch): Compute proper instruction length using
11025 compact_hreg_operand.
11026 * config/arc/predicates.md (compact_hreg_operand): New predicate.
11028 2017-04-25 Richard Biener <rguenther@suse.de>
11030 PR middle-end/80509
11031 * passes.c (pass_manager::pass_manager): Initialize
11032 m_name_to_pass_map.
11034 2017-04-25 Richard Biener <rguenther@suse.de>
11036 PR tree-optimization/79201
11037 * tree-ssa-sink.c (statement_sink_location): Handle calls.
11039 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11042 * config/s390/vector.md: Split MEM->GPR vector moves for
11043 non-s_operand addresses.
11045 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11048 * config/s390/predicates.md (reload_const_wide_int_operand): New
11050 * config/s390/s390.md ("movti"): Remove d/P alternative.
11051 ("movti_bigconst"): New pattern definition.
11053 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11056 * s390-protos.h (s390_expand_cs_hqi): Removed.
11057 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
11058 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
11059 modes as well as CCZ1mode and CCZmode.
11060 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
11061 signature of s390_emit_compare_and_swap.
11062 (s390_expand_cs_hqi): Likewise, make static.
11063 (s390_expand_cs_tdsi): Generate an explicit compare before trying
11064 compare-and-swap, in some cases.
11065 (s390_expand_cs): Wrapper function.
11066 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
11068 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
11069 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
11070 patterns for small and large integers. Forbid symref memory operands.
11071 Move expander to s390.c. Require cc register.
11072 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
11073 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
11074 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
11075 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
11076 symref memory operands. Remove CC mode and call s390_match_ccmode
11078 ("atomic_exchange<mode>"): Allow and implement all integer modes.
11080 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11082 * config/s390/s390.md (define_peephole2): New peephole to help
11083 combining the load-and-test pattern with volatile memory.
11085 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11087 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
11088 with CCZmode for TARGET_Z196.
11090 2017-04-25 Jakub Jelinek <jakub@redhat.com>
11092 PR rtl-optimization/80501
11093 * combine.c (make_compound_operation_int): Set subreg_code to SET
11094 even for AND with mask of the sign bit of mode.
11096 PR rtl-optimization/80500
11097 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
11098 sum's initial value.
11100 2017-04-25 Julian Brown <julian@codesourcery.com>
11101 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11103 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
11105 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
11107 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
11109 2017-04-25 Julian Brown <julian@codesourcery.com>
11110 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11112 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
11113 (thunderx2t99_sha): New Reservation.
11115 2017-04-25 Julian Brown <julian@codesourcery.com>
11116 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11118 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
11119 type for 1-element load.
11121 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
11123 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
11125 2017-04-24 Martin Jambor <mjambor@suse.cz>
11127 PR tree-optimization/80293
11128 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
11129 char arrays not totally scalarizable if it is false.
11130 (analyze_all_variable_accesses): Pass correct value in the new
11131 parameter. Add a statistics counter.
11133 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
11135 PR middle-end/79931
11136 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
11138 2017-04-24 Richard Biener <rguenther@suse.de>
11140 PR tree-optimization/80494
11141 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
11142 out for complex types.
11144 2017-04-24 Richard Biener <rguenther@suse.de>
11146 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
11147 * tree-ssa-sccvn.c (print_scc): Print SCC size.
11148 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
11149 (DFS): Adjust and never fail.
11150 (sccvn_dom_walker::fail): Remove.
11151 (sccvn_dom_walker::before_dom_children): Adjust.
11152 (run_scc_vn): Likewise and never fail.
11153 * tree-ssa-pre.c (pass_pre::execute): Adjust.
11154 (pass_fre::execute): Likewise.
11156 2017-04-24 Richard Biener <rguenther@suse.de>
11158 PR tree-optimization/79725
11159 * tree-ssa-sink.c (statement_sink_location): Return whether
11160 failure reason was zero uses. Move that check later.
11161 (sink_code_in_bb): Deal with zero uses by removing the stmt
11164 2017-04-24 Richard Biener <rguenther@suse.de>
11167 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
11168 pointer-based references.
11170 2017-04-24 Richard Biener <rguenther@suse.de>
11173 * pass_manager.h (pass_manager::operator new): Remove.
11174 (pass_manager::operator delete): Likewise.
11175 * passes.c (pass_manager::operator new): Remove.
11176 (pass_manager::operator delete): Likewise.
11177 (pass_manager::pass_manager): Zero individual pass members.
11179 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
11182 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
11183 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
11184 Check "XEXP (src, 1)" operand here.
11185 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
11186 Check "XEXP (src, 1)" operand here.
11187 (dimode_scalar_chain::make_vector_copies): Detect count register
11188 of a shift instruction. Zero extend count register from QImode
11189 to DImode to satisfy vector shift pattern count operand predicate.
11190 Substitute vector shift count operand with a DImode copy.
11191 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
11194 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
11196 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
11197 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
11198 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
11199 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
11200 (UNSPEC_NOREX_MEM): Remove definition.
11202 2017-04-21 Richard Biener <rguenther@suse.de>
11204 PR tree-optimization/79547
11205 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
11206 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
11207 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
11208 without any constraints.
11210 2017-04-21 Richard Biener <rguenther@suse.de>
11212 PR tree-optimization/78847
11213 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
11215 2017-04-21 Richard Biener <rguenther@suse.de>
11217 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
11218 (build_distinct_type_copy): Likewise.
11219 (build_variant_type_copy): Likewise.
11220 * tree.c (build_qualified_type): Pass down mem-stat info.
11221 (build_distinct_type_copy): Likewise.
11222 (build_variant_type_copy): Likewise.
11224 2017-04-21 Richard Biener <rguenther@suse.de>
11226 PR tree-optimization/80237
11227 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
11229 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
11230 for a simplified result.
11232 2016-04-21 Richard Biener <rguenther@suse.de>
11234 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
11235 sth as strict as a simple_iv but a chrec without symbols and an
11236 operand defined in the loop we are peeling (and not some subloop).
11237 (propagate_constants_for_unrolling): Propagate all constants.
11239 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
11242 * config/i386/i386.c (print_reg): Remove assert for disalowed
11243 regno values, call output_operand_lossage instead.
11245 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
11248 * config/i386/constraints.md (Yc): New register constraint.
11249 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
11250 Use Yc constraint for alternative 2 of operand 0. Remove
11251 preferred_for_speed attribute.
11253 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
11255 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
11256 lastprivate clauses in SIMT case.
11258 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
11260 * doc/invoke.texi (-Wextra-semi): Document new warning option.
11262 2017-04-20 Richard Biener <rguenther@suse.de>
11264 PR tree-optimization/57796
11265 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
11266 as N scalar stores.
11267 (vect_model_load_cost): Cost gathers as N scalar loads.
11269 2017-04-20 Richard Biener <rguenther@suse.de>
11271 * ggc-page.c (ggc_allocated_p): Rename to ...
11272 (safe_lookup_page_table_entry): ... this and return the lookup
11274 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
11276 2017-04-20 Richard Biener <rguenther@suse.de>
11278 PR tree-optimization/80453
11279 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
11280 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
11281 from the conditions.
11282 (vn_phi_eq): Pass them down.
11283 (vn_phi_lookup): Record them.
11284 (vn_phi_insert): Likewise.
11286 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
11288 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
11289 uninitialized variable warning to avoid buffer overrun.
11291 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
11294 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
11295 is suppressed for '{ 0 }' in C.
11297 2017-04-20 Jakub Jelinek <jakub@redhat.com>
11299 * BASE-VER: Set to 8.0.0.
11301 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
11303 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
11304 priority .init_array and .fini_array section with SECTION_NOTYPE
11307 2017-04-20 Jakub Jelinek <jakub@redhat.com>
11309 PR middle-end/80423
11310 * tree.h (build_array_type): Add typeless_storage default argument.
11311 * tree.c (type_cache_hasher::equal): Also compare
11312 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
11313 (build_array_type): Add typeless_storage argument, set
11314 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
11316 (build_nonshared_array_type): Adjust build_array_type_1 caller.
11317 (build_array_type): Likewise. Add typeless_storage argument.
11319 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
11320 Jakub Jelinek <jakub@redhat.com>
11322 PR tree-optimization/80426
11323 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
11324 operation on symbolic operands, also compute the overflow for the
11325 invariant part when the operation degenerates into a negation.
11327 2017-04-19 Jakub Jelinek <jakub@redhat.com>
11330 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
11331 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
11334 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
11336 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
11339 * config/avr/avr.c (tree.h): Include it.
11340 (cgraph.h): Include it.
11341 (avr_encode_section_info): Don't warn for uninitialized progmem
11342 variable if it's just an alias.
11344 2017-04-19 Richard Biener <rguenther@suse.de>
11347 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
11348 when needed by AutoPGO.
11350 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
11353 * doc/lto.texi: Remove an extra 'that'.
11355 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
11357 PR rtl-optimization/80429
11358 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
11359 are only used in debug insns.
11361 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
11362 Vladimir Makarov <vmakarov@redhat.com>
11364 * config/sparc/predicates.md (input_operand): Add comment. Return
11365 true for any memory operand when LRA is in progress.
11366 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
11368 2017-04-18 Jeff Law <law@redhat.com>
11371 * mips.md ({return,simple_return}_internal): Do not overwrite
11374 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11376 PR tree-optimization/80443
11377 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
11378 instead of adding 1, subtract -1 and similarly instead of subtracting
11381 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
11383 PR rtl-optimization/80357
11384 * haifa-sched.c (tmp_bitmap): New variable.
11385 (model_recompute): Handle duplicate use records.
11386 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
11387 (free_global_sched_pressure_data): Free it.
11389 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
11392 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
11393 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
11394 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
11395 instead of SYSTEM_HEADER_DIR.
11397 2017-04-18 Jeff Law <law@redhat.com>
11399 PR middle-end/80422
11400 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
11401 predecessors after walking up the insn chain.
11403 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11406 * dwarf2out.c (modified_type_die): Try harder not to emit internal
11407 sizetype type into debug info.
11409 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
11412 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
11413 unneeded test for TARGET_UPPER_REGS_SF.
11414 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
11416 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11419 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
11420 instead of gsi_after_labels.
11422 2017-04-18 Jeff Law <law@redhat.com>
11424 * regcprop.c (maybe_mode_change): Avoid creating copies of the
11428 2017-04-13 Jeff Law <law@redhat.com>
11429 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11430 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11432 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
11435 * config/avr/avr.c (intl.h): Include it.
11436 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
11438 2017-04-18 Martin Liska <mliska@suse.cz>
11440 PR gcov-profile/78783
11441 * gcov-tool.c (gcov_output_files): Validate that destination
11442 file is either removed by the tool or by a user.
11444 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
11445 Guy Benyei <guybe@mellanox.com>
11447 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
11448 block, and do not negate it, the stored id is already negative.
11450 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
11452 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
11454 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
11457 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
11458 masks of options that should be turned off if the VSX vector
11459 options are turned off.
11460 (OTHER_P8_VECTOR_MASKS): Likewise.
11461 (OTHER_VSX_VECTOR_MASKS): Likewise.
11462 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
11463 rs6000_disable_incompatible_switches to validate no type switches
11465 (rs6000_incompatible_switch): New function to disallow turning on
11466 other vector options if -mno-vsx, -mno-power8-vector, or
11467 -mno-power9-vector are specified.
11469 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11471 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
11473 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11475 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
11476 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
11477 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
11478 (ARG_POINTER_CFA_OFFSET): Likewise.
11480 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11482 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
11483 conditions to take advantage of various optimizations.
11485 2017-04-13 Jeff Law <law@redhat.com>
11487 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11488 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11489 (zero_extendsidi2_dext): Likewise.
11491 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11494 * fold-const.c (fold_ternary_loc): Revert
11495 use op0 instead of fold_convert_loc (loc, type, arg0) part of
11498 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
11500 PR rtl-optimization/80343
11501 * lra-remat.c (update_scratch_ops): Assign original hard reg to
11502 new scratch pseudo.
11504 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
11507 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
11508 to ubsan_encode_value.
11510 2017-04-13 Jeff Law <law@redhat.com>
11512 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
11513 appearing in DEBUG_INSNs.
11515 2017-04-13 Martin Liska <mliska@suse.cz>
11517 PR gcov-profile/80413
11518 * gcov-io.c (gcov_write_string): Copy to buffer just when
11519 allocated size is greater than zero.
11521 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11524 * dwarf2out.c (decls_for_scope): Ignore declarations of
11525 current_function_decl in BLOCK_NONLOCALIZED_VARS.
11527 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
11530 * ipa-visibility.c (non_local_p): Fix typos.
11531 (localize_node): When localizing symbol in same comdat group,
11532 dissolve the group only when we know external symbols are going
11534 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
11536 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11538 PR tree-optimization/79390
11539 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
11540 order does not result in usable sequence, retry with reversed operand
11546 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
11547 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
11548 op0 instead of fold_convert_loc (loc, type, arg0).
11550 2017-04-12 Jeff Law <law@redhat.com>
11552 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
11553 has a delay slot in the generated code.
11555 * config/cris/cris.md (cris_preferred_reload_class): Return
11556 GENNONACR_REGS rather than GENERAL_REGS.
11558 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11561 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
11562 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
11563 signedness of the result type.
11565 2017-04-12 Richard Biener <rguenther@suse.de>
11566 Jeff Law <law@redhat.com>
11568 PR tree-optimization/80359
11569 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
11570 trim stores to TARGET_MEM_REFs.
11572 2017-04-12 Richard Biener <rguenther@suse.de>
11574 PR tree-optimization/79390
11575 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
11576 threading case even more.
11578 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
11581 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
11582 for quad_address_p for TImode, instead of just not indexed_address.
11584 2017-04-12 Richard Biener <rguenther@suse.de>
11585 Bernd Edlinger <bernd.edlinger@hotmail.de>
11587 PR middle-end/79671
11588 * alias.c (component_uses_parent_alias_set_from): Handle
11589 TYPE_TYPELESS_STORAGE.
11590 (get_alias_set): Likewise.
11591 * tree-core.h (tree_type_common): Add typeless_storage flag.
11592 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
11593 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
11594 for types containing members with TYPE_TYPELESS_STORAGE.
11595 (place_field): Likewise.
11596 (layout_type): Likewise for ARRAY_TYPE.
11597 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
11598 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
11599 TYPE_TYPELESS_STORAGE.
11600 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
11602 2017-04-12 Jakub Jelinek <jakub@redhat.com>
11605 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
11606 first argument to type.
11608 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11612 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
11613 CONST0_RTX (mode) rather than const0_rtx where appropriate.
11614 (rs6000_expand_binop_builtin): Likewise.
11615 (rs6000_expand_ternop_builtin): Likewise; also add missing
11616 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
11617 vshasigma built-ins.
11618 * doc/extend.texi: Document that vec_xxpermdi's third argument
11619 must be a constant.
11621 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
11623 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
11624 Use shift_const cost parameter when calculating gain of STV shifts.
11626 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
11628 PR rtl-optimization/70478
11629 * lra-constraints.c (process_alt_operands): Check memory for
11630 disfavoring memory insn operand.
11632 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11634 PR middle-end/80100
11635 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
11636 left shift in unsigned HOST_WIDE_INT type.
11638 PR rtl-optimization/80385
11639 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
11640 (not (neg X)) into (plus X -1) for complex or non-integral modes.
11643 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
11644 if they have any depend clauses.
11646 2017-04-11 Martin Liska <mliska@suse.cz>
11649 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
11650 * ipa-split.c (split_function): Create a local comdat symbol
11651 if caller is in a comdat group.
11653 2017-04-11 Martin Liska <mliska@suse.cz>
11656 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
11659 2017-04-11 Martin Sebor <msebor@redhat.com>
11661 PR middle-end/80364
11662 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
11663 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
11665 (directive::set_width, directive::set_precision, format_character):
11667 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
11670 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
11673 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
11674 conflict, set target->arch_name instead of target->cpu_name.
11676 2017-04-11 Richard Biener <rguenther@suse.de>
11678 PR tree-optimization/80374
11679 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
11680 build_zero_cst, remove fold_convertible_p check again.
11682 2017-04-11 Martin Liska <mliska@suse.cz>
11685 * ubsan.c (instrument_object_size): Do not instrument register
11688 2017-04-11 Jakub Jelinek <jakub@redhat.com>
11691 * config/i386/i386-builtin-types.def
11692 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
11693 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
11694 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
11695 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
11696 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
11697 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
11698 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
11699 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
11700 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
11701 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
11702 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
11703 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
11704 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
11705 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
11706 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
11707 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
11708 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
11709 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
11710 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
11711 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
11712 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
11713 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
11714 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
11715 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
11716 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
11717 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
11718 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
11719 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
11720 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
11721 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
11722 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
11723 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
11724 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
11725 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
11726 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
11727 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
11728 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
11729 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
11730 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
11731 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
11732 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
11733 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
11734 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
11735 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
11736 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
11737 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
11739 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
11740 flag to second_arg_count, handle 4 argument function type _COUNT
11741 aliases, handle second_arg_count on second argument rather than last.
11743 2017-04-10 Jeff Law <law@redhat.com>
11745 PR tree-optimization/80374
11746 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
11747 record anything if we can not convert integer_zero_node to the
11750 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
11753 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11754 Enhance special handling given to the TARGET_P9_MINMAX option in
11755 relation to certain other options.
11757 2017-04-10 Bin Cheng <bin.cheng@arm.com>
11759 PR tree-optimization/80153
11760 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
11761 remove POINTER_PLUS_EXPR's base part directly, rather than through
11764 2017-04-10 Richard Biener <rguenther@suse.de>
11765 Bin Cheng <bin.cheng@arm.com>
11767 PR tree-optimization/80153
11768 * tree-affine.c (aff_combination_to_tree): Get base pointer from
11769 the first element of pointer type aff_tree. Build result expr in
11771 (add_elt_to_tree): Convert to type unconditionally. Remove other
11772 fold_convert calls.
11773 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
11774 (rewrite_use_nonlinear_expr): Check invariant using iv information.
11776 2017-04-10 Richard Biener <rguenther@suse.de>
11778 * tree-ssa-structalias.c (find_func_aliases): Properly handle
11781 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
11783 PR rtl-optimization/70478
11784 * lra-constraints.c (curr_small_class_check): New.
11785 (update_and_check_small_class_inputs): New.
11786 (process_alt_operands): Update curr_small_class_check. Disfavor
11787 alternative insn memory operands. Check available regs for small
11790 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
11793 * config/mips/mips.opt (-mvirt): Update description.
11794 * doc/invoke.texi (-mvirt): Likewise.
11796 2017-04-10 Richard Biener <rguenther@suse.de>
11798 PR middle-end/80362
11799 * fold-const.c (fold_binary_loc): Look at unstripped ops when
11800 looking for NEGATE_EXPR in -A / -B to A / B folding.
11802 2017-04-10 Martin Liska <mliska@suse.cz>
11804 PR gcov-profile/80224
11805 * gcov.c (print_usage): Fix usage string.
11806 (get_gcov_intermediate_filename): Remove.
11807 (output_gcov_file): Use both for normal and intermediate format.
11808 (generate_results): Do not initialize special file for
11809 intermediate format.
11811 2017-04-10 Richard Biener <rguenther@suse.de>
11813 PR tree-optimization/80304
11814 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
11817 2017-04-10 Nathan Sidwell <nathan@acm.org>
11820 * config/rs6000/rs6000.c (rs6000_vector_type): New.
11821 (rs6000_init_builtins): Use it.
11823 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11825 * config/arm/arm.md (<mrc>): Add mode to SET source.
11826 (<mrrc>): Likewise.
11828 2017-04-10 Richard Biener <rguenther@suse.de>
11830 PR middle-end/80344
11831 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
11833 2017-04-10 Jakub Jelinek <jakub@redhat.com>
11836 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
11837 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
11838 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
11839 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
11840 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
11841 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
11842 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
11843 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
11844 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
11845 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
11846 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
11847 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
11848 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
11849 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
11850 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
11851 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
11852 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
11853 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
11854 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
11855 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
11856 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
11857 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
11858 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
11860 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
11862 PR rtl-optimization/70478
11863 * lra-constraints.c: Reverse the last patch.
11865 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
11867 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
11868 Add comment for WCHAR_T.
11870 2017-04-08 Martin Liska <mliska@suse.cz>
11873 2017-04-07 Martin Liska <mliska@suse.cz>
11876 * ipa-split.c (split_function): Add function part to a same comdat
11879 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11882 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
11884 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
11886 * rs6000/rs6000.c (vec_load_pendulum): Rename...
11887 (vec_pairing): ...to this.
11888 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
11889 (rs6000_sched_init): Adjust for name change.
11890 (struct rs6000_sched_context): Likewise.
11891 (rs6000_init_sched_context): Likewise.
11892 (rs6000_set_sched_context): Likewise.
11894 2017-04-07 Jakub Jelinek <jakub@redhat.com>
11900 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
11902 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
11903 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
11904 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
11906 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
11908 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
11910 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
11912 PR rtl-optimization/70703
11913 * ira-color.c (update_conflict_hard_regno_costs): Use
11914 int64_t instead of HOST_WIDE_INT.
11916 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
11918 PR rtl-optimization/70478
11919 * lra-constraints.c (process_alt_operands): Disfavor alternative
11920 insn memory operands.
11922 2017-04-07 Jeff Law <law@redhat.com>
11924 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
11925 CALL and NOTE_INSN_CALL_ARG_LOCATION.
11927 2017-04-07 Martin Liska <mliska@suse.cz>
11930 * config/aarch64/aarch64.c (aarch64_process_target_attr):
11931 Show error message instead of an ICE.
11933 2017-04-07 Martin Liska <mliska@suse.cz>
11936 * ipa-split.c (split_function): Add function part to a same comdat
11939 2017-04-07 Richard Biener <rguenther@suse.de>
11941 PR middle-end/80341
11942 * tree.c (get_unwidened): Also handle ! for_type case for
11944 * convert.c (do_narrow): Split out from ...
11945 (convert_to_integer_1): ... here. Do not pass final truncation
11946 type to get_unwidened for TRUNC_DIV_EXPR.
11948 2017-04-07 Richard Biener <rguenther@suse.de>
11950 * tree-affine.c (wide_int_ext_for_comb): Take type rather
11952 (aff_combination_const): Adjust.
11953 (aff_combination_scale): Likewise.
11954 (aff_combination_add_elt): Likewise.
11955 (aff_combination_add_cst): Likewise.
11956 (aff_combination_convert): Likewise.
11957 (add_elt_to_tree): Likewise. Remove unused argument.
11958 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
11960 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
11962 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
11964 * config/arm/arm.c (arm_default_short_enums): Use
11965 ARM_DEFAULT_SHORT_ENUMS.
11966 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
11968 2017-04-06 Jakub Jelinek <jakub@redhat.com>
11971 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
11972 members with redundant out-of-class redeclaration.
11974 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
11977 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
11978 * config/i386/i386.md (*zero_extendsidi2):
11979 Add (?*x,*x) and (?*v,*v) alternatives.
11981 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
11984 * config/i386/i386.c (ix86_expand_builtin)
11985 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
11986 mode from insn data. Convert operands to insn operand mode.
11987 Copy operands that don't satisfy insn predicate to a register.
11989 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
11991 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
11994 2017-04-06 Richard Biener <rguenther@suse.de>
11996 PR tree-optimization/80334
11997 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
11998 preserve alignment of accesses.
12000 2017-04-06 Richard Biener <rguenther@suse.de>
12002 PR tree-optimization/80262
12003 * tree-sra.c (build_ref_for_offset): Preserve address-space
12005 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
12006 Drop useless address-space information on MEM_REF offsets.
12008 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
12010 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
12012 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
12014 PR rtl-optimization/70703
12015 * ira-color.c (update_conflict_hard_regno_costs): Use
12016 HOST_WIDE_INT instead of long.
12018 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
12021 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
12022 not defined for x86_64 target. Add -mmmx target option when __SSE2__
12024 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
12025 for x86_64 target. Handle -m3dnowa option.
12027 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
12029 PR rtl-optimization/70703
12030 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
12031 (update_conflict_hard_regno_costs): Use long instead of unsigned
12032 arithmetic for cost calculation.
12034 2017-04-05 Jakub Jelinek <jakub@redhat.com>
12035 Bernd Edlinger <bernd.edlinger@hotmail.de>
12038 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
12041 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
12044 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
12045 ptr_mode with Pmode throughout.
12046 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
12047 into probe_stack_range and use DImode.
12049 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
12052 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
12053 call_eh_return is true.
12055 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12057 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
12058 Initialize last_match_fntype_index.
12060 2017-04-05 Jakub Jelinek <jakub@redhat.com>
12063 * tree-nvr.c: Include internal-fn.h.
12064 (pass_return_slot::execute): Ignore internal calls without
12067 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12068 Richard Biener <rguenther@suse.de>
12071 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
12072 captures used multiple times, except for the last use.
12073 * generic-match-head.c: Include gimplify.h.
12075 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12077 PR tree-optimization/79390
12078 * target.h (struct noce_if_info): Declare.
12079 * targhooks.h (default_noce_conversion_profitable_p): Declare.
12080 * target.def (noce_conversion_profitable_p): New target hook.
12081 * ifcvt.h (struct noce_if_info): New type, moved from ...
12082 * ifcvt.c (struct noce_if_info): ... here.
12083 (noce_conversion_profitable_p): Renamed to ...
12084 (default_noce_conversion_profitable_p): ... this. No longer
12086 (noce_try_store_flag_constants, noce_try_addcc,
12087 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
12088 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
12089 instead of noce_conversion_profitable_p.
12090 * config/i386/i386.c: Include ifcvt.h.
12091 (ix86_option_override_internal): Don't override
12092 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
12093 (ix86_noce_conversion_profitable_p): New function.
12094 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
12095 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
12096 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
12097 * doc/tm.texi: Regenerated.
12099 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12101 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
12104 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
12107 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
12108 instructions for small multiply cores.
12110 2017-04-04 Jeff Law <law@redhat.com>
12112 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
12114 (mips_expand_vec_perm_const): Initialize elements in orig_perm
12115 that are not set by the loop over the elements.
12117 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12120 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
12121 int mode, convert_modes it to mode as unsigned, otherwise use
12122 lowpart_subreg to mode rather than SImode.
12123 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
12124 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
12125 Use DImode instead of SImode for the shift count operand.
12126 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
12129 2017-04-04 Richard Biener <rguenther@suse.de>
12131 PR middle-end/80281
12132 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
12133 arithmetic done for the negate or the plus. Simplify.
12134 (A - (-B) -> A + B): Likewise.
12135 * fold-const.c (split_tree): Make sure to not negate pointers.
12137 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
12139 PR rtl-optimization/60818
12140 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
12141 a compare of comparisons with the thing compared if this results
12142 in a different machine mode.
12144 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
12146 * alias.c (base_alias_check): Fix typo in comment.
12147 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
12148 * cgraphunit.c (symbol_table::compile): Likewise.
12149 * collect2.c (maybe_run_lto_and_relink): Likewise.
12150 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
12151 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
12152 * config/avr/avr.c (avr_map_op_t): Likewise.
12153 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
12154 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
12155 * config/epiphany/epiphany.md (movcc): Likewise.
12156 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
12157 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
12159 * config/mips/mips.c (mips_save_restore_reg): Likewise.
12160 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
12161 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
12162 * config/sh/sh.c (sh_rtx_costs): Likewise.
12163 * fold-const.c (fold_truth_andor): Likewise.
12164 * genautomata.c (collapse_flag): Likewise.
12165 * gengtype.h (struct type::u::s): Likewise.
12166 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
12167 * input.c (FORMAT_AMOUNT): Likewise.
12168 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
12169 (known_aggs_to_agg_replacement_list): Likewise.
12170 * ipa-inline-analysis.c: Likewise.
12171 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
12172 * ipa-polymorphic-call.c
12173 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
12174 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
12175 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
12177 * modulo-sched.c (apply_reg_moves): Likewise.
12178 * omp-expand.c (build_omp_regions_1): Likewise.
12179 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
12180 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
12181 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
12182 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
12183 * value-prof.c: Likewise.
12184 * var-tracking.c (val_reset): Likewise.
12186 2017-04-03 Richard Biener <rguenther@suse.de>
12188 PR tree-optimization/80275
12189 * fold-const.c (split_address_to_core_and_offset): Handle
12192 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
12194 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
12195 descriptors is at least equal to that of functions.
12197 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
12199 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
12201 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
12204 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
12205 (mov<IMOD4:mode>): New expander.
12206 (*mov<IMOD4:mode>_internal): New insn and split pattern.
12208 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
12210 PR rtl-optimization/79405
12211 * fwprop.c (propagations_left): New variable.
12212 (forward_propagate_into): Decrement it.
12213 (fwprop_init): Initialize it.
12214 (fw_prop): If the variable has reached zero, stop propagating.
12215 (fwprop_addr): Ditto.
12217 2017-03-31 Jakub Jelinek <jakub@redhat.com>
12220 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
12221 a FUNCTION_DECL, pass it as decl instead of origin to
12224 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
12226 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
12229 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
12232 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
12233 TARGET_VSX_SMALL_INTEGER.
12235 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12237 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
12238 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
12240 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
12242 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
12243 extraction from odd-numbered MSA register.
12245 2017-03-31 Jakub Jelinek <jakub@redhat.com>
12247 PR middle-end/80173
12248 * expmed.c (store_bit_field_1): Don't attempt to create
12249 a word subreg out of hard registers wider than word if they
12250 have HARD_REGNO_NREGS of 1 for their mode.
12252 PR middle-end/80163
12253 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
12254 conversions to integer types wider than word and pointer.
12257 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
12258 (rtx_equal_for_cselib_p): Pass 0 to it.
12259 * cselib.c (cselib_hasher::equal): Likewise.
12260 (rtx_equal_for_cselib_1): Add depth argument. If depth
12261 is 128, don't look up VALUE locs and punt. Increment
12262 depth in recursive calls when walking VALUE locs.
12264 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
12266 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
12267 (make_gcov_file_name): Use the canonical path name for generating
12269 (read_line): Fix handling of files with ascii null bytes.
12271 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
12273 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
12274 to initialise a vector register instead
12275 of using a const_int.
12277 2017-03-30 Jakub Jelinek <jakub@redhat.com>
12279 PR translation/80189
12280 * gimplify.c (omp_default_clause): Use %qs instead of %s in
12281 diagnostic messages.
12283 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
12286 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
12287 (dfp_diex_<mode>): Update mode of operand 1.
12288 * doc/extend.texi (dxex, dxexq): Document change to return type.
12289 (diex, diexq): Document change to argument type.
12291 2017-03-30 Martin Jambor <mjambor@suse.cz>
12294 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
12295 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
12296 it reflects the signature changes performed at the callee side.
12297 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
12298 to cgraph_build_function_type_skip_args.
12299 (build_function_decl_skip_args): Adjust call to the above function.
12301 2017-03-30 Jakub Jelinek <jakub@redhat.com>
12304 * config/i386/sse.md
12305 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
12306 register as dest whenever it is a MEM not rtx_equal_p to the
12307 corresponding dup operand, and when forcing into reg move the
12308 reg into the memory afterwards.
12309 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
12310 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
12311 for the force_reg mode.
12312 (avx512vl_vextractf128<mode>): Use register as dest either
12313 always when a MEM, or when it is a MEM not rtx_equal_p to the
12314 corresponding dup operand, or even not when it is a CONST_VECTOR
12315 depending on the mode and lo vs. hi.
12316 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
12318 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
12319 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
12320 Likewise. Require that operands[2] is even.
12321 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
12322 Remove extraneous parens. Require that operands[2] is a multiple
12324 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
12325 operands[0] is a MEM if <mask_applied>, the predicates/constraints
12326 disallow memory then.
12328 2017-03-30 Richard Biener <rguenther@suse.de>
12330 PR tree-optimization/77498
12331 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
12332 to non-constants over backedges.
12334 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
12336 PR rtl-optimization/80233
12337 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
12338 as last_combined_insn. Do not test for BARRIER_P separately.
12340 2017-03-29 Andreas Schwab <schwab@suse.de>
12343 * calls.c (prepare_call_address): Convert funexp to Pmode before
12344 copying to temp reg.
12346 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12348 PR tree-optimization/80158
12349 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
12350 Handle possible future case of more than one alternate
12352 (replace_rhs_if_not_dup): Likewise.
12353 (replace_one_candidate): Likewise.
12355 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
12357 PR rtl-optimization/80193
12358 * ira.c (ira): Do not check allocation for LRA.
12360 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
12362 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
12363 (nvptx_output_simt_exit): Declare.
12364 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
12365 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
12366 (init_softstack_frame): Move initialization of crtl->is_leaf to...
12367 (nvptx_declare_function_name): ...here. Emit declaration of local
12368 memory space buffer for omp_simt_enter insn.
12369 (nvptx_output_unisimt_switch): New.
12370 (nvptx_output_softstack_switch): New.
12371 (nvptx_output_simt_enter): New.
12372 (nvptx_output_simt_exit): New.
12373 * config/nvptx/nvptx.h (struct machine_function): New fields
12374 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
12375 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
12376 (UNSPECV_SIMT_EXIT): Ditto.
12377 (omp_simt_enter_insn): New insn.
12378 (omp_simt_enter): New expansion.
12379 (omp_simt_exit): New insn.
12380 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
12382 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
12383 (expand_GOMP_SIMT_ENTER_ALLOC): New.
12384 (expand_GOMP_SIMT_EXIT): New.
12385 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
12386 (GOMP_SIMT_ENTER_ALLOC): Ditto.
12387 (GOMP_SIMT_EXIT): Ditto.
12388 * target-insns.def (omp_simt_enter): New insn.
12389 (omp_simt_exit): Ditto.
12390 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
12392 (lower_rec_simd_input_clauses): Implement SIMT privatization.
12393 (lower_rec_input_clauses): Likewise.
12394 (lower_lastprivate_clauses): Handle SIMT privatization.
12396 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
12397 (ompdevlow_adjust_simt_enter): New.
12398 (find_simtpriv_var_op): New.
12399 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
12400 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
12402 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
12403 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
12404 (copy_decl_for_dup_finish): Ditto.
12406 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
12408 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
12411 * config/i386/i386.c (ix86_option_override_internal): Always
12412 allow -mpreferred-stack-boundary=3 for 64-bit targets.
12414 2017-03-28 Bin Cheng <bin.cheng@arm.com>
12416 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
12418 2017-03-28 Bin Cheng <bin.cheng@arm.com>
12420 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
12421 mark new edge's irreducible flag accordign to it.
12422 (vect_do_peeling): Check loop preheader edge's irreducible flag
12423 and pass it to function slpeel_add_loop_guard.
12425 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
12427 PR tree-optimization/80218
12428 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
12429 Update block frequencies and counts.
12431 2017-03-28 Richard Biener <rguenther@suse.de>
12433 PR tree-optimization/78644
12434 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
12435 of a simplification result we may not use it at all.
12437 2017-03-28 Richard Biener <rguenther@suse.de>
12440 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
12441 without arguments, generate default definition of a SSA name.
12443 2017-03-28 Richard Biener <rguenther@suse.de>
12445 PR middle-end/80222
12446 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
12447 TYPE_REF_CAN_ALIAS_ALL references.
12448 * fold-const.c (fold_indirect_ref_1): Likewise.
12450 2017-03-28 Martin Liska <mliska@suse.cz>
12453 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
12454 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
12456 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12457 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
12459 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
12460 (EXTRA_SPECS): Define.
12461 (SUBTARGET_EXTRA_SPECS): Likewise.
12462 (SUBTARGET_CPP_SPEC): Likewise.
12463 * config/arc/elf.h (EXTRA_SPECS): Renamed to
12464 SUBTARGET_EXTRA_SPECS.
12465 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
12467 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12469 * config/arc/simdext.md (vst64_insn): Update pattern.
12470 (vld32wh_insn): Likewise.
12471 (vld32wl_insn): Likewise.
12472 (vld64_insn): Likewise.
12473 (vld32_insn): Likewise.
12475 2017-03-28 Marek Polacek <polacek@redhat.com>
12478 * fold-const.c (fold_comparison): Use protected_set_expr_location
12479 instead of SET_EXPR_LOCATION.
12481 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
12483 * tree.c (add_expr): Avoid name lookup warning.
12485 2017-03-27 Jeff Law <law@redhat.com>
12487 PR tree-optimization/80216
12488 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
12489 function name. Limit recursion depth.
12490 (record_temporary_equivalences): Corresponding changes.
12492 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
12494 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
12497 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12500 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
12502 * cfgcleanup.c (reg_note_cfa_p): New array.
12503 (insns_have_identical_cfa_notes): New function.
12504 (old_insns_match_p): Don't cross-jump in between /f
12505 and non-/f instructions. If both i1 and i2 are frame related,
12506 verify all CFA notes, their order and content.
12508 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
12511 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
12512 HImode and SImode with zero extend to DImode to one insn.
12513 (bswap<mode>2_extenddi): Likewise.
12514 (bswapsi2_extenddi): Likewise.
12515 (bswaphi2_extendsi): Likewise.
12516 (bswaphi2): Combine bswap HImode and SImode into one insn.
12517 Separate memory insns from swapping register.
12518 (bswapsi2): Likewise.
12519 (bswap<mode>2): Likewise.
12520 (bswaphi2_internal): Delete, no longer used.
12521 (bswapsi2_internal): Likewise.
12522 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
12523 store, and gpr<-gpr swap insns.
12524 (bswap<mode>2_store): Likewise.
12525 (bswaphi2_reg): Register only splitter, combine with the splitter.
12526 (bswaphi2 splitter): Likewise.
12527 (bswapsi2_reg): Likewise.
12528 (bswapsi2 splitter): Likewise.
12529 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
12530 the insns into load, store, and register/register insns.
12531 (bswapdi2_ldbrx): Likewise.
12532 (bswapdi2_load): Likewise.
12533 (bswapdi2_store): Likewise.
12534 (bswapdi2_reg): Likewise.
12536 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
12538 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
12539 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
12541 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12544 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
12546 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
12547 special handling for target option conflicts between dform
12548 options (-mpower9-dform, -mpower9-dform-vector,
12549 -mpower9-dform-scalar) and -mno-direct-move.
12551 2017-03-27 Richard Biener <rguenther@suse.de>
12553 PR tree-optimization/80181
12554 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
12556 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12558 * config/arc/predicates.md (move_double_src_operand): Replace the
12559 call to move_double_src_operand with a call to address_operand.
12561 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12563 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
12564 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
12565 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
12567 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12569 * config/arc/predicates.md (long_immediate_loadstore_operand):
12570 Consider scaled addresses cases.
12572 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
12574 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
12575 restored when in interrupt.
12576 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
12577 doesn't have delay slot.
12579 2017-03-27 Richard Biener <rguenther@suse.de>
12582 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
12583 inlined thunk clones.
12585 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12588 * asan.c (instrument_derefs): Copy over last operand from
12589 original COMPONENT_REF to the new COMPONENT_REF with
12590 DECL_BIT_FIELD_REPRESENTATIVE.
12591 * ubsan.c (instrument_object_size): Likewise.
12593 2017-03-27 Richard Biener <rguenther@suse.de>
12595 PR tree-optimization/80170
12596 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
12597 sure DR/SCEV didnt fold in constants we do not see when looking
12598 at the reference base alignment.
12600 2017-03-27 Richard Biener <rguenther@suse.de>
12602 PR middle-end/80171
12603 * gimple-fold.c (fold_ctor_reference): Properly guard against
12604 NULL return value from canonicalize_constructor_val.
12606 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
12609 * config/i386/i386.c (ix86_expand_builtin)
12610 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
12611 flags reg setting and flags reg using instructions.
12612 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
12613 clobbering instructions to zero extend op2.
12615 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
12617 * doc/install.texi (Configuration) <--with-aix-soname>:
12618 Update link to AIX ld.
12620 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
12622 PR rtl-optimization/80160
12623 PR rtl-optimization/80159
12624 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
12625 reg_alternate_class into account.
12627 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
12630 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
12631 to consider in curr_insn_transform.
12633 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12635 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
12636 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
12637 and emit_mode_inner.
12639 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12641 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
12642 argument to the overloaded builtin variants. Use the new flag to
12643 deprecate certain builtin variants.
12644 * config/s390/s390-builtin-types.def: Add new builtin types.
12645 * config/s390/s390-builtins.h: Support new flags field for
12646 overloaded builtins.
12647 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
12648 (s390_macro_to_expand): Enable vector float data type.
12649 (s390_cpu_cpp_builtins_internal): Indicate support of the new
12650 builtins by incrementing the __VEC__ version number.
12651 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
12653 (s390_resolve_overloaded_builtin): Emit error messages depending
12654 on the builtin flags.
12655 * config/s390/s390.c (s390_expand_builtin): Support additional
12656 flags argument. Change error message to match the messages
12657 emitted in s390-c.c.
12658 * config/s390/s390.md: New UNSPEC_* constants.
12659 (op_type): Add new instruction types.
12660 * config/s390/vecintrin.h: Add new builtins and test data class
12662 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
12663 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
12664 (VEC_INEXACT, VEC_NOINEXACT): New constants.
12665 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
12666 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
12667 ("vec_mergel<mode>"): V_HW -> VEC_HW.
12669 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
12670 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
12671 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
12672 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
12674 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
12675 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
12676 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
12677 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
12679 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
12680 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
12681 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
12682 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
12683 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
12684 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
12685 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
12687 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
12688 ("vec_scatter_element<V_HW_4:mode>_DI")
12689 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
12690 ("vec_fpint<mode>", "vflls")
12691 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
12692 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
12693 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
12694 ("*vec_cmphe<mode>_cc"): ... these.
12696 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
12697 mode constant instead of magic value.
12699 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12701 * config/s390/s390.c (s390_expand_vec_compare): Support other
12702 vector floating point modes than just V2DF.
12703 (s390_expand_vcond): Likewise.
12704 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
12705 (s390_cannot_change_mode_class): Prevent mode changes between TF
12706 and V1TF in vector registers.
12707 * config/s390/s390.md (DF, SF): New mode attributes.
12708 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
12709 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
12710 SFmode support for VRs.
12711 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
12713 (VFT, VF_HW): New mode iterators.
12714 (vw, sdx): New mode attributes.
12715 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
12716 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
12717 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
12718 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
12719 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
12720 also the new vector floating point modes. Renaming to ...
12722 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
12723 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
12724 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
12725 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
12726 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
12727 ("vec_unordered<mode>"): ... these.
12729 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
12730 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
12731 ("*vec_extendv2df"): New insn definitions.
12733 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12735 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
12736 ("mulditi3_2", "*muldi3_sign"): New patterns.
12737 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
12738 rename the pattern definition.
12740 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12742 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
12744 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
12746 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12748 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
12749 instruction if possible.
12750 * config/s390/vector.md (vec_halfnumelts): New mode
12752 ("*vec_vllezlf<mode>"): New pattern.
12754 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12756 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
12757 ("popcountv4si2", "popcountv2di2"): Rename to ...
12758 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
12759 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
12761 ("popcount<mode>2_vxe"): New pattern.
12763 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12765 * common/config/s390/s390-common.c (processor_flags_table): Add
12767 * config.gcc: Add arch12.
12768 * config/s390/driver-native.c (s390_host_detect_local_cpu):
12769 Default to arch12 for unknown CPU model numbers.
12770 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
12771 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
12772 PROCESSOR_max sanity check.
12773 * config/s390/s390-opts.h (enum processor_type): Add
12775 * config/s390/s390.c (processor_table): Add arch12.
12776 (s390_expand_builtin): Add check for B_VXE flag.
12777 (s390_issue_rate): Add PROCESSOR_ARCH12.
12778 (s390_get_sched_attrmask): Likewise.
12779 (s390_get_unit_mask): Likewise.
12780 (s390_sched_score): Enable z13 scheduling for arch12.
12781 (s390_sched_reorder): Likewise.
12782 (s390_sched_variable_issue): Likewise.
12783 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
12785 (s390_tune_attr): Use z13 scheduling also for arch12.
12786 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
12787 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
12788 (TARGET_VXE_P): New macros.
12789 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
12790 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
12791 * config/s390/s390.opt: Add arch12 as processor_type.
12793 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12795 * config/s390/s390.md
12796 ("fixuns_truncdddi2", "fixuns_trunctddi2")
12797 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
12798 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
12800 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
12801 Rename expanders to ...
12803 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
12804 ("fixuns_truncdddi2_emu"): ... these.
12806 ("fixuns_trunc<mode>si2_emu"): New expander.
12808 ("*fixuns_truncdfdi2_z13"): Rename to ...
12809 ("*fixuns_truncdfdi2_vx"): ... this.
12811 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12813 * config/s390/2964.md: Remove the single element vector compare
12814 instructions which are no longer used.
12815 * config/s390/s390.c (s390_select_ccmode): Remove handling of
12817 (s390_canonicalize_comparison): Remove handling of DFmode
12819 (s390_expand_vec_compare_scalar): Remove function.
12820 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
12821 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
12823 ("*cmp<mode>_ccs"): Add wfcdb instruction.
12825 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12827 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
12829 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
12830 will anyway by matched by mov<mode>_64dfp.
12832 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12834 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
12835 vlef/vstef. Add missing operand to vleif.
12837 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12839 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
12840 pair for all vector types with 64 bit elements.
12841 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
12842 * config/s390/vector.md (V_HW_64): ... here.
12843 (V_128_NOSINGLE): New mode iterator.
12844 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
12845 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
12846 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
12847 ("*vec_load_pairv2di"): Change to ...
12848 ("*vec_load_pair<mode>"): ... this one.
12850 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12852 * config/s390/constraints.md: Add comments.
12853 (jKK): Reject element sizes > 8 bytes.
12854 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
12856 * config/s390/s390.md: Add the s_operand checks formerly in
12857 s390_split_ok_p to various splitters where they are still
12859 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
12860 for 128 bit vectors. Plus two splitters.
12862 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12864 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
12867 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12870 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
12871 error if the boundary argument is not constant.
12873 2017-03-24 Jakub Jelinek <jakub@redhat.com>
12875 PR rtl-optimization/80112
12876 * loop-doloop.c (doloop_condition_get): Don't check condition
12877 if cmp isn't SET with IF_THEN_ELSE src.
12879 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12881 PR tree-optimization/80158
12882 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
12883 replacing a candidate statement, also replace it for the
12884 candidate's alternate interpretation.
12885 (replace_rhs_if_not_dup): Likewise.
12886 (replace_one_candidate): Likewise.
12888 2017-03-24 Richard Biener <rguenther@suse.de>
12890 PR tree-optimization/80167
12891 * graphite-isl-ast-to-gimple.c
12892 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
12894 (translate_isl_ast_to_gimple::get_rename): Likewise.
12896 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
12898 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
12899 handling of certain combinations of target options, including the
12900 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
12901 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
12903 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12906 * config/arm/arm.md (*load_multiple): Add reload_completed to
12907 matching condition.
12909 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12910 Richard Biener <rguenth@suse.de>
12912 PR tree-optimization/79908
12913 PR tree-optimization/80136
12914 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
12915 been cast away, gimplify_and_add suffices.
12917 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
12919 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
12921 2017-03-23 Richard Biener <rguenther@suse.de>
12923 PR tree-optimization/80032
12924 * gimplify.c (gimple_push_cleanup): Forced unconditional
12925 cleanups still have to go to the conditional_cleanups
12928 2017-03-22 Jakub Jelinek <jakub@redhat.com>
12930 PR tree-optimization/80072
12931 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
12933 (next_operand_entry_id): Change type to unsigned int.
12934 (sort_by_operand_rank): Make sure to return the right return value
12935 even if unsigned fields are bigger than INT_MAX.
12936 (struct oecount): Change cnt and id type to unsigned int.
12937 (oecount_hasher::equal): Formatting fix.
12938 (oecount_cmp): Make sure to return the right return value
12939 even if unsigned fields are bigger than INT_MAX.
12940 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
12943 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
12944 TREE_READONLY on result if writing it more than once.
12947 * doc/invoke.texi (-fsanitize=thread): Document that with
12948 -fnon-call-exceptions atomics are not able to throw
12952 * tsan.c: Include tree-eh.h.
12953 (instrument_builtin_call): Call maybe_clean_eh_stmt or
12954 maybe_clean_or_replace_eh_stmt where needed.
12955 (instrument_memory_accesses): Add cfg_changed argument.
12956 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
12957 if it returned true.
12958 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
12960 PR rtl-optimization/63191
12961 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
12962 wrapper function, moved the whole old content into ...
12963 (ix86_delegitimize_address_1): ... this. New inline function.
12964 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
12965 true as last argument instead of ix86_delegitimize_address.
12967 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
12969 * config/aarch64/aarch64.c (generic_branch_cost): Copy
12970 cortexa57_branch_cost.
12972 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
12974 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
12976 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12979 * doc/md.texi (Constraints): Document wA constraint.
12980 * config/rs6000/constraints.md (wA): New.
12981 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
12982 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
12983 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
12984 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
12986 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
12989 * gimplify.c (is_oacc_declared): New function.
12990 (oacc_default_clause): Use it to set default flags for acc declared
12991 variables inside parallel regions.
12992 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
12993 declared variables.
12994 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
12995 declare attribute to any decl as necessary.
12997 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
13000 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
13001 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
13002 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
13003 (arm_arch_lpae): This.
13004 * config/arm/arm.c (arm_arch7ve): Rename into ...
13005 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
13006 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
13009 2017-03-22 Martin Liska <mliska@suse.cz>
13012 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
13013 error message instead of an ICE.
13015 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13017 * doc/extend.texi (6.11 Additional Floating Types): Revise.
13019 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
13021 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
13023 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
13026 2017-03-21 Martin Sebor <msebor@redhat.com>
13028 * doc/extend.texi: Use "cannot" instead of "can't."
13029 * doc/hostconfig.texi: Same.
13030 * doc/install.texi: Same.
13031 * doc/invoke.texi: Same.
13032 * doc/loop.texi: Same.
13033 * doc/md.texi: Same.
13034 * doc/objc.texi: Same.
13035 * doc/rtl.texi: Same.
13036 * doc/tm.texi: Same.
13037 * doc/tm.texi.in: Same.
13038 * doc/trouble.texi: Same.
13040 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
13043 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
13044 (collect_checksum_attributes): Set it.
13045 (die_checksum_ordered): Use it.
13047 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13049 PR tree-optimization/79908
13050 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
13051 change: For a VA_ARG whose LHS has been cast away, use
13052 force_gimple_operand to construct the side effects.
13054 2017-03-21 David Malcolm <dmalcolm@redhat.com>
13056 PR translation/80001
13057 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
13058 more amenable to translation.
13059 (oacc_loop_auto_partitions): Likewise.
13061 2017-03-21 Marek Polacek <polacek@redhat.com>
13062 Martin Sebor <msebor@redhat.com>
13064 PR tree-optimization/80109
13065 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
13066 on INTEGRAL_TYPE_P.
13068 2017-03-21 Jakub Jelinek <jakub@redhat.com>
13069 Segher Boessenkool <segher@kernel.crashing.org>
13072 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
13073 check reg_used_between_p between insn and one of succ or succ2
13074 depending on if succ is artificial insn not inserted into insn
13077 2017-03-21 Martin Liska <mliska@suse.cz>
13079 PR gcov-profile/80081
13080 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
13081 * doc/gcc.texi: Include gcov-dump stuff.
13082 * doc/gcov-dump.texi: New file.
13084 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
13086 PR rtl-optimization/79150
13087 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
13088 conditional jump, if the jump is the last insn of the loop.
13090 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13091 Richard Biener <rguenth@suse.de>
13093 PR tree-optimization/79908
13094 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
13095 been cast away, use force_gimple_operand to construct the side
13098 2017-03-21 Martin Liska <mliska@suse.cz>
13100 PR libfortran/79956
13101 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
13104 2017-03-21 Brad Spengler <spender@grsecurity.net>
13107 * plugin.c (htab_hash_plugin): New function.
13108 (add_new_plugin): Use it and adjust.
13109 (parse_plugin_arg_opt): Adjust.
13110 (init_one_plugin): Likewise.
13112 2017-03-21 Richard Biener <rguenther@suse.de>
13114 PR tree-optimization/80032
13115 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
13116 if set force the cleanup to happen unconditionally.
13117 (gimplify_target_expr): Push inserted clobbers with force_uncond
13118 to avoid them being removed by control-dependent DCE.
13120 2017-03-21 Richard Biener <rguenther@suse.de>
13122 PR tree-optimization/80122
13123 * tree-inline.c (copy_bb): Do not expans va-arg packs or
13124 va_arg_pack_len when the inlined call stmt requires pack
13126 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
13128 2017-03-21 Jakub Jelinek <jakub@redhat.com>
13131 * tsan.c (instrument_builtin_call): If the memory model argument
13132 is not a constant, assume it is valid.
13135 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
13138 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
13140 PR rtl-optimization/79910
13141 * combine.c (can_combine_p): Do not allow combining an I0 or I1
13142 if its dest is used by an insn before I2 (other than the combined
13143 insns themselves, which are properly handled already).
13145 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
13148 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
13150 * combine.c (record_used_regs): New static function.
13151 (try_combine): Handle situations where there is an additional
13152 instruction between I2 and I3 which needs to have a LOG_LINK
13156 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
13158 * combine.c (try_combine): Delete redundant i1 test. Call
13159 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
13161 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13164 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
13165 alternatives 13/14.
13167 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13169 PR tree-optimization/80054
13170 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
13171 the optimization if a PHI or any of its arguments is not dominated
13172 by the candidate's basis. Use gphi* rather than gimple* as
13174 (replace_profitable_candidates): Clean up a gimple* variable that
13175 should be a gphi* variable.
13177 2017-03-20 Martin Sebor <msebor@redhat.com>
13180 * doc/extend.texi (attribute constructor): Document present limitation.
13182 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
13185 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
13186 __POWER9_VECTOR__ #ifdef control, change template definition to
13187 use Power9-specific built-in function.
13188 (vec_any_eq): Likewise.
13189 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
13190 to control outcomes from this test.
13191 (vector_ae_<mode>p): For VEC_F modes, likewise.
13193 2017-03-20 Ian Lance Taylor <iant@google.com>
13195 * config/i386/i386.c (ix86_function_regparm): Save an extra
13196 register for -fsplit-stack with DECL_STATIC_CHAIN.
13198 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
13201 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
13202 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
13204 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
13206 * config/riscv/riscv.c (riscv_print_operand): Use "fence
13208 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
13211 2017-03-20 Marek Polacek <polacek@redhat.com>
13214 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
13216 2017-03-20 Richard Biener <rguenther@suse.de>
13218 PR tree-optimization/80113
13219 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
13220 allocate extra SSA name for PHI def.
13221 (add_close_phis_to_outer_loops): Likewise.
13222 (add_close_phis_to_merge_points): Likewise.
13223 (copy_loop_close_phi_args): Likewise.
13224 (copy_cond_phi_nodes): Likewise.
13226 2017-03-20 Martin Liska <mliska@suse.cz>
13228 PR middle-end/79753
13229 * tree-chkp.c (chkp_build_returned_bound): Do not build
13230 returned bounds for a LHS that's not a BOUNDED_P type.
13232 2017-03-20 Martin Liska <mliska@suse.cz>
13236 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
13237 COMPLEX_CST and VECTOR_CST.
13239 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13242 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
13243 target operand. A new splitter adds the clobber statement in case
13244 the target operand is dead anyway.
13246 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
13248 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
13249 to age-old versions of binutils and glibc.
13251 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
13253 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
13255 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
13257 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
13259 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
13261 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
13262 requirement for binutils 2.13.
13264 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
13266 * combine.c (try_combine): Delete redundant i1 test. Call
13267 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
13269 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
13271 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
13272 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
13274 <riscv64-*-elf>: Re-arrange section
13275 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
13276 <riscv32-*-linux>: Likewise.
13277 <riscv64-*-elf>: Likewise
13278 <riscv64-*-linux>: Likewise.
13280 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
13283 * aarch64.opt(verbose-cost-dump): Fix typo.
13285 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
13288 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
13289 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
13291 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
13293 * reload.c (find_reloads): When reloading a nonoffsettable address,
13294 use RELOAD_OTHER for it and its address reloads.
13296 PR rtl-optimization/79910
13297 * combine.c (record_used_regs): New static function.
13298 (try_combine): Handle situations where there is an additional
13299 instruction between I2 and I3 which needs to have a LOG_LINK
13302 2017-03-17 Jeff Law <law@redhat.com>
13304 PR tree-optimization/71437
13305 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
13306 conditional in the hash table first.
13307 (vrp_dom_walker::before_dom_children): Extract condition from
13308 ASSERT_EXPR. Record condition, its inverion and any implied
13309 conditions as well.
13311 2017-03-17 Marek Polacek <polacek@redhat.com>
13312 Markus Trippelsdorf <markus@trippelsdorf.de>
13314 PR tree-optimization/80079
13315 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
13318 2017-03-17 Richard Biener <rguenther@suse.de>
13320 PR middle-end/80075
13321 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
13322 Properly verify the LHS before the RHS possibly claims to be
13324 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
13327 2017-03-17 Martin Jambor <mjambor@suse.cz>
13329 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
13330 (List of -O2 options): Likewise.
13331 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
13334 2017-03-17 Tom de Vries <tom@codesourcery.com>
13336 * gcov-dump.c (print_usage): Print bug_report_url.
13338 2017-03-17 Richard Biener <rguenther@suse.de>
13340 PR middle-end/80050
13341 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
13342 (parser::peek): Likewise.
13344 2017-03-17 Richard Biener <rguenther@suse.de>
13346 PR tree-optimization/80048
13347 * sese.c (free_sese_info): Properly release rename_map and
13348 copied_bb_map elements.
13350 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
13352 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
13353 Add linked-list forward and backlinks. Insert on
13354 construction, remove on destruction.
13355 (class pass_store_merging): Add m_stores_head field.
13356 (pass_store_merging::terminate_and_process_all_chains):
13357 Iterate over m_stores_head list.
13358 (pass_store_merging::terminate_all_aliasing_chains):
13360 (pass_store_merging::execute): Check for debug stmts first.
13361 Push new chains onto the m_stores_head stack.
13363 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
13366 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
13367 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
13368 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
13370 2017-03-16 Jeff Law <law@redhat.com>
13372 PR tree-optimization/71437
13373 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
13374 member function. Implementation moved into after_dom_children
13375 member function and into the threader's thread_outgoing_edges
13377 (dom_opt_dom_walker::after_dom_children): Simplify by moving
13378 some code into new thread_outgoing_edges.
13379 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
13380 definition. Simplify marker handling (do it here). Assume we always
13381 have the available expression and the const/copies tables.
13382 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
13384 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
13385 * tree-vrp.c (equiv_stack): No longer file scoped.
13386 (vrp_dom_walker): New class.
13387 (vrp_dom_walker::before_dom_children): New member function.
13388 (vrp_dom_walker::after_dom_children): Likewise.
13389 (identify_jump_threads): Setup domwalker. Use it rather than
13390 walking edges in a random order by hand. Simplify setup/finalization.
13391 (finalize_jump_threads): Remove.
13392 (vrp_finalize): Do not call identify_jump_threads here.
13393 (execute_vrp): Do it here instead and call thread_through_all_blocks
13396 PR tree-optimization/71437
13397 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
13399 (simplify_stmt_for_jump_threading): Add basic_block argument. All
13401 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
13402 (dom_opt_dom_walker::thread_across_edge): Remove
13403 handle_dominating_asserts argument. All callers changed.
13404 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
13405 changes. Remove calls to lhs_of_dominating_assert. Other
13406 uses of handle_dominating_asserts turn into unconditional code
13407 (simplify_control_stmt_condition_1): Likewise.
13408 (simplify_control_stmt_condition): Likewise.
13409 (thread_through_normal_block, thread_across_edge): Likewise.
13410 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
13411 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
13412 object if it is not an SSA_NAME.
13413 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
13414 before calling into the VRP specific simplifiers.
13415 (identify_jump_threads): Remove handle_dominating_asserts
13418 2017-03-16 Jakub Jelinek <jakub@redhat.com>
13421 * tree-diagnostic.c (default_tree_printer): No longer static.
13422 * tree-diagnostic.h (default_tree_printer): New prototype.
13424 2017-03-16 Tamar Christina <tamar.christina@arm.com>
13426 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
13427 Change ins into fmov.
13429 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13431 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
13432 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
13433 Use h_con constraint for operand 1.
13434 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
13435 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
13437 2017-03-15 Jeff Law <law@redhat.com>
13439 PR tree-optimization/71437
13440 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
13441 (record_temporary_equivalences): Use it.
13443 PR tree-optimization/71437
13444 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
13445 tree-ssa-scopedtables.
13446 (lookup_avail_expr, build_and_record_new_cond): Likewise.
13447 (record_conditions, record_cond, vuse_eq): Likewise.
13448 (record_edge_info): Adjust to API tweak of record_conditions.
13449 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
13450 (record_temporary_equivalences, optimize_stmt): Likewise.
13451 (eliminate_redundant_computations): Likewise.
13452 (record_equivalences_from_stmt): Likewise.
13453 * tree-ssa-scopedtables.c: Include options.h and params.h.
13454 (vuse_eq): New function, moved from tree-ssa-dom.c
13455 (build_and_record_new_cond): Likewise.
13456 (record_conditions): Likewise. Accept vector of conditions rather
13457 than edge_equivalence structure for first argument.
13458 for the first argument.
13459 (avail_exprs_stack::lookup_avail_expr): New member function, moved
13460 from tree-ssa-dom.c.
13461 (avail_exprs_stack::record_cond): Likewise.
13462 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
13463 from tree-ssa-dom.c.
13464 (avail_exprs_stack): Add new member functions lookup_avail_expr
13466 (record_conditions): Declare.
13468 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
13471 * lra-constraints.c (process_alt_operands): Increase reject for
13472 reloading an input/output operand.
13474 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
13477 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
13478 insns to convert from signed/unsigned char/short to IEEE 128-bit
13480 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
13482 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
13485 * config/i386/i386.c (ix86_vector_duplicate_value): Create
13486 subreg of inner mode for values already in registers.
13488 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
13490 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
13491 iteration reg is used after the loop.
13493 2017-03-14 Martin Sebor <msebor@redhat.com>
13495 PR tree-optimization/79800
13496 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
13497 precision in negative-positive range.
13498 (format_floating): Call non-const overload with adjusted precision.
13500 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
13503 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
13506 2017-03-14 Martin Sebor <msebor@redhat.com>
13508 PR middle-end/80020
13509 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
13510 * builtins.def (aligned_alloc): Use it.
13513 * Makefile.in (GTFILES): Add calls.c.
13514 * calls.c: Include "gt-calls.h".
13516 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
13518 PR rtl-optimization/79728
13519 * regs.h (struct target_regs): New field
13520 x_contains_allocatable_regs_of_mode.
13521 (contains_allocatable_regs_of_mode): New macro.
13522 * reginfo.c (init_reg_sets_1): Initialize it, and change
13523 contains_reg_of_mode so it includes global regs as well.
13524 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
13525 rather than contains_regs_of_mode.
13527 2017-03-14 Martin Liska <mliska@suse.cz>
13529 * doc/invoke.texi: Document options that can't be combined with
13530 -fcheck-pointer-bounds.
13532 2017-03-14 Martin Liska <mliska@suse.cz>
13534 PR middle-end/79831
13535 * doc/invoke.texi (-Wchkp): Document the option.
13537 2017-03-14 Martin Liska <mliska@suse.cz>
13539 * Makefile.in: Install gcov-dump.
13541 2017-03-14 Martin Liska <mliska@suse.cz>
13543 * multiple_target.c (expand_target_clones): Bail out for
13544 an invalid attribute.
13546 2017-03-14 Richard Biener <rguenther@suse.de>
13548 * alias.c (struct alias_set_entry): Pack properly.
13549 * cfgloop.h (struct loop): Likewise.
13550 * cse.c (struct set): Likewise.
13551 * ipa-utils.c (struct searchc_env): Likewise.
13552 * loop-invariant.c (struct invariant): Likewise.
13553 * lra-remat.c (struct cand): Likewise.
13554 * recog.c (struct change_t): Likewise.
13555 * rtl.h (struct address_info): Likewise.
13556 * symbol-summary.h (function_summary): Likewise.
13557 * tree-loop-distribution.c (struct partition): Likewise.
13558 * tree-object-size.c (struct object_size_info): Likewise.
13559 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
13560 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
13561 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
13562 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
13563 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
13564 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
13565 (struct _stmt_vec_info): Likewise.
13567 2017-03-14 Martin Liska <mliska@suse.cz>
13570 * multiple_target.c (create_dispatcher_calls): Check that
13571 a target can create a function dispatcher.
13573 2017-03-14 Martin Liska <mliska@suse.cz>
13576 * multiple_target.c (expand_target_clones): Drop local.local
13577 flag for default implementation.
13579 2017-03-14 Richard Biener <rguenther@suse.de>
13581 PR tree-optimization/80030
13582 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
13584 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
13586 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
13587 gcc_fallthrough() instead of __attribute__((fallthrough));
13589 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13591 * doc/gcc.texi: Remove "up" link to (DIR).
13592 * doc/gccint.texi: Ditto.
13594 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
13596 * doc/install.texi (Specific) <avr>: Remove reference to
13599 2017-03-13 Jeff Law <law@redhat.com>
13601 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
13602 attribute rather than comments.
13604 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
13605 match_scratch operand is highest.
13607 2017-03-13 Martin Liska <mliska@suse.cz>
13609 PR middle-end/78339
13610 * ipa-pure-const.c (warn_function_noreturn): If the declarations
13611 is a CHKP clone, use original declaration.
13613 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13615 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
13616 (arc_conditional_register_usage): Use a different allocation order
13617 when optimizing for size.
13618 * common/config/arc/arc-common.c (arc_option_optimization_table):
13619 Section anchors default on when optimizing for size.
13621 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13623 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
13625 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13627 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
13628 * config/arc/arc.md (cpu_facility): Add cd variant.
13629 (*movqi_insn): Add code density variant.
13630 (*movhi_insn): Likewise.
13631 (*movqi_insn): Likewise.
13632 (*addsi3_mixed): Likewise.
13633 (subsi3_insn): Likewise.
13635 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13637 * config/arc/arc.md (movsi_cond_exec): Update constraint.
13639 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
13641 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
13642 expressions with MINUS and UNARY ops.
13644 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13647 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
13649 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
13650 between vec_select and vector argument.
13651 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
13652 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
13653 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
13654 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
13655 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
13656 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
13658 2017-03-13 Richard Biener <rguenther@suse.de>
13661 * params.def (vect-max-peeling-for-alignment): Fix typo.
13663 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13665 * doc/install.texi (Specific) <mips-*-*>: Remove description of
13666 issue that only occurred with binutils below 2.18.
13668 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13670 * doc/install.texi (Specific) <cris-axis-elf>: No longer
13671 refer to binutils 2.11/2.12 minimum.
13673 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
13675 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
13676 ftp.kernel.org and simplify binutils requirement.
13678 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
13680 * doc/invoke.texi (Warning Options): Fix spelling of link-time
13682 (Optimize Options): Ditto. Also remove redundancy.
13684 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13686 PR translation/79848
13687 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
13689 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
13690 to G_ to avoid double translation.
13692 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13694 PR translation/79923
13695 * auto-profile.c (get_combined_location): Convert leading
13696 character of diagnostics to lower case and remove trailing period.
13697 (read_profile): Likewise for various diagnostics.
13698 * config/arm/arm.c (arm_option_override): Remove trailing period
13699 from various diagnostics.
13700 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
13701 (msp430_expand_delay_cycles): Likewise.
13703 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13706 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
13707 full command-line argument, rather than just "str".
13708 (aarch64_validate_march): Likewise.
13709 (aarch64_validate_mtune): Likewise.
13711 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
13713 PR rtl-optimization/78911
13714 * lra-assigns.c (must_not_spill_p): New function.
13715 (spill_for): Use it.
13717 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13719 PR tree-optimization/79981
13720 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
13721 ATOMIC_COMPARE_EXCHANGE ifn result.
13722 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
13723 IFN_ATOMIC_COMPARE_EXCHANGE.
13725 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13728 * opts.c (parse_sanitizer_options): Add missing question mark to
13729 "did you mean" message.
13731 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13733 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
13735 (VMULEUH_UNS): Likewise.
13736 (VMULOUB_UNS): Likewise.
13737 (VMULOUH_UNS): Likewise.
13738 * config/rs6000/rs6000.c (builtin_function_type): Remove
13739 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
13741 2017-03-10 David Malcolm <dmalcolm@redhat.com>
13744 * read-rtl-function.c (function_reader::read_rtx_operand): Update
13745 x with result of extra_parsing_for_operand_code_0.
13746 (function_reader::extra_parsing_for_operand_code_0): Convert
13747 return type from void to rtx, returning x. When reading
13748 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
13749 larger size containing struct block_symbol.
13751 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
13753 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
13754 -mfloat128-hardware without -m64.
13756 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
13759 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
13760 entries to the case statement that marks unsigned arguments to
13761 overloaded functions.
13763 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
13765 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
13766 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
13768 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
13771 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
13772 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
13774 2017-03-10 Martin Liska <mliska@suse.cz>
13778 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
13779 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
13782 2017-03-10 Olivier Hainque <hainque@adacore.com>
13784 * tree-switch-conversion (array_value_type): Start by resetting
13785 candidate type to it's main variant.
13787 2017-03-10 Jakub Jelinek <jakub@redhat.com>
13789 PR rtl-optimization/79909
13790 * combine.c (try_combine): Use simplify_replace_rtx on individual
13791 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
13792 of the whole CALL_INSN_FUNCTION_USAGE.
13794 PR tree-optimization/79972
13795 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
13796 get_range_info on SSA_NAMEs. Formatting fixes.
13798 2017-03-10 Richard Biener <rguenther@suse.de>
13799 Jakub Jelinek <jakub@redhat.com>
13801 PR tree-optimization/77975
13802 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
13803 edge to be constant.
13804 (get_val_for): For constant x return it. Formatting fix.
13805 (loop_niter_by_eval): Avoid pointless looping if the next iteration
13806 would use the same bases as the current one.
13808 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13810 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
13811 instead of vec_select for V1TImode.
13812 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
13814 (VSX_LE_128): Add V1TI to this mode iterator.
13815 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
13816 (*vsx_le_perm_store_<mode>): Likewise.
13817 (pre-reload splitter for VSX stores): Likewise.
13818 (post-reload splitter for VSX stores): Likewise.
13819 (*vsx_xxpermdi2_le_<mode>): Likewise.
13820 (*vsx_lxvd2x2_le_<mode>): Likewise.
13821 (*vsx_stxvd2x2_le_<mode>): Likewise.
13823 2017-03-09 Michael Eager <eager@eagercon.com>
13825 Correct failures with --enable-checking=yes,rtl.
13827 * config/microblaze/microblaze.c (microblaze_expand_shift):
13828 Replace GET_CODE test with CONST_INT_P and INTVAL test with
13829 test for const0_rtx.
13830 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
13831 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
13833 2017-03-09 Richard Biener <rguenther@suse.de>
13835 PR tree-optimization/79977
13836 * graphite-scop-detection.c (scop_detection::merge_sese):
13837 Handle the case of extra exits to blocks dominating the entry.
13839 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
13841 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
13844 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
13846 PR rtl-optimization/79949
13847 * lra-constraints.c (process_alt_operands): Check memory when
13848 trying to predict a cycle. Print about the overall increase.
13850 2017-03-09 Richard Biener <rguenther@suse.de>
13852 PR middle-end/79971
13853 * gimple-expr.c (useless_type_conversion_p): Preserve
13854 TYPE_SATURATING for fixed-point types.
13856 2017-03-09 Richard Biener <rguenther@suse.de>
13859 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
13860 alignment of BLKmode params.
13862 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13865 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
13866 (VALL_NO_V2Q): Likewise.
13868 * config/aarch64/aarch64-simd.md
13869 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
13871 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
13872 VALL_NO_V2Q mode iterator.
13873 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
13875 2017-03-09 Martin Liska <mliska@suse.cz>
13877 PR tree-optimization/79631
13878 * tree-chkp-opt.c (chkp_is_constant_addr): Call
13879 tree_int_cst_sign_bit just for INTEGER constants.
13881 2017-03-09 Martin Liska <mliska@suse.cz>
13885 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
13888 2017-03-09 Marek Polacek <polacek@redhat.com>
13891 * tree.c (inchash::add_expr): Handle TREE_VEC.
13893 2017-03-09 Martin Liska <mliska@suse.cz>
13896 (chkp_narrow_size_and_offset): New function.
13897 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
13898 (void chkp_parse_bit_field_ref): New function.
13899 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
13900 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
13902 2017-03-09 Martin Liska <mliska@suse.cz>
13905 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
13906 (chkp_find_bounds_1): Remove gcc_unreachable.
13908 2017-03-09 Jakub Jelinek <jakub@redhat.com>
13911 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
13912 BUILT_IN_SYNC*, determine the access type from the size suffix and
13913 always build a MEM_REF with that type. Handle forgotten
13914 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
13917 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
13918 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
13919 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
13920 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
13921 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
13922 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
13923 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
13924 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
13925 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
13926 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
13927 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
13928 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
13929 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
13930 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
13931 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
13932 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
13933 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
13934 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
13935 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
13936 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
13937 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
13938 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
13939 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
13940 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
13941 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
13942 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
13943 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
13944 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
13945 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
13946 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
13947 definitions outside of __OPTIMIZE__ guarded section.
13950 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
13951 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
13952 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
13953 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
13956 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13958 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
13959 ("vfenez<mode>"): Add missing constraints.
13961 2017-03-08 Martin Sebor <msebor@redhat.com>
13964 * config/nds32/nds32.c (nds32_option_override):
13965 Fix misspelled diagnostic.
13967 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13970 * gimplify.c (gimplify_omp_for): Replace index var in outer
13971 taskloop statement with an artificial variable and add
13972 OMP_CLAUSE_PRIVATE clause for it.
13974 2017-03-08 Richard Biener <rguenther@suse.de>
13976 PR tree-optimization/79955
13977 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
13978 for accesses that are completely outside of the variable.
13980 2017-03-08 Andrew Haley <aph@redhat.com>
13982 PR tree-optimization/79943
13983 * tree-ssa-loop-split.c (compute_new_first_bound): When
13984 calculating the new upper bound, (END-BEG) should be added, not
13987 2017-03-08 Jakub Jelinek <jakub@redhat.com>
13989 * config/avr/avr.md (setmemhi): Make sure match_dup
13990 operand number comes before match_scratch.
13992 2017-03-08 Richard Biener <rguenther@suse.de>
13994 PR tree-optimization/79920
13995 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
13996 with ncopies == 1 to ...
13997 (vect_transform_slp_perm_load): ... here. Properly compute
13998 all element loads by iterating VF times over the group. Do
13999 not handle ncopies (computed in a broken way) in
14000 vect_create_mask_and_perm.
14002 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14005 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
14006 is a uniform vector, use uniform_vector_p return value instead of
14007 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
14009 2017-03-07 Marek Polacek <polacek@redhat.com>
14011 PR middle-end/79809
14012 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
14013 (alloca_call_type): Likewise.
14015 2017-03-07 Martin Liska <mliska@suse.cz>
14017 * gcov.c (process_args): Put comment to correct location.
14019 2017-03-07 Martin Liska <mliska@suse.cz>
14021 PR middle-end/68270
14022 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
14023 Use array_at_struct_end_p instead of DECL_CHAIN (field).
14024 (chkp_narrow_bounds_for_field): Likewise.
14025 (chkp_parse_array_and_component_ref): Pass one more argument to
14028 2017-03-07 Richard Biener <rguenther@suse.de>
14030 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
14033 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
14035 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
14036 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
14038 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14041 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
14042 to end of description.
14043 (PARAM_MAX_STORES_TO_MERGE): Likewise.
14045 2017-03-07 Jakub Jelinek <jakub@redhat.com>
14047 PR rtl-optimization/79901
14048 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
14050 (*avx512f_<code><mode>3<mask_name>): ... this.
14051 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
14052 iterator instead of VI8_AVX2_AVX512BW.
14054 PR rtl-optimization/79901
14055 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
14056 min/max expander, expand it using expand_vec_cond_expr.
14059 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
14062 2017-03-06 Jakub Jelinek <jakub@redhat.com>
14065 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
14066 to void * for PCH reasons.
14067 * dwarf2out.c (output_loc_operands, output_die): Cast
14068 v.val_vec.array to unsigned char *.
14070 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
14073 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
14076 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
14078 PR rtl-optimization/79571
14079 * lra-constraints.c (process_alt_operands): Calculate static
14080 reject and subtract it from overall when only addresses will be
14083 2017-03-06 Julia Koval <julia.koval@intel.com>
14086 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
14087 incoming stack boundary to 128 for 64-bit targets.
14089 2017-03-06 Richard Biener <rguenther@suse.de>
14091 PR tree-optimization/79894
14092 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
14093 to NULL after folding it.
14095 2017-03-06 Richard Biener <rguenther@suse.de>
14097 PR tree-optimization/79824
14098 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
14099 check disabling peeling for gaps.
14101 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
14103 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
14104 attributes): Document gettimeofday.
14106 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
14108 * config/s390/s390.c (s390_option_override_internal): Set
14109 PARAM_MIN_VECT_LOOP_BOUND
14111 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
14113 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
14114 * config/s390/s390.md: Likewise.
14116 2017-03-06 Jakub Jelinek <jakub@redhat.com>
14119 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
14120 (<avx2_avx512>_perm<mode>): Rename to ...
14121 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
14123 (<avx512>_perm<mode>_mask): Rename to ...
14124 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
14126 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
14127 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
14128 instead of VI8F_256_512.
14129 (avx512f_perm<mode>): New define_expand.
14130 (avx512f_perm<mode>_mask): Likewise.
14131 (avx512f_perm<mode>_1<mask_name>): New define_insn.
14132 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
14134 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14136 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
14137 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
14139 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
14141 2017-03-06 Martin Liska <mliska@suse.cz>
14144 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
14145 when having a SSA NAME w/o VAR_DECL assigned to it.
14147 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14149 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
14150 msa_dpsub_<su>_d): Fix MODE for vec_select.
14152 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14154 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
14156 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
14158 2017-03-06 Richard Biener <rguenther@suse.de>
14160 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
14161 * plugin.c (register_plugin_info): Likewise.
14162 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
14164 2017-03-05 Jakub Jelinek <jakub@redhat.com>
14166 * config/i386/sse.md (sse_storehps, sse_storelps,
14167 avx_<castmode><avxsizesuffix>_<castmode>,
14168 avx512f_<castmode><avxsizesuffix>_<castmode>,
14169 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
14170 in condition that at least one operand is not a MEM.
14172 2017-03-03 Jakub Jelinek <jakub@redhat.com>
14174 PR middle-end/79805
14175 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
14176 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
14178 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
14179 gimple_call_nothrow_p flag based on whether original builtin can throw.
14180 If it can, emit following stmts on the fallthrough edge.
14181 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
14182 don't create new bb if inserting just debug stmts on the edge, try to
14183 insert them on the fallthru bb or just reset debug stmts.
14185 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
14188 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
14189 restore recog_data (including the operand rtxes inside it) around
14190 the call to get_insn_template.
14192 2017-03-03 Martin Sebor <msebor@redhat.com>
14194 PR tree-optimization/79699
14195 * context.c (context::~context): Free MPFR caches to avoid
14196 a memory leak on program exit.
14198 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14200 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
14201 Use wide_int::ulow () instead of .elt (0).
14203 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
14205 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
14206 (*pushxf): Limit oF constraint to 32bit targets and add oC
14207 constraint for 64bit targets.
14208 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
14209 (*pushdf): Change rmF constraint to rmC.
14211 2017-03-03 Martin Liska <mliska@suse.cz>
14213 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
14214 Remove unused variable.
14216 2017-03-03 Jakub Jelinek <jakub@redhat.com>
14219 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
14220 is a memory operand, increase num_memory.
14221 (ix86_expand_args_builtin): Likewise.
14223 2017-03-03 Jan Hubicka <jh@suse.cz>
14226 * ipa-devirt.c (maybe_record_node): Properly handle
14227 __cxa_pure_virtual visibility.
14229 2017-03-03 Martin Liska <mliska@suse.cz>
14231 PR tree-optimization/79803
14232 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
14234 (pass_loop_prefetch::execute): Disabled optimization if an
14235 assumption about L1 cache size is not met.
14237 2017-03-03 Martin Liska <mliska@suse.cz>
14239 PR rtl-optimization/79574
14240 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
14241 (hash_scan_set): Likewise.
14242 (dump_hash_table): Likewise.
14243 (hoist_code): Likewise.
14245 2017-03-03 Richard Biener <rguenther@suse.de>
14247 * fixed-value.c (fixed_from_string): Restore use of elt (1)
14248 in place of uhigh ().
14249 (fixed_convert_from_real): Likewise.
14251 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
14254 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
14256 2017-03-03 Richard Biener <rguenther@suse.de>
14258 PR middle-end/79818
14259 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
14260 TYPE_OVERFLOW_UNDEFINED check.
14262 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14264 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
14266 (vector_ae_<mode>_p): Likewise.
14267 (vector_nez_<mode>_p): Likewise.
14268 (vector_ne_v2di_p): Likewise.
14269 (vector_ae_v2di_p): Likewise.
14270 (vector_ne_<mode>_p): Likewise.
14271 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
14273 (vsx_tsqrt<mode>2_fe): Likewise.
14275 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
14278 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
14280 2017-03-02 Jakub Jelinek <jakub@redhat.com>
14282 PR rtl-optimization/79780
14283 * cprop.c (one_cprop_pass): When second and further conditional trap
14284 in a single basic block is turned into an unconditional trap, turn it
14285 into a deleted note to avoid RTL verification failures.
14287 2017-03-02 Richard Biener <rguenther@suse.de>
14289 * fold-const.c (const_binop): Use ulow () instead of elt (0).
14291 2017-03-02 Richard Biener <rguenther@suse.de>
14293 PR tree-optimization/79345
14295 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
14296 param and abort the walk, returning -1 if it is hit.
14297 (walk_aliased_vdefs): Take a limit param and pass it on.
14298 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
14299 defaulting to 0 and return a signed int.
14300 * tree-ssa-uninit.c (struct check_defs_data): New struct.
14301 (check_defs): New helper.
14302 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
14303 about uninitialized memory.
14304 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
14305 bogus uninitialized warning.
14306 (fixed_convert_from_real): Likewise.
14308 2017-03-02 Bin Cheng <bin.cheng@arm.com>
14310 PR tree-optimization/66768
14311 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
14312 iv_use if base object can't be determined.
14314 2017-03-02 Jakub Jelinek <jakub@redhat.com>
14316 PR tree-optimization/79345
14317 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
14318 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
14319 (get_pattern_stats): Initialize it.
14320 * genemit.c (gen_expand): Verify match_scratch numbers come after
14321 match_operand/match_dup numbers.
14322 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
14323 match_scratch numbers.
14324 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
14326 * config/s390/s390.md (trunctdsd2): Likewise.
14328 2017-03-02 Richard Biener <rguenther@suse.de>
14330 * wide-int.h (wide_int_storage::operator=): Implement in terms
14333 2017-03-02 Richard Biener <rguenther@suse.de>
14335 PR tree-optimization/79777
14336 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
14337 the to insert expression to sth existing.
14339 2017-03-01 Martin Sebor <msebor@redhat.com>
14341 PR middle-end/79692
14342 * gimple-ssa-sprintf.c
14343 (directive::known_width_and_precision): New function.
14344 (format_integer): Use it.
14345 (get_mpfr_format_length): Consider the full range of precision
14346 when computing %g output with the # flag. Set the likely byte
14347 count to 3 rather than 1 when precision is indeterminate.
14348 (format_floating): Correct the lower bound of precision.
14350 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14352 * doc/invoke.texi: Document default code model for 64-bit Linux.
14354 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14357 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
14358 udiv rather than div since input pattern is unsigned.
14360 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
14362 * config/i386/i386.c (print_reg): Warn for values of
14363 unsupported size in integer register.
14365 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
14368 * config/rs6000/predicates.md (current_file_function_operand): Do
14369 not allow self calls to be local if the function is replaceable.
14371 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
14374 * config/rs6000/altivec.h (vec_ctz and others): Change the
14375 preprocessor macro that controls conditional compilation from
14376 _ARCH_PWR9 to __POWER9_VECTOR__.
14377 (vec_all_ne): Change parameterization of __altivec_scalar_pred
14378 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
14379 control (instead of _ARCH_PWR9 control) so that template
14380 definition uses power9-specific function.
14381 (vec_any_eq): Likewise.
14382 (vec_all_ne): Change macro definition to use a power9-specific
14383 expansion under #ifdef __POWER9_VECTOR__ control (instead of
14384 _ARCH_PWR9 control).
14385 (vec_any_eq) Likewise.
14386 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
14387 expansion for CMPNEF to remove support for xvcmpnesp instruction.
14388 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
14389 support for xvcmpnedp instruction.
14390 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
14391 macro expansion so that Power9 implementation of vec_all_ne does
14392 not use the AltiVec predicate framework.
14393 (VCMPNEH_P): Likewise.
14394 (VCMPNEW_P): Likewise.
14395 (VCMPNED_P): Likewise.
14396 (VCMPNEFP_P): Likewise.
14397 (VCMPNEDP_P): Likewise.
14398 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
14399 implementation of vec_any_eq to not use AltiVec predicate
14401 (VCMPAEH_P): Likewise.
14402 (VCMPAEW_P): Likewise.
14403 (VCMPAED_P): Likewise.
14404 (VCMPAEFP_P): Likewise.
14405 (VCMPAEDP_P): Likewise.
14406 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
14407 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
14408 not use the AltiVec predicate framework.
14409 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
14410 of vec_any_eq to not use AltiVec predicate framework.
14411 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
14412 support for predefined __POWER9_VECTOR__ macro to indicate that
14413 Power9 instruction selection is enabled.
14414 (altivec_overloaded_builtins): Remove extraneous
14415 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
14416 function argument types RS6000_BTI_bool_V16QI and
14417 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
14418 entry for overloaded function argument types RS6000_BTI_bool_V4SI
14419 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
14420 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
14421 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
14422 Power9 for implementations of vec_cmpne. Change the signature for
14423 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
14424 (representing vec_all_ne) to remove the previously described first
14425 argument of type RS6000_BTI_INTSI, as this was an artifact of
14426 reliance on the AltiVec predicate framework, which is no longer
14427 used in the implementation of these functions. Add
14428 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
14429 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
14430 since, unlike the AltiVec predicate framework implementation, we
14431 do not share function descriptors between vec_alle and vec_anyeq.
14432 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
14433 set of modes that receive special treatment even when
14434 TARGET_P9_VECTOR is true. The special treatment emits code that
14435 does not depend on Power9 instructions.
14436 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
14437 define_expand to not rely on AltiVec predicate framework.
14438 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
14440 (vector_ne_v2di_p): Change this define_expand to not rely on
14441 AltiVec predicate framework.
14442 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
14444 (vector_ne_<mode>_p): Change this define_expand to not rely on
14445 AltiVec predicate framework.
14446 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
14448 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
14449 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
14450 define_insn pattern.
14451 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
14452 define_insn pattern because the xvcmpne<VSs>. instruction is not
14454 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
14455 instruction is not supported.
14457 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14459 * config/nvptx/nvptx.c: Include intl.h.
14461 2017-03-01 Martin Jambor <mjambor@suse.cz>
14464 * ipa-prop.h (ipa_bits): Removed field known.
14465 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
14466 to pointers. Adjusted their comments to warn about their sharing.
14467 (ipcp_transformation_summary): Change bits to a vector of pointers.
14468 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
14469 (ipa_get_ipa_bits_for_value): Declare.
14470 * tree-vrp.h (value_range): Mark as GTY((for_user)).
14471 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
14472 (ipa_bits_hash_table): Likewise.
14473 (ipa_vr_ggc_hash_traits): Likewise.
14474 (ipa_vr_hash_table): Likewise.
14475 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
14476 being pointers and vr_known being removed.
14477 (ipa_set_jf_unknown): Likewise.
14478 (ipa_get_ipa_bits_for_value): New function.
14479 (ipa_set_jfunc_bits): Likewise.
14480 (ipa_get_value_range): New overloaded functions.
14481 (ipa_set_jfunc_vr): Likewise.
14482 (ipa_compute_jump_functions_for_edge): Use the above functions to
14483 construct bits and vr parts of jump functions.
14484 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
14485 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14487 (ipcp_grow_transformations_if_necessary): Also allocate
14488 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14490 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
14491 them. Fix too long lines.
14492 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
14493 vr_known being removed.
14494 (ipa_read_jump_function): Use new setter functions to construct bits
14495 and vr parts of jump functions or set them to NULL.
14496 (write_ipcp_transformation_info): Adjust for bits being pointers.
14497 (read_ipcp_transformation_info): Likewise.
14498 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
14500 Include gt-ipa-prop.h.
14501 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
14503 (ipcp_store_bits_results): Likewise.
14504 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
14505 Do not write to existing jump functions but use a temporary instead.
14507 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14510 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
14511 attempt to use its first operand as BIT_FIELD_REF base.
14513 2017-03-01 Richard Biener <rguenther@suse.de>
14515 PR middle-end/79721
14516 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
14517 interpolating formula in wrapping arithmetic.
14518 (chrec_apply): Convert chrec_evaluate return value to wanted type.
14520 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14522 PR tree-optimization/79734
14523 * tree-vect-generic.c (expand_vector_condition): Optimize
14524 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
14525 Handle VEC_COND_EXPR where comparison has different inner width from
14526 type's inner width.
14528 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
14530 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
14531 markup, and similar issues. Remove @opindex entries for things
14532 that aren't options. Add missing -mmpy-option entries.
14534 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14536 PR tree-optimization/79737
14537 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
14538 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
14539 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
14540 instead of byte_size. Formatting fix.
14541 (shift_bytes_in_array_right): Formatting fix.
14543 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
14546 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
14547 condition on optimize for the leaf function test.
14549 2017-02-28 Martin Liska <mliska@suse.cz>
14552 * read-rtl-function.c (function_reader::handle_unknown_directive):
14553 Bail out when one uses -flto.
14555 2017-02-28 Martin Liska <mliska@suse.cz>
14557 * common.opt: Replace space with tabular for options of <number>
14559 * config/i386/i386.opt: Show <number> value for
14560 -mlarge-data-threshold.
14561 * opts.c (print_filtered_help): Do not display number in hexadecimal
14564 2017-02-28 Martin Liska <mliska@suse.cz>
14566 * common.opt: Fix --help=option -Q for options which are of
14569 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
14571 * config/i386/i386.c (print_reg): Error out for values
14572 of 8-bit size in invalid integer register.
14574 2017-02-28 Martin Sebor <msebor@redhat.com>
14576 PR tree-optimization/79691
14577 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
14579 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14582 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
14583 gcc_unreachable with output_operand_lossage.
14585 2017-02-28 Richard Biener <rguenther@suse.de>
14587 PR tree-optimization/79740
14588 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
14590 (visit_nary_op): Insert the nary into the hashtable if we
14591 pattern-matched sth.
14592 * tree-ssa-pre.c (eliminate_insert): Robustify.
14594 2017-02-28 Richard Biener <rguenther@suse.de>
14596 PR middle-end/79731
14597 * fold-const.c (decode_field_reference): Reject out-of-bound
14600 2017-02-28 Jakub Jelinek <jakub@redhat.com>
14602 * config/i386/i386.c: Include intl.h.
14603 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
14604 instead of just cond ? "..." : "...".
14605 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
14606 * coverage.c (read_counts_file): Likewise.
14607 * omp-offload.c: Include intl.h.
14608 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
14609 of just cond ? "..." : "...".
14610 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
14611 of just cond ? "..." : "...".
14613 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
14616 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
14618 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
14619 'tune for' CPU name.
14620 * config/arm/arm-cpu-data.h: Regenerated.
14622 2017-02-28 Richard Biener <rguenther@suse.de>
14624 PR tree-optimization/79732
14625 * tree-inline.c (expand_call_inline): Do not shadow var.
14627 2017-02-28 Richard Biener <rguenther@suse.de>
14629 PR tree-optimization/79723
14630 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
14631 address-space properly.
14633 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
14635 * doc/optinfo.texi (Optimization groups): Fix option used for
14637 * doc/invoke.texi (-fopt-info): Document "omp".
14638 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
14639 (OPTGROUP_ALL): Add OPTGROUP_OMP.
14640 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
14641 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
14642 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
14644 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
14646 * dumpfile.c (optgroup_options): Instead of "openmp", associate
14647 OPTGROUP_OMP with "omp".
14649 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
14652 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
14653 for arithmetic shift of unsigned V2DI.
14655 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
14657 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
14658 arc/linux.h headers.
14659 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
14660 (LINK_SPEC): Likewise.
14661 (ARC_TLS_EXTRA_START_SPEC): Likewise.
14662 (EXTRA_SPECS): Likewise.
14663 (STARTFILE_SPEC): Likewise.
14664 (ENDFILE_SPEC): Likewise.
14665 (LIB_SPEC): Likewise.
14666 (TARGET_SDATA_DEFAULT): Likewise.
14667 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
14668 (MULTILIB_DEFAULTS): Likewise.
14669 (DWARF2_UNWIND_INFO): Likewise.
14670 * config/arc/big.h: New file.
14671 * config/arc/elf.h: Likewise.
14672 * config/arc/linux.h: Likewise.
14673 * config/arc/t-uClibc: Remove.
14675 2017-02-27 Bin Cheng <bin.cheng@arm.com>
14677 PR tree-optimization/77536
14678 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
14679 (tree_transform_and_unroll_loop): Use above function to compute the
14680 estimated niter of unrolled loop and use it when scaling profile.
14681 Also use count info rather than frequency if it's non-zero.
14682 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
14683 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
14684 (vect_transform_loop): Call above function.
14686 2017-02-27 Richard Biener <rguenther@suse.de>
14688 PR tree-optimization/45397
14689 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
14690 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
14691 (visit_nary_op): Add pattern matching for CSEing sign-changed
14692 or truncated operations with wider ones.
14694 2017-02-27 Richard Biener <rguenther@suse.de>
14696 PR tree-optimization/79690
14697 * tree-vect-stmts.c (vectorizable_store): Use vector type
14698 built from the DR with address-space.
14700 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
14702 * doc/invoke.texi (Optimize Options): Refine the description
14703 of asan-use-after-return.
14705 2017-02-25 Alan Modra <amodra@gmail.com>
14707 PR rtl-optimization/79584
14708 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
14709 base, not ad->base_term, the reg within base. Remove assertion
14710 that ad->base == ad->base_term. Replace gen_int_mode using
14711 bogus mode with const0_rtx.
14713 2017-02-25 Jakub Jelinek <jakub@redhat.com>
14715 PR middle-end/79396
14716 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
14717 FMA_EXPR like tcc_binary or tcc_unary.
14719 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
14722 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
14724 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
14725 (output_loc_operands): Handle DW_OP_call_ref and
14726 DW_OP_GNU_variable_value.
14727 (struct variable_value_struct): New type.
14728 (struct variable_value_hasher): Likewise.
14729 (variable_value_hash): New variable.
14730 (string_types): Remove.
14731 (copy_loc_descr): New function.
14732 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
14733 (prepend_loc_descr_to_each): New function.
14734 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
14735 instead of add_loc_descr_to_each if the first argument is single
14736 location list and the second has multiple.
14737 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
14738 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
14739 when looking for variable value which doesn't have other location info.
14740 (loc_list_from_tree): Formatting fix.
14741 (gen_array_type_die): Simplify DW_AT_string_length handling.
14742 (adjust_string_types): Remove.
14743 (gen_subprogram_die): Don't call adjust_string_types nor test/set
14744 string_types. Call resolve_variable_values.
14745 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
14746 (resolve_addr_in_expr): Likewise. Add A argument.
14747 (copy_deref_exprloc): Remove deref argument. Adjust for the
14748 original expression being DW_OP_GNU_variable_value with optionally
14749 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
14750 optionally after it.
14751 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
14752 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
14753 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
14754 (variable_value_hasher::hash, variable_value_hasher::equal): New
14756 (resolve_variable_value_in_expr, resolve_variable_value,
14757 resolve_variable_values, note_variable_value_in_expr,
14758 note_variable_value): New functions.
14759 (dwarf2out_early_finish): Call note_variable_value on all toplevel
14762 2017-02-24 Jakub Jelinek <jakub@redhat.com>
14765 * opts.h (handle_generated_option): Add GENERATED_P argument.
14766 * opts-common.c (handle_option): Adjust function comment.
14767 (handle_generated_option): Add GENERATED_P argument, pass it to
14769 (control_warning_option): Pass false to handle_generated_option
14771 * opts.c (maybe_default_option): Pass true to handle_generated_option
14773 * optc-gen.awk: Likewise.
14775 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14777 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
14778 a REG, look at the REG it is a SUBREG of.
14779 (splitter for cmpeqsi_t): Ditto.
14781 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14783 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
14784 the special USEs with the pattern of the insn, not the insn itself.
14786 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
14789 * doc/invoke.texi: Document -mload-store-pairs.
14791 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14792 Sandra Loosemore <sandra@codesourcery.com>
14794 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
14795 argument isn't a CONST_INT.
14796 (nios2_alternate_compare_const): Assert op is a CONST_INT.
14797 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
14798 (nios2_validate_compare): Bypass alternate compare logic if *op2
14799 is not a CONST_INT.
14800 (ldstwm_operation_p): Return false if first_base is not a REG or
14801 if first_offset is not a CONST_INT.
14803 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14805 * config/cris/cris.md: Use correct operand in a define_peephole2.
14807 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14809 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
14811 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
14813 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
14814 this_insn if it is an INSN or JUMP_INSN.
14815 (force_offsettable): Look at base, not at addr.
14816 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
14817 on things that aren't necessarily CONST_INTs.
14819 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
14821 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
14822 -mfpmath=sse is the default also for x86-32 targets with SSE2
14823 instruction set when @option{-ffast-math} is enabled
14825 2017-02-24 Jeff Law <law@redhat.com>
14827 PR rtl-optimizatoin/79286
14828 * ira.c (update_equiv_regs): Drop may_trap_p exception to
14831 2017-02-24 Richard Biener <rguenther@suse.de>
14833 PR tree-optimization/79389
14834 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
14837 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
14839 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
14840 function comment to reflect reality.
14841 (loop_exits_before_overflow): Fix typo in function description.
14843 2017-02-24 Richard Biener <rguenther@suse.de>
14845 PR tree-optimization/79389
14846 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
14847 properly that a threading opportunity exists. Detect conditional
14848 copy/constant propagation opportunities.
14850 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
14852 * config/visium/visium.md (type): Add trap.
14853 (b): New mode attribute.
14854 (*btst): Rename into...
14855 (*btst<mode>): ...this and adjust.
14856 (*cbranchsi4_btst_insn): Rename into...
14857 (*cbranch<mode>4_btst_insn): ...this and adjust.
14858 (trap): New define_insn.
14860 2017-02-23 Jakub Jelinek <jakub@redhat.com>
14862 PR tree-optimization/79389
14863 * ifcvt.c (struct noce_if_info): Add rev_cond field.
14864 (noce_reversed_cond_code): New function.
14865 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
14866 reversed_comparison_code. Formatting fix.
14867 (noce_try_store_flag): Test rev_cond != NULL in addition to
14868 reversed_comparison_code.
14869 (noce_try_store_flag_constants): Likewise.
14870 (noce_try_store_flag_mask): Likewise.
14871 (noce_try_addcc): Use rev_cond if non-NULL instead of
14872 reversed_comparison_code.
14873 (noce_try_cmove_arith): Likewise. Formatting fixes.
14874 (noce_try_minmax, noce_try_abs): Clear rev_cond.
14875 (noce_find_if_block): Initialize rev_cond.
14876 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
14877 instead of false as last argument never attempt to reverse it
14880 2017-02-23 Bin Cheng <bin.cheng@arm.com>
14882 PR tree-optimization/79663
14883 * tree-predcom.c (combine_chains): Process refs in reverse order
14884 only for ZERO length chains, and add explaining comment.
14886 2017-02-23 Jeff Law <law@redhat.com>
14888 PR tree-optimization/79578
14889 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
14890 in call to operand_equal_p.
14892 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
14895 * config/i386/cpuid.h: Fix another undefined behavior.
14897 2017-02-23 Richard Biener <rguenther@suse.de>
14899 PR tree-optimization/79683
14900 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
14901 vector types for data-refs.
14903 2017-02-23 Martin Liska <mliska@suse.cz>
14905 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
14907 2017-02-23 Jakub Jelinek <jakub@redhat.com>
14909 PR middle-end/79665
14910 * internal-fn.c (get_range_pos_neg): Moved to ...
14911 * tree.c (get_range_pos_neg): ... here. No longer static.
14912 * tree.h (get_range_pos_neg): New prototype.
14913 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
14914 are known to be in between 0 and signed maximum inclusive, try to
14915 expand both unsigned and signed divmod and use the cheaper one from
14918 2017-02-22 Jeff Law <law@redhat.com>
14920 PR tree-optimization/79578
14921 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
14922 to compare base operands.
14924 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
14927 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
14928 gpc_reg_operand instead of fpr_reg_operand.
14930 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
14932 * config/mips/mips.c (mips_return_in_memory): Force FP
14933 vector types to be returned in memory for o32 ABI.
14935 2017-02-22 Jakub Jelinek <jakub@redhat.com>
14937 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
14938 instead of DW_TAG_member for static data member declarations and don't
14939 set no_linkage_name for static inline data members.
14940 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
14943 2017-02-22 Martin Liska <mliska@suse.cz>
14945 * doc/invoke.texi: Replace inequality signs with square brackets
14948 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14950 PR tree-optimization/68644
14951 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
14953 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
14956 * lra-constraints.c (simplify_operand_subreg): Handle
14957 WORD_REGISTER_OPERATIONS targets.
14959 2017-02-22 Jakub Jelinek <jakub@redhat.com>
14962 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
14963 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
14964 elimination by swapping fld*.
14966 2017-02-22 Richard Biener <rguenther@suse.de>
14968 PR tree-optimization/79673
14969 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
14970 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
14971 irrelevant address-space qualifiers and avoiding a
14972 ADDR_SPACE_CONVERT_EXPR from fold_convert.
14974 2017-02-22 Richard Biener <rguenther@suse.de>
14976 PR tree-optimization/79666
14977 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
14978 to not symbolically negate if that may introduce undefined
14981 2017-02-22 Martin Liska <mliska@suse.cz>
14984 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
14985 * data-streamer-out.c (streamer_write_gcov_count_stream):
14987 * value-prof.c (stream_out_histogram_value): Make assert more
14988 precise based on type of counter.
14990 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
14993 * config/i386/i386.md (standard_x87sse_constant_load splitter):
14994 Use nonimmediate_operand instead of memory_operand for operand 1.
14995 (float-extend standard_x87sse_constant_load splitter): Ditto.
14997 2017-02-21 Jeff Law <law@redhat.com>
14999 PR tree-optimization/79621
15000 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
15001 blocks with edges to themselves.
15003 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15006 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
15007 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
15008 Use gimple_call_builtin_p.
15011 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
15012 on temporarily removed DEBUG_INSNs.
15014 PR tree-optimization/79649
15015 * tree-loop-distribution.c (classify_partition): Give up on
15016 non-generic address space loads/stores.
15018 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
15020 * doc/loop.texi (Loop manipulation): Remove nonexistent
15021 tree_ssa_loop_version from the documentation.
15022 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
15024 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15027 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
15028 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
15029 * config/rs6000/rs6000.c: Include except.h.
15030 (rs6000_expand_split_stack_prologue): Call
15031 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
15033 2017-02-21 Martin Jambor <mjambor@suse.cz>
15036 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
15037 have been analyzed.
15039 2017-02-21 Martin Jambor <mjambor@suse.cz>
15041 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
15042 for backward compatibility only.
15043 * doc/invoke.texi (Option Summary): Remove all references to
15044 -fipa-cp-alignment.
15046 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
15050 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15052 * lra-constraints.c (curr_insn_transform): Handle
15053 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
15055 2017-02-21 Martin Liska <mliska@suse.cz>
15057 * config/i386/i386.opt: Replace -masm-dialect with -masm.
15059 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
15061 PR translation/79638
15062 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
15064 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
15067 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
15068 (arm_function_ok_for_sibcall): Return false for an indirect call by
15069 descriptor if all the argument registers are used.
15070 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
15071 alignment of the function.
15073 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15075 PR tree-optimization/61441
15076 * simplify-rtx.c (simplify_const_unary_operation): For
15077 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
15078 the sNaN unmodified.
15080 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
15082 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
15083 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
15084 instead of SYSTEM_HEADER_DIR.
15086 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
15087 Martin Liška <mliska@suse.cz>
15089 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
15090 Fix typos and grammar, use active voice, and clarify.
15092 2017-02-20 Marek Polacek <polacek@redhat.com>
15094 PR middle-end/79537
15095 * gimplify.c (gimplify_expr): Handle unused *&&L;.
15098 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
15100 2017-02-20 Jakub Jelinek <jakub@redhat.com>
15103 * config/i386/i386.c (ix86_expand_builtin): Handle
15104 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
15105 ix86_builtins_isa[fcode].isa as a requirement of those
15106 flags and any other flag in the bitmask.
15107 (ix86_init_mmx_sse_builtins): Use 0 instead of
15108 ~OPTION_MASK_ISA_64BIT as mask.
15109 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
15110 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
15111 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
15112 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
15114 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15117 * lra-constraints.c (split_reg): Check requested split mode
15118 is supported by the register.
15120 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15122 * lra-constraints.c (simplify_operand_subreg): Remove early
15125 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15128 * lra-constraints.c (curr_insn_transform): Tighten condition
15129 for converting SUBREG reloads from OP_OUT to OP_INOUT.
15131 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15134 * lra-constraints.c (curr_insn_transform): Handle
15135 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
15137 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
15140 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
15142 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
15144 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
15147 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
15150 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15152 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
15155 2017-02-18 Jakub Jelinek <jakub@redhat.com>
15157 * final.c (last_columnnum, override_columnnum): New variables.
15158 (final_start_function): Set last_columnnum, pass it to begin_prologue
15159 hook and pass 0 to dwarf2out_begin_prologue.
15160 (final_scan_insn): Update override_columnnum. Pass last_columnnum
15161 to source_line debug hook.
15162 (notice_source_line): Compute last_columnnum and for debug_column_info
15163 return true on column changes.
15164 * debug.h (struct gcc_debug_hooks): Add column argument to
15165 source_line and begin_prologue hooks.
15166 (debug_nothing_int_charstar_int_bool): Remove prototype.
15167 (debug_nothing_int_int_charstar,
15168 debug_nothing_int_int_charstar_int_bool): New prototypes.
15169 (dwarf2out_begin_prologue): Add column argument.
15170 * debug.c (do_nothing_debug_hooks): Adjust source_line and
15171 begin_prologue hooks.
15172 (debug_nothing_int_charstar_int_bool): Remove.
15173 (debug_nothing_int_int_charstar,
15174 debug_nothing_int_int_charstar_int_bool): New functions.
15175 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
15176 through to dwarf2out_source_line.
15177 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
15178 (dwarf2out_source_line): Add column argument, emit it if requested.
15179 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
15181 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
15182 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
15183 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
15184 through to dwarf2out_begin_prologue.
15185 (vmsdbgout_source_line): Add column argument, pass it through to
15186 dwarf2out_source_line.
15187 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
15188 dbxout_source_line caller.
15189 (dbxout_source_line): Add column argument.
15191 * common.opt (gno-column-info, gcolumn-info): New options.
15192 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
15193 (check_die): Also test for multiple DW_AT_decl_column attributes.
15194 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
15195 DW_AT_decl_column if requested.
15196 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
15198 (gen_variable_die): Likewise.
15199 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
15200 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
15203 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
15204 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
15205 (ix86_handle_option): Handle OPT_m3dnowa.
15206 * doc/invoke.texi (-m3dnowa): Document.
15207 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
15208 -m3dnowa instead of -m3dnow -march=athlon.
15211 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
15212 instead of gcc_assert for K, r and R code checks. Formatting fixes.
15214 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15217 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
15218 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
15219 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
15220 generator for vsx_xxpermdi_<mode>_be.
15221 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
15222 force big-endian semantics.
15223 (vsx_xxpermdi_<mode>_be): New define_expand with same
15224 implementation as previous version of vsx_xxpermdi_<mode>.
15226 2017-02-17 Jakub Jelinek <jakub@redhat.com>
15228 PR tree-optimization/79327
15229 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
15230 variable, its initialization and use.
15232 2017-02-17 Julia Koval <julia.koval@intel.com>
15234 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
15235 (OPTION_MASK_ISA_PKU_UNSET): New.
15236 (ix86_handle_option): Handle -mrdpid.
15237 * config/i386/cpuid.h (bit_RDPID): New.
15238 * config/i386/driver-i386.c (host_detect_local_cpu):
15239 Detect RDPID feature.
15240 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
15241 * config/i386/i386-c.c (ix86_target_macros_internal):
15243 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
15244 (ix86_valid_target_attribute_inner_p): Add "rdpid".
15245 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
15246 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
15247 * config/i386/i386.md (define_insn "rdpid"): New.
15248 * config/i386/i386.opt Add -mrdpid.
15249 * config/i386/immintrin.h (_rdpid_u32): New.
15251 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
15253 PR rtl-optimization/79541
15254 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
15255 instead of transforming it into USE.
15257 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
15259 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
15260 If HONOR_SNANS (SFmode) force the input to a register.
15261 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
15262 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
15263 an frsp or similar insn.
15265 2017-02-17 Martin Liska <mliska@suse.cz>
15267 PR rtl-optimization/79577
15268 * params.def (selsched-max-sched-times): Increase minimum to 1.
15270 2017-02-17 Martin Liska <mliska@suse.cz>
15272 PR rtl-optimization/79574
15273 * gcse.c (want_to_gcse_p): Prevent integer overflow.
15275 2017-02-17 Martin Liska <mliska@suse.cz>
15277 PR tree-optimization/79529
15278 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
15279 ssa_defined_default_def_p to handle cases which are implicitly
15281 * tree-ssa.c (ssa_defined_default_def_p): New function.
15282 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
15283 which are implicitly defined.
15284 * tree-ssa.h (ssa_defined_default_def_p): Declare.
15286 2017-02-17 Richard Biener <rguenther@suse.de>
15288 PR middle-end/79576
15289 * params.def (max-ssa-name-query-depth): Limit to 10.
15291 2017-02-17 Richard Biener <rguenther@suse.de>
15293 PR tree-optimization/79552
15294 * tree-ssa-structalias.c (visit_loadstore): Properly verify
15297 2017-02-17 Richard Biener <rguenther@suse.de>
15300 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
15302 2017-02-17 Marek Polacek <polacek@redhat.com>
15304 PR middle-end/79536
15305 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
15306 (fold_negate_expr): New wrapper.
15308 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
15310 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
15311 Correct terminology and de-emphasize pre-standard behavior.
15313 2017-02-16 Alan Modra <amodra@gmail.com>
15315 PR rtl-optimization/79286
15316 * ira.c (def_dominates_uses): New function.
15317 (update_equiv_regs): Don't create an equivalence for insns that
15318 may trap where the register def does not dominate the use.
15320 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
15322 PR rtl-optimization/78127
15323 * lra.c (lra): Call lra_eliminate before finish the loop after
15326 2017-02-16 Richard Biener <rguenther@suse.de>
15328 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
15330 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
15331 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
15332 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
15333 (isl_val_int_from_wi): New function.
15334 (extract_affine_gmp): Rename to ...
15335 (extract_affine_wi): ... this, take a widest_int.
15336 (extract_affine_int): Just wrap extract_affine_wi.
15337 (add_param_constraints): Use isl_val_int_from_wi.
15338 (add_loop_constraints): Likewise, and extract_affine_wi.
15340 2017-02-15 Jeff Law <law@redhat.com>
15342 PR middle-end/79521
15343 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
15344 ira_init_register_move_cost_if_necessary.
15346 2017-02-15 Martin Sebor <msebor@redhat.com>
15348 PR middle-end/32003
15349 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
15350 removed in a prior commit.
15352 2017-02-15 Bin Cheng <bin.cheng@arm.com>
15354 PR tree-optimization/79347
15355 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
15356 counters during peeling.
15358 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
15360 * Makefile.in (site.exp): Remove "set ISLVER".
15362 2017-02-15 Jakub Jelinek <jakub@redhat.com>
15365 * real.c (real_from_integer): Call real_convert even for decimal.
15367 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
15370 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
15372 2017-02-14 Andrew Pinski <apinski@cavium.com>
15374 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
15375 cores and change the partno/implementer to be correct.
15376 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
15377 the 'B" as the implementer.
15378 * config/aarch64/aarch64-tune.md: Regenerate.
15380 2017-02-14 Carl Love <cel@us.ibm.com>
15382 * config/rs6000/rs6000.c: Add case statement entry to make the
15383 xvcvuxdsp built-in argument unsigned.
15384 * config/rs6000/vsx.md: Fix the source and return operand types so they
15385 match the instruction definitions from the ISA document. Fix typo
15386 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
15389 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
15392 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
15393 member early_clobber_alts.
15394 * lra-lives.c (reg_early_clobber_p): New.
15395 (process_bb_lives): Use it.
15396 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
15397 (debug_operand_data): Initialize early_clobber_alts.
15398 (setup_operand_alternative): Set up early_clobber_alts.
15399 (collect_non_operand_hard_regs): Ditto. Pass early clobber
15400 alternatives to new_insn_reg.
15401 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
15403 (lra_update_insn_regno_info): Pass the new arg.
15405 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15407 PR middle-end/79505
15408 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
15409 (new_oacc_loop_raw): Don't clear already cleared fields.
15412 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
15413 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
15414 _mm512_prefetch_i64gather_ps): New inline functions and macros.
15416 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
15419 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
15421 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
15424 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
15425 the extra instruction to the right place to store 128-bit constant
15428 2017-02-14 Martin Sebor <msebor@redhat.com>
15430 PR middle-end/79448
15431 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
15432 warning for strings of unknown length.
15434 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
15436 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
15438 2017-02-14 Jeff Law <law@redhat.com>
15441 * ira-costs.c (scan_one_insn): Initialize register move costs
15442 for pseudos seen in USE/CLOBBER insns.
15444 PR tree-optimization/79095
15445 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
15446 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
15447 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
15448 if the operands are known to be not equal, then the resulting range
15450 (intersect_ranges): If the new range is ~[0,0] and the old range is
15451 wide, then prefer ~[0,0].
15452 * tree-vrp.c (overflow_comparison_p_1): New function.
15453 (overflow_comparison_p): New function.
15454 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
15455 if NAME is used in an overflow test.
15456 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
15457 overflow check that can be expressed as an equality test, then adjust
15458 ops to be that equality test.
15460 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15462 * config/s390/s390-builtin-types.def: Remove flags argument.
15463 * config/s390/s390.c (s390_init_builtins): Likewise.
15465 2017-02-14 Martin Liska <mliska@suse.cz>
15467 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
15468 vector. Fix trailing white spaces.
15470 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
15472 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
15475 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15477 PR rtl-optimization/68664
15478 * config/arm/arm.c (arm_sched_can_speculate_insn):
15479 New function. Declare prototype.
15480 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15482 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15484 PR rtl-optimization/68664
15485 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
15487 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15489 2017-02-14 Amit Pawar <amit.pawar@amd.com>
15491 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
15492 max skip bytes for function, loop and jump.
15494 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15496 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
15497 ABS_EXPR for gimple dump.
15499 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15502 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
15504 PR tree-optimization/79408
15505 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
15506 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
15507 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
15508 also if rhs1 is INTEGER_CST.
15510 2017-02-14 Richard Biener <rguenther@suse.de>
15512 PR middle-end/79432
15513 * tree-into-ssa.c (insert_phi_nodes): When the function can
15514 have abnormal edges rewrite SSA names with broken use-def
15515 dominance out of SSA and register them for PHI insertion.
15517 2017-02-13 Martin Sebor <msebor@redhat.com>
15519 PR middle-end/79496
15520 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
15521 clearing info.nowrite flag when snprintf size argument is a range.
15523 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15525 * cprop.c (cprop_jump): Add missing space in string literal.
15526 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
15527 (get_constraint_for_component_ref): Likewise.
15528 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
15529 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
15530 * lra-constraints.c (process_alt_operands): Likewise.
15531 * ipa-inline.c (inline_small_functions): Likewise.
15532 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
15533 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15534 * trans-mem.c (diagnose_tm_1_op): Likewise.
15535 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
15536 (grid_parallel_clauses_gridifiable): Likewise.
15538 * config/nvptx/mkoffload.c (process): Add space in between
15541 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
15542 "MOD4_SSE_REGS" and "ALL_REGS".
15544 * spellcheck.c (test_data): Add , in between "foo" and "food".
15546 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15549 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
15550 boundary crossing check and subsequent code generation agree.
15552 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15554 * config/aarch64/aarch64.c (has_memory_op): Delete.
15555 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
15558 2017-02-13 Jakub Jelinek <jakub@redhat.com>
15560 PR rtl-optimization/79388
15561 PR rtl-optimization/79450
15562 * combine.c (distribute_notes): When removing TEM_INSN for which
15563 corresponding dest has last value recorded, invalidate that last
15566 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15568 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
15569 of explicit '@'. Add missing assembly comment marker on branch costs
15572 2017-02-13 Nathan Sidwell <nathan@acm.org>
15574 * gengtype-lex.l (<in_struct>): Add '/'.
15576 2017-02-13 Martin Liska <mliska@suse.cz>
15579 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
15581 2017-02-13 Richard Biener <rguenther@suse.de>
15583 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
15585 * configure: Re-generate.
15586 * config.in: Likewise.
15587 * graphite-dependences.c: Simplify as if
15588 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
15589 * graphite-isl-ast-to-gimple.c: Likewise.
15590 * graphite-optimize-isl.c: Likewise.
15591 * graphite-poly.c: Likewise.
15592 * graphite-sese-to-poly.c: Likewise.
15593 * graphite.h: Likewise.
15594 * toplev.c: Include isl/version.h and use isl_version () for
15595 printing the ISL version.
15596 * doc/install.texi: Update ISL requirement.
15598 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15600 * doc/standards.texi (Standards): Update reference to
15603 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
15605 * doc/extend.texi (Named Address Spaces): sourceware.org now
15607 * doc/install.texi (Binaries): Ditto.
15610 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15612 * doc/cpp.texi: Replace "stringify"/"stringification" with C
15613 standard terminology "stringize"/"stringizing" throughout.
15614 * doc/cppinternals.texi: Likewise.
15616 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
15618 * doc/extend.texi: Fix some spelling mistakes and typos.
15619 * doc/invoke.texi: Likewise.
15621 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15624 * params.def (inline-min-speedup) Change from 10 to 8.
15626 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15628 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
15631 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
15634 * ipa-inline-analysis.c (get_minimal_bb): New function.
15635 (record_modified): Use it.
15636 (remap_edge_change_prob): Handle also ancestor functions.
15638 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
15640 * doc/contrib.texi (Contributors): Remove broken link into
15641 the Mauve CVS repository.
15643 2017-02-11 Jakub Jelinek <jakub@redhat.com>
15645 PR middle-end/79454
15646 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
15647 result computation whenever lhs doesn't have vector mode, not
15648 just when it has BLKmode.
15650 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15652 * doc/makefile.texi (profiledbootstrap): Refer to the
15653 installation instructions only in textual form.
15655 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15658 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
15660 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
15662 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
15663 (Specific): Update mingw-w64 reference.
15665 (Specific): Remove broken link to Renesas RX processor.
15667 2017-02-10 Richard Biener <rguenther@suse.de>
15669 * toplev.c (process_options): Do not mention obsolete graphite
15670 options when printing sorry message about missing graphite support.
15671 Mention -floop-nest-optimize.
15673 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
15675 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
15676 (vtst_p16): Likewise.
15677 (vtstq_p8): Likewise.
15678 (vtstq_p16): Likewise.
15680 (vtstq_p64): Likewise.
15681 * config/arm/arm_neon.h (vgetq_lane_p64): New.
15682 (vset_lane_p64): New.
15683 (vsetq_lane_p64): New.
15685 2017-02-10 Jakub Jelinek <jakub@redhat.com>
15687 PR tree-optimization/79411
15688 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
15689 stmt operands are SSA_NAMEs used in abnormal phis.
15690 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
15693 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15696 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
15699 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
15701 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
15703 2017-02-09 Jakub Jelinek <jakub@redhat.com>
15705 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
15709 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
15710 not arbitrary TREE_CONSTANT.
15713 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
15714 "omp declare target link" attribute unless is_global_var.
15715 * omp-offload.c (find_link_var_op): Likewise.
15717 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
15718 Chung-Lin Tang <cltang@codesourcery.com>
15720 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
15722 (gimplify_adjust_omp_clauses): Don't delete TILE.
15723 (gimplify_omp_for): Deal with TILE.
15724 * internal-fn.c (expand_GOACC_TILE): New function.
15725 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
15727 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
15728 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
15730 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
15731 avoid DIV for outermost collapse var.
15732 (expand_oacc_for): Insert tile element loop as needed. Adjust.
15733 Remove out of date comments, fix whitespace.
15734 * omp-general.c (omp_extract_for_data): Deal with tiling.
15735 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
15736 adjust OLF_DIM_BASE value.
15737 (struct omp_for_data): Add tiling field.
15738 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
15739 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
15740 for auto loops. Remove default auto determining, moved to
15741 oacc_loop_fixed_partitions.
15742 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
15743 stmts, add e_mask field.
15744 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
15745 (oacc_thread_numbers): Use oacc_dim_call.
15746 (oacc_xform_tile): New.
15747 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
15748 (finish_oacc_loop): Adjust for ifns vector.
15749 (oacc_loop_discover_walk): Append loop abstraction sites to list,
15750 add case for GOACC_TILE fns.
15751 (oacc_loop_xform_loop): Delete.
15752 (oacc_loop_process): Iterate over call list directly, and add
15753 handling for GOACC_TILE fns.
15754 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
15756 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
15757 vector partitioning to outer loops. Assign 2 partitions to loops
15758 when available. Add TILE handling.
15759 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
15760 (execite_oacc_device_lower): Process GOACC_TILE fns,
15761 ignore unknown specs.
15762 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
15763 * tree.c (omp_clause_num_ops): Adjust TILE ops.
15764 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
15766 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
15768 * configure.ac (ACX_BUGURL): Update.
15769 * configure: Regenerate.
15771 2017-02-09 Richard Biener <rguenther@suse.de>
15773 PR tree-optimization/69823
15774 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
15775 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
15777 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15779 * config/arc/arc-c.def: Add __NPS400__ definition.
15780 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
15781 (TARGET_NPS400): Define.
15783 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
15785 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
15787 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
15788 pointer, arch_info.
15789 (arc_cpu_types): Fill the arch_info field with a pointer into the
15790 arc_arch_types table.
15791 (arc_selected_cpu): Declare.
15792 * config/arc/arc.c (arc_selected_cpu): Make global.
15793 (arc_selected_arch): Delete.
15794 (arc_base_cpu): Delete.
15795 (arc_override_options): Remove references to deleted variables,
15796 update access to arch information.
15797 (ARC_OPT): Update access to arch information.
15798 (ARC_OPTX): Likewise.
15799 * config/arc/arc.h (arc_base_cpu): Remove declaration.
15800 (TARGET_ARC600): Update access to arch information.
15801 (TARGET_ARC601): Likewise.
15802 (TARGET_ARC700): Likewise.
15803 (TARGET_EM): Likewise.
15804 (TARGET_HS): Likewise.
15805 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
15808 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
15811 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
15812 condition/operands for integer GE/LE/GEU/LEU operations.
15814 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
15816 PR translation/79397
15817 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
15820 2017-02-08 Martin Jambor <mjambor@suse.cz>
15823 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
15824 whether allocation happened.
15825 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
15826 nothing was allocated.
15828 2017-02-08 Jakub Jelinek <jakub@redhat.com>
15830 PR tree-optimization/79408
15831 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
15832 constant, but SSA_NAME with a known integer range, use the minimum
15833 of that range instead of op1 to determine if modulo can be replaced
15834 with its first operand.
15836 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15838 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
15840 2017-02-08 Richard Biener <rguenther@suse.de>
15842 PR tree-optimization/71824
15843 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
15844 Check all loops contained in the merged region.
15846 2017-02-07 Andrew Pinski <apinski@cavium.com>
15848 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
15850 2017-02-07 Andrew Pinski <apinski@cavium.com>
15852 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
15853 (thunderxt88): Likewise.
15854 (thunderxt81): Disable LSE and change v8.1 to v8.
15855 (thunderxt83): Likewise.
15857 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15858 Richard Biener <rguenther@suse.de>
15860 PR middle-end/79399
15861 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
15862 type from int to size_t.
15863 * ira-costs.c (struct_costs_size): Change type from int to size_t.
15865 2017-02-07 Jakub Jelinek <jakub@redhat.com>
15867 PR rtl-optimization/79386
15868 * cprop.c (bypass_conditional_jumps): Initialize
15869 bypass_last_basic_block already before splitting bbs after
15870 unconditional traps...
15871 (bypass_conditional_jumps): ... rather than here.
15874 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
15875 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
15876 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
15877 fix -masm=intel patterns.
15879 2017-02-07 Richard Biener <rguenther@suse.de>
15881 PR tree-optimization/79256
15882 PR middle-end/79278
15883 * builtins.c (get_object_alignment_2): Use min_align_of_type
15884 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
15885 and ADJUST_FIELD_ALIGN.
15887 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
15889 * doc/tm.texi: Regenerate.
15890 * stor-layout.c (layout_decl): Adjust.
15891 (update_alignment_for_field): Likewise.
15892 (place_field): Likewise.
15893 (min_align_of_type): Likewise.
15894 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
15895 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
15896 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
15897 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
15898 * config/frv/frv.c (frv_adjust_field_align): Likewise.
15899 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
15900 * config/i386/i386.c (x86_field_alignment): Likewise.
15901 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
15902 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
15903 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
15904 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
15905 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
15906 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
15910 2017-01-30 Richard Biener <rguenther@suse.de>
15912 PR tree-optimization/79256
15913 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
15914 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
15917 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
15919 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
15920 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
15921 builtins to SImode and emit a zero-extend, if necessary.
15923 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15925 * docs/invoke.texi (RISC-V Options): Alphabetize.
15927 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15929 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
15932 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
15934 * config/riscv/riscv.c: New file.
15935 * common/config/riscv/riscv-common.c: Likewise.
15936 * config.gcc: Likewise.
15937 * config/riscv/constraints.md: Likewise.
15938 * config/riscv/elf.h: Likewise.
15939 * config/riscv/generic.md: Likewise.
15940 * config/riscv/linux.h: Likewise.
15941 * config/riscv/multilib-generator: Likewise.
15942 * config/riscv/peephole.md: Likewise.
15943 * config/riscv/pic.md: Likewise.
15944 * config/riscv/predicates.md: Likewise.
15945 * config/riscv/riscv-builtins.c: Likewise.
15946 * config/riscv/riscv-c.c: Likewise.
15947 * config/riscv/riscv-ftypes.def: Likewise.
15948 * config/riscv/riscv-modes.def: Likewise.
15949 * config/riscv/riscv-opts.h: Likewise.
15950 * config/riscv/riscv-protos.h: Likewise.
15951 * config/riscv/riscv.h: Likewise.
15952 * config/riscv/riscv.md: Likewise.
15953 * config/riscv/riscv.opt: Likewise.
15954 * config/riscv/sync.md: Likewise.
15955 * config/riscv/t-elf-multilib: Likewise.
15956 * config/riscv/t-linux: Likewise.
15957 * config/riscv/t-linux-multilib: Likewise.
15958 * config/riscv/t-riscv: Likewise.
15959 * configure.ac: Likewise.
15960 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
15961 Waterman as RISC-V maintainers.
15962 * doc/install.texi: Add RISC-V entries.
15963 * doc/invoke.texi: Add RISC-V options section.
15964 * doc/md.texi: Add RISC-V constraints section.
15965 * configure: Regenerated.
15967 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
15970 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
15971 false values to be constant vectors with all 0 or all 1 bits set.
15972 (vcondu<mode><mode>): Likewise.
15973 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
15975 (fpmask_comparison_operator): Update comment.
15976 (vecint_comparison_operator): New predicate.
15977 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
15978 vector conditionals when the true and false values are constant
15979 vectors with all 0 bits or all 1 bits set.
15981 2017-02-06 Martin Sebor <msebor@redhat.com>
15983 PR tree-optimization/79376
15984 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
15986 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
15988 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
15989 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
15990 to simplify split condition.
15992 2017-02-06 Jakub Jelinek <jakub@redhat.com>
15994 * omp-expand.c (oxpand_omp_atomic_fetch_op,
15995 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
15998 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
16000 PR rtl-optimization/68664
16001 * target.def (can_speculate_insn): New hook.
16002 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
16003 * doc/tm.texi: Regenerate.
16004 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
16005 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
16006 (rs6000_sched_can_speculate_insn): New function.
16008 2017-02-06 Jakub Jelinek <jakub@redhat.com>
16010 PR tree-optimization/79284
16011 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
16012 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
16013 vectorizable_mask_load_store, vectorizable_operation,
16014 vect_is_simple_cond, get_same_sized_vectype): Use it instead
16015 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
16016 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
16017 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
16018 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
16019 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
16020 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
16021 is_gimple_assign (stmt). Replace another such test with
16022 is_gimple_assign (stmt).
16024 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
16027 * config/avr/avr.c (rtl-iter.h): Include it.
16028 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
16029 (avr_legitimate_combined_insn): ...and implementation.
16031 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
16033 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
16034 * config/s390/s390.c (s390_const_operand_ok)
16035 (s390_canonicalize_comparison, s390_extract_part)
16036 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
16037 (s390_contiguous_bitmask_p, s390_rtx_costs)
16038 (legitimize_pic_address): Likewise.
16039 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
16040 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
16041 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
16042 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
16043 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
16045 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
16047 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
16048 REGNO($0) == REGNO($1).
16050 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16052 * config/s390/linux.h(SIZE_TYPE): Add comment.
16054 2017-02-06 Julian Brown <julian@codesourcery.com>
16055 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16056 Virendra Pathak <virendra.pathak@broadcom.com>
16058 * config/aarch64/aarch64-cores.def: Change the scheduler
16060 * config/aarch64/aarch64.md: Include thunderx2t99.md.
16061 * config/aarch64/thunderx2t99.md: New file.
16063 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16065 * doc/standards.texi (Go Language): Update link to language
16068 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
16070 * tree-eh.c (lower_resx): Sanitize profile.
16071 (cleanup_empty_eh_move_lp): Likewise.
16073 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
16076 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
16078 * cfgloopmanip.h (loop_version): Update prototype.
16079 * modulo-sched.c (sms_schedule): Update call of loop_version.
16080 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
16081 * tree-parloops.c (gen_parallel_loop): Likewise.
16082 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
16083 * tree-ssa-loop-split.c (split_loop): Likewise.
16084 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
16085 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
16087 2017-02-05 Martin Liska <mliska@suse.cz>
16090 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
16092 (print_operand_address): Initialize a struct to zero.
16094 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16096 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
16097 garbage collector only in textual form.
16099 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16101 * doc/extend.texi (x86 specific memory model extensions for
16102 transactional memory): Simplify a phrase.
16104 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
16107 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
16108 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
16109 (atomic_storedi_1): Likewise.
16111 2017-02-04 Jakub Jelinek <jakub@redhat.com>
16113 PR tree-optimization/79338
16114 * tree-parloops.c (gather_scalar_reductions): Don't call
16115 vect_analyze_loop_form for loop->inner before destroying loop's
16118 2017-02-03 Martin Sebor <msebor@redhat.com>
16120 PR tree-optimization/79327
16121 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
16122 when precision has resulted in leading zeros.
16123 (format_integer): Adjust the likely counter to assume an unknown
16124 argument that may be zero is non-zero.
16126 2017-02-03 Jason Merrill <jason@redhat.com>
16129 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
16130 avoid copying non-taken branch.
16132 2017-02-03 Jakub Jelinek <jakub@redhat.com>
16134 PR tree-optimization/79340
16135 * tree-vect-loop.c (vectorizable_reduction): Release
16136 vec_defs elements after safe_splicing them into other vectors.
16139 PR tree-optimization/79327
16140 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
16141 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
16143 (format_integer): Use wide_int_to_tree instead of build_int_cst
16144 + to_?hwi. If argmin is NULL, just set argmin and argmax to
16145 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
16146 of shortest and longest sequence.
16148 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
16150 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
16151 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
16153 2017-02-03 Walter Lee <walt@tilera.com>
16156 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
16157 after initial stackframe link reg save.
16158 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
16160 2017-02-03 Jakub Jelinek <jakub@redhat.com>
16163 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
16164 wu for stxssp alternative.
16166 2017-02-03 Martin Sebor <msebor@redhat.com>
16168 PR tree-optimization/79352
16169 * gimple-fold.c (get_range_strlen): Add argument.
16170 (get_range_strlen): Change return type to bool.
16171 (get_maxval_strlen): Pass in a dummy argument.
16172 * gimple-fold.h (get_range_strlen): Change return type to bool.
16173 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
16174 * tree.h (array_at_struct_end_p): Add argument.
16175 * tree.c (array_at_struct_end_p): Handle it.
16177 2017-02-03 Martin Liska <mliska@suse.cz>
16180 * multiple_target.c (create_dispatcher_calls): Redirect edge
16181 from a caller of a dispatcher.
16182 (expand_target_clones): Make the clones local.
16183 (ipa_target_clone): Do both target clones and resolvers.
16184 (ipa_dispatcher_calls): Remove the pass.
16185 (pass_dispatcher_calls::gate): Likewise.
16186 (make_pass_dispatcher_calls): Likewise.
16187 * passes.def (pass_target_clone): Put as very first IPA early
16190 2017-02-03 Martin Liska <mliska@suse.cz>
16192 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
16193 in case of a function with ifunc attribute.
16195 2017-02-03 Martin Liska <mliska@suse.cz>
16197 * cgraph.c (cgraph_node::dump): Dump function version info.
16198 * symtab.c (symtab_node::dump_base): Add missing new line.
16200 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
16202 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
16203 (ifcombine_ifandif): Use it.
16205 2017-02-03 Martin Liska <mliska@suse.cz>
16207 * doc/invoke.texi: Document default value for
16208 use-after-scope-direct-emission-threshold.
16210 2017-02-03 Martin Liska <mliska@suse.cz>
16212 PR tree-optimization/79339
16213 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
16214 (format_floating): Likewise.
16216 2017-02-03 Martin Liska <mliska@suse.cz>
16219 * ipa-prop.c (ipa_node_params_t::insert): Remove current
16221 (ipa_node_params_t::remove): Likewise.
16222 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
16223 initialization from removed ipa_node_params_t::insert.
16224 (ipa_node_params::~ipa_node_params): Move from removed
16225 ipa_node_params_t::release.
16226 * symbol-summary.h (symbol_summary::m_released): New member.
16227 Do not release a summary twice. Do not allow to call finalizer
16228 for types of a summary that live in GGC memory.
16230 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16232 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
16235 2017-02-02 Martin Sebor <msebor@redhat.com>
16237 PR middle-end/79275
16238 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
16239 (format_string): Tighten up the range of output for non-constant
16240 strings and correct the expected range for wide non-constant strings.
16242 2017-02-02 Martin Sebor <msebor@redhat.com>
16244 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
16246 PR middle-end/32003
16247 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
16249 (-fdump-tree-@var): Add to index and document how to come up
16250 with pass-specific option and dump file names.
16251 (-fdump-passes): Clarify where to look for output.
16253 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
16255 PR middle-end/77445
16256 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
16257 statistics of the analyzed path; allow threading for speed when
16258 any of BBs along the path are optimized for speed.
16260 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
16262 PR middle-end/78468
16263 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
16264 settings of the virtual registers.
16267 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
16269 * explow.c (get_dynamic_stack_size): Take known alignment of stack
16270 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
16273 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16275 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
16276 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
16278 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16280 * config/s390/s390.md: Add missing comments with the expanded
16282 * config/s390/vector.md: Likewise.
16283 * config/s390/vx-builtins.md: Likewise.
16285 2017-02-02 Jakub Jelinek <jakub@redhat.com>
16288 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
16289 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
16290 conditions on a single line.
16292 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16294 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
16295 __S390_VX__ to __VX__.
16297 2017-02-01 Andrew Pinski <apinski@cavium.com>
16299 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
16300 stmt_info to record_stmt_cost.
16301 (vect_get_known_peeling_cost): Pass stmt_info if known to
16303 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
16304 cpu_vector_cost field into
16305 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
16306 field into vec_int_stmt_cost and vec_fp_stmt_cost.
16307 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
16308 splitting of scalar_stmt_cost and vec_stmt_cost.
16309 (thunderx_vector_cost): Likewise.
16310 (cortexa57_vector_cost): LIkewise.
16311 (exynosm1_vector_cost): Likewise.
16312 (xgene1_vector_cost): Likewise.
16313 (thunderx2t99_vector_cost): Improve after the splitting of the two
16315 (aarch64_builtin_vectorization_cost): Update for the splitting of
16316 scalar_stmt_cost and vec_stmt_cost.
16318 2017-02-01 Torvald Riegel <triegel@redhat.com>
16319 Richard Henderson <rth@redhat.com>
16321 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
16322 conditional on existance of a fast atomic load.
16323 * optabs-query.c (can_atomic_load_p): New function.
16324 * optabs-query.h (can_atomic_load_p): Declare it.
16325 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
16326 no fast atomic load is available for the particular size of access.
16327 (expand_atomic_compare_and_swap): Likewise.
16328 (expand_atomic_load): Likewise.
16329 (expand_atomic_store): Likewise.
16330 (expand_atomic_fetch_op): Likewise.
16331 * testsuite/lib/target-supports.exp
16332 (check_effective_target_sync_int_128): Remove x86 because it provides
16333 no fast atomic load.
16334 (check_effective_target_sync_int_128_runtime): Likewise.
16336 2017-02-01 Richard Biener <rguenther@suse.de>
16338 * graphite.c: Include tree-vectorizer.h for find_loop_location.
16339 (graphite_transform_loops): Provide opt-info for optimized nests.
16340 * tree-parloop.c (parallelize_loops): Provide opt-info for
16341 parallelized loops.
16343 2017-02-01 Richard Biener <rguenther@suse.de>
16345 PR middle-end/79315
16346 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
16347 was not set before.
16349 2017-02-01 Richard Biener <rguenther@suse.de>
16351 PR tree-optimization/71824
16352 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
16353 Verify the loops are valid in the merged SESE region.
16354 (scop_detection::can_represent_loop_1): Check analyzing the
16355 evolution of the number of iterations in the region succeeds.
16357 2017-01-31 Ian Lance Taylor <iant@golang.org>
16359 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
16360 REG_ARGS_SIZE note to 32-bit push insns and call insn.
16362 2017-01-31 David Malcolm <dmalcolm@redhat.com>
16364 PR preprocessor/79210
16365 * input.c (get_substring_ranges_for_loc): Replace line_width
16366 assertion with error-handling.
16368 2017-01-31 Richard Biener <rguenther@suse.de>
16370 PR tree-optimization/77318
16371 * graphite-sese-to-poly.c (extract_affine): Fix assert.
16372 (create_pw_aff_from_tree): Take loop parameter.
16373 (add_condition_to_pbb): Pass loop of the condition to
16374 create_pw_aff_from_tree.
16376 2017-01-31 Jakub Jelinek <jakub@redhat.com>
16378 * config/s390/s390.c (s390_asan_shadow_offset): New function.
16379 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
16381 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
16385 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
16387 (convert_int_to_float128): Likewise.
16388 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
16389 (convert_int_to_float128): Likewise.
16390 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
16391 (UNSPEC_IEEE128_CONVERT): Likewise.
16392 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
16393 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
16394 Use local variables for IBM extended format.
16395 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
16396 (fix_trunc<mode>si2_fprs): Likewise.
16397 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
16398 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
16399 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
16400 to know that we can now have integers of all sizes in vector
16402 (fix<uns>_<mode>di2_hw): Likewise.
16403 (float<uns>_<mode>si2_hw): Likewise.
16404 (fix_<mode>si2_hw): Likewise.
16405 (fixuns_<mode>si2_hw): Likewise.
16406 (float<uns>_<mode>di2_hw): Likewise.
16407 (float_<mode>di2_hw): Likewise.
16408 (float_<mode>si2_hw): Likewise.
16409 (floatuns_<mode>di2_hw): Likewise.
16410 (floatuns_<mode>si2_hw): Likewise.
16411 (xscvqp<su>wz_<mode>): Delete, no longer used.
16412 (xscvqp<su>dz_<mode>): Likewise.
16413 (xscv<su>dqp_<mode>): Likewise.
16414 (ieee128_mfvsrd_64bit): Likewise.
16415 (ieee128_mfvsrd_32bit): Likewise.
16416 (ieee128_mfvsrwz): Likewise.
16417 (ieee128_mtvsrw): Likewise.
16418 (ieee128_mtvsrd_64bit): Likewise.
16419 (ieee128_mtvsrd_32bit): Likewise.
16421 2017-01-31 Martin Liska <mliska@suse.cz>
16424 * ipa-prop.c (ipa_free_all_node_params): Call release method
16425 instead of ~sumbol_summary to not to trigger double times
16428 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
16430 PR tree-optimization/71691
16431 * bitmap.h (class auto_bitmap): New.
16432 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
16433 is_maybe_undefined instead of ssa_undefined_value_p.
16435 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16437 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
16438 __S390_ARCH_LEVEL__ to __ARCH__.
16440 2017-01-31 Jakub Jelinek <jakub@redhat.com>
16442 PR tree-optimization/79267
16443 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
16444 if should_remove_lhs_p is true.
16446 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
16449 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
16450 (add_alignment_attribute): New.
16451 (base_type_die): Add alignment attribute.
16452 (subrange_type_die): Likewise.
16453 (modified_type_die): Likewise.
16454 (gen_array_type_die): Likewise.
16455 (gen_descr_array_type_die: Likewise.
16456 (gen_enumeration_type_die): Likewise.
16457 (gen_subprogram_die): Likewise.
16458 (gen_variable_die): Likewise.
16459 (gen_field_die): Likewise.
16460 (gen_ptr_to_mbr_type_die): Likewise.
16461 (gen_struct_or_union_type_die): Likewise.
16462 (gen_subroutine_type_die): Likewise.
16463 (gen_typedef_die): Likewise.
16464 (base_type_cmp): Compare alignment attribute.
16466 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16469 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
16470 (setb_unsigned) New pattern for setb with CCUNS.
16471 * config/rs6000/rs6000.c (expand_block_compare): Use a different
16472 subfc./subfe sequence to avoid overflow problems. Generate a
16473 shorter sequence with cmpld/setb for power9.
16474 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
16475 for generating subfc. instruction.
16476 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
16477 now uses this instruction.
16479 2017-01-30 Ian Lance Taylor <iant@google.com>
16482 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
16483 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
16485 2017-01-30 Martin Sebor <msebor@redhat.com>
16487 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
16488 Move constant to the right of a relational operator.
16489 (get_mpfr_format_length, format_character, format_string): Ditto.
16490 (should_warn_p, maybe_warn): Same.
16492 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
16494 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
16497 * asan.c (get_translation_unit_decl): Remove function.
16498 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
16500 2017-01-30 Martin Liska <mliska@suse.cz>
16502 PR gcov-profile/79259
16503 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
16504 -fprofile-generate.
16506 2017-01-30 Martin Liska <mliska@suse.cz>
16509 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
16510 Initialize variables with NULL value.
16512 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
16515 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
16517 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
16519 2017-01-30 Richard Biener <rguenther@suse.de>
16521 PR tree-optimization/79276
16522 * tree-vrp.c (process_assert_insertions): Properly adjust common
16523 when removing a duplicate.
16525 * gcc.dg/torture/pr79276.c: New testcase.
16527 2017-01-30 Richard Biener <rguenther@suse.de>
16529 PR tree-optimization/79256
16530 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
16531 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
16533 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
16535 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
16538 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
16539 ("*r<noxa>sbg_<mode>_sll_bitmask")
16540 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
16541 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
16542 Use contiguous_bitmask_nowrap_operand.
16544 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16547 * config/rs6000/altivec.h (vec_xl): Revise #define.
16548 (vec_xst): Likewise.
16550 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
16552 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
16554 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
16556 PR rtl-optimization/79194
16557 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
16558 traps before call to bypass_conditional_jumps.
16560 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16562 PR tree-optimization/71374
16563 * lra-constraints.c (check_conflict_input_operands): New.
16564 (match_reload): Use it.
16566 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
16569 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
16570 account to calculate conflict_set.
16572 2017-01-27 Bin Cheng <bin.cheng@arm.com>
16574 PR rtl-optimization/78559
16575 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
16576 other_insn in combine.
16578 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16580 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
16581 uint16_type_node for BT_UINT16.
16583 2017-01-27 David Malcolm <dmalcolm@redhat.com>
16585 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
16586 "RTL Tests" to menu.
16587 (GIMPLE Tests): New node.
16588 (RTL Tests): New node.
16590 2017-01-27 Richard Biener <rguenther@suse.de>
16592 PR tree-optimization/79245
16593 * tree-loop-distribution.c (distribute_loop): Apply cost
16594 modeling also to detected patterns.
16596 2017-01-27 Richard Biener <rguenther@suse.de>
16598 PR tree-optimization/71433
16599 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
16600 (compare_assert_loc): New function.
16601 (process_assert_insertions): Sort and optimize assert locations
16602 to remove duplicates and push down identical assertions on
16603 edges to their destination block.
16605 2017-01-27 Richard Biener <rguenther@suse.de>
16607 PR tree-optimization/79244
16608 * tree-vrp.c (remove_range_assertions): Forcefully propagate
16609 out SSA names even if abnormal.
16611 2017-01-27 Jakub Jelinek <jakub@redhat.com>
16613 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
16614 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
16615 instead of MPFR_RNDN.
16617 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
16620 * arm.c (arm_option_override): Don't call build_target_option_node
16621 until after doing all option overrides.
16622 (arm_valid_target_attribute_tree): Likewise.
16624 2017-01-27 Martin Liska <mliska@suse.cz>
16626 * doc/invoke.texi (-fprofile-arcs): Document profiling support
16627 for {cd}tors and C++ {cd}tors.
16629 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
16631 * config/s390/s390.md ("*setmem_long_and")
16632 ("*setmem_long_and_31z"): Use zero_extend instead of and.
16634 2017-01-26 Martin Sebor <msebor@redhat.com>
16636 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
16639 2017-01-26 Martin Sebor <msebor@redhat.com>
16641 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
16642 HAVE_DFmode before using XFmode or DFmode.
16643 (parse_directive): Avoid using the z length modifier to avoid
16644 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
16646 PR middle-end/78703
16647 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
16648 to accept adjustment as an array.
16649 (get_int_range): New function.
16650 (struct directive): Make width and prec arrays.
16651 (directive::set_width, directive::set_precision): Call get_int_range.
16652 (format_integer, format_floating): Handle width and precision ranges.
16653 (format_string, parse_directive): Same.
16655 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16658 * dwarf2out.c (generate_skeleton_bottom_up): For children with
16659 comdat_type_p set, just clone them, but keep the children in the
16663 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
16664 which have direct callers with -fvar-tracking-assignments enabled
16666 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
16667 inside of type units.
16669 2017-01-26 Martin Sebor <msebor@redhat.com>
16671 PR middle-end/78703
16672 * gimple-ssa-sprintf.c (struct result_range): Add likely and
16674 (struct format_result): Replace number_chars, number_chars_min,
16675 and number_chars_max with a single member of struct result_range.
16677 (format_result::operator+=): Adjust.
16678 (struct fmtresult): Remove bounded. Handle likely and unlikely
16680 (fmtresult::adjust_for_width_or_precision): New function.
16681 (fmtresult:type_max_digits): New function.
16682 (bytes_remaining): Handle likely and unlikely counters.
16683 (min_bytes_remaining): Remove.
16684 (format_percent): Simplify.
16685 (format_integer, format_floating): Set likely and unlikely counters.
16686 (get_string_length, format_character, format_string): Same.
16687 (format_plain, should_warn_p): New function.
16688 (maybe_warn): Call should_warn_p. Update diagnostic messages
16689 and handle those for all directives, including plain strings.
16690 (format_directive): Handle likely and unlikely counters.
16691 Remove unnecessary quoting from diagnostics. Add an informational
16693 (add_bytes): Remove.
16694 (pass_sprintf_length::compute_format_length): Simplify.
16695 (try_substitute_return_value): Handle likely and unlikely counters.
16697 2017-01-26 Carl Love <cel@us.ibm.com>
16699 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
16700 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
16702 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
16705 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
16706 endianess for subregs into account.
16707 * lra-constraints.c (lra_constraints): Do risky transformations
16708 always on the first iteration.
16709 * lra-lives.c (check_pseudos_live_through_calls): Add arg
16710 last_call_used_reg_set.
16711 (process_bb_lives): Define and use last_call_used_reg_set.
16712 * lra.c (lra): Always continue after lra_constraints on the first
16715 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
16717 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
16719 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
16721 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16723 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
16724 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
16725 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
16726 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
16727 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
16728 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
16729 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
16730 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
16731 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
16733 2017-01-26 Marek Polacek <polacek@redhat.com>
16736 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
16737 for the third operand.
16739 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16741 PR middle-end/79236
16742 * omp-low.c (struct omp_context): Add simt_stmt field.
16743 (scan_omp_for): Return omp_context *.
16744 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
16745 context to the _simt_ SIMD stmt.
16746 (lower_omp_for): For combined SIMD with sibling _simt_
16747 SIMD, make sure to use the same decls in _looptemp_
16748 clauses as in the sibling.
16750 2017-01-26 David Sherwood <david.sherwood@arm.com>
16752 PR middle-end/79212
16753 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
16756 2017-01-26 Jakub Jelinek <jakub@redhat.com>
16759 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
16760 emit fld b; fld a; if possible.
16762 * brig-builtins.def: Update copyright years.
16763 * config/arm/arm_acle_builtins.def: Update copyright years.
16765 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16768 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
16769 constraint instead of o for the stxsd instruction.
16771 2017-01-25 Carl Love <cel@us.ibm.com>
16773 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
16774 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
16776 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
16778 * doc/invoke.texi (C++ Dialect Options): Fix typo.
16780 2017-01-25 Richard Biener <rguenther@suse.de>
16782 PR tree-optimization/69264
16783 * target.def (vector_alignment_reachable): Improve documentation.
16784 * doc/tm.texi: Regenerate.
16785 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
16787 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
16788 earlier changes with respect to TYPE_USER_ALIGN.
16789 (vector_alignment_reachable_p): Likewise. Improve dumping.
16791 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16794 * config/arm/arm.md (xordi3): Force constant operand into a register
16797 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16799 * doc/invoke.texi (-fstore-merging): Correct default optimization
16800 levels at which it is enabled.
16801 (-O): Move -fstore-merging from list to...
16804 2017-01-25 Richard Biener <rguenther@suse.de>
16807 * omp-expand.c: Include debug.h.
16808 (expand_omp_taskreg): Make sure to generate early debug before
16809 outlining anything from a function.
16810 (expand_omp_target): Likewise.
16811 (grid_expand_target_grid_body): Likewise.
16813 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
16816 * asan.c (get_translation_unit_decl): New function.
16817 (asan_add_global): Extract modules file name from globals
16818 TRANSLATION_UNIT_DECL name.
16820 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
16823 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
16824 for long calls with APCS frame and VFP.
16826 2017-01-24 David Malcolm <dmalcolm@redhat.com>
16828 * cfg.c (original_copy_tables_initialized_p): New function.
16829 * cfg.h (original_copy_tables_initialized_p): New decl.
16830 * cfgrtl.c (relink_block_chain): Guard the call to
16831 free_original_copy_tables with a call to
16832 original_copy_tables_initialized_p.
16833 * cgraph.h (symtab_node::native_rtl_p): New decl.
16834 * cgraphunit.c (symtab_node::native_rtl_p): New function.
16835 (symtab_node::needed_p): Don't assert for early assembly output
16836 for __RTL functions.
16837 (cgraph_node::finalize_function): Set "force_output" for __RTL
16839 (cgraph_node::analyze): Bail out early for __RTL functions.
16840 (analyze_functions): Update assertion to support __RTL functions.
16841 (cgraph_node::expand): Bail out early for __RTL functions.
16842 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
16844 * function.h (struct function): Update comment for field
16846 * gimple-expr.c: Include "tree-pass.h".
16847 (gimple_has_body_p): Return false for __RTL functions.
16848 * Makefile.in (OBJS): Add run-rtl-passes.o.
16849 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
16851 (gcc::pass_manager::get_clean_slate): New accessor.
16852 * passes.c: Include "insn-addr.h".
16853 (should_skip_pass_p): Add logging. Update logic for running
16854 "expand" to be compatible with both __GIMPLE and __RTL. Guard
16855 property-provider override so it is only done for gimple passes.
16857 (skip_pass): New function.
16858 (execute_one_pass): Call skip_pass when skipping passes.
16859 * read-md.c (md_reader::read_char): Support filtering
16860 the input to a subset of line numbers.
16861 (md_reader::md_reader): Initialize fields
16862 m_first_line and m_last_line.
16863 (md_reader::read_file_fragment): New function.
16864 * read-md.h (md_reader::read_file_fragment): New decl.
16865 (md_reader::m_first_line): New field.
16866 (md_reader::m_last_line): New field.
16867 * read-rtl-function.c (function_reader::create_function): Only
16868 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
16869 curr_properties. Set DECL_INITIAL to a dummy block.
16870 (read_rtl_function_body_from_file_range): New function.
16871 * read-rtl-function.h (read_rtl_function_body_from_file_range):
16873 * run-rtl-passes.c: New file.
16874 * run-rtl-passes.h: New file.
16876 2017-01-24 Jeff Law <law@redhat.com>
16878 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
16881 2017-01-24 Bin Cheng <bin.cheng@arm.com>
16883 PR tree-optimization/79159
16884 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
16885 (record_nonwrapping_iv): Improve boundary using above function if no
16886 value range information.
16888 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
16889 Martin Jambor <mjambor@suse.cz>
16891 * brig-builtins.def: New file.
16892 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
16893 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
16894 (DEF_HSAIL_SAT_BUILTIN): Likewise.
16895 (DEF_HSAIL_INTR_BUILTIN): Likewise.
16896 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
16897 * builtin-types.def (BT_INT8): New.
16898 (BT_INT16): Likewise.
16899 (BT_UINT8): Likewise.
16900 (BT_UINT16): Likewise.
16901 (BT_FN_ULONG): Likewise.
16902 (BT_FN_UINT_INT): Likewise.
16903 (BT_FN_UINT_ULONG): Likewise.
16904 (BT_FN_UINT_LONG): Likewise.
16905 (BT_FN_UINT_PTR): Likewise.
16906 (BT_FN_ULONG_PTR): Likewise.
16907 (BT_FN_INT8_FLOAT): Likewise.
16908 (BT_FN_INT16_FLOAT): Likewise.
16909 (BT_FN_UINT32_FLOAT): Likewise.
16910 (BT_FN_UINT16_FLOAT): Likewise.
16911 (BT_FN_UINT8_FLOAT): Likewise.
16912 (BT_FN_UINT64_FLOAT): Likewise.
16913 (BT_FN_UINT16_UINT32): Likewise.
16914 (BT_FN_UINT32_UINT16): Likewise.
16915 (BT_FN_UINT16_UINT16_UINT16): Likewise.
16916 (BT_FN_INT_PTR_INT): Likewise.
16917 (BT_FN_UINT_PTR_UINT): Likewise.
16918 (BT_FN_LONG_PTR_LONG): Likewise.
16919 (BT_FN_ULONG_PTR_ULONG): Likewise.
16920 (BT_FN_VOID_UINT64_UINT64): Likewise.
16921 (BT_FN_UINT8_UINT8_UINT8): Likewise.
16922 (BT_FN_INT8_INT8_INT8): Likewise.
16923 (BT_FN_INT16_INT16_INT16): Likewise.
16924 (BT_FN_INT_INT_INT): Likewise.
16925 (BT_FN_UINT_FLOAT_UINT): Likewise.
16926 (BT_FN_FLOAT_UINT_UINT): Likewise.
16927 (BT_FN_ULONG_UINT_UINT): Likewise.
16928 (BT_FN_ULONG_UINT_PTR): Likewise.
16929 (BT_FN_ULONG_ULONG_ULONG): Likewise.
16930 (BT_FN_UINT_UINT_UINT): Likewise.
16931 (BT_FN_VOID_UINT_PTR): Likewise.
16932 (BT_FN_UINT_UINT_PTR: Likewise.
16933 (BT_FN_UINT32_UINT64_PTR): Likewise.
16934 (BT_FN_INT_INT_UINT_UINT): Likewise.
16935 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
16936 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
16937 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
16938 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
16939 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
16940 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
16941 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
16942 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
16943 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
16944 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
16945 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
16946 * doc/frontends.texi: List BRIG FE.
16947 * doc/install.texi (Testing): Add BRIG tesring requirements.
16948 * doc/invoke.texi (Overall Options): Mention BRIG.
16949 * doc/standards.texi (Standards): Doucment BRIG HSA version.
16951 2017-01-24 Richard Biener <rguenther@suse.de>
16953 PR translation/79208
16954 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
16956 2017-01-24 Martin Jambor <mjambor@suse.cz>
16959 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
16960 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
16961 and known_contexts.
16963 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
16965 PR middle-end/79123
16966 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
16967 casts from signed to unsigned really don't have a range.
16969 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
16971 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
16972 GMP_RNDx for compatiblity.
16974 2017-01-24 Martin Liska <mliska@suse.cz>
16977 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
16978 that would prevent us to call alloca with -1 as argument.
16980 2017-01-24 Jakub Jelinek <jakub@redhat.com>
16982 * dwarf2out.c (output_compilation_unit_header, output_file_names):
16983 Avoid -Wformat-security warning.
16985 2017-01-23 Andrew Pinski <apinski@cavium.com>
16987 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
16990 2017-01-23 Martin Sebor <msebor@redhat.com>
16992 PR middle-end/78703
16993 * gimple-ssa-sprintf.c (warn_level): New global.
16994 (format_integer): Use it here and throughout the rest of the file.
16995 Use the same switch to compute sign as base.
16996 (maybe_warn): New function.
16997 (format_directive): Factor out warnings into maybe_warn.
16998 Add debugging output. Use warn_level.
16999 (add_bytes): Use warn_level.
17000 (pass_sprintf_length::compute_format_length): Add debugging output.
17001 (try_substitute_return_value): Same.
17002 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
17004 PR middle-end/78703
17005 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
17006 (struct fmtresult, format_integer, format_floating): Adjust.
17007 (fmtresult::fmtresult): Set max correctly in two argument ctor.
17008 (get_string_length, format_string,format_directive): Same.
17009 (pass_sprintf_length::compute_format_length): Same.
17010 (try_substitute_return_value): Simplify slightly.
17012 PR middle-end/78703
17013 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
17014 (fmtresult::operator+=): Outlined.
17015 (struct fmtresult): Add ctors.
17016 (struct conversion_spec): Rename...
17017 (struct directive): ...to this. Add and remove data members.
17018 (directive::set_width, directive::set_precision): New functions.
17019 (format_percent): Use fmtresult ctor.
17020 (get_width_and_precision): Remove.
17021 (format_integer): Make naming changes. Avoid computing width and
17023 (format_floating): Same. Adjust indentation.
17024 (format_character, format_none): New functions.
17025 (format_string): Moved character handling to format_character.
17026 (format_directive): Remove arguments, change return type.
17027 (parse_directive): New function.
17028 (pass_sprintf_length::compute_format_length): Move directive
17029 parsing to parse_directive.
17031 2017-01-23 Jakub Jelinek <jakub@redhat.com>
17033 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
17034 (assign_assembler_name_if_needed): ... this.
17035 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
17036 (assign_assembler_name_if_needed): ... this.
17037 (free_lang_data_in_cgraph): Adjust callers.
17038 * cgraphunit.c (cgraph_node::analyze): Likewise.
17039 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
17042 2017-01-23 Richard Biener <rguenther@suse.de>
17044 PR tree-optimization/79088
17045 PR tree-optimization/79188
17046 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
17047 resetting loop bounds after last path deletion. Reset loop
17048 bounds of the target loop, make code match the comments.
17049 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
17050 Make sure loops need no fixups.
17052 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
17054 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
17055 exponent support with double type for first argument.
17056 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
17057 type returned by __builtin_vec_extract_sig,
17058 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
17059 functions from "vector int" to "vector unsigned int" or from
17060 "vector long long int" to "vector unsigned long long int".
17061 Changed type returned by __builtin_vec_extract_exp,
17062 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
17063 functions from "vector int" to "vector unsigned int" or from
17064 "vector long long int" to "vector unsigned long long int".
17065 Changed return type of __builtin_vec_test_data_class,
17066 __builtin_vec_test_data_class_sp, and
17067 __builtin_vec_test_data_class_dp from "vector int" to
17068 "vector bool int" or from "vector long long int" to "vector bool
17069 long long int" and changed second argument type from "unsigned
17070 int" to "int". Added new overloaded function forms "vector float
17071 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
17072 "vector float __builtin_vec_insert_exp_sp (vector float, vector
17073 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
17074 double, vector unsigned long long int)" and "vector double
17075 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
17076 long int)". Changed return type of
17077 __builtin_scalar_test_data_class and
17078 __builtin_scalar_test_data_class_sp and
17079 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
17080 int" and changed second argument from "unsigned int" to "int".
17081 Changed type returned by __builtin_scalar_test_neg,
17082 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
17083 from "int" to "bool int". Added new overloaded function form
17084 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
17085 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
17086 exponent double-precision with floating point first argument.
17087 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
17088 documentation of scalar_test_data_class, scalar_test_neg,
17089 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
17090 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
17091 vec_test_data_class built-in functions to reflect refinements in
17092 their type signatures.
17094 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
17096 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
17098 (aarch64_elf_asm_destructor): Likewise.
17100 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
17102 PR rtl-optimization/78634
17103 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
17104 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
17105 * ifcvt.c (noce_try_cmove): Add missing cost check.
17107 PR rtl-optimization/71724
17108 * combine.c (if_then_else_cond): Look for situations where it is
17109 beneficial to undo the work of one of the recursive calls.
17111 2017-01-23 Bin Cheng <bin.cheng@arm.com>
17113 PR tree-optimization/70754
17114 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
17115 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
17116 combined stmt before it if not NULL.
17117 (combine_chains): Process refs reversely and compute dominance point
17120 2017-01-23 Martin Liska <mliska@suse.cz>
17122 PR tree-optimization/79196
17123 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
17124 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
17126 (strlen_optimize_stmt): Call the renamed function.
17128 2017-01-23 Michael Matz <matz@suse.de>
17130 PR tree-optimization/78384
17131 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
17133 2017-01-23 Richard Biener <rguenther@suse.de>
17135 PR tree-optimization/79186
17136 * tree-vrp.c (register_new_assert_for): Make sure we've seen
17137 both incoming edges before moving an assert.
17139 2017-01-23 Martin Jambor <mjambor@suse.cz>
17141 * ipa-prop.c (load_from_param_1): Removed.
17142 (load_from_unmodified_param): Bits from load_from_param_1 put back
17144 (load_from_param): Removed.
17145 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
17146 with stmt. Reverted back to use of load_from_unmodified_param.
17148 2017-01-23 Martin Jambor <mjambor@suse.cz>
17151 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
17152 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
17153 field a pointer to garbage collected vector, mark lattices and
17154 ipcp_orig_node with GTY((skip)).
17155 (ipa_get_param_count): Adjust to descriptors being a pointer.
17156 (ipa_get_param): Likewise.
17157 (ipa_get_type): Likewise.
17158 (ipa_get_param_move_cost): Likewise.
17159 (ipa_set_param_used): Likewise.
17160 (ipa_get_controlled_uses): Likewise.
17161 (ipa_set_controlled_uses): Likewise.
17162 (ipa_is_param_used): Likewise.
17163 (ipa_node_params_t): Move into garbage collector. New methods insert
17165 (ipa_node_params_sum): Annotate wth GTY(()).
17166 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
17168 (ipa_load_from_parm_agg): Adjust declaration.
17169 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
17170 * ipa-profile.c (ipa_profile): Likewise.
17171 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
17172 (ipa_populate_param_decls): Make descriptors parameter garbage
17174 (ipa_dump_param): Adjust to descriptors being a pointer.
17175 (ipa_alloc_node_params): Likewise.
17176 (ipa_initialize_node_params): Likewise.
17177 (load_from_param_1): Make descriptors parameter garbage collected.
17178 (load_from_unmodified_param): Likewise.
17179 (load_from_param): Likewise.
17180 (ipa_load_from_parm_agg): Likewise.
17181 (ipa_node_params::~ipa_node_params): Removed.
17182 (ipa_free_all_node_params): Remove call to delete operator.
17183 (ipa_node_params_t::insert): New.
17184 (ipa_node_params_t::remove): Likewise.
17185 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
17186 copy known_csts and known_contexts vectors.
17187 (ipa_read_node_info): Adjust to descriptors being a pointer.
17188 (ipcp_modif_dom_walker): Make m_descriptors field garbage
17190 (ipcp_transform_function): Make descriptors variable garbage
17193 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
17195 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
17196 * config/i386/avx512dqintrin.h: Ditto.
17197 * config/i386/avx512fintrin.h: Ditto.
17198 * config/i386/i386.c: Handle new builtins.
17199 * config/i386/i386-builtin.def: Add new builtins.
17200 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
17201 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
17203 2017-01-23 Jakub Jelinek <jakub@redhat.com>
17204 Martin Liska <mliska@suse.cz>
17206 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
17207 * asan.c (asan_expand_poison_ifn): Support stores and use
17208 appropriate ASAN report function.
17209 * internal-fn.c (expand_ASAN_POISON_USE): New function.
17210 * internal-fn.def (ASAN_POISON_USE): Declare.
17211 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
17212 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
17213 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
17214 ASAN_POISON calls w/o LHS.
17215 * tree-ssa.c (execute_update_addresses_taken): Create clobber
17216 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
17217 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
17218 * gimplify.c (asan_poison_variables): Add attribute
17219 use_after_scope_memory to variables that really needs to live
17221 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
17222 having the attribute.
17224 2017-01-23 Martin Liska <mliska@suse.cz>
17226 * asan.c (create_asan_shadow_var): New function.
17227 (asan_expand_poison_ifn): Likewise.
17228 * asan.h (asan_expand_poison_ifn): New declaration.
17229 * internal-fn.c (expand_ASAN_POISON): Likewise.
17230 * internal-fn.def (ASAN_POISON): New builtin.
17231 * sanopt.c (pass_sanopt::execute): Expand
17232 asan_expand_poison_ifn.
17233 * tree-inline.c (copy_decl_for_dup_finish): Make function
17235 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
17236 * tree-ssa.c (is_asan_mark_p): New function.
17237 (execute_update_addresses_taken): Rewrite local variables
17238 (identified just by use-after-scope as addressable) into SSA.
17240 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
17242 * doc/install.texi (Specific): opensource.apple.com uses https
17243 now. Remove trailing slash.
17245 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
17247 * README.Portability: Remove note on an Irix compatibility issue.
17249 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
17251 * gcov.c (INCLUDE_ALGORITHM): Define.
17252 (INCLUDE_VECTOR): Define.
17253 No longer include <vector> and <algorithm> directly.
17255 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
17257 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
17259 * doc/invoke.texi (Code Gen Options): Ditto.
17261 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
17264 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
17266 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
17268 rtl-optimization/79125
17269 * cprop.c (local_cprop_pass): Handle cases where we make an
17270 unconditional trap.
17272 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
17276 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
17277 read from, for big endian.
17279 2017-01-20 Jiong Wang <jiong.wang@arm.com>
17281 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
17282 register pauth builtins for LP64 only.
17284 2017-01-20 Marek Polacek <polacek@redhat.com>
17287 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
17290 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
17292 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
17294 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
17295 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
17296 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
17298 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17301 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
17302 in big-endian mode when they are not a single duplicated value.
17304 2017-01-20 Richard Biener <rguenther@suse.de>
17306 * BASE-VER: Bump to 7.0.1.
17308 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
17310 * omp-low.c (omplow_simd_context): New struct. Use it...
17311 (lower_rec_simd_input_clauses): ...here and...
17312 (lower_rec_input_clauses): ...here to hold common data. Adjust all
17313 references to idx, lane, max_vf, is_simt.
17315 2017-01-20 Graham Markall <graham.markall@embecosm.com>
17317 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
17320 2017-01-20 Martin Jambor <mjambor@suse.cz>
17322 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
17323 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
17325 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
17326 (GTFILES): Rename hsa.c to hsa-common.c.
17327 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
17328 * hsa-dump.c: Likewise.
17329 * hsa-gen.c: Likewise.
17330 * hsa-regalloc.c: Likewise.
17331 * ipa-hsa.c: Likewise.
17332 * omp-expand.c: Likewise.
17333 * omp-low.c: Likewise.
17334 * toplev.c: Likewise.
17336 2017-01-20 Marek Polacek <polacek@redhat.com>
17339 * doc/invoke.texi: Document -Wduplicated-branches.
17340 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
17341 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
17342 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
17343 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
17344 return 0 only when not OEP_LEXICOGRAPHIC.
17345 (fold_build_cleanup_point_expr): Use the expression
17346 location when building CLEANUP_POINT_EXPR.
17347 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
17348 * tree.c (add_expr): Handle error_mark_node.
17350 2017-01-20 Martin Liska <mliska@suse.cz>
17353 * tree-profile.c (init_ic_make_global_vars): Do not call
17355 (gimple_init_gcov_profiler): Likewise.
17357 2017-01-20 Martin Liska <mliska@suse.cz>
17360 * cgraph.h (maybe_create_reference): Remove argument and
17362 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
17364 * ipa-cp.c (create_specialized_node): Likewise.
17365 * symtab.c (symtab_node::maybe_create_reference): Handle
17366 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
17368 2017-01-20 Martin Liska <mliska@suse.cz>
17370 * read-rtl-function.c (function_reader::create_function): Use
17371 build_decl instread of build_decl_stat.
17373 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
17375 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
17376 * config/i386/avx512dqintrin.h: Ditto.
17377 * config/i386/avx512fintrin.h: Ditto.
17378 * config/i386/i386-builtin-types.def: Add new types.
17379 * config/i386/i386.c: Handle new types.
17380 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
17381 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
17382 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
17383 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
17384 (__builtin_ia32_kshiftridi): New.
17385 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
17387 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
17391 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
17392 define to rs6000_init_stack_protect_guard.
17393 (rs6000_init_stack_protect_guard): New function.
17395 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
17396 Yunqiang Su <yunqiang.su@imgtec.com>
17398 * config.gcc (supported_defaults): Add madd4.
17399 (with_madd4): Add validation.
17400 (all_defaults): Add madd4.
17401 * config/mips/mips.opt (mmadd4): New option.
17402 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17404 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17406 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
17407 (ISA_HAS_FUSED_MADD4): Likewise.
17408 * doc/invoke.texi (-mmadd4): Document the new option.
17409 * doc/install.texi (--with-madd4): Document the new option.
17411 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17413 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
17414 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
17415 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
17416 (aarch64_init_pauth_hint_builtins): New.
17417 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
17418 (aarch64_expand_builtin): Expand new builtins.
17420 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17422 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
17423 * combine-stack-adj.c (no_unhandled_cfa): Handle
17424 REG_CFA_TOGGLE_RA_MANGLE.
17425 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
17426 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
17427 info for return address signing.
17428 (aarch64_expand_epilogue): Likewise.
17430 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17432 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
17433 * config/aarch64/aarch64-protos.h
17434 (aarch64_return_address_signing_enabled): New declaration.
17435 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
17437 (aarch64_expand_prologue): Sign return address before it's pushed onto
17439 (aarch64_expand_epilogue): Authenticate return address fetched from
17441 (aarch64_override_options): Sanity check for ILP32 and ISA level.
17442 (aarch64_attributes): New function attributes for "sign-return-address".
17443 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
17444 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
17445 ("*do_return"): Generate combined instructions according to key index.
17446 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
17447 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
17449 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
17450 * config/aarch64/aarch64.opt (msign-return-address=): New.
17451 * doc/extend.texi (AArch64 Function Attributes): Documents
17452 "sign-return-address=".
17453 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
17455 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
17457 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
17458 overall option summary.
17460 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17462 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
17463 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
17464 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
17465 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
17467 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
17469 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
17470 -mpower9-minmax by default for -mcpu=power9.
17471 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
17472 128-bit floating point.
17474 2017-01-20 Alan Modra <amodra@gmail.com>
17476 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
17477 optimizing for size.
17479 2017-01-20 Alan Modra <amodra@gmail.com>
17482 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
17483 for strcmp and strncmp from corresponding builtin decl.
17485 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17487 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
17488 instead of i386/rtems-64.h.
17489 * config/i386/rtems-64.h: Remove.
17491 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17495 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
17497 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
17499 2017-01-19 Tamar Christina <tamar.christina@arm.com>
17501 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
17502 Change int to HOST_WIDE_INT.
17503 * config/aarch64/aarch64-protos.h
17504 (aarch64_simd_gen_const_vector_dup): Likewise.
17505 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
17507 2017-01-19 David Malcolm <dmalcolm@redhat.com>
17509 * langhooks-def.h (lhd_type_for_size): New decl.
17510 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
17511 * langhooks.c (lhd_type_for_size): New function, taken from
17514 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
17516 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
17517 define_bypass for CR latency.
17518 (power9-cracked-alu): Update bypass latency and remove power9-branch.
17519 (power9-alu2): Add define_bypass for CR latency.
17521 (power9-mul): Update insn latency.
17522 (power9-mul-compare): Update insn latency, bypass latency and remove
17525 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17527 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
17529 * config/aarch64/aarch64.md
17530 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
17531 aarch64_nopcrelative_literal_loads.
17532 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
17534 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
17536 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
17537 TARGET_LOONGSON_3A.
17538 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
17540 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
17543 * config.gcc (supported_defaults): Add lxc1-sxc1.
17544 (with_lxc1_sxc1): Add validation.
17545 (all_defaults): Add lxc1-sxc1.
17546 * config/mips/mips.opt (mlxc1-sxc1): New option.
17547 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17549 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17550 __mips_no_lxc1_sxc1.
17551 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
17552 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
17553 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
17555 2017-01-19 Richard Biener <rguenther@suse.de>
17557 PR tree-optimization/72488
17558 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
17559 sure to restore SSA info.
17560 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
17562 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
17564 PR rtl-optimization/79121
17565 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
17566 of the inner type when shifting an extended value.
17568 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
17571 * symtab.c (symtab_node::equal_address_to): Fix comparing of
17572 interposable aliases.
17574 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
17577 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
17578 Use the evmergelohi instruction.
17579 (mov_si<mode>_e500_subreg4_2_le): Likewise.
17580 (mov_sitf_e500_subreg8_2_be): Likewise.
17581 (mov_sitf_e500_subreg12_2_le): Likewise.
17582 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
17583 (mov_si<mode>_e500_subreg4_2_be): Likewise.
17584 (mov_sitf_e500_subreg8_2_le): Likewise.
17585 (mov_sitf_e500_subreg12_2_be): Likewise.
17587 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17589 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
17590 attribute from vecsimple to vecperm.
17591 (altivec_vbpermq2): Likewise.
17593 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17596 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
17598 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17599 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
17600 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
17601 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
17602 case where N arg is SIZE_MAX.
17603 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
17604 (cmpstrsi): Add pattern.
17606 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17608 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17609 __builtin_vec_revb builtins.
17610 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
17611 built-in functions to support generation of the ISA 3.0 XXBR<x>
17612 vector byte reverse instructions.
17613 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
17614 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
17615 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
17616 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
17617 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
17618 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
17619 (P9V_BUILTIN_VEC_REVB): Likewise.
17620 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
17621 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
17622 (p9_xxbrq_v16qi): Likewise.
17623 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
17624 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
17625 (p9_xxbrh_v8hi): Likewise.
17626 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
17627 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
17628 vec_revb built-in functions.
17630 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
17632 PR rtl-optimization/78952
17633 * config/i386/i386.md (any_extract): New code iterator.
17634 (*insvqi_2): Use any_extract for source operand.
17635 (*insvqi_3): Use any_shiftrt for source operand.
17637 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
17639 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
17641 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
17643 2017-01-18 Matthias Klose <doko@ubuntu.com>
17645 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
17647 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17649 * config/rs6000/altivec.h (vec_bperm): Change #define.
17650 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
17651 (altivec_vbpermq2): New define_insn.
17652 (altivec_vbpermd): Likewise.
17653 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
17654 function interface.
17655 (VBPERMD): Likewise.
17656 (VBPERM): New polymorphic function interface.
17657 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
17658 Add entries for P9V_BUILTIN_VEC_VBPERM.
17659 * doc/extend.texi: Add interfaces for vec_bperm.
17661 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17663 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
17664 first letter of error messages.
17665 (s390_resolve_overloaded_builtin): Likewise.
17666 * config/s390/s390.c (s390_expand_builtin): Likewise.
17667 (s390_invalid_arg_for_unprototyped_fn): Likewise.
17668 (s390_valid_target_attribute_inner_p): Likewise.
17669 * config/s390/s390.md ("tabort"): Likewise.
17671 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
17673 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
17674 (ISA_AVOID_DIV_HILO): New macro.
17675 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
17676 (ISA_HAS_DDIV): Likewise.
17678 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17680 * doc/invoke.texi (fabi-version): Correct number of occurrences.
17682 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17684 * doc/invoke.texi (fabi-version): Spelling fix.
17686 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17689 * doc/invoke.texi (fabi-version): Mention mangling fix for
17692 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
17695 * doc/invoke.texi (fabi-version): Document discriminator mangling.
17697 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
17700 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
17701 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
17703 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
17705 (stack_protect_test): Ditto.
17706 * config/rs6000/rs6000.opt (mstack-protector-guard=,
17707 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
17709 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
17710 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
17711 -mstack-protector-guard-offset=.
17712 (RS/6000 and PowerPC Options): Ditto.
17714 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
17716 * config/i386/i386.h (MASK_CLASS_P): New define.
17717 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
17718 there are no registers from different register sets also when
17719 mask registers are used. Update function comment.
17720 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
17721 to (*k/*r) and (*k/*km) alternatives.
17723 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
17725 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
17726 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
17727 (EH_RETURN_HANDLER_RTX): New define.
17728 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
17729 Force frame pointer in EH return functions.
17730 (aarch64_expand_epilogue): Add barrier for eh_return.
17731 (aarch64_final_eh_return_addr): Remove.
17732 (aarch64_eh_return_handler_rtx): New function.
17733 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
17735 (aarch64_eh_return_handler_rtx): New prototype.
17737 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17739 * config/rs6000/altivec.h (vec_rlmi): New #define.
17740 (vec_vrlnm): Likewise.
17741 (vec_rlnm): Likewise.
17742 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
17743 (UNSPEC_VRLNM): Likewise.
17744 (VIlong): New mode iterator.
17745 (altivec_vrl<VI_char>mi): New define_insn.
17746 (altivec_vrl<VI_char>nm): Likewise.
17747 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
17749 (VRLDNM): Likewise.
17750 (RLNM): New polymorphic function entry.
17751 (VRLWMI): New monomorphic function entry.
17752 (VRLDMI): Likewise.
17753 (RLMI): New polymorphic function entry.
17754 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
17755 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
17756 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
17759 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17762 * dwarf2out.c (field_byte_offset): Restore the
17763 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
17764 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
17765 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
17768 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17771 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
17773 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17776 * dwarf2out.c (add_data_member_location_attribute): For constant
17777 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
17778 instead of DW_AT_data_member_location, DW_AT_bit_offset and
17779 DW_AT_byte_size attributes.
17781 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
17783 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
17784 after forcing to constant memory when the code model is medium.
17786 2017-01-17 Julia Koval <julia.koval@intel.com>
17789 * config/i386/avx512fintrin.h
17790 (_mm512_i32gather_ps): Change __addr type to void const*.
17791 (_mm512_mask_i32gather_ps): Ditto.
17792 (_mm512_i32gather_pd): Ditto.
17793 (_mm512_mask_i32gather_pd): Ditto.
17794 (_mm512_i64gather_ps): Ditto.
17795 (_mm512_mask_i64gather_ps): Ditto.
17796 (_mm512_i64gather_pd): Ditto.
17797 (_mm512_mask_i64gather_pd): Ditto.
17798 (_mm512_i32gather_epi32): Ditto.
17799 (_mm512_mask_i32gather_epi32): Ditto.
17800 (_mm512_i32gather_epi64): Ditto.
17801 (_mm512_mask_i32gather_epi64): Ditto.
17802 (_mm512_i64gather_epi32): Ditto.
17803 (_mm512_mask_i64gather_epi32): Ditto.
17804 (_mm512_i64gather_epi64): Ditto.
17805 (_mm512_mask_i64gather_epi64): Ditto.
17806 (_mm512_i32scatter_ps): Change __addr type to void*.
17807 (_mm512_mask_i32scatter_ps): Ditto.
17808 (_mm512_i32scatter_pd): Ditto.
17809 (_mm512_mask_i32scatter_pd): Ditto.
17810 (_mm512_i64scatter_ps): Ditto.
17811 (_mm512_mask_i64scatter_ps): Ditto.
17812 (_mm512_i64scatter_pd): Ditto.
17813 (_mm512_mask_i64scatter_pd): Ditto.
17814 (_mm512_i32scatter_epi32): Ditto.
17815 (_mm512_mask_i32scatter_epi32): Ditto.
17816 (_mm512_i32scatter_epi64): Ditto.
17817 (_mm512_mask_i32scatter_epi64): Ditto.
17818 (_mm512_i64scatter_epi32): Ditto.
17819 (_mm512_mask_i64scatter_epi32): Ditto.
17820 (_mm512_i64scatter_epi64): Ditto.
17821 (_mm512_mask_i64scatter_epi64): Ditto.
17822 * config/i386/avx512pfintrin.h
17823 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
17824 (_mm512_mask_prefetch_i32gather_ps): Ditto.
17825 (_mm512_mask_prefetch_i64gather_pd): Ditto.
17826 (_mm512_mask_prefetch_i64gather_ps): Ditto.
17827 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
17828 (_mm512_prefetch_i32scatter_ps): Ditto.
17829 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
17830 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
17831 (_mm512_prefetch_i64scatter_pd): Ditto.
17832 (_mm512_prefetch_i64scatter_ps): Ditto.
17833 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
17834 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
17835 * config/i386/avx512vlintrin.h
17836 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
17837 (_mm_mmask_i32gather_ps): Ditto.
17838 (_mm256_mmask_i32gather_pd): Ditto.
17839 (_mm_mmask_i32gather_pd): Ditto.
17840 (_mm256_mmask_i64gather_ps): Ditto.
17841 (_mm_mmask_i64gather_ps): Ditto.
17842 (_mm256_mmask_i64gather_pd): Ditto.
17843 (_mm_mmask_i64gather_pd): Ditto.
17844 (_mm256_mmask_i32gather_epi32): Ditto.
17845 (_mm_mmask_i32gather_epi32): Ditto.
17846 (_mm256_mmask_i32gather_epi64): Ditto.
17847 (_mm_mmask_i32gather_epi64): Ditto.
17848 (_mm256_mmask_i64gather_epi32): Ditto.
17849 (_mm_mmask_i64gather_epi32): Ditto.
17850 (_mm256_mmask_i64gather_epi64): Ditto.
17851 (_mm_mmask_i64gather_epi64): Ditto.
17852 (_mm256_i32scatter_ps): Change __addr type to void*.
17853 (_mm256_mask_i32scatter_ps): Ditto.
17854 (_mm_i32scatter_ps): Ditto.
17855 (_mm_mask_i32scatter_ps): Ditto.
17856 (_mm256_i32scatter_pd): Ditto.
17857 (_mm256_mask_i32scatter_pd): Ditto.
17858 (_mm_i32scatter_pd): Ditto.
17859 (_mm_mask_i32scatter_pd): Ditto.
17860 (_mm256_i64scatter_ps): Ditto.
17861 (_mm256_mask_i64scatter_ps): Ditto.
17862 (_mm_i64scatter_ps): Ditto.
17863 (_mm_mask_i64scatter_ps): Ditto.
17864 (_mm256_i64scatter_pd): Ditto.
17865 (_mm256_mask_i64scatter_pd): Ditto.
17866 (_mm_i64scatter_pd): Ditto.
17867 (_mm_mask_i64scatter_pd): Ditto.
17868 (_mm256_i32scatter_epi32): Ditto.
17869 (_mm256_mask_i32scatter_epi32): Ditto.
17870 (_mm_i32scatter_epi32): Ditto.
17871 (_mm_mask_i32scatter_epi32): Ditto.
17872 (_mm256_i32scatter_epi64): Ditto.
17873 (_mm256_mask_i32scatter_epi64): Ditto.
17874 (_mm_i32scatter_epi64): Ditto.
17875 (_mm_mask_i32scatter_epi64): Ditto.
17876 (_mm256_i64scatter_epi32): Ditto.
17877 (_mm256_mask_i64scatter_epi32): Ditto.
17878 (_mm_i64scatter_epi32): Ditto.
17879 (_mm_mask_i64scatter_epi32): Ditto.
17880 (_mm256_i64scatter_epi64): Ditto.
17881 (_mm256_mask_i64scatter_epi64): Ditto.
17882 (_mm_i64scatter_epi64): Ditto.
17883 (_mm_mask_i64scatter_epi64): Ditto.
17884 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
17885 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
17886 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
17887 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
17888 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
17889 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
17890 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
17891 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
17892 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
17893 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
17894 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
17895 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
17896 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
17897 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
17898 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
17899 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
17900 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
17901 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
17902 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
17903 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
17904 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
17905 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
17906 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
17907 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
17908 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
17909 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
17910 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
17911 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
17912 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
17913 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
17914 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
17915 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
17916 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
17917 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
17918 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
17919 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
17920 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
17921 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
17922 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
17923 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
17924 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
17925 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
17926 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
17927 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
17928 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
17929 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
17930 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
17931 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
17932 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
17933 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
17934 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
17935 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
17936 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
17937 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
17938 definitions accordingly.
17940 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
17941 Kuan-Lin Chen <kuanlinchentw@gmail.com>
17944 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
17947 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
17950 * ira-conflicts.c (ira_build_conflicts): Update total conflict
17951 hard regs for inner regno.
17953 2017-01-17 Martin Liska <mliska@suse.cz>
17956 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
17957 assumption and add comment.
17959 2017-01-17 Nathan Sidwell <nathan@acm.org>
17961 * ipa-visibility.c (localize_node): New function, broken out of ...
17962 (function_and_variable_visibility): ... here. Call it.
17964 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
17966 PR middle-end/77445
17967 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
17968 correctly set frequency of oudgoing edge.
17969 (duplicate_thread_path): Fix profile updating.
17971 2017-01-17 Jakub Jelinek <jakub@redhat.com>
17974 * configure.ac: Add GCC_BASE_VER.
17975 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
17976 version from BASE-VER file.
17977 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
17978 (gcc.o): Depend on $(BASEVER).
17979 * common.opt (dumpfullversion): New option.
17980 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
17981 * doc/invoke.texi: Document -dumpfullversion.
17982 * doc/install.texi: Document --with-gcc-major-version-only.
17983 * configure: Regenerated.
17985 2017-01-17 Richard Biener <rguenther@suse.de>
17987 PR tree-optimization/71433
17988 * tree-vrp.c (register_new_assert_for): Merge same asserts
17989 on all incoming edges.
17990 (process_assert_insertions_for): Handle insertions at the
17993 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
17995 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
17996 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
17998 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
18001 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
18004 2017-01-17 Alan Modra <amodra@gmail.com>
18007 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
18008 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
18009 symbolic stack limit when pic.
18011 2017-01-16 Martin Sebor <msebor@redhat.com>
18013 PR tree-optimization/78608
18014 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
18016 2017-01-16 Jeff Law <law@redhat.com>
18019 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
18020 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
18021 for several include directories that may be relative to sysroot.
18022 * config/i386/x-mingw32 (gplus_includedir): Define.
18023 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
18024 (native_system_includedir): Likewise.
18025 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
18026 override if TARGET_SYSTEM_ROOT is defined.
18027 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
18029 PR tree-optimization/79090
18030 PR tree-optimization/33562
18031 PR tree-optimization/61912
18032 PR tree-optimization/77485
18033 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
18034 and computed trims into the dump file.
18036 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
18038 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
18040 2017-01-16 Jakub Jelinek <jakub@redhat.com>
18043 * gimplify.c (gimplify_init_constructor): If want_value and
18044 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
18048 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
18049 sequence. Formatting fixes.
18050 (doloop_optimize): Formatting fixes.
18053 * gcc.c (debug_level_greater_than_spec_func): New function.
18054 (static_spec_functions): Add debug-level-gt spec function.
18055 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
18057 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
18058 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
18059 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
18060 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
18061 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
18062 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
18064 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
18066 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
18067 QImode fixups to general and mask registers only.
18069 2017-01-16 Carl Love <cel@us.ibm.com>
18071 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
18072 for built-in functions
18073 vector signed char vec_nabs (vector signed char)
18074 vector signed short vec_nabs (vector signed short)
18075 vector signed int vec_nabs (vector signed int)
18076 vector signed long long vec_nabs (vector signed long long)
18077 vector float vec_nabs (vector float)
18078 vector double vec_nabs (vector double)
18079 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
18081 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
18082 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
18083 * doc/extend.texi: Update the documentation file for the new built-in
18086 2017-01-16 Martin Sebor <msebor@redhat.com>
18088 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
18091 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18093 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
18094 UNSPEC_VSX__XXSPLTD to require special splat handling.
18096 2017-01-16 David Malcolm <dmalcolm@redhat.com>
18099 * system.h: Poison strndup.
18101 2017-01-16 Alan Modra <amodra@gmail.com>
18104 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
18107 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
18109 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
18111 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
18113 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
18114 call recog here. Assert that INSN_CODE (insn) is non-negative.
18116 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
18119 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
18121 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
18122 in the currently scheduled RTL fragment.
18124 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
18126 PR rtl-optimization/78751
18127 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
18130 2017-01-14 Jeff Law <law@redhat.com>
18132 PR tree-optimization/79090
18133 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
18134 variable length stores.
18135 (compute_trims): Delete dead assignment to *trim_tail.
18136 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
18139 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
18141 PR rtl-optimization/78626
18142 PR rtl-optimization/78727
18143 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
18144 of a block, and split such blocks after everything else is finished.
18146 2017-01-14 Alan Modra <amodra@gmail.com>
18149 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
18150 target legitimate_combined_insn.
18151 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
18152 (rs6000_legitimate_combined_insn): New function.
18153 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
18155 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
18156 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
18157 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
18159 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
18161 * doc/frontends.texi (G++ and GCC): Remove references to Java.
18163 2017-01-13 Jeff Law <law@redhat.com>
18165 PR tree-optimization/33562
18166 PR tree-optimization/61912
18167 PR tree-optimization/77485
18168 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
18170 (delete_dead_assignment): Likewise.
18171 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
18172 statement to delete_dead_call and delete_dead_assignment.
18174 2017-01-13 David Malcolm <dmalcolm@redhat.com>
18177 * substring-locations.c (format_warning_va): Strengthen case 1 so
18178 that both endpoints of the substring must be within the format
18179 range for just the substring to be printed.
18181 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
18183 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
18184 * config/i386/i386.c (ix86_target_string): Add missing options
18185 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
18186 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
18187 flags_other and ix86_target_other to flags2_other. Display unknown
18189 (ix86_valid_target_attribute_inner_p): Add missing options and
18190 reorder options by implied ISAs, as in ix86_target_string.
18192 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
18194 * hash-table.h (hash_table::too_empty_p): New function.
18195 (hash_table::expand): Use it.
18196 (hash_table::traverse): Likewise.
18197 (hash_table::empty_slot): Use sizeof (value_type) instead of
18198 sizeof (PTR) to convert bytes to elements. Shrink the table
18199 if the current size is excessive for the current number of
18202 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
18204 * ira-costs.c (record_reg_classes): Break from the inner loop
18205 early once alt_fail is known to be true. Update outer loop
18206 handling accordingly.
18208 2017-01-13 Jeff Law <law@redhat.com>
18210 * tree-ssa-dse.c (decrement_count): New function.
18211 (increment_start_addr, maybe_trim_memstar_call): Likewise.
18212 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
18213 when we know the partially dead statement is a mem* function.
18215 PR tree-optimization/61912
18216 PR tree-optimization/77485
18217 * tree-ssa-dse.c: Include expr.h.
18218 (maybe_trim_constructor_store): New function.
18219 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
18221 PR tree-optimization/33562
18222 PR tree-optimization/61912
18223 PR tree-optimization/77485
18224 * doc/invoke.texi: Document new dse-max-object-size param.
18225 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
18226 * tree-ssa-dse.c: Include params.h.
18227 (dse_store_status): New enum.
18228 (initialize_ao_ref_for_dse): New, partially extracted from
18230 (valid_ao_ref_for_dse, normalize_ref): New.
18231 (setup_live_bytes_from_ref, compute_trims): Likewise.
18232 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
18233 (maybe_trim_partially_dead_store): Likewise.
18234 (maybe_trim_complex_store): Likewise.
18235 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
18236 Track what bytes live from the original store. Return tri-state
18237 for dead, partially dead or live.
18238 (dse_dom_walker): Add constructor, destructor and new private members.
18239 (delete_dead_call, delete_dead_assignment): New extracted from
18241 (dse_optimize_stmt): Make a member of dse_dom_walker.
18242 Use initialize_ao_ref_for_dse.
18244 PR tree-optimization/33562
18245 PR tree-optimization/61912
18246 PR tree-optimization/77485
18247 * sbitmap.h (bitmap_count_bits): Prototype.
18248 (bitmap_clear_range, bitmap_set_range): Likewise.
18249 * sbitmap.c (bitmap_clear_range): New function.
18250 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
18252 2017-01-13 Martin Liska <mliska@suse.cz>
18255 * function.c (set_cfun): Add new argument force.
18256 * function.h (set_cfun): Likewise.
18257 * ipa-inline-transform.c (inline_call): Use the function when
18258 strict alising from is dropped for function we inline to.
18260 2017-01-13 Richard Biener <rguenther@suse.de>
18262 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
18263 for dumping GIMPLE INTEGER_CSTs.
18265 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18267 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
18268 to 201112L since C++17.
18270 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
18273 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
18274 if -fsanitize=kernel-address is present.
18276 2017-01-13 Richard Biener <rguenther@suse.de>
18278 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
18279 as _Literal ( type ) number in case usual suffixes do not
18280 preserve all information.
18282 2017-01-13 Richard Biener <rguenther@suse.de>
18284 PR tree-optimization/77283
18285 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
18286 and ssa-iterators.h.
18287 (is_feasible_trace): Implement a cost model based on joiner
18290 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
18293 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
18294 char or short to __float128/_Float128 directly.
18296 2017-01-12 Martin Sebor <msebor@redhat.com>
18298 to -Wformat-overflow.
18299 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
18300 (min_bytes_remaining): Same.
18301 (get_string_length): Same.
18302 (format_string): Same.
18303 (format_directive): Same.
18305 (pass_sprintf_length::handle_gimple_call): Same.
18307 2017-01-12 Jakub Jelinek <jakub@redhat.com>
18309 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
18310 info.nowrite calls with no lhs that can't throw. Return bool
18311 whether gsi_remove has been called or not.
18312 (pass_sprintf_length::handle_gimple_call): Return bool whether
18313 try_substitute_return_value called gsi_remove. Formatting fix.
18314 (pass_sprintf_length::execute): Don't use gsi_remove if
18315 handle_gimple_call returned true.
18318 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
18319 be removed due to side-effects, don't remove following barrier nor
18320 turn the successor edge into fallthru edge.
18322 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18325 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
18326 element-reversing loads and stores as not swappable.
18328 2017-01-12 Nathan Sidwell <nathan@acm.org>
18329 Nicolai Stange <nicstange@gmail.com>
18331 * combine.c (try_combine): Don't ignore result of overlap checking
18332 loop. Combine overlap & asm check into single loop.
18334 2017-01-12 Richard Biener <rguenther@suse.de>
18336 * tree-pretty-print.c (dump_generic_node): Provide -gimple
18337 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
18339 2017-01-12 Richard Biener <rguenther@suse.de>
18341 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
18342 and TS_TARGET_OPTION directly derive from TS_BASE.
18343 * tree-core.h (tree_optimization_option): Derive from tree_base.
18344 (tree_target_option): Likewise.
18346 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
18348 * config/i386/i386.c (memory_address_length): Increase len
18349 only when rip_relative_addr_p returns false.
18351 2017-01-11 Julia Koval <julia.koval@intel.com>
18353 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
18354 (OPTION_MASK_ISA_SGX_SET): New.
18355 (ix86_handle_option): Handle OPT_msgx.
18356 * config.gcc: Added sgxintrin.h.
18357 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
18358 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
18359 * config/i386/i386.c (ix86_target_string): Add -msgx.
18361 (ix86_option_override_internal): Handle new options.
18362 (ix86_valid_target_attribute_inner_p): Add sgx.
18363 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
18364 * config/i386/i386.opt: Add msgx.
18365 * config/i386/sgxintrin.h: New file.
18366 * config/i386/x86intrin.h: Add sgxintrin.h.
18368 2017-01-11 Jakub Jelinek <jakub@redhat.com>
18371 * fold-const.c (maybe_nonzero_address): Return 1 for function
18373 (tree_single_nonzero_warnv_p): Don't handle function local objects
18377 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
18380 2017-01-11 David Malcolm <dmalcolm@redhat.com>
18383 * opts.c: Include "spellcheck.h"
18384 (struct string_fragment): New struct.
18385 (struct edit_distance_traits<const string_fragment &>): New
18387 (get_closest_sanitizer_option): New function.
18388 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
18390 2017-01-11 Jakub Jelinek <jakub@redhat.com>
18392 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
18394 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
18395 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
18396 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
18397 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
18398 for initial die_offset if dwarf_split_debug_info.
18399 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
18400 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
18402 (output_skeleton_debug_sections): Formatting fix. Use
18403 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
18404 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
18406 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
18408 * config/arm/cortex-a53.md: Add bypasses for
18409 cortex_a53_r2f_cvt.
18410 (cortex_a53_r2f): Only use for transfers.
18411 (cortex_a53_f2r): Likewise.
18412 (cortex_a53_r2f_cvt): Add reservation for conversions.
18413 (cortex_a53_f2r_cvt): Likewise.
18415 2017-01-11 Tamar Christina <tamar.christina@arm.com>
18417 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
18418 to all inlined functions, change static to extern.
18420 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
18423 * config/arm/arm.c (legitimize_pic_address): Handle reference to
18425 (arm_assemble_integer): Likewise.
18427 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
18429 * config.gcc: Use new awk script to check CPU, FPU and architecture
18430 parameters for --with-... options.
18431 * config/arm/parsecpu.awk: New file
18432 * config/arm/arm-cpus.in: New file.
18433 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
18435 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
18437 * config/arm/t-arm: Update dependency rules.
18438 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
18439 of processing .def files.
18440 * config/arm/genopt.sh: Deleted.
18441 * config/arm/gentune.sh: Deleted.
18442 * config/arm/arm-cores.def: Deleted.
18443 * config/arm/arm-arches.def: Deleted.
18444 * config/arm/arm-fpus.def: Deleted.
18445 * config/arm/arm-tune.md: Regenerated.
18446 * config/arm/arm-tables.opt: Regenerated.
18447 * config/arm/arm-cpu.h: New generated file.
18448 * config/arm/arm-cpu-data.h: New generated file.
18449 * config/arm/arm-cpu-cdata.h: New generated file.
18451 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
18454 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
18456 (input_varpool_node): Unpack dynamically_initialized bit.
18458 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
18460 PR rtl-optimization/79032
18461 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
18462 the alignment of the adjusted memory reference against that of MODE,
18463 instead of the alignment of the original memory reference.
18465 2017-01-11 Martin Jambor <mjambor@suse.cz>
18467 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
18469 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
18470 decorated functions.
18472 2017-01-11 Richard Biener <rguenther@suse.de>
18474 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
18475 set range/nonnull info for PHI results. Do not set it on
18476 stmts marked for removal.
18478 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
18480 * expr.c (store_field): In the bitfield case, fetch the return value
18481 from the registers before applying a single big-endian adjustment.
18482 Always do a final load for a BLKmode value not larger than a word.
18484 2017-01-10 David Malcolm <dmalcolm@redhat.com>
18487 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18488 that we correctly handle column numbers greater than
18489 LINE_MAP_MAX_COLUMN_NUMBER.
18491 2017-01-10 Martin Sebor <msebor@redhat.com>
18493 PR middle-end/78245
18494 * gimple-ssa-sprintf.c (get_destination_size): Call
18495 {init,fini}object_sizes.
18496 * tree-object-size.c (addr_object_size): Adjust.
18497 (pass_through_call): Adjust.
18498 (pass_object_sizes::execute): Adjust.
18499 * tree-object-size.h (fini_object_sizes): Declare.
18501 2017-01-10 Martin Sebor <msebor@redhat.com>
18503 PR tree-optimization/78775
18504 * builtins.c (get_size_range): Move...
18505 * calls.c: ...to here.
18506 (alloc_max_size): Accept zero argument.
18507 (operand_signed_p): Remove.
18508 (maybe_warn_alloc_args_overflow): Call get_size_range.
18509 * calls.h (get_size_range): Declare.
18511 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
18513 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
18514 from TI's devices.csv file as of September 2016.
18515 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
18517 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
18519 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
18520 * doc/invoke.texi: Likewise.
18521 * doc/md.texi: Likewise.
18522 * doc/objc.texi: Likewise.
18524 2017-01-10 Joshua Conner <joshconner@google.com>
18526 * config/arm/fuchsia-elf.h: New file.
18527 * config/fuchsia.h: New file.
18528 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
18529 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
18531 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
18533 2016-01-10 Richard Biener <rguenther@suse.de>
18535 PR tree-optimization/79034
18536 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18537 Propagate out degenerate PHIs in the joiner.
18539 2017-01-10 Martin Liska <mliska@suse.cz>
18541 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
18542 (sort_congruence_classes_by_decl_uid): Likewise.
18543 (sort_congruence_class_groups_by_decl_uid): Likewise.
18544 (sem_item_optimizer::merge_classes): Sort class, groups in these
18545 classes and members in the groups by DECL_UID of declarations.
18546 This would make merge operations stable.
18548 2017-01-10 Martin Liska <mliska@suse.cz>
18550 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
18551 usage of m_classes_vec.
18552 (sem_item_optimizer::~sem_item_optimizer): Likewise.
18553 (sem_item_optimizer::get_group_by_hash): Likewise.
18554 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
18555 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
18556 (sem_item_optimizer::verify_classes): Likewise.
18557 (sem_item_optimizer::process_cong_reduction): Likewise.
18558 (sem_item_optimizer::dump_cong_classes): Likewise.
18559 (sem_item_optimizer::merge_classes): Likewise.
18560 * ipa-icf.h (congruence_class_hash): Rename from
18561 congruence_class_group_hash. Remove declaration of m_classes_vec.
18563 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
18565 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
18566 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
18567 * config.gcc: Add avx512vpopcntdqintrin.h.
18568 * config/i386/avx512vpopcntdqintrin.h: New.
18569 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
18570 * config/i386/i386-builtin-types.def: Add new types.
18571 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
18572 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
18573 __builtin_ia32_vpopcountq_v8di_mask): New.
18574 * config/i386/i386-c.c (ix86_target_macros_internal): Define
18575 __AVX512VPOPCNTDQ__.
18576 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
18577 (PTA_AVX512VPOPCNTDQ): Define.
18578 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
18579 TARGET_AVX512VPOPCNTDQ_P): Define.
18580 * config/i386/i386.opt: Add mavx512vpopcntdq.
18581 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
18582 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
18584 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18586 PR middle-end/77484
18587 * predict.def (PRED_CALL): Set to 67.
18589 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
18591 * expr.c (store_field): In the bitfield case, if the value comes from
18592 a function call and is of an aggregate type returned in registers, do
18593 not modify the field mode; extract the value in all cases if the mode
18594 is BLKmode and the size is not larger than a word.
18596 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
18599 * config/i386/cpuid.h: Fix undefined behavior.
18601 2017-01-04 Jeff Law <law@redhat.com>
18603 PR tree-optimization/79007
18604 PR tree-optimization/67955
18605 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
18606 conservative for pt.null when flag_non_call_exceptions is on.
18608 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18610 PR translation/79019
18611 PR translation/79020
18612 * params.def (PARAM_INLINE_MIN_SPEEDUP,
18613 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
18614 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
18616 * config/avr/avr.opt (maccumulate-args): Likewise.
18617 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
18618 * common.opt (freport-bug): Likewise.
18619 * cif-code.def (CIF_FINAL_ERROR): Likewise.
18620 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
18621 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
18622 translatable string.
18623 * config/i386/i386.c (function_value_32): Likewise.
18624 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
18625 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
18627 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
18628 * common/config/msp430/msp430-common.c (msp430_handle_option):
18630 * symtab.c (symtab_node::verify_base): Likewise.
18631 * opts.c (set_debug_level): Likewise.
18632 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
18633 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
18634 missing whitespace to translatable strings.
18635 * config/avr/avr.md (bswapsi2): Fix typo in comment.
18636 * config/sh/superh.h: Likewise.
18637 * config/i386/xopintrin.h: Likewise.
18638 * config/i386/znver1.md: Likewise.
18639 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
18640 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
18641 * double-int.h (struct double_int): Likewise.
18642 * double-int.c (div_and_round_double): Likewise.
18643 * wide-int.cc: Likewise.
18644 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
18645 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
18646 * cfgcleanup.c (crossjumps_occured): Renamed to ...
18647 (crossjumps_occurred): ... this.
18648 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
18651 PR tree-optimization/78899
18652 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
18653 returning bool return struct loop *, NULL for failure and the new
18655 (versionable_outer_loop_p): Don't version outer loop if it has
18656 dont_vectorized bit set.
18657 (tree_if_conversion): When versioning outer loop, ensure
18658 tree_if_conversion is performed also on the inner loop of the
18659 non-vectorizable outer loop copy.
18660 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
18661 LOOP_VECTORIZED in inner loop of the scalar outer loop and
18662 prevent vectorization of it.
18663 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
18664 the outer loop vectorization of the non-scalar version is attempted
18665 before vectorization of the inner loop in scalar version. If
18666 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
18667 vectorization of its inner loop.
18668 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
18669 has 2 inner loops, rename also on edges from bb whose single pred
18670 is outer_loop->header. Fix typo in function comment.
18672 2017-01-09 Martin Sebor <msebor@redhat.com>
18675 * asan.c (asan_emit_stack_protection): Increase local buffer size
18676 to avoid snprintf truncation warning.
18678 2017-01-09 Andrew Pinski <apinski@cavium.com>
18680 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
18681 to reference thunderx2t99 for the tuning structure
18682 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
18684 (thunderx2t99_extra_costs): This.
18685 * config/aarch64/aarch64-tune.md: Regenerate.
18686 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
18687 (vulcan_addrcost_table): This.
18688 (vulcan_regmove_cost): Rename to ...
18689 (thunderx2t99_regmove_cost): This.
18690 (vulcan_vector_cost): Rename to ...
18691 (thunderx2t99_vector_cost): this.
18692 (vulcan_branch_cost): Rename to ...
18693 (thunderx2t99_branch_cost): This.
18694 (vulcan_tunings): Rename to ...
18695 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
18696 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
18698 2017-01-09 Martin Jambor <mjambor@suse.cz>
18702 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
18703 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
18704 (propagate_vr_accross_jump_function): Use the above function for all
18705 value range computations for pass-through jump functions and type
18706 converasion from explicit value range values.
18707 (ipcp_propagate_stage): Do not attempt to deduce types of formal
18708 parameters from TYPE_ARG_TYPES.
18709 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
18710 (ipa_write_node_info): Stream type of the actual argument.
18711 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
18713 2017-01-09 Martin Liska <mliska@suse.cz>
18716 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
18717 (lookup_compiler): Do not show error message with have_E.
18719 2017-01-09 Jakub Jelinek <jakub@redhat.com>
18721 PR tree-optimization/78938
18722 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
18723 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
18724 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
18725 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
18728 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18730 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
18733 2017-01-09 Richard Biener <rguenther@suse.de>
18735 PR tree-optimization/78997
18736 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
18737 name condition properly.
18739 2017-01-09 Richard Biener <rguenther@suse.de>
18742 * dwarf2out.c (is_cxx): New overload with context.
18743 (is_naming_typedef_decl): Use it.
18745 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18747 * invoke.texi (Option Summary): Correct spacing in option lists
18748 and add line breaks to fix over-long lines.
18750 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18752 PR middle-end/17660
18754 * extend.texi (Common Variable Attributes): Add xref to GCC
18755 Internals manual to explain mode attribute keywords.
18757 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
18760 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
18761 and Preprocessor Options.
18762 (Options for Linking): Document -pthread here....
18763 (RS/6000 and PowerPC Options): ...not here.
18764 (Solaris 2 Options): ...or here.
18765 * doc/cppopts.texi: Document -pthread.
18767 2017-01-08 Martin Sebor <msebor@redhat.com>
18769 PR middle-end/77708
18770 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
18771 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
18772 New member functions.
18773 (format_directive): Used them.
18775 (pass_sprintf_length::handle_gimple_call): Same.
18776 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
18777 to avoid truncation for any argument.
18778 (extract_affine_mul): Same.
18779 * tree.c (get_file_function_name): Same.
18781 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
18783 PR middle-end/77484
18784 * predict.def (PRED_INDIR_CALL): Set to 86.
18786 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18788 PR preprocessor/54124
18789 * doc/cppopts.texi: Reformat -d subtable to list the full name
18790 of the options. Add cross-reference to the docs for the general
18791 compiler -d options.
18792 * doc/invoke.texi (Developer Options): Add cross-reference to the
18793 preprocessor-specific -d option documentation.
18795 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18797 PR preprocessor/13498
18798 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
18799 redudant material, and reflect new command-line options.
18800 (System Headers): Likewise.
18802 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
18804 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
18805 -isystem, and -idirafter. Copy-edit.
18806 * doc/cppopts.texi: Copy-edit. Remove contradiction about
18807 default for -ftrack-macro-expansion. Delete obsolete and
18808 badly-formatted implementation details about -fdebug-cpp output.
18809 * doc/cppwarnopts.texi: Copy-edit.
18811 2017-01-07 David Malcolm <dmalcolm@redhat.com>
18814 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18815 that the transition from a max line width >= 1<<10 to narrower
18816 lines works correctly.
18818 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
18820 * doc/options.texi (PerFunction): New.
18821 * opt-functions.awk (switch_flags): Map both Optimization and
18822 PerFunction to CL_OPTIMIZATION.
18823 * opth-gen.awk: Test for PerFunction flag along with
18825 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
18826 it only when the latter is present. Skip those that don't in
18827 the hash function generator.
18828 * common.opt (fvar-tracking): Mark as PerFunction instead of
18830 (fvar-tracking-assignments): Likewise.
18831 (fvar-tracking-assignments-toggle): Likewise.
18832 (fvar-tracking-uninit): Likewise.
18834 2017-01-07 Jakub Jelinek <jakub@redhat.com>
18836 PR translation/79018
18837 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
18840 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
18843 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
18844 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
18845 TARGET_LONG_JUMP_TABLE_OFFSETS.
18846 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
18847 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
18848 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
18849 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
18850 * config/m68k/m68k.md (tablejump expander): Likewise.
18851 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
18852 TARGET_LONG_JUMP_TABLE_OFFSETS.
18853 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
18854 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
18856 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
18857 David Holsgrove <david.holsgrove@xilinx.com>
18859 * common/config/microblaze/microblaze-common.c
18860 (TARGET_EXCEPT_UNWIND_INFO): Remove.
18861 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
18863 * config/microblaze/microblaze.c (microblaze_must_save_register)
18864 (microblaze_expand_epilogue, microblaze_return_addr): Handle
18866 (microblaze_eh_return): New function.
18867 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
18868 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
18869 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
18870 * config/microblaze/microblaze.md (eh_return): New pattern.
18872 2017-01-06 Jakub Jelinek <jakub@redhat.com>
18874 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
18875 GCC_DIAGNOSTIC_STRINGIFY): Define.
18877 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
18879 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18881 * config/arm/arm.md (<mcrr>): New.
18883 * config/arm/arm.c (arm_arch5te): New.
18884 (arm_option_override): Set arm_arch5te.
18885 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
18887 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
18888 (arm_mcrr_qualifiers): ... this. New.
18889 (MRRC_QUALIFIERS): Define to...
18890 (arm_mrrc_qualifiers): ... this. New.
18891 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
18893 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
18894 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
18895 (MRRCI, mrrc, MRRC): New.
18896 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
18897 VUNSPEC_MRRC2): New.
18899 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18901 * config/arm/arm.md (<mcr>): New.
18903 * config/arm/arm.c (arm_coproc_builtin_available): Add
18904 support for mcr, mrc, mcr2 and mrc2.
18905 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
18906 (arm_mcr_qualifiers): ... this. New.
18907 (MRC_QUALIFIERS): Define to ...
18908 (arm_mrc_qualifiers): ... this. New.
18909 (MCR_QUALIFIERS): Define to ...
18910 (arm_mcr_qualifiers): ... this. New.
18911 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
18913 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
18914 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
18915 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
18916 VUNSPEC_MRC2): New.
18918 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18920 * config/arm/arm.md (*ldc): New.
18924 * config/arm/arm.c (arm_coproc_builtin_available): Add
18925 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
18926 (arm_coproc_ldc_stc_legitimate_address): New.
18927 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
18928 'qualifier_const_pointer'.
18929 (LDC_QUALIFIERS): Define to...
18930 (arm_ldc_qualifiers): ... this. New.
18931 (STC_QUALIFIERS): Define to...
18932 (arm_stc_qualifiers): ... this. New.
18933 * config/arm/arm-protos.h
18934 (arm_coproc_ldc_stc_legitimate_address): New.
18935 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
18936 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
18937 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
18938 stc2, stcl, stc2l): New.
18939 * config/arm/constraints.md (Uz): New.
18940 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
18941 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
18942 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
18943 VUNSPEC_STC2L): New.
18945 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18947 * config/arm/arm.md (<cdp>): New.
18948 * config/arm/arm.c (neon_const_bounds): Rename this ...
18949 (arm_const_bounds): ... this.
18950 (arm_coproc_builtin_available): New.
18951 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
18952 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
18953 (CDP_QUALIFIERS): Define to...
18954 (arm_cdp_qualifiers): ... this. New.
18956 (arm_expand_builtin_args): Add case for 6 arguments.
18957 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
18958 (arm_const_bounds): ... this.
18959 (arm_coproc_builtin_available): New.
18960 * config/arm/arm_acle.h (__arm_cdp): New.
18962 * config/arm/arm_acle_builtins.def (cdp): New.
18964 * config/arm/iterators.md (CDPI,CDP,cdp): New.
18965 * config/arm/neon.md: Rename all 'neon_const_bounds' to
18966 'arm_const_bounds'.
18967 * config/arm/types.md (coproc): New.
18968 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
18969 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
18970 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
18971 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
18973 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18975 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
18976 (UBINOP_QUALIFIERS): New.
18978 (acle_builtin_data): New. Change comment.
18979 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
18980 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
18981 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
18982 arm_acle_builtins.def.
18983 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
18984 (arm_init_acle_builtins): New.
18985 (CRC32_BUILTIN): Remove.
18986 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
18987 crc32cb, crc32ch and crc32cw.
18988 (arm_init_crc32_builtins): Remove.
18989 (arm_init_builtins): Use arm_init_acle_builtins rather
18990 than arm_init_crc32_builtins.
18991 (arm_expand_acle_builtin): New.
18992 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
18993 * config/arm/arm_acle_builtins.def: New.
18995 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
18997 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
18998 (arm_builtin_datum): ... this.
18999 (arm_init_neon_builtin): Rename to ...
19000 (arm_init_builtin): ... this. Add a new parameters PREFIX
19001 and USE_SIG_IN_NAME.
19002 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
19003 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
19004 'arm_builtin_datum'.
19005 (arm_init_vfp_builtins): Likewise.
19006 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
19007 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
19008 (arm_expand_neon_args): Rename to ...
19009 (arm_expand_builtin_args): ... this. Rename builtin_arg
19010 enum values and differentiate between ARG_BUILTIN_MEMORY
19011 and ARG_BUILTIN_NEON_MEMORY.
19012 (arm_expand_neon_builtin_1): Rename to ...
19013 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
19014 values, arm_expand_builtin_args and add bool parameter NEON.
19015 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
19016 (arm_expand_vfp_builtin): Likewise.
19017 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
19019 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19021 PR middle-end/77484
19022 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
19023 * predict.c (tree_estimate_probability_bb): Reverse direction of
19024 polymorphic call predictor.
19026 2017-01-06 David Malcolm <dmalcolm@redhat.com>
19028 * passes.c (execute_one_pass): Split out pass-skipping logic into...
19029 (determine_pass_name_match): ...this new function and...
19030 (should_skip_pass_p): ...this new function.
19032 2017-01-06 Nathan Sidwell <nathan@acm.org>
19034 * ipa-visibility.c (function_and_variable_visibility): Reformat
19035 comments and long lines. Remove extrneous if.
19036 * symtab.c (symtab_node::make_decl_local): Fix code format.
19037 (symtab_node::set_section_for_node): Fix comment typo.
19039 2017-01-06 Martin Liska <mliska@suse.cz>
19042 * lra-constraints.c: Rename invariant to lra_invariant.
19043 * predict.c (set_even_probabilities): Initialize e to NULL.
19045 2017-01-05 Martin Sebor <msebor@redhat.com>
19047 PR tree-optimization/78910
19048 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
19049 (format_integer): Correct off-by-one error in the handling
19050 of precision with negative numbers in signed conversions..
19052 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
19054 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
19056 2017-01-05 Jakub Jelinek <jakub@redhat.com>
19058 PR tree-optimization/71016
19059 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
19060 factor_out_conditional_conversion. Formatting fix.
19061 (factor_out_conditional_conversion): Add cond_stmt argument.
19062 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
19063 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
19066 2017-01-05 David Malcolm <dmalcolm@redhat.com>
19068 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
19069 read-rtl-function.o, and selftest-rtl.o.
19070 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
19071 (selftest::aarch64_test_loading_full_dump): New function.
19072 (selftest::aarch64_run_selftests): New function.
19073 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
19074 selftest::aarch64_run_selftests.
19075 * config/i386/i386.c
19076 (selftest::ix86_test_loading_dump_fragment_1): New function.
19077 (selftest::ix86_test_loading_call_insn): New function.
19078 (selftest::ix86_test_loading_full_dump): New function.
19079 (selftest::ix86_test_loading_unspec): New function.
19080 (selftest::ix86_run_selftests): Call the new functions.
19081 * emit-rtl.c (maybe_set_max_label_num): New function.
19082 * emit-rtl.h (maybe_set_max_label_num): New decl.
19083 * function.c (instantiate_decls): Guard call to
19084 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
19085 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
19087 * gensupport.c (gen_reader::gen_reader): Pass "false"
19088 for new "compact" param of rtx_reader.
19089 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
19090 rather than an empty string for NULL strings.
19091 * read-md.c: Potentially include config.h rather than bconfig.h.
19092 Wrap include of errors.h with #ifdef GENERATOR_FILE.
19093 (have_error): New global, copied from errors.c.
19094 (md_reader::read_name): Rename to...
19095 (md_reader::read_name_1): ...this, adding "out_loc" param,
19096 and converting "missing name or number" to returning false, rather
19098 (md_reader::read_name): Reimplement in terms of read_name_1.
19099 (md_reader::read_name_or_nil): New function.
19100 (md_reader::read_string): Handle "(nil)" by returning NULL.
19101 (md_reader::md_reader): Add new param "compact".
19102 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
19103 (md_reader::read_file): New method.
19104 * read-md.h (md_reader::md_reader): Add new param "compact".
19105 (md_reader::read_file): New method.
19106 (md_reader::is_compact): New accessor.
19107 (md_reader::read_name): Convert return type from void to file_location.
19108 (md_reader::read_name_or_nil): New decl.
19109 (md_reader::read_name_1): New decl.
19110 (md_reader::m_compact): New field.
19111 (noop_reader::noop_reader): Pass "false" for new "compact" param
19113 (rtx_reader::rtx_reader): Add new "compact" param.
19114 (rtx_reader::read_rtx_operand): Make virtual and convert return
19115 type from void to rtx.
19116 (rtx_reader::read_until): New decl.
19117 (rtx_reader::handle_any_trailing_information): New virtual function.
19118 (rtx_reader::postprocess): New virtual function.
19119 (rtx_reader::finalize_string): New virtual function.
19120 (rtx_reader::m_in_call_function_usage): New field.
19121 (rtx_reader::m_reuse_rtx_by_id): New field.
19122 * read-rtl-function.c: New file.
19123 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
19124 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
19125 (selftest::verify_three_block_rtl_cfg): New decl.
19126 * read-rtl-function.h: New file.
19127 * read-rtl.c: Potentially include config.h rather than bconfig.h.
19128 For host, include function.h, memmodel.h, and emit-rtl.h.
19129 (one_time_initialization): New function.
19130 (struct compact_insn_name): New struct.
19131 (compact_insn_names): New array.
19132 (find_code): Handle insn codes in compact dumps.
19133 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
19134 (bind_subst_iter_and_attr): Likewise.
19135 (add_condition_to_string): Likewise.
19136 (add_condition_to_rtx): Likewise.
19137 (apply_attribute_uses): Likewise.
19138 (add_current_iterators): Likewise.
19139 (apply_iterators): Likewise.
19140 (initialize_iterators): Guard usage of apply_subst_iterator with
19141 #ifdef GENERATOR_FILE.
19142 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
19143 (md_reader::read_mapping): Likewise.
19144 (add_define_attr_for_define_subst): Likewise.
19145 (add_define_subst_attr): Likewise.
19146 (read_subst_mapping): Likewise.
19147 (check_code_iterator): Likewise.
19148 (rtx_reader::read_rtx): Likewise. Move one-time initialization
19150 (one_time_initialization): New function.
19151 (rtx_reader::read_until): New method.
19152 (read_flags): New function.
19153 (parse_reg_note_name): New function.
19154 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
19155 Handle reuse_rtx ids.
19156 Wrap iterator lookup within #ifdef GENERATOR_FILE.
19157 Add parsing support for RTL dumps, mirroring the special-cases in
19158 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
19159 values, and calling handle_any_trailing_information.
19160 (rtx_reader::read_rtx_operand): Convert return type from void
19161 to rtx, returning return_rtx. Handle case 'e'. Call
19162 finalize_string on XSTR and XTMPL fields.
19163 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
19164 "(nil)" values were omitted. Call the postprocess vfunc on the
19166 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
19167 class ctor. Initialize m_in_call_function_usage. Call
19168 one_time_initialization.
19169 * rtl-tests.c (selftest::test_uncond_jump): Call
19170 set_new_first_and_last_insn.
19171 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
19172 * selftest-rtl.c: New file.
19173 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
19174 (selftest::get_insn_by_uid): New decl.
19175 * selftest-run-tests.c (selftest::run_tests): Call
19176 read_rtl_function_c_tests.
19177 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
19178 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
19181 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
19183 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
19184 operands in a special way. Assert that pos+len <= mode precision.
19186 2017-01-05 Jakub Jelinek <jakub@redhat.com>
19188 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
19189 3 argument Alias with unlimited for the negative form.
19190 (fno-vect-cost-model): Removed.
19192 2017-01-05 Martin Liska <mliska@suse.cz>
19194 * hsa-gen.c (gen_hsa_divmod): New function.
19195 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
19197 2017-01-05 Martin Liska <mliska@suse.cz>
19200 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
19203 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19205 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
19206 small constant length operands.
19208 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19210 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
19211 between loop iterations.
19213 2017-01-05 Martin Liska <mliska@suse.cz>
19216 * gimplify.c (gimplify_decl_expr): Compare to
19217 asan_poisoned_variables instread of checking flags.
19218 (gimplify_target_expr): Likewise.
19219 (gimplify_expr): Likewise.
19220 (gimplify_function_tree): Conditionally initialize
19221 asan_poisoned_variables.
19223 2017-01-04 Jeff Law <law@redhat.com>
19225 PR tree-optimizatin/78812
19226 * rtl.h (contains_mem_rtx_p): Prototype.
19227 * ifcvt.c (containts_mem_rtx_p): Move from here to...
19228 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
19229 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
19230 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
19231 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
19233 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19235 * input.c (assert_char_at_range): Default-initialize actual_range.
19237 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19239 * df-scan.c (df_ref_create_structure): Make regno unsigned,
19240 to match the caller.
19242 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19244 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
19245 insns after final jump in test to emit dummy move.
19247 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19249 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
19250 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
19252 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19254 * multiple_target.c (create_dispatcher_calls): Init e_next.
19255 * tree-ssa-loop-split.c (split_loop): Init border.
19256 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
19259 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
19264 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
19265 (altivec_register_operand): Do not return true if the operand
19266 contains a SUBREG mixing SImode and SFmode.
19267 (vsx_register_operand): Likewise.
19268 (vsx_reg_sfsubreg_ok): New predicate.
19269 (vfloat_operand): Do not return true if the operand contains a
19270 SUBREG mixing SImode and SFmode.
19271 (vint_operand): Likewise.
19272 (vlogical_operand): Likewise.
19273 (gpc_reg_operand): Likewise.
19274 (int_reg_operand): Likewise.
19275 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
19276 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
19277 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
19279 (rs6000_emit_move_si_sf_subreg): New helper function.
19280 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
19281 fixup SUBREGs involving SImode and SFmode.
19282 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
19283 numbers for the new peephole2 optimization.
19284 (peephole2 for SFmode unions): New peephole2 to optimize cases in
19285 the GLIBC math library that do AND/IOR/XOR operations on single
19286 precision floating point.
19287 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
19288 target macros to say whether we need to avoid SUBREGs mixing
19290 (TARGET_ALLOW_SF_SUBREG): Likewise.
19291 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
19292 (UNSPEC_SI_FROM_SF): Likewise.
19293 (iorxor): Change spacing.
19294 (and_ior_xor): New iterator for AND, IOR, and XOR.
19295 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
19296 (movdi_from_sf_zero_ext): Likewise.
19297 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
19298 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
19299 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
19300 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
19301 (fms<mode>4): Likewise.
19302 (fnma<mode>4): Likewise.
19303 (fnms<mode>4): Likewise.
19304 (nfma<mode>4): Likewise.
19305 (nfms<mode>4): Likewise.
19307 2017-01-04 Marek Polacek <polacek@redhat.com>
19310 * doc/invoke.texi: Document -Wpointer-compare.
19312 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19314 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
19317 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
19318 descriptions for -gdwarf-5 and emit them as uleb128 instead of
19321 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
19324 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
19325 documentation of the powerpc_popcntb_ok attribute.
19326 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19327 code to issue warning messages if a requested CPU configuration is
19328 not supported by the binary (assembler and loader) toolchain.
19329 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
19330 made to define a built-in function that has been disabled.
19331 (paired_init_builtins): Add assertion to prevent ICE if attempt is
19332 made to define a built-in function that has been disabled.
19333 (altivec_init_builtins): Add comment explaining why definition
19334 of the DST built-in functions is not preceded by an assertion
19335 check. Add assertions to prevent ICE if attempts are made to
19336 define an altivec predicate or an abs* built-in function that has
19338 (htm_init_builtins): Add comment explaining why definition of the
19339 htm built-in functions is not preceded by an assertion check.
19341 2017-01-04 Jeff Law <law@redhat.com>
19343 PR tree-optimizatin/67955
19344 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
19345 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
19346 the points-to solution does not include pt_null. Use DECL_PT_UID
19349 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
19351 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
19352 Use gen_int_mode instead of gen_lopwart for const_int operands.
19354 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19356 PR tree-optimization/71563
19357 * match.pd: Simplify X << Y into X if Y is known to be 0 or
19358 out of range value - has low bits known to be zero.
19360 2017-01-04 Alan Modra <amodra@gmail.com>
19362 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
19363 * configure: Regenerate.
19364 * config.in: Regenerate.
19366 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19369 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
19370 a substring of the message, but strcmp with the whole message. Ifdef
19371 ENABLE_NLS, translate the message first using dgettext.
19373 2017-01-03 Jeff Law <law@redhat.com>
19375 PR tree-optimizatin/78856
19376 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
19377 (mark_threaded_blocks): Remove code to truncate thread paths that
19378 cross multiple loop headers. Instead invalidate the cached loop
19379 iteration information and handle case of a thread path walking
19380 into an irreducible region.
19382 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
19385 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
19386 assertions. Add support for doing the signbit if the IEEE 128-bit
19387 floating point value is in a GPR.
19388 * config/rs6000/rs6000.md (Fsignbit): Delete.
19389 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
19390 Update the length attribute if the value is in a GPR.
19391 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
19392 the sign or zero extension instruction, since the value is always 0/1.
19393 (signbit<mode>2_dm2): Delete using <Fsignbit>.
19396 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
19397 extracting SImode to a GPR register so that we can generate a
19398 store, limit the vector to be in a traditional Altivec register
19399 for the vextuwrx instruction.
19401 2017-01-03 Ian Lance Taylor <iant@google.com>
19403 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
19405 2017-01-03 Martin Sebor <msebor@redhat.com>
19407 PR tree-optimization/78696
19408 * gimple-ssa-sprintf.c (format_floating): Correct handling of
19409 precision. Use MPFR for %f for greater fidelity. Correct handling
19411 (pass_sprintf_length::compute_format_length): Set width and precision
19412 specified by asrerisk to void_node for vararg functions.
19413 (try_substitute_return_value): Adjust dump output.
19415 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
19417 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
19419 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
19421 * doc/invoke.texi (SPARC options): Document -mlra as the default.
19422 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
19423 -mlra/-mno-lra was passed to the compiler.
19425 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
19427 PR rtl-optimization/65618
19428 * emit-rtl.c (try_split): Move initialization of "before" and
19429 "after" to just before the call to emit_insn_after_setloc.
19431 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
19433 * doc/md.texi (Standard Names): Remove reference to Java frontend.
19435 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
19437 * dwarf2out.c (gen_enumeration_type_die): When
19438 -gno-strict-dwarf, add a DW_AT_encoding attribute.
19440 2017-01-03 Jakub Jelinek <jakub@redhat.com>
19442 PR tree-optimization/78965
19443 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
19444 Change first argument from const call_info & to call_info &. For %n
19445 set info.nowrite to false.
19447 PR middle-end/78901
19448 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
19449 possibly throwing calls.
19451 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
19452 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
19453 and fns handling, rather than in a separate case SSA_NAME.
19455 2017-01-02 Jeff Law <law@redhat.com>
19457 * config/darwin-driver.c (darwin_driver_init): Const-correctness
19458 fixes for first_period and second_period variables.
19460 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
19463 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
19464 (*insvqi_1): New insn pattern.
19465 (*insvqi_1_mem_rex64): Ditto.
19466 (*insvqi_2): Ditto.
19467 (*insvqi_3): Rename from *insvqi.
19469 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
19471 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
19473 * doc/cfg.texi (Edges): Remove reference to Java.
19474 (Maintaining the CFG): Ditto.
19476 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19478 PR middle-end/77674
19479 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
19480 transparent aliases.
19482 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19484 PR middle-end/77484
19485 * predict.def (PRED_CALL): Update hitrate.
19486 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
19487 * predict.c (tree_estimate_probability_bb): Split CALL predictor
19488 into direct/indirect/polymorphic variants.
19490 2017-01-01 Jakub Jelinek <jakub@redhat.com>
19492 Update copyright years.
19494 * gcc.c (process_command): Update copyright notice dates.
19495 * gcov-dump.c (print_version): Ditto.
19496 * gcov.c (print_version): Ditto.
19497 * gcov-tool.c (print_version): Ditto.
19498 * gengtype.c (create_file): Ditto.
19499 * doc/cpp.texi: Bump @copying's copyright year.
19500 * doc/cppinternals.texi: Ditto.
19501 * doc/gcc.texi: Ditto.
19502 * doc/gccint.texi: Ditto.
19503 * doc/gcov.texi: Ditto.
19504 * doc/install.texi: Ditto.
19505 * doc/invoke.texi: Ditto.
19507 Copyright (C) 2017 Free Software Foundation, Inc.
19509 Copying and distribution of this file, with or without modification,
19510 are permitted in any medium without royalty provided the copyright
19511 notice and this notice are preserved.