1 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2 Alan Hayward <alan.hayward@arm.com>
3 David Sherwood <david.sherwood@arm.com>
5 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
6 * regcprop.c (maybe_mode_change): Likewise.
7 * reload1.c (alter_reg): Likewise.
9 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11 * inchash.h (inchash::hash::add_wide_int): New function.
12 * lto-streamer-out.c (hash_tree): Use it.
14 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
16 * inchash.h (inchash::hash::add_wide_int): Rename to...
17 (inchash::hash::add_hwi): ...this.
18 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
19 (polymorphic_call_target_hasher::hash): Likewise.
20 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
21 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
22 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
23 * lto-streamer-out.c (hash_tree): Likewise.
24 * optc-save-gen.awk: Likewise.
25 * tree.c (add_expr): Likewise.
27 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
30 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
31 for ordered inequality comparisons even with TARGET_IEEE_FP.
33 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
36 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
37 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
38 Expand with cmp<dwi>_doubleword.
40 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
42 * extend.texi: Add x86 specific to 'nocf_check' attribute.
44 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
45 specific to -fcf-protection option.
47 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
49 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
50 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
51 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
52 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
53 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
54 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
55 (extra_objs): Add cet.o for Linux/x86 targets.
56 (tmake_file): Add i386/t-cet for Linux/x86 targets.
57 * config/i386/cet.c: New file.
58 * config/i386/cetintrin.h: Likewise.
59 * config/i386/t-cet: Likewise.
60 * config/i386/cpuid.h (bit_SHSTK): New.
62 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
63 pass IBT and SHSTK bits.
64 * config/i386/i386-builtin-types.def
65 (VOID_FTYPE_UNSIGNED_PVOID): New.
66 (VOID_FTYPE_UINT64_PVOID): Likewise.
67 * config/i386/i386-builtin.def: Add CET intrinsics.
68 * config/i386/i386-c.c (ix86_target_macros_internal): Add
69 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
70 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
71 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
73 * config/i386/i386.c (rest_of_insert_endbranch): New.
74 (pass_data_insert_endbranch): Likewise.
75 (pass_insert_endbranch): Likewise.
76 (make_pass_insert_endbranch): Likewise.
77 (ix86_notrack_prefixed_insn_p): Likewise.
78 (ix86_target_string): Add -mibt, -mshstk flags.
79 (ix86_option_override_internal): Add flag_cf_protection
81 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
82 (ix86_print_operand): Add 'notrack' prefix output.
83 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
84 (ix86_expand_builtin): Expand CET intrinsics.
85 (x86_output_mi_thunk): Add 'endbranch' instruction.
86 * config/i386/i386.h (TARGET_IBT): New.
87 (TARGET_IBT_P): Likewise.
88 (TARGET_SHSTK): Likewise.
89 (TARGET_SHSTK_P): Likewise.
90 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
91 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
92 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
93 (builtin_setjmp_setup): New pattern.
94 (builtin_longjmp): Likewise.
95 (rdssp<mode>): Likewise.
96 (incssp<mode>): Likewise.
97 (saveprevssp): Likewise.
99 (wrss<mode>): Likewise.
100 (wruss<mode>): Likewise.
101 (setssbsy): Likewise.
102 (clrssbsy): Likewise.
103 (nop_endbr): Likewise.
104 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
106 * config/i386/immintrin.h: Include <cetintrin.h>.
107 * config/i386/linux-common.h
108 (file_end_indicate_exec_stack_and_cet): New prototype.
109 (TARGET_ASM_FILE_END): New.
111 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
113 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
115 2017-10-20 Jakub Jelinek <jakub@redhat.com>
117 * config/i386/i386.md (isa): Remove fma_avx512f.
118 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
119 <avx512>_fmadd_<mode>_mask3<round_name>,
120 <avx512>_fmsub_<mode>_mask<round_name>,
121 <avx512>_fmsub_<mode>_mask3<round_name>,
122 <avx512>_fnmadd_<mode>_mask<round_name>,
123 <avx512>_fnmadd_<mode>_mask3<round_name>,
124 <avx512>_fnmsub_<mode>_mask<round_name>,
125 <avx512>_fnmsub_<mode>_mask3<round_name>,
126 <avx512>_fmaddsub_<mode>_mask<round_name>,
127 <avx512>_fmaddsub_<mode>_mask3<round_name>,
128 <avx512>_fmsubadd_<mode>_mask<round_name>,
129 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
130 (*vec_widen_umult_even_v16si<mask_name>,
131 *vec_widen_smult_even_v16si<mask_name>): Likewise.
132 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
134 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
136 * extend.texi: Add 'nocf_check' documentation.
137 * gimple.texi: Add second parameter to
138 gimple_build_call_from_tree.
139 * invoke.texi: Add -fcf-protection documentation.
140 * rtl.texi: Add REG_CALL_NOTRACK documenation.
142 2017-10-20 Richard Biener <rguenther@suse.de>
144 PR tree-optimization/82473
145 * tree-vect-loop.c (vectorizable_reduction): Properly get at
146 the largest input type.
148 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
150 * c-attribs.c (handle_nocf_check_attribute): New function.
151 (c_common_attribute_table): Add 'nocf_check' handling.
152 * gimple-parser.c: Add second argument NULL to
153 gimple_build_call_from_tree.
154 * attrib.c (comp_type_attributes): Check nocf_check attribute.
155 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
157 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
158 * common.opt: Add fcf-protection flag.
159 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
160 * flag-types.h: Add enum cf_protection_level.
161 * gimple.c (gimple_build_call_from_tree): Add second parameter.
162 Add 'nocf_check' attribute propagation to gimple call.
163 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
164 (gimple_build_call_from_tree): Update prototype.
165 (gimple_call_nocf_check_p): New function.
166 (gimple_call_set_nocf_check): Likewise.
167 * gimplify.c: Add second argument to gimple_build_call_from_tree.
168 * ipa-icf.c: Add nocf_check attribute in statement hash.
169 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
170 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
171 * toplev.c (process_options): Add flag_cf_protection handling.
173 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
175 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
177 2017-10-20 Richard Biener <rguenther@suse.de>
179 PR tree-optimization/82603
180 * tree-if-conv.c (predicate_mem_writes): Make sure to only
181 remove false predicated stores.
183 2017-10-20 Richard Biener <rguenther@suse.de>
185 * graphite-isl-ast-to-gimple.c
186 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
187 Remove return value and simplify, dump copied stmt after lhs
189 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
190 Reduce dump verbosity.
191 (gsi_insert_earliest): Likewise.
192 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
193 * graphite.c (print_global_statistics): Adjust dumping.
194 (print_graphite_scop_statistics): Likewise.
195 (print_graphite_statistics): Do not dump loops here.
196 (graphite_transform_loops): But here.
198 2017-10-20 Nicolas Roche <roche@adacore.com>
200 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
201 * configure: Regenerate.
203 2017-10-20 Jakub Jelinek <jakub@redhat.com>
206 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
207 functions when optimizing replace GIMPLE_RETURN stmts with
208 calls to __builtin_unreachable ().
211 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
212 for -fsanitize=thread link of executables.
213 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
217 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
219 (<shift_insn><mode>3<mask_name>): Change the last of the 3
220 define_insns for logical vector shifts to use VI248_AVX512BW
221 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
222 condition, useless isa and prefix attributes. Change the first
223 2 of these define_insns to ...
224 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
225 define_insn for avx512vl.
226 (<shift_insn><mode>3): ... and this, new define_insn without
227 masking for non-avx512vl.
230 * config/i386/sse.md (*andnot<mode>3,
231 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
232 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
233 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
234 not applied use empty suffix even for TARGET_AVX512VL.
235 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
236 is applied, supply evex,evex or evex,evex,evex instead of just
239 2017-10-20 Julia Koval <julia.koval@intel.com>
241 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
242 (OPTION_MASK_ISA_GFNI_UNSET): New.
243 (ix86_handle_option): Handle OPT_mgfni.
244 * config/i386/cpuid.h (bit_GFNI): New.
245 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
246 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
247 * config/i386/i386.c (ix86_target_string): Add -mgfni.
248 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
249 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
250 * config/i386/i386.opt: Add mgfni.
252 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
254 * config/msp430/msp430.c (msp430_option_override): Disable
255 -fdelete-null-pointer-checks.
256 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
258 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
260 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
261 of x87 and SSE instructions.
263 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
265 * asan.c (create_cond_insert_point): Do not update edge count.
266 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
267 (afdo_propagate_circuit): Likewise.
268 (afdo_calculate_branch_prob): Likewise.
269 (afdo_annotate_cfg): Likewise.
270 * basic-block.h (struct edge_def): Remove count.
271 (edge_def::count): New accessor.
272 * bb-reorder.c (rotate_loop): Update.
273 (find_traces_1_round): Update.
274 (connect_traces): Update.
275 (sanitize_hot_paths): Update.
276 * cfg.c (unchecked_make_edge): Update.
277 (make_single_succ_edge): Update.
278 (check_bb_profile): Update.
279 (dump_edge_info): Update.
280 (update_bb_profile_for_threading): Update.
281 (scale_bbs_frequencies_int): Update.
282 (scale_bbs_frequencies_gcov_type): Update.
283 (scale_bbs_frequencies_profile_count): Update.
284 (scale_bbs_frequencies): Update.
285 * cfganal.c (connect_infinite_loops_to_exit): Update.
286 * cfgbuild.c (compute_outgoing_frequencies): Update.
287 (find_many_sub_basic_blocks): Update.
288 * cfgcleanup.c (try_forward_edges): Update.
289 (try_crossjump_to_edge): Update
290 * cfgexpand.c (expand_gimple_cond): Update
291 (expand_gimple_tailcall): Update
292 (construct_exit_block): Update
293 * cfghooks.c (verify_flow_info): Update
294 (redirect_edge_succ_nodup): Update
296 (make_forwarder_block): Update
297 (duplicate_block): Update
298 (account_profile_record): Update
299 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
300 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
301 * cfgloopmanip.c (scale_loop_profile): Update.
303 (lv_adjust_loop_entry_edge): Update.
304 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
305 (force_nonfallthru_and_redirect): Update.
306 (purge_dead_edges): Update.
307 (rtl_flow_call_edges_add): Update.
308 * cgraphunit.c (init_lowered_empty_function): Update.
309 (cgraph_node::expand_thunk): Update.
310 * gimple-pretty-print.c (dump_probability): Update.
311 (dump_edge_probability): Update.
312 * gimple-ssa-isolate-paths.c (isolate_path): Update.
313 * haifa-sched.c (sched_create_recovery_edges): Update.
314 * hsa-gen.c (convert_switch_statements): Update.
315 * ifcvt.c (dead_or_predicable): Update.
316 * ipa-inline-transform.c (inline_transform): Update.
317 * ipa-split.c (split_function): Update.
318 * ipa-utils.c (ipa_merge_profiles): Update.
319 * loop-doloop.c (add_test): Update.
320 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
321 * lto-streamer-in.c (input_cfg): Update.
322 (input_function): Update.
323 * lto-streamer-out.c (output_cfg): Update.
324 * modulo-sched.c (sms_schedule): Update.
325 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
326 * predict.c (maybe_hot_edge_p): Update.
327 (unlikely_executed_edge_p): Update.
328 (probably_never_executed_edge_p): Update.
329 (dump_prediction): Update.
330 (drop_profile): Update.
331 (propagate_unlikely_bbs_forward): Update.
332 (determine_unlikely_bbs): Update.
333 (force_edge_cold): Update.
334 * profile.c (compute_branch_probabilities): Update.
335 * reg-stack.c (better_edge): Update.
336 * shrink-wrap.c (handle_simple_exit): Update.
337 * tracer.c (better_p): Update.
338 * trans-mem.c (expand_transaction): Update.
339 (split_bb_make_tm_edge): Update.
340 * tree-call-cdce.c: Update.
341 * tree-cfg.c (gimple_find_sub_bbs): Update.
342 (gimple_split_edge): Update.
343 (gimple_duplicate_sese_region): Update.
344 (gimple_duplicate_sese_tail): Update.
345 (gimple_flow_call_edges_add): Update.
346 (insert_cond_bb): Update.
347 (execute_fixup_cfg): Update.
348 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
349 * tree-complex.c (expand_complex_div_wide): Update.
350 * tree-eh.c (lower_resx): Update.
351 (unsplit_eh): Update.
352 (cleanup_empty_eh_move_lp): Update.
353 * tree-inline.c (copy_edges_for_bb): Update.
354 (freqs_to_counts): Update.
355 (copy_cfg_body): Update.
356 * tree-ssa-dce.c (remove_dead_stmt): Update.
357 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
358 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
359 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
360 (unloop_loops): Update.
361 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
362 * tree-ssa-loop-split.c (connect_loops): Update.
363 (split_loop): Update.
364 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
365 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
366 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
367 * tree-ssa-reassoc.c (branch_fixup): Update.
368 * tree-ssa-tail-merge.c (replace_block_by): Update.
369 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
370 (compute_path_counts): Update.
371 (update_profile): Update.
372 (recompute_probabilities): Update.
373 (update_joiner_offpath_counts): Update.
374 (estimated_freqs_path): Update.
375 (freqs_to_counts_path): Update.
376 (clear_counts_path): Update.
377 (ssa_fix_duplicate_block_edges): Update.
378 (duplicate_thread_path): Update.
379 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
380 (case_bit_test_cmp): Update.
381 (collect_switch_conv_info): Update.
382 (gen_inbound_check): Update.
383 (do_jump_if_equal): Update.
384 (emit_cmp_and_jump_insns): Update.
385 * tree-tailcall.c (decrease_profile): Update.
386 (eliminate_tail_call): Update.
387 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
388 (vect_do_peeling): Update.
389 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
390 * ubsan.c (ubsan_expand_null_ifn): Update.
391 (ubsan_expand_ptr_ifn): Update.
392 * value-prof.c (gimple_divmod_fixed_value): Update.
393 (gimple_mod_pow2): Update.
394 (gimple_mod_subtract): Update.
396 (gimple_stringop_fixed_value): Update.
398 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
401 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
403 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
405 PR rtl-optimization/82395
406 * ira-color.c (allocno_priority_compare_func): Fix comparison step
407 based on non_spilled_static_chain_regno_p.
409 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
411 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
412 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
413 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
415 2017-10-19 Martin Sebor <msebor@redhat.com>
417 PR tree-optimization/82596
418 * tree.c (array_at_struct_end_p): Handle STRING_CST.
420 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
422 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
423 (get_mem_refs_of_builtin_call): Likewise.
424 * builtins.c (expand_builtin_apply): Adjust call to
425 allocate_dynamic_stack_space.
426 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
427 the third argument to allocate_dynamic_stack_space, otherwise -1.
428 (expand_builtin): Deal with all alloca variants.
429 (is_inexpensive_builtin): Likewise.
430 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
431 * calls.c (special_function_p): Deal with all alloca variants.
432 (initialize_argument_information): Adjust call to
433 allocate_dynamic_stack_space.
434 (expand_call): Likewise.
435 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
436 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
437 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
438 use it for the stack usage computation.
439 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
440 * function.c (gimplify_parameters): Call build_alloca_call_expr.
441 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
442 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
443 (in_loop_p): Remove first argument and useless check.
444 (pass_walloca::execute): Remove useless test and adjust call to above.
445 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
446 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
447 (gimplify_call_expr): Deal with all alloca variants.
448 * hsa-gen.c (gen_hsa_alloca): Likewise.
449 (gen_hsa_insns_for_call): Likewise.
450 * ipa-pure-const.c (special_builtin_state): Likewise.
451 * tree-chkp.c (chkp_build_returned_bound): Likewise.
452 * tree-object-size.c (alloc_object_size): Likewise.
453 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
454 (call_may_clobber_ref_p_1): Likewise.
455 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
456 (ccp_fold_stmt): Likewise.
457 (optimize_stack_restore): Likewise.
458 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
459 (mark_all_reaching_defs_necessary_1): Likewise.
460 (propagate_necessity): Likewise.
461 (eliminate_unnecessary_stmts): Likewise.
462 * tree.c (build_common_builtin_nodes): Build
463 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
464 (build_alloca_call_expr): New function.
465 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
466 (CASE_BUILT_IN_ALLOCA): Likewise.
467 (build_alloca_call_expr): Declare.
468 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
470 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
473 * dwarf2out.c (new_die_raw): New static inline function.
474 (new_die): Use it to create the DIE.
475 (add_AT_external_die_ref): Likewise.
476 (clone_die): Likewise.
477 (clone_as_declaration): Likewise.
478 (dwarf2out_vms_debug_main_pointer): Likewise.
479 (base_type_die): Likewise. Remove early return for corner cases.
480 Do not call add_pubtype on the DIE here.
481 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
482 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
483 typedefs for base types with DW_AT_endianity. Make sure a DIE with
484 native order exists for base types, attach the DIE manually and call
485 add_pubtype on it. Do not equate a reverse order DIE to the type.
487 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
489 * config/arm/arm.c (align_ok_ldrd_strd): New function.
490 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
492 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
494 2017-10-19 Jakub Jelinek <jakub@redhat.com>
496 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
497 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
498 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
499 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
500 * opts.c (sanitizer_opts): Add builtin.
501 * ubsan.c (instrument_builtin): New function.
502 (pass_ubsan::execute): Call it.
503 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
504 * doc/invoke.texi: Document -fsanitize=builtin.
506 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
507 builtins, store max (log2 (align), 0) into uchar field instead of
508 align into uptr field.
509 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
510 store uchar 0 field instead of uptr 0 field.
511 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
512 instead of passing one address of struct with 2 locations pass
513 two addresses of structs with 1 location each.
514 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
515 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
516 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
517 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
518 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
519 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
520 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
521 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
523 2017-10-19 Martin Liska <mliska@suse.cz>
526 * file-find.c (remove_prefix): Remove.
527 * file-find.h (remove_prefix): Likewise.
528 * gcc-ar.c: Remove smartness of lookup.
530 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
532 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
533 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
534 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
536 2017-10-19 Jakub Jelinek <jakub@redhat.com>
539 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
540 (setcc + and to xor + setcc): New peephole2.
542 2017-10-19 Tom de Vries <tom@codesourcery.com>
544 * doc/sourcebuild.texi (Test Directives, Variants of
545 dg-require-support): Add dg-require-stack-size.
547 2017-10-19 Martin Liska <mliska@suse.cz>
550 * gimplify.c (gimplify_decl_expr): Do not instrument variables
551 that have a large alignment.
552 (gimplify_target_expr): Likewise.
554 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
556 PR rtl-optimization/82602
557 * ira.c (rtx_moveable_p): Return false for volatile asm.
559 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
562 * config/i386/i386-modes.def (CCGZ): New CC mode.
563 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
564 * config/i386/predicates.md (ix86_comparison_operator):
566 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
567 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
568 with double-word subtraction.
569 (put_condition_code): Handle CCGZmode.
571 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
573 * wide-int.cc (debug (const wide_int &)): New.
574 (debug (const wide_int *)): New.
575 (debug (const widest_int &)): New.
576 (debug (const widest_int *)): New.
578 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
581 * lra-constraints.c (curr_insn_transform): Use non-input operand
582 instead of output one for matched reload.
584 2017-10-18 Bin Cheng <bin.cheng@arm.com>
586 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
587 (tree-ssa-loop-ivopts.h): New header file.
588 (struct builtin_info): New fields.
589 (classify_builtin_1): Compute and record base and offset parts for
590 memset builtin partition by calling strip_offset.
591 (offset_cmp, fuse_memset_builtins): New functions.
592 (finalize_partitions): Fuse adjacent memset partitions by calling
594 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
595 Expose the interface.
596 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
598 2017-10-18 Bin Cheng <bin.cheng@arm.com>
600 PR tree-optimization/82574
601 * tree-loop-distribution.c (find_single_drs): New parameter. Check
602 that data reference must be executed exactly once per iteration
603 against the outermost loop in nest.
604 (classify_partition): Update call to above function.
606 2017-10-18 Richard Biener <rguenther@suse.de>
608 PR tree-optimization/82591
609 * graphite.c (graphite_transform_loops): Move code gen message
611 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
612 Here. Handle scop_to_isl_ast failing.
613 (scop_to_isl_ast): Limit the number of ISL operations.
615 2017-10-18 Richard Biener <rguenther@suse.de>
617 * graphite-isl-ast-to-gimple.c
618 (translate_isl_ast_to_gimple::set_rename): Simplify.
619 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
620 (graphite_copy_stmts_from_block): ... here.
621 (copy_bb_and_scalar_dependences): Simplify.
622 (add_parameters_to_ivs_params): Canonicalize.
623 (generate_entry_out_of_ssa_copies): Simplify.
624 * graphite-sese-to-poly.c (extract_affine_name): Simplify
625 by passing in ISL dimension.
626 (parameter_index_in_region_1): Rename to ...
627 (parameter_index_in_region): ... this.
628 (extract_affine): Adjust assert, pass down parameter index.
629 (add_param_constraints): Use range-info when available.
630 (build_scop_context): Adjust.
631 * sese.c (new_sese_info): Adjust.
632 (free_sese_info): Likewise.
633 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
634 Remove unused typedefs.
635 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
637 2017-10-18 Martin Liska <mliska@suse.cz>
639 * combine.c (simplify_compare_const): Add gcc_fallthrough.
641 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
643 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
644 (s390_sched_init): Do not reset s390_sched_state if we entered the
645 current basic block via a fallthru edge and all others are unlikely.
647 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
649 * config/s390/s390.c (NUM_SIDES): New variable.
650 (LONGRUNNING_THRESHOLD): New variable.
651 (LATENCY_FACTOR): New variable.
652 (s390_sched_score): Decrease score for long-running instructions on
654 (s390_sched_variable_issue): Perform bookkeeping for long-running
657 2017-10-18 Richard Biener <rguenther@suse.de>
659 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
660 Simplify with removal of the parameter rename map.
661 (set_rename): Likewise.
662 (should_copy_to_new_region): Likewise.
663 (graphite_copy_stmts_from_block): Likewise.
664 (copy_bb_and_scalar_dependences): Remove initialization of
665 unused copied_bb_map.
667 (copy_internal_parameters): Likewise.
668 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
669 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
671 (parameter_index_in_region_1): Rename to ...
672 (assign_parameter_index_in_region): ... this. Assert we have
673 a parameter we handle.
674 (scan_tree_for_params): Adjust.
675 * sese.h (parameter_rename_map_t): Remove.
676 (struct sese_info_t): Remove unused parameter_rename_map and
677 copied_bb_map members.
678 * sese.c (new_sese_info): Adjust.
679 (free_sese_info): Likewise.
681 2017-10-18 Martin Liska <mliska@suse.cz>
684 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
687 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
689 * doc/invoke.texi (ffunction-sections and fdata-sections):
692 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
694 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
695 the use statement can throw internally.
697 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
699 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
700 any RTX present on the RHS of a SET.
701 * compare-elim.c (try_eliminate_compare): Restore comment.
703 2017-10-17 Jakub Jelinek <jakub@redhat.com>
705 * langhooks.h (struct lang_hooks): Document that tree_size langhook
706 may be also called on tcc_type nodes.
707 * langhooks.c (lhd_tree_size): Likewise.
709 2017-10-17 David Malcolm <dmalcolm@redhat.com>
711 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
712 format_warning_at_substring.
713 (maybe_warn): Convert source_range * param to a location_t. Pass
714 UNKNOWN_LOCATION rather than NULL to fmtwarn.
715 (format_directive): Remove code to extract source_ranges and
716 source_range * in favor of just a location_t.
717 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
719 * substring-locations.c (format_warning_va): Convert
720 source_range * param to a location_t.
721 (format_warning_at_substring): Likewise.
722 * substring-locations.h (format_warning_va): Likewise.
723 (format_warning_at_substring): Likewise.
725 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
727 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
729 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
730 and scatter/gather ops.
732 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
733 vec_gather_load and vec_scatter_store.
734 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
735 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
736 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
737 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
738 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
739 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
741 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
743 * reg-stack.c (compare_for_stack_reg): Add bool argument.
744 Detect FTST instruction and handle its register pops. Only pop
745 second operand if can_pop_second_op is true.
746 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
747 set can_pop_second_op to false in the compare_for_stack_reg call.
749 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
750 output_fp_compare for stack register operands.
751 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
752 instructions here. Do not emit stack register pops here. Assert
753 that FCOMPP pops next to top stack register. Rewrite function.
755 2017-10-17 Nathan Sidwell <nathan@acm.org>
758 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
759 use DECL_ASSEMBLER_NAME_RAW.
762 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
765 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
766 Wilco Dijkstra <wilco.dijkstra@arm.com>
768 * builtins.c (expand_builtin_update_setjmp_buf): Add a
769 converstion to Pmode from the buf_addr.
771 2017-10-17 Richard Biener <rguenther@suse.de>
773 * graphite-dependences.c (scop_get_reads_and_writes): Change
774 output parameters to references.
776 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
778 PR 71026/tree-optimization
779 * fold-const.c (distribute_real_division): Removed.
780 (fold_binary_loc): Remove calls to distribute_real_divison.
782 2017-10-17 Richard Biener <rguenther@suse.de>
784 * graphite-scop-detection.c
785 (scop_detection::stmt_has_simple_data_refs_p): Always use
786 the full nest as region.
787 (try_generate_gimple_bb): Likewise.
788 * sese.c (scalar_evolution_in_region): Simplify now that
789 SCEV can handle instantiation in regions.
790 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
791 in the non-loop part of a function if requested.
793 2017-10-17 Richard Biener <rguenther@suse.de>
795 PR tree-optimization/82563
796 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
798 (graphite_regenerate_ast_isl): Call it.
799 * graphite-scop-detection.c (build_scops): Remove entry edge split.
801 2017-10-17 Jakub Jelinek <jakub@redhat.com>
803 PR tree-optimization/82549
804 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
805 Formatting fixes. Instead of calling make_bit_field_ref with negative
808 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
810 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
811 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
812 _mm_maskz_reduce_ss): New.
813 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
814 __builtin_ia32_reducess_mask): Ditto..
815 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
816 * config/i386/sse.md (reduces<mode>): Renamed to ...
817 (reduces<mode><mask_scalar_name>): ... this.
818 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}): Changed
820 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
821 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
823 2017-10-16 David Malcolm <dmalcolm@redhat.com>
825 * Makefile.in (OBJS): Add unique-ptr-tests.o.
826 * selftest-run-tests.c (selftest::run_tests): Call
827 selftest::unique_ptr_tests_cc_tests.
828 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
829 * unique-ptr-tests.cc: New file.
831 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
834 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
836 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
837 (make_hard_regno_born, make_hard_regno_dead): Update
838 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
840 2017-10-16 Jeff Law <law@redhat.com>
842 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
844 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
846 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
848 2017-10-16 Olivier Hainque <hainque@adacore.com>
850 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
851 with_cpu if we were configured for an e500v2 target cpu name.
853 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
855 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
856 * doc/invoke.texi: Document +nodsp as a valid extension for
859 2017-10-16 Martin Liska <mliska@suse.cz>
861 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
862 (test_set_range): Likewise.
863 (test_range_functions): Rename to ...
864 (test_bit_in_range): ... this.
865 (sbitmap_c_tests): Add new test.
867 2017-10-16 Tamar Christina <tamar.christina@arm.com>
869 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32): New.
870 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
871 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
873 2017-10-16 Tamar Christina <tamar.christina@arm.com>
875 * config/aarch64/aarch64-builtins.c
876 (aarch64_types_quadopu_lane_qualifiers): New.
877 (TYPES_QUADOPU_LANE): New.
878 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
879 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
880 (aarch64_<sur>dot_laneq<vsi2qi>): New.
881 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
882 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
883 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
884 (Vdottype, DOTPROD): New.
885 (sur): Add SDOT and UDOT.
887 2017-10-16 Tamar Christina <tamar.christina@arm.com>
889 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
890 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
891 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Add TARGET_DOTPROD.
892 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
893 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75): Enable TARGET_DOTPROD.
894 (cortex-a75.cortex-a55): Likewise.
895 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
897 2017-10-16 Tamar Christina <tamar.christina@arm.com>
899 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
900 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS): New.
901 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane): new.
902 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
903 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
904 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
905 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
906 * config/arm/types.md (neon_dot, neon_dot_q): New.
907 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
909 2017-10-16 Tamar Christina <tamar.christina@arm.com>
911 * config/arm/arm.h (TARGET_DOTPROD): New.
912 * config/arm/arm.c (arm_arch_dotprod): New.
913 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
914 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
915 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
916 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
917 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
918 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
919 * doc/invoke.texi (armv8.2-a): Document dotprod
921 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
923 * i386.c (ix86_vec_cost): New function.
924 (ix86_rtx_costs): Handle vector operations better.
925 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
926 * x86-tune-costs.h: Add new costs to all tables.
928 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
930 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
932 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
933 divsd, sqrtss and sqrtsd
934 * x86-tune-costs.h: Add new entries to all costs.
935 (znver1_cost): Fix to match real instruction latencies.
937 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
938 Michael Collison <michael.collison@arm.com>
940 * compare-elim.c: Include emit-rtl.h.
941 (can_merge_compare_into_arith): New function.
942 (try_validate_parallel): Likewise.
943 (try_merge_compare): Likewise.
944 (try_eliminate_compare): Call the above when no previous clobber
946 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
949 2017-10-14 Jakub Jelinek <jakub@redhat.com>
953 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
954 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
958 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
959 to be any operand_equal_p operands. For & (B - 1) require
960 B to be power of 2. Recognize
961 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
963 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
966 * optabs.c (expand_memory_blockage): Fix call of
967 targetm.have_memory_blockage.
969 2017-10-14 Jakub Jelinek <jakub@redhat.com>
972 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
973 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
974 objects to extra_objs instead of overwriting it.
976 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
978 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
979 Use any_fp_register_operand as operand[3] predicate. Simplify
980 equality test for operands[2] and operands[4] memory location.
981 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
982 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
983 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
984 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
985 any_fp_register_operand as operand[1] predicate. Simplify
986 equality test for operands[0] and operands[3] memory location.
987 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
988 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
989 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
991 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
993 * target-insns.def: Add memory_blockage.
994 * optabs.c (expand_memory_blockage): New function.
995 (expand_asm_memory_barrier): Rename ...
996 (expand_asm_memory_blockage): ... to this.
997 (expand_mem_thread_fence): Call expand_memory_blockage
998 instead of expand_asm_memory_barrier.
999 (expand_mem_singnal_fence): Ditto.
1000 (expand_atomic_load): Ditto.
1001 (expand_atomic_store): Ditto.
1002 * doc/md.texi (Standard Pattern Names For Generation):
1003 Document memory_blockage instruction pattern.
1005 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
1007 * config/rl78/rl78.c (rl78_emit_libcall): New function.
1008 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
1009 * config/rl78/rl78.md: New define_expand "adddi3".
1011 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
1013 * cfghooks.c (verify_flow_info): Disable check that all probabilities
1016 2017-10-13 Jeff Law <law@redhat.com>
1018 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
1020 2017-10-13 Jakub Jelinek <jakub@redhat.com>
1023 * internal-fn.c (expand_mul_overflow): If both operands have
1024 the same highpart of -1 or 0 and the topmost bit of lowpart
1025 is different, overflow is if res <= 0 rather than res < 0.
1027 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
1029 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
1030 TARGET_P9_VECTOR code for unaligned_load case.
1032 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
1034 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
1036 2017-10-13 Nathan Sidwell <nathan@acm.org>
1038 * tree-core.h (tree_contains_struct): Make bool.
1039 * tree.c (tree_contains_struct): Likewise.
1040 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
1041 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
1042 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
1043 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
1044 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
1046 2017-10-13 Richard Biener <rguenther@suse.de>
1048 * graphite-isl-ast-to-gimple.c
1049 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
1050 parameters and dominance check.
1051 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
1052 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
1053 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
1054 Do not update SSA form here or do intermediate IL verification.
1055 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
1056 (graphite_initialize): Remove check on the number of loops in
1057 the function and inline into graphite_transform_loops.
1058 (graphite_finalize): Inline into graphite_transform_loops.
1059 (graphite_transform_loops): Perform SSA update and IL verification
1061 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
1063 2017-10-13 Richard Biener <rguenther@suse.de>
1065 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
1066 graphite_expression_type_precision): Avoid global constructor
1068 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
1069 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
1070 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
1071 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
1072 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
1073 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
1075 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
1078 * config/i386/i386.h (ix86_red_zone_size): New.
1079 * config/i386/i386.md (push peephole2s): Replace
1080 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
1082 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1083 Alan Hayward <alan.hayward@arm.com>
1084 David Sherwood <david.sherwood@arm.com>
1086 * combine.c (can_change_dest_mode): Reject changes in
1087 REGMODE_NATURAL_SIZE.
1089 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1090 Alan Hayward <alan.hayward@arm.com>
1091 David Sherwood <david.sherwood@arm.com>
1093 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
1094 (expand_debug_source_expr): Likewise.
1095 * combine.c (combine_simplify_rtx): Likewise.
1096 * cse.c (fold_rtx): Likewise.
1097 * fwprop.c (canonicalize_address): Likewise.
1098 * targhooks.c (default_shift_truncation_mask): Likewise.
1100 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1101 Alan Hayward <alan.hayward@arm.com>
1102 David Sherwood <david.sherwood@arm.com>
1104 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
1105 (widened_mode): Likewise.
1106 (expand_unop): Likewise.
1107 * ree.c (transform_ifelse): Likewise.
1108 (merge_def_and_ext): Likewise.
1109 (combine_reaching_defs): Likewise.
1110 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1112 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1113 Alan Hayward <alan.hayward@arm.com>
1114 David Sherwood <david.sherwood@arm.com>
1116 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
1117 * combine.c (gen_lowpart_for_combine): Likewise.
1118 * dwarf2out.c (rtl_for_decl_location): Likewise.
1119 * final.c (alter_subreg): Likewise.
1120 * rtlhooks.c (gen_lowpart_general): Likewise.
1121 (gen_lowpart_if_possible): Likewise.
1123 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1124 Alan Hayward <alan.hayward@arm.com>
1125 David Sherwood <david.sherwood@arm.com>
1127 * calls.c (expand_call): Use subreg_lowpart_offset.
1128 * cse.c (cse_insn): Likewise.
1129 * regcprop.c (copy_value): Likewise.
1130 (copyprop_hardreg_forward_1): Likewise.
1132 2017-10-13 Jakub Jelinek <jakub@redhat.com>
1135 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
1136 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
1137 =Q constraints to +Q and into insn condition add check
1138 that operands[0] and operands[1] are equal.
1139 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
1140 =Q constraints to +Q and into insn condition add check
1141 that operands[0] is equal to either operands[1] or operands[2].
1144 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
1145 instead of handling MINUS_EXPR twice (once for each argument),
1146 canonicalize operand order and handle just once, use rtype where
1147 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
1150 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
1151 any values of __C while still being pattern recognizable as a simple
1154 2017-10-13 Richard Biener <rguenther@suse.de>
1156 PR tree-optimization/82451
1158 2017-10-02 Richard Biener <rguenther@suse.de>
1160 PR tree-optimization/82355
1161 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
1162 a mapping for the enclosing loop but avoid generating one for
1164 (copy_bb_and_scalar_dependences): Remove premature codegen
1165 error on PHIs in blocks duplicated into multiple places.
1166 * graphite-scop-detection.c
1167 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
1168 in the region use it as loop and nest to analyze the DR in.
1169 (try_generate_gimple_bb): Likewise.
1170 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
1171 (add_loop_constraints): For blocks in a loop not in the region
1172 create a dimension with a single iteration.
1173 * sese.h (gbb_loop_at_index): Remove assert.
1175 * cfgloop.c (loop_preheader_edge): For the loop tree root
1176 return the single successor of the entry block.
1177 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1178 Reset the SCEV hashtable and niters.
1179 * graphite-scop-detection.c
1180 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
1181 assert that we only have POLYNOMIAL_CHREC that vary in loops
1182 contained in the region.
1183 (scop_detection::graphite_can_represent_expr): Adjust.
1184 (scop_detection::stmt_has_simple_data_refs_p): For loops
1185 not in the region set loop to NULL. The nest is now the
1186 entry edge to the region.
1187 (try_generate_gimple_bb): Likewise.
1188 * sese.c (scalar_evolution_in_region): Adjust for
1189 instantiate_scev change.
1190 * tree-data-ref.h (graphite_find_data_references_in_stmt):
1191 Make nest parameter the edge into the region.
1192 (create_data_ref): Likewise.
1193 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
1194 entry edge into a region and adjust instantiate_scev calls.
1195 (create_data_ref): Likewise.
1196 (graphite_find_data_references_in_stmt): Likewise.
1197 (find_data_references_in_stmt): Pass the loop preheader edge
1198 from the nest argument.
1199 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
1200 parameter the edge into the region.
1201 (instantiate_parameters): Use the loop preheader edge as entry.
1202 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
1204 (get_instantiated_value_entry): Make instantiate_below parameter
1205 the edge into the region.
1206 (instantiate_scev_name): Likewise. Adjust dominance checks,
1207 when we cannot use loop-based instantiation instantiate by
1208 walking use-def chains.
1209 (instantiate_scev_poly): Adjust.
1210 (instantiate_scev_binary): Likewise.
1211 (instantiate_scev_convert): Likewise.
1212 (instantiate_scev_not): Likewise.
1213 (instantiate_array_ref): Remove.
1214 (instantiate_scev_3): Likewise.
1215 (instantiate_scev_2): Likewise.
1216 (instantiate_scev_1): Likewise.
1217 (instantiate_scev_r): Do not blindly handle N-operand trees.
1218 Do not instantiate array-refs. Handle all constants and invariants.
1219 (instantiate_scev): Make instantiate_below parameter
1220 the edge into the region.
1221 (resolve_mixers): Use the loop preheader edge for the region
1222 parameter to instantiate_scev_r.
1223 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
1225 2017-10-13 Richard Biener <rguenther@suse.de>
1227 PR tree-optimization/82525
1228 * graphite-isl-ast-to-gimple.c
1229 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
1231 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
1232 Fail code generation when we cannot represent the isl integer.
1233 (binary_op_to_tree): Elide modulo operations that are no-ops
1234 in the type we code generate. Remove now superfluous code
1237 2017-10-13 Richard Biener <rguenther@suse.de>
1239 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
1240 (scop_detection::harmful_loop_in_region): Remove premature
1241 IV type restriction.
1242 (scop_detection::graphite_can_represent_scev): We can handle
1243 pointer IVs just fine.
1245 2017-10-13 Alan Modra <amodra@gmail.com>
1247 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
1248 "Clobbers and Scratch Registers". Add paragraph on
1249 alternative to clobbers for scratch registers and OpenBLAS
1252 2017-10-13 Alan Modra <amodra@gmail.com>
1254 * doc/extend.texi (Clobbers): Correct vax example. Delete old
1255 example of a memory input for a string of known length. Move
1256 commentary out of table. Add a number of new examples
1257 covering array memory inputs.
1259 2017-10-12 Martin Liska <mliska@suse.cz>
1261 PR tree-optimization/82493
1262 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
1263 (test_range_functions): New function.
1264 (sbitmap_c_tests): Likewise.
1265 * selftest-run-tests.c (selftest::run_tests): Run new tests.
1266 * selftest.h (sbitmap_c_tests): New function.
1268 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
1270 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1272 * config/rs6000/amo.h: Fix spacing issue.
1274 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1277 * config/i386/i386.md (*ashl<mode>3_mask_1,
1278 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
1279 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
1282 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
1284 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
1285 (profile_probability): Set max_probability
1286 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
1288 * profile-count.c (profile_probability::differs_from_p): Do not
1289 rely on max_probaiblity == 10000
1291 2017-10-12 Jeff Law <law@redhat.com>
1293 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
1296 2017-10-12 Martin Sebor <msebor@redhat.com>
1300 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
1301 (handle_alias_pairs): Call it.
1302 * common.opt (-Wattribute-alias): New option.
1303 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
1304 * doc/invoke.texi (-Wattribute-alias): Document.
1306 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
1309 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
1311 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1313 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1314 (make_hard_regno_born, make_hard_regno_dead): Update
1315 bb_killed_pseudos and bb_gen_pseudos.
1317 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
1319 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
1321 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
1323 * config/alpha/alpha.c (alpha_split_conditional_move):
1324 Use std::swap instead of manually swapping.
1325 (alpha_stdarg_optimize_hook): Ditto.
1326 (alpha_canonicalize_comparison): Ditto.
1328 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1330 * tree-loop-distribution.c (struct builtin_info): New struct.
1331 (struct partition): Refactor fields into struct builtin_info.
1332 (partition_free): Free struct builtin_info.
1333 (build_size_arg_loc, build_addr_arg_loc): Delete.
1334 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
1335 information from struct builtin_info.
1336 (find_single_drs): New function refactored from classify_partition.
1337 Also moved builtin validity checks to this function.
1338 (compute_access_range, alloc_builtin): New functions.
1339 (classify_builtin_st, classify_builtin_ldst): New functions.
1340 (classify_partition): Refactor code into functions find_single_drs,
1341 classify_builtin_st and classify_builtin_ldst.
1342 (distribute_loop): Don't do runtime alias check when distributing
1344 (find_seed_stmts_for_distribution): New function.
1345 (pass_loop_distribution::execute): Refactor code finding seed
1346 stmts into above function. Support distribution for the innermost
1347 two-level loop nest. Adjust dump information.
1349 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1351 * tree-loop-distribution.c: Adjust the general comment.
1352 (NUM_PARTITION_THRESHOLD): New macro.
1353 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
1354 (classify_partition): Skip builtin pattern of loop nest's inner loop.
1355 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
1356 in call to build_partition_graph.
1357 (finalize_partitions): New parameter. Make loop distribution more
1358 conservative by fusing more partitions.
1359 (distribute_loop): Don't do runtime alias check in case of loop nest
1361 (find_seed_stmts_for_distribution): New function.
1362 (prepare_perfect_loop_nest): New function.
1363 (pass_loop_distribution::execute): Refactor code finding seed stmts
1364 and loop nest into above functions. Support loop nest distribution.
1365 Adjust dump information accordingly.
1367 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1369 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
1370 and set PTYPE_SEQUENTIAL for merged partition.
1372 2017-10-12 Richard Biener <rguenther@suse.de>
1374 PR tree-optimization/69728
1376 2017-09-19 Richard Biener <rguenther@suse.de>
1378 PR tree-optimization/69728
1379 * graphite-sese-to-poly.c (schedule_error): New global.
1380 (add_loop_schedule): Handle empty domain by failing the
1382 (build_original_schedule): Handle schedule_error.
1384 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
1385 domain by returning an unchanged schedule.
1387 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1389 * genrecog.c (validate_pattern): For VEC_SELECT verify that
1390 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
1392 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
1394 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
1397 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1400 * expr.c (store_field): Don't optimize away bitsize == 0 store
1401 from CALL_EXPR with addressable return type.
1403 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
1405 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
1406 * config/rs6000/rs6000.md (sel): Delete mode attribute.
1407 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
1408 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
1409 TARGET_ISEL instead of TARGET_ISEL<sel>.
1411 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
1413 * config/rs6000/rs6000.c
1414 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
1416 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
1418 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
1420 (reg_or_cint_operand): Fix comment.
1421 (reg_or_zero_operand): New predicate.
1422 * config/rs6000/rs6000-protos.h (output_isel): Delete.
1423 * config/rs6000/rs6000.c (output_isel): Delete.
1424 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
1425 instead of reg_or_cint_operand. Output instruction directly (not via
1427 (isel_unsigned_<mode>): Ditto.
1428 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
1429 gpc_reg_operand. Add an instruction alternative for this. Output
1430 instruction directly.
1431 (*isel_reversed_unsigned_<mode>): Ditto.
1433 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1435 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
1436 (TARGET_CANONICALIZE_COMPARISON): Define.
1438 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
1441 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
1442 Check whether the dest is REG before adding REG_EQUIV note.
1444 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
1447 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1449 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1450 (make_hard_regno_born, make_hard_regno_dead): Update
1451 bb_killed_pseudos and bb_gen_pseudos.
1453 2017-10-11 Nathan Sidwell <nathan@acm.org>
1455 * incpath.h (enum incpath_kind): Name enum, prefix values.
1456 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
1457 * incpath.c (heads, tails): Use INC_MAX.
1458 (add_env_var_paths, add_standard_paths): Use incpath_kind.
1459 (merge_include_chains, split_quote_chain,
1460 register_include_chains): Update incpath_kind names.
1461 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
1462 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
1464 (add_framework_path, darwin_register_objc_includes): Likewise.
1465 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
1467 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1469 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
1470 Do not use float_operator operator predicate.
1471 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
1472 * config/i386/predicates.md (float_operator): Remove predicate.
1474 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1476 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
1477 (*jccxf_i387): Ditto.
1478 (*jcc<mode>_i387): Ditto.
1479 (*jccu<mode>_i387): Ditto.
1480 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
1481 (*jcc_*_i387 splitters): Remove.
1482 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
1483 * config/i386/i386.c (ix86_split_fp_branch): Remove.
1484 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
1487 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
1489 * profile-count.h (slow_safe_scale_64bit): New function.
1490 (safe_scale_64bit): New inline.
1491 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
1492 * profile-count.c: Include wide-int.h
1493 (slow_safe_scale_64bit): New.
1495 2017-10-11 Nathan Sidwell <nathan@acm.org>
1497 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
1498 HAS_DECL_ASSEMBLER_NAME_P.
1499 * gimple-expr.c (gimple_decl_printable_name: Check
1500 HAS_DECL_ASSEMBLER_NAME_P too.
1501 * ipa-utils.h (type_in_anonymous_namespace_p): Check
1502 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
1503 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
1504 * passes.c (rest_of_decl_compilation): Check
1505 HAS_DECL_ASSEMBLER_NAME_P too.
1506 * recog.c (verify_changes): Likewise.
1507 * tree-pretty-print.c (dump_decl_name): Likewise.
1508 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
1510 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
1511 (SET_DECL_ASSEMBLER_NAME): Use it.
1512 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
1513 (COPY_DECL_ASSEMBLER_NAME): Likewise.
1514 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
1516 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
1518 * config.gcc (i386, x86_64): Add extra objects.
1519 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
1520 (ix86_min_insn_size): Declare.
1521 (ix86_issue_rate): Declare.
1522 (ix86_adjust_cost): Declare.
1523 (ia32_multipass_dfa_lookahead): Declare.
1524 (ix86_macro_fusion_p): Declare.
1525 (ix86_macro_fusion_pair_p): Declare.
1526 (ix86_bd_has_dispatch): Declare.
1527 (ix86_bd_do_dispatch): Declare.
1528 (ix86_core2i7_init_hooks): Declare.
1529 (ix86_atom_sched_reorder): Declare.
1530 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
1531 (COSTS_N_BYTES): Move to x86-tune-costs.h.
1532 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
1533 (rip_relative_addr_p): Rename to ...
1534 (ix86_rip_relative_addr_p): ... this one; export.
1535 (memory_address_length): Update.
1536 (ix86_issue_rate): Move to x86-tune-sched.c.
1537 (ix86_flags_dependent): Move to x86-tune-sched.c.
1538 (ix86_agi_dependent): Move to x86-tune-sched.c.
1539 (exact_dependency_1): Move to x86-tune-sched.c.
1540 (exact_store_load_dependency): Move to x86-tune-sched.c.
1541 (ix86_adjust_cost): Move to x86-tune-sched.c.
1542 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
1543 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
1544 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
1545 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
1546 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
1547 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
1548 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
1549 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
1550 (min_insn_size): Rename to ...
1551 (ix86_min_insn_size): ... this one; export.
1552 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
1553 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
1554 (core2i7_first_cycle_multipass_backtrack): Move to
1555 x86-tune-sched-core.c.
1556 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
1557 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
1558 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
1559 (ix86_avoid_jump_mispredicts): Update.
1560 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
1561 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
1562 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
1563 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
1564 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
1565 (MAX_INSN): Move to ix86-tune-sched-bd.c.
1566 (MAX_IMM): Move to ix86-tune-sched-bd.c.
1567 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
1568 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
1569 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
1570 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
1571 (MAX_STORE): Move to ix86-tune-sched-bd.c.
1572 (BIG): Move to ix86-tune-sched-bd.c.
1573 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
1574 (enum insn_path): Move to ix86-tune-sched-bd.c.
1575 (get_mem_group): Move to ix86-tune-sched-bd.c.
1576 (is_cmp): Move to ix86-tune-sched-bd.c.
1577 (dispatch_violation): Move to ix86-tune-sched-bd.c.
1578 (is_branch): Move to ix86-tune-sched-bd.c.
1579 (is_prefetch): Move to ix86-tune-sched-bd.c.
1580 (init_window): Move to ix86-tune-sched-bd.c.
1581 (allocate_window): Move to ix86-tune-sched-bd.c.
1582 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
1583 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
1584 (process_end_window): Move to ix86-tune-sched-bd.c.
1585 (allocate_next_window): Move to ix86-tune-sched-bd.c.
1586 (find_constant): Move to ix86-tune-sched-bd.c.
1587 (get_num_immediates): Move to ix86-tune-sched-bd.c.
1588 (has_immediate): Move to ix86-tune-sched-bd.c.
1589 (get_insn_path): Move to ix86-tune-sched-bd.c.
1590 (get_insn_group): Move to ix86-tune-sched-bd.c.
1591 (count_num_restricted): Move to ix86-tune-sched-bd.c.
1592 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
1593 (add_insn_window): Move to ix86-tune-sched-bd.c.
1594 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
1595 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
1596 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
1597 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
1598 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
1599 (do_dispatch): Move to ix86-tune-sched-bd.c.
1600 (has_dispatch): Move to ix86-tune-sched-bd.c.
1601 * config/i386/t-i386: Add new object files.
1602 * config/i386/x86-tune-costs.h: New file.
1603 * config/i386/x86-tune-sched-atom.c: New file.
1604 * config/i386/x86-tune-sched-bd.c: New file.
1605 * config/i386/x86-tune-sched-core.c: New file.
1606 * config/i386/x86-tune-sched.c: New file.
1608 2017-10-11 Liu Hao <lh_mouse@126.com>
1610 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
1611 the generic version below instead.
1612 (should_colorize): Recognize Windows consoles as terminals
1614 * pretty-print.c [__MINGW32__] (write_all): New function.
1615 [__MINGW32__] (find_esc_head): Likewise.
1616 [__MINGW32__] (find_esc_terminator): Likewise.
1617 [__MINGW32__] (eat_esc_sequence): Likewise.
1618 [__MINGW32__] (mingw_ansi_fputs): New function that handles
1620 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
1623 2017-10-11 Richard Biener <rguenther@suse.de>
1625 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
1626 Properly call analyze_scalar_evolution with the loop of the stmt.
1628 2017-10-11 Richard Biener <rguenther@suse.de>
1630 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
1631 * tree-core.h (tree_base): Add chrec_var union member.
1632 * tree.h (CHREC_VAR): Remove.
1633 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
1634 * tree-chrec.h (build_polynomial_chrec): Adjust.
1635 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
1636 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
1638 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
1640 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
1641 * match.pd: ... here.
1642 ((T) X == (T) Y): Relax condition.
1644 2017-10-11 Bin Cheng <bin.cheng@arm.com>
1646 PR tree-optimization/82472
1647 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
1649 (break_alias_scc_partitions): Update postorder number.
1651 2017-10-11 Martin Liska <mliska@suse.cz>
1654 * opts.c (parse_no_sanitize_attribute): Do not use error_value
1656 * opts.h (parse_no_sanitize_attribute): Remove last argument.
1658 2017-10-11 Martin Liska <mliska@suse.cz>
1660 * print-rtl.c (print_insn): Move declaration of idbuf
1661 to same scope as name.
1663 2017-10-11 Martin Liska <mliska@suse.cz>
1668 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
1671 2017-10-11 Martin Liska <mliska@suse.cz>
1674 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
1677 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
1679 * config.gcc (default_gnu_indirect_function): Default to yes for
1680 arm*-*-linux* with glibc.
1682 2017-10-11 Richard Biener <rguenther@suse.de>
1684 * tree-scalar-evolution.c (get_scalar_evolution): Handle
1685 default-defs and types we do not want to analyze.
1686 (interpret_loop_phi): Replace unreachable code with an assert.
1687 (compute_scalar_evolution_in_loop): Remove and inline ...
1688 (analyze_scalar_evolution_1): ... here, replacing condition with
1689 what makes the intent clearer. Remove handling of cases
1690 get_scalar_evolution now handles.
1692 2017-10-10 Jim Wilson <wilson@tuliptree.org>
1694 PR rtl-optimization/81434
1695 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
1696 comment for main loop. In sched_group_found if, also add checks for
1697 pass and min_cost_group.
1699 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
1701 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
1702 (rs6000_insn_cost): New function.
1703 * config/rs6000/rs6000.md (cost): New attribute.
1705 2017-10-10 Jakub Jelinek <jakub@redhat.com>
1706 H.J. Lu <hongjiu.lu@intel.com>
1710 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
1711 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
1712 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
1713 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
1714 to OPTION_MASK_ISA_AVX512VL - builtins that have both
1715 OPTION_MASK_ISA_MMX and some other bit set require both
1716 mmx and the ISAs without the mmx bit.
1717 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
1718 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
1719 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
1720 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
1721 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
1722 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
1723 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
1724 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
1725 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
1726 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
1727 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
1728 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
1729 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
1730 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
1731 Add OPTION_MASK_ISA_MMX.
1733 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
1735 * config.gcc (armv7*-*-freebsd*): New target.
1736 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
1738 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
1740 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
1741 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
1744 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
1746 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
1747 a template parameter.
1748 (WIDE_INT_REF_FOR): Update accordingly.
1749 * tree.h (wi::int_traits <const_tree>): Delete.
1750 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
1751 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
1752 (wi::tree_to_wide_ref): New typedef.
1753 (wi::to_wide): New function.
1754 * calls.c (get_size_range): Use wi::to_wide when operating on
1756 * cgraph.c (cgraph_node::create_thunk): Likewise.
1757 * config/i386/i386.c (ix86_data_alignment): Likewise.
1758 (ix86_local_alignment): Likewise.
1759 * dbxout.c (stabstr_O): Likewise.
1760 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
1761 * expr.c (const_vector_from_tree): Likewise.
1762 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
1763 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
1764 (fold_negate_expr_1, int_const_binop_1, const_binop)
1765 (fold_convert_const_int_from_real, optimize_bit_field_compare)
1766 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
1767 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
1768 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
1769 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
1770 (fold_not_const, round_up_loc): Likewise.
1771 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
1772 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
1773 (alloca_call_type): Likewise.
1774 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
1775 * godump.c (go_output_typedef): Likewise.
1776 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
1777 * internal-fn.c (get_min_precision): Likewise.
1778 * ipa-cp.c (ipcp_store_vr_results): Likewise.
1779 * ipa-polymorphic-call.c
1780 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1781 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
1782 (ipa_modify_call_arguments): Likewise.
1783 * match.pd: Likewise.
1784 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
1785 * print-tree.c (print_node_brief, print_node): Likewise.
1786 * stmt.c (expand_case): Likewise.
1787 * stor-layout.c (layout_type): Likewise.
1788 * tree-affine.c (tree_to_aff_combination): Likewise.
1789 * tree-cfg.c (group_case_labels_stmt): Likewise.
1790 * tree-data-ref.c (dr_analyze_indices): Likewise.
1791 (prune_runtime_alias_test_list): Likewise.
1792 * tree-dump.c (dequeue_and_dump): Likewise.
1793 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
1794 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
1795 * tree-pretty-print.c (dump_generic_node): Likewise.
1796 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
1797 (simple_iv_with_niters): Likewise.
1798 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1799 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
1800 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
1801 * tree-ssa-loop-niter.c (split_to_var_and_offset)
1802 (refine_value_range_using_guard, number_of_iterations_ne_max)
1803 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
1804 (get_cst_init_from_scev, record_nonwrapping_iv)
1805 (scev_var_range_cant_overflow): Likewise.
1806 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
1807 * tree-ssa-pre.c (compute_avail): Likewise.
1808 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1809 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
1810 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
1811 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
1812 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
1813 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
1814 (dump_case_nodes, try_switch_expansion): Likewise.
1815 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
1816 (vect_do_peeling): Likewise.
1817 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
1818 * tree-vect-stmts.c (vectorizable_load): Likewise.
1819 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
1820 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
1821 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
1822 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
1823 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
1824 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
1825 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
1826 (evrp_dom_walker::before_dom_children): Likewise.
1827 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
1828 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
1829 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
1830 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
1831 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
1832 (get_range_pos_neg): Likewise.
1833 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
1834 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
1835 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
1836 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
1837 * config/avr/avr.c (avr_fold_builtin): Likewise.
1838 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
1839 * config/msp430/msp430.c (msp430_attr): Likewise.
1840 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
1841 * config/powerpcspe/powerpcspe-c.c
1842 (altivec_resolve_overloaded_builtin): Likewise.
1843 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
1844 (rs6000_expand_ternop_builtin): Likewise.
1845 * config/rs6000/rs6000-c.c
1846 (altivec_resolve_overloaded_builtin): Likewise.
1847 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
1848 (rs6000_expand_ternop_builtin): Likewise.
1849 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
1851 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1853 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
1854 when copying loop nest with only one inner loop.
1856 2017-10-10 Richard Biener <rguenther@suse.de>
1858 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
1859 blocks if SCEV is active.
1860 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
1862 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
1863 (scev_initialize): Assert we are not yet initialized.
1865 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1867 * tree-loop-distribution.c (generate_loops_for_partition): Remove
1868 inner loop's exit stmt by making it always exit the loop, otherwise
1869 we would generate an infinite empty loop.
1871 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1873 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
1874 renaming variables in new preheader if it's deleted.
1876 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1878 * tree-loop-distribution.c (struct partition): Remove unused field
1879 loops of the structure.
1880 (partition_alloc, partition_free): Ditto.
1881 (build_rdg_partition_for_vertex): Ditto.
1883 2017-10-09 Jeff Law <law@redhat.com>
1885 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
1886 return type to match prototype and documentation.
1888 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1890 * config/rs6000/rs6000.c (processor_costs): Move to ...
1891 * config/rs6000/rs6000.h: ... here.
1892 (rs6000_cost): Declare.
1894 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
1896 * except.c (setjmp_fn): New global variable.
1897 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
1898 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
1899 if DONT_USE_BUILTIN_SETJMP is defined.
1901 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1903 * target.def (insn_cost): New hook.
1904 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
1905 * doc/tm.texi: Regenerate.
1906 * rtlanal.c (insn_cost): Use the new hook.
1908 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1910 * combine.c (combine_validate_cost): Compute the new insn_cost,
1911 not just pattern_cost.
1912 (try_combine): Adjust comment.
1914 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1916 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
1918 * combine.c (uid_insn_cost): Adjust comment.
1919 (combine_validate_cost): Adjust comment. Use pattern_cost instead
1921 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
1922 * dse.c (find_shift_sequence): Ditto.
1923 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
1924 (bb_valid_for_noce_process_p): Use pattern_cost.
1925 * rtl.h (insn_rtx_cost): Delete.
1926 (pattern_cost): New prototype.
1927 (insn_cost): New prototype.
1928 * rtlanal.c (insn_rtx_cost): Rename to...
1929 (pattern_cost): ... this.
1932 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
1934 * config/i386/i386.md (*jcc_2): Remove insn pattern.
1935 (*jcc<mode>_0_r_i387): Ditto.
1936 (*jccxf_r_i387): Ditto.
1937 (*jcc<mode>_r_i387): Ditto.
1938 (*jccu<mode>_r_i387): Ditto.
1939 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
1940 (*jcc): Rename from *jcc_1.
1942 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1944 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
1945 deferred rescans after the lvx/stvx recombination pre-pass.
1947 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
1949 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
1950 memory operation instruction support.
1951 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
1952 (rs6000-ibm-aix[789]*): Likewise.
1953 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
1954 Document new functions.
1956 2017-10-09 Richard Biener <rguenther@suse.de>
1958 PR tree-optimization/82397
1959 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
1960 equality only for semantically equal trees.
1962 2017-10-09 Richard Biener <rguenther@suse.de>
1964 PR tree-optimization/82449
1965 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
1966 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
1967 allow constant addresses.
1968 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
1971 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1973 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
1976 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1979 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
1982 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1985 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
1987 2017-10-09 Jakub Jelinek <jakub@redhat.com>
1990 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
1991 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
1993 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
1995 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
1996 (WI_BINARY_PREDICATE_RESULT): Likewise.
1997 (wi::binary_traits::operator_result): New type.
1998 (wi::binary_traits::predicate_result): Likewise.
1999 (generic_wide_int::operator~, unary generic_wide_int::operator-)
2000 (generic_wide_int::operator==, generic_wide_int::operator!=)
2001 (generic_wide_int::operator&, generic_wide_int::and_not)
2002 (generic_wide_int::operator|, generic_wide_int::or_not)
2003 (generic_wide_int::operator^, generic_wide_int::operator+
2004 (binary generic_wide_int::operator-, generic_wide_int::operator*):
2006 (operator~, unary operator-, operator==, operator!=, operator&)
2007 (operator|, operator^, operator+, binary operator-, operator*): New
2009 * expr.c (get_inner_reference): Use wi::bit_and_not.
2010 * fold-const.c (fold_binary_loc): Likewise.
2011 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
2012 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
2013 (bit_value_binop): Likewise.
2014 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
2015 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
2016 (extract_range_from_binary_expr_1): Likewise.
2017 (masked_increment): Likewise.
2018 (simplify_bit_ops_using_ranges): Likewise.
2020 2017-10-09 Martin Jambor <mjambor@suse.cz>
2023 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
2024 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
2025 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
2027 (hsa_fixup_mov_insn_type): New function.
2028 (hsa_op_with_type::get_in_type): Use it.
2029 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
2030 immediates in an assert.
2031 (hsa_op_with_type::extend_int_to_32bit): New method.
2032 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
2033 types. Convert to dest type if necessary.
2034 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
2035 (reg_for_gimple_ssa): Pass false as min32int to
2036 hsa_type_for_scalar_tree_type.
2037 (gen_hsa_addr): Fixup type when creating addresable temporary.
2038 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
2039 (gen_hsa_unary_operation): Extend operands and convert to dest type if
2040 necessary. Call hsa_fixup_mov_insn_type.
2041 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
2042 extend operands and convert to dest type if necessary.
2043 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
2044 to dest type if necessary.
2045 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
2046 if conversion nt necessary and size matches.
2047 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
2048 to dest type if necessary.
2049 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
2050 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
2052 (gen_hsa_clrsb): Likewise.
2053 (gen_hsa_ffs): Likewise.
2054 (gen_hsa_divmod): Extend operands and convert to dest type if
2056 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
2058 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
2060 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
2061 Remove empty default arguments. Use a brace block as output
2063 (conditional return): Ditto.
2065 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
2066 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
2068 (group_ending_nop): Ditto.
2069 (doloop_end): Ditto.
2070 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
2071 (splitters for those): Ditto.
2073 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
2075 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
2076 a conditional jump (and the compare for it) so that pc_rtx is the
2078 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
2079 for the deleted and renamed ctr<mode>_internal[234] patterns.
2080 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
2081 Delete second conditional return pattern.
2082 (ctr<mode>_internal2): Delete this second bdnz pattern.
2083 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
2084 (ctr<mode>_internal4): Delete this second bdz pattern.
2086 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
2088 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
2089 (always_initialized_rtx_for_ssa_name_p): New predicate.
2090 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
2091 (finish_out_of_ssa): Free new field of SA.
2092 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
2093 * tree-ssa-coalesce.c: Include tree-ssa.h.
2094 (get_parm_default_def_partitions): Remove extern keyword.
2095 (get_undefined_value_partitions): New function.
2096 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
2097 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
2099 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
2101 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
2103 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
2105 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
2107 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
2108 for targets that preffer 128bit.
2110 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
2112 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
2114 2017-10-08 Olivier Hainque <hainque@adacore.com>
2116 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
2117 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
2119 (thumb_set_return_address): Likewise.
2121 2017-10-08 Olivier Hainque <hainque@adacore.com>
2123 * common/config/arm/arm-common.c (arm_except_unwind_info):
2124 Handle DWARF2_UNWIND_INFO.
2126 2017-10-07 Michael Collison <michael.collison@arm.com>
2128 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
2131 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
2133 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
2134 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
2135 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
2136 defined, force the creation of a new block for a dispatch label.
2138 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
2140 * invoke.texi (Wsuggest-attribute=cold): Document.
2141 * common.opt (Wsuggest-attribute=cold): New
2142 * ipa-pure-const.c (warn_function_cold): New function.
2143 * predict.c (compute_function_frequency): Use it.
2144 * predict.h (warn_function_cold): Declare.
2146 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
2148 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
2151 2017-10-06 Martin Liska <mliska@suse.cz>
2153 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
2154 keyword for member functions.
2155 (struct sanopt_tree_couple): New struct.
2156 (struct sanopt_tree_couple_hash): New function.
2157 (struct sanopt_ctx): Add new hash_map.
2158 (has_dominating_ubsan_ptr_check): New function.
2159 (record_ubsan_ptr_check_stmt): Likewise.
2160 (maybe_optimize_ubsan_ptr_ifn): Likewise.
2161 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
2162 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
2164 2017-10-06 Sudakshina Das <sudi.das@arm.com>
2167 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
2168 aarch64_simd_valid_immediate on CONST_VECTORs.
2169 (aarch64_reg_or_bic_imm): Likewise.
2171 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
2173 PR rtl-optimization/82396
2174 * haifa-sched.c (ready_sort_real): Disable qsort checking.
2176 2017-10-06 Sebastian Pop <sebpop@gmail.com>
2178 * graphite-dependences.c (scop_get_reads): Move code to...
2179 (scop_get_must_writes): Move code to...
2180 (scop_get_may_writes): Move code to...
2181 (scop_get_reads_and_writes): ... here.
2182 (scop_get_dependences): Call scop_get_reads_and_writes.
2184 2017-10-06 Jakub Jelinek <jakub@redhat.com>
2186 PR tree-optimization/82434
2187 * fold-const.h (can_native_encode_type_p,
2188 can_native_encode_string_p): Remove.
2189 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
2190 don't encode anything, just return what would be otherwise returned.
2191 (native_encode_fixed, native_encode_complex, native_encode_vector):
2193 (native_encode_string): Likewise. Inline by hand
2194 can_native_encode_string_p.
2195 (can_native_encode_type_p): Remove.
2196 (can_native_encode_string_p): Remove.
2197 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
2198 STRING_CSTs using can_native_encode_string_p, test all
2199 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
2200 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
2201 argument from native_encode_expr.
2202 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
2203 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
2206 2017-10-06 Richard Biener <rguenther@suse.de>
2208 PR tree-optimization/82397
2209 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
2210 operand_equal_p but rely on data_ref_compare_tree for detecting
2212 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
2213 to match up with dr_group_sort_cmp.
2215 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2218 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
2220 * config/s390/s390-builtin-types.def: Regenerate.
2222 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2225 * config/s390/s390-builtin-types.def: Regenerate.
2226 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
2227 Change flag from B_VXE to B_VX.
2228 (s390_vec_min_dbl): Remove B_VXE flag.
2230 2017-10-06 Richard Biener <rguenther@suse.de>
2232 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
2233 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
2234 translate_isl_ast_to_gimple::is_valid_rename,
2235 translate_isl_ast_to_gimple::get_rename,
2236 translate_isl_ast_to_gimple::get_def_bb_for_const,
2237 translate_isl_ast_to_gimple::get_new_name,
2238 translate_isl_ast_to_gimple::collect_all_ssa_names,
2239 translate_isl_ast_to_gimple::copy_loop_phi_args,
2240 translate_isl_ast_to_gimple::collect_all_ssa_names,
2241 translate_isl_ast_to_gimple::copy_loop_phi_args,
2242 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
2243 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
2244 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
2245 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
2246 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
2247 translate_isl_ast_to_gimple::copy_cond_phi_args,
2248 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
2249 translate_isl_ast_to_gimple::edge_for_new_close_phis,
2250 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
2251 translate_isl_ast_to_gimple::rename_uses,
2252 translate_isl_ast_to_gimple::rename_all_uses): Remove.
2253 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
2254 (set_rename_for_each_def): Likewise.
2255 (graphite_copy_stmts_from_block): Handle debug stmt resetting
2256 here. Handle rewriting SCEV analyzable uses here.
2257 (copy_bb_and_scalar_dependences): Generate code for PHI
2259 (graphite_regenerate_ast_isl): Adjust.
2260 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
2261 (add_write, add_read): New functions.
2262 (build_cross_bb_scalars_def): Use it and simplify.
2263 (build_cross_bb_scalars_use): Likewise.
2264 (graphite_find_cross_bb_scalar_vars): Inline into...
2265 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
2266 simulating out-of-SSA. Compute liveout and add dependencies.
2267 (build_scops): Force an empty entry block.
2268 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
2270 (sese_build_liveouts): Declare.
2271 (sese_trivially_empty_bb_p): Likewise.
2272 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
2273 compute liveout and debug_liveout.
2274 (sese_bad_liveouts_use): Remove.
2275 (sese_reset_debug_liveouts_bb): Likewise.
2276 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
2277 (sese_build_liveouts): Build liveout and debug_liveout and store
2279 (new_sese_info): Adjust.
2280 (free_sese_info): Likewise.
2281 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
2282 do not build liveout here.
2283 (move_sese_in_condition): Adjust region entry.
2284 (scev_analyzable_p): Match up with chrec_apply requirements.
2285 (sese_trivially_empty_bb_p): New.
2286 * tree-into-ssa.c (get_reaching_def): Properly support generating
2287 default-defs for incremental rewrite of anonymous names.
2289 2017-10-06 Richard Biener <rguenther@suse.de>
2291 * graphite-sese-to-poly.c (extract_affine): For casts increasing
2292 precision do not perform modulo reduction.
2294 2017-10-06 Richard Biener <rguenther@suse.de>
2296 PR tree-optimization/82436
2297 * tree-vect-slp.c (vect_supported_load_permutation_p): More
2298 conservatively choose the vectorization factor when checking
2299 whether we can perform the required load permutation.
2300 (vect_transform_slp_perm_load): Assert when we may not fail.
2302 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
2304 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
2305 message for incompatible -msdata=* and -mcall-* options.
2307 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2309 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
2310 rate for post-reload scheduling.
2312 2017-10-05 Tamar Christina <tamar.christina@arm.com>
2314 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
2316 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2318 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
2319 to improve monte carlo in scimark.
2321 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2323 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
2324 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
2325 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
2326 pentium4_cost, nocona_cost): Set reassociation width to 1.
2327 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
2328 width to 2 for fp operations and 1 otherwise.
2329 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
2331 (atom_cost): Set reassociation width to 2.
2332 (slm_cost, generic_cost): Set fp reassociation width
2333 to 2 and 1 otherwise.
2334 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
2335 (core_cost): Set fp reassociation width to 4 and vector to 2.
2336 (ix86_reassociation_width): Rewrite using cost table; special case
2337 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
2338 and TARGET_AVX128_OPTIMAL.
2339 * config/i386/i386.h (processor_costs): Add
2340 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
2341 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
2342 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
2343 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
2344 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
2345 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
2347 2017-10-05 Nathan Sidwell <nathan@acm.org>
2349 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
2351 2017-10-05 Tamar Christina <tamar.christina@arm.com>
2353 * config/arm/arm.c (arm_test_fpu_data): New.
2354 (arm_run_selftests): Call arm_test_fpu_data.
2356 2017-10-04 Nathan Sidwell <nathan@acm.org>
2358 * toplev.c (toplev::main): Remove excess parens on pretty_printer
2360 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
2362 2017-10-04 Sudakshina Das <sudi.das@arm.com>
2364 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
2365 check type for aarch64_simd_valid_immediate.
2366 (aarch64_output_simd_mov_immediate): Update prototype.
2367 (aarch64_simd_valid_immediate): Update prototype.
2368 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
2369 support for ORR-immediate.
2370 (and<mode>3): modified pattern to add support for BIC-immediate.
2371 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
2372 now checks for valid immediate for BIC and ORR based on new enum
2374 (aarch64_output_simd_mov_immediate): Function now used to output
2375 BIC/ORR imm as well based on new enum argument.
2376 * config/aarch64/constraints.md (Do): New vector immediate constraint.
2378 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
2379 (aarch64_reg_or_bic_imm): Likewise.
2381 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2383 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
2384 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
2386 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
2390 PR rtl-optimization/82396
2391 * haifa-sched.c (autopref_multipass_init): Simplify
2393 (autopref_rank_data): Simplify sort order.
2394 * sched-int.h (autopref_multipass_data_): Remove
2395 multi_mem_insn_p, min_offset and max_offset.
2397 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2399 * doc/sourcebuild.texi: Document vect_peeling_profitable.
2401 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2403 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
2406 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2408 * doc/sourcebuild.texi: Document vect_long_mult.
2410 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
2412 PR tree-optimization/82413
2413 * fold-const.c (build_range_check): Use widest_int when comparing
2414 the maximum ETYPE value with HIGH.
2416 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
2418 PR rtl-optimization/82396
2419 * haifa-sched.c (autopref_multipass_init): Simplify
2421 (autopref_rank_data): Simplify sort order.
2422 * sched-int.h (autopref_multipass_data_): Remove
2423 multi_mem_insn_p, min_offset and max_offset.
2425 2017-10-04 Jakub Jelinek <jakub@redhat.com>
2427 PR tree-optimization/82381
2428 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
2429 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
2432 PR tree-optimization/82374
2433 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
2434 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
2435 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
2436 current_function_decl to the new decl.
2438 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
2440 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
2441 helper macro for IEEE float128 hardware built-in functions.
2442 (SQRTF128_ODD): Add built-in functions with the round-to-odd
2444 (TRUNCF128_ODD): Likewise.
2445 (ADDF128_ODD): Likewise.
2446 (SUBF128_ODD): Likewise.
2447 (MULF128_ODD): Likewise.
2448 (DIVF128_ODD): Likewise.
2449 (FMAF128_ODD): Likewise.
2450 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
2451 UNSPEC_TRUNC_ROUND_TO_ODD.
2452 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
2453 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
2454 floating point round to odd instructions.
2455 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
2456 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
2457 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
2458 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
2459 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
2460 (trunc<mode>sf2_hw): Change the truncate with round to odd
2461 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
2462 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
2463 to odd hardware instructions.
2464 (sub<mode>3_odd): Likewise.
2465 (mul<mode>3_odd): Likewise.
2466 (div<mode>3_odd): Likewise.
2467 (sqrt<mode>2_odd): Likewise.
2468 (fma<mode>4_odd): Likewise.
2469 (fms<mode>4_odd): Likewise.
2470 (nfma<mode>4_odd): Likewise.
2471 (nfms<mode>4_odd): Likewise.
2472 (trunc<mode>df2_odd): Change the truncate with round to odd
2473 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
2475 * doc/extend.texi (PowerPC built-in functions): Update documentation
2476 for existing IEEE float128-bit built-in functions. Add built-in
2477 functions that generate the IEEE 128-bit floating point round to
2480 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
2482 PR rtl-optimization/77729
2483 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
2484 to (X&(C1&~C2))|C2 transformations.
2486 2017-10-03 Martin Jambor <mjambor@suse.cz>
2488 PR tree-optimization/82363
2489 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
2490 mismatch, mark lacc written regardless of racc.
2492 2017-10-03 Jakub Jelinek <jakub@redhat.com>
2494 PR tree-optimization/82381
2495 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
2496 stmt_to_insert nor wheather SSA_NAMEs are default defs.
2497 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
2498 fallthrough into reassoc_stmt_dominates_stmt_p.
2501 * combine.c (combine_instructions): Don't combine in unreachable
2504 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
2507 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
2508 function to not use the have_cpu variable. Do not set cpu_index,
2509 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
2511 (rs6000_valid_attribute_p): Remove duplicate initializations of
2512 old_optimize and func_optimize.
2513 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
2514 (rs6000_activate_target_options): Make global.
2515 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
2518 2017-10-02 Jakub Jelinek <jakub@redhat.com>
2520 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
2521 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
2522 Set *poffset to 0 and *psize and *pmax_size to -1 if
2523 *poffset + *psize overflows in HOST_WIDE_INT.
2525 PR tree-optimization/82387
2526 PR tree-optimization/82388
2527 PR tree-optimization/82389
2528 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
2529 instead of live_bytes non-NULL.
2531 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
2534 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
2537 2017-10-02 Richard Biener <rguenther@suse.de>
2539 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
2540 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
2541 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
2543 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
2545 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
2546 requested precision matches the type's.
2547 * calls.c (alloc_max_size): Calculate the new candidate size as
2548 a widest_int and use wi::to_widest when comparing it with the
2549 current candidate size.
2550 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
2551 zero rather than integer_zero_node.
2552 * match.pd: Check for a no-op conversion before using wi::add
2553 rather than after. Use tree_to_uhwi when summing small shift
2554 counts into an unsigned int.
2556 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
2557 Alan Hayward <alan.hayward@arm.com>
2558 David Sherwood <david.sherwood@arm.com>
2561 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
2562 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
2563 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
2564 POINTER_AND_FP_REGS.
2566 2017-10-02 Richard Biener <rguenther@suse.de>
2568 PR tree-optimization/82355
2569 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
2570 a mapping for the enclosing loop but avoid generating one for
2572 (copy_bb_and_scalar_dependences): Remove premature codegen
2573 error on PHIs in blocks duplicated into multiple places.
2574 * graphite-scop-detection.c
2575 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
2576 in the region use it as loop and nest to analyze the DR in.
2577 (try_generate_gimple_bb): Likewise.
2578 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
2579 (add_loop_constraints): For blocks in a loop not in the region
2580 create a dimension with a single iteration.
2581 * sese.h (gbb_loop_at_index): Remove assert.
2583 2017-10-01 Kevin Buettner <kevinb@redhat.com>
2585 * omp-expand.c (adjust_context_scope): New function.
2586 (expand_parallel_call): Call adjust_context_scope.
2588 2017-10-01 Jeff Law <law@redhat.com>
2590 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
2591 dom_opt_dom_walker class with direct access to private members.
2592 Add comments. Call test_for_singularity.
2593 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
2594 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
2595 m_dummy_cond anymore.
2596 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
2598 (pass_dominator:execute): Build the dummy_cond here and pass it
2599 to the dom_opt_dom_walker ctor.
2600 (test_for_singularity): New function.
2602 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
2603 Maya Rashish <coypu@sdf.org>
2605 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
2606 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
2607 (alpha*-*-netbsd*) Use nbsd_tm_file.
2608 (arm*-*-netbsdelf*) Likewise.
2609 (i[34567]86-*-netbsdelf*) Likewise.
2610 (x86_64-*-netbsd*) Likewise.
2611 (mips*-*-netbsd*) Likewise.
2612 (powerpc-*-netbsd*) Likewise.
2613 (sh*-*-netbsd*) Likewise.
2614 (sparc-*-netbsdelf*) Likewise.
2615 (sparc64-*-netbsd*) Likewise.
2616 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
2618 (vax-*-netbsdelf*) Likewise.
2619 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
2620 (UINT_FAST8_TYPE) Likewise.
2621 (INT_FAST16_TYPE) Check CHAR_FAST16.
2622 (UINT_FAST16_TYPE) Likewise.
2624 2017-09-30 Jakub Jelinek <jakub@redhat.com>
2627 * config/i386/i386.md
2628 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
2629 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
2630 *divmodsi4_zext_2): New define_insn_and_split.
2631 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
2632 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
2633 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
2634 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
2635 New define_insn_and_split.
2636 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
2637 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
2638 operands[1] having DImode when mode is SImode.
2640 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
2641 always SImode for DIV and MOD in REG_EQUAL notes.
2643 2017-09-29 Yury Gribov <tetra2005@gmail.com>
2646 * match.pd: Fix handling of NaNs in pattern.
2648 2017-09-29 Jeff Law <law@redhat.com>
2650 * sbitmap.c (bitmap_bit_in_range_p): New function.
2651 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
2652 * tree-ssa-dse.c (live_bytes_read): New function.
2653 (dse_classify_store): Ignore reads of dead bytes.
2655 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
2656 typos and whitespace errors.
2657 * config/i386/predicates.md (address_no_seg_operand): Likewise.
2658 * config/s390/s390.c (s390_emit_prologue): Likewise.
2660 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
2663 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
2664 with a symbol for LRA.
2666 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
2668 PR rtl-optimization/82338
2669 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
2671 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
2673 * genmodes.c (calc_wider_mode): Suppress qsort macro.
2674 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
2675 (qsort_chk): Declare.
2676 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
2677 (qsort_chk): New function.
2679 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2681 PR tree-optimization/82337
2682 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
2683 phi definition if the PHI result appears in an abnormal PHI.
2684 (find_basis_for_base_expr): Don't record a basis if the LHS of the
2685 basis appears in an abnormal PHI.
2687 2017-09-29 Richard Biener <rguenther@suse.de>
2689 * graphite-isl-ast-to-gimple.c
2690 (translate_isl_ast_to_gimple::set_codegen_error): New function.
2691 (binary_op_to_tree): Use it.
2692 (get_rename_from_scev): Likewise.
2693 (copy_loop_phi_nodes): Likewise.
2694 (copy_bb_and_scalar_dependences): Likewise.
2695 (translate_pending_phi_nodes): Likewise.
2697 2017-09-29 Jakub Jelinek <jakub@redhat.com>
2700 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
2701 for movabsq $(i32 << shift), r64.
2703 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
2705 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
2706 index when encoding %esp as %rsp to avoid 0x67 prefix.
2708 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
2710 * config/i386/i386.md (*movsf_internal, *movdf_internal):
2711 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
2713 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
2715 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
2716 Extensions with more than 16 double VFP registers.
2717 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
2718 to_clear_mask and all code related to it. Replace the remaining
2719 entry by a sbitmap and adapt code accordingly.
2721 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
2723 * brig-builtins.def: Change pure attributes to const.
2725 2017-09-28 Joseph Myers <joseph@codesourcery.com>
2727 * config.gcc (default_gnu_indirect_function): Default to yes for
2728 sparc*-*-linux* with glibc.
2730 2017-09-28 Joseph Myers <joseph@codesourcery.com>
2732 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
2733 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
2734 when creating .init_array and .fini_array sections with priority
2737 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
2740 * config/aarch64/aarch64.c
2741 (aarch64_builtin_support_vector_misalignment): Always return false
2742 when misalignment is unknown.
2744 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
2746 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
2747 this function to return false if the definition used by the swap
2748 instruction is artificial, or if the memory address from which the
2749 constant value is loaded is not represented by a base address held
2750 in a register or if the base address register is a frame or stack
2751 pointer. Additionally, return false if the base address of the
2752 loaded constant is a SYMBOL_REF but is not considered to be a
2754 (replace_swapped_load_constant): New function.
2755 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
2756 loaded constant vector with a load of a swapped constant vector.
2758 2017-09-27 Carl Love <cel@us.ibm.com>
2760 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
2761 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
2762 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
2765 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
2767 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
2768 first, always call autopref_rank_data otherwise.
2770 2017-09-27 Richard Biener <rguenther@suse.de>
2772 * graphite-scop-detection.c (find_scop_parameters): Move
2773 loop bound handling ...
2774 (gather_bbs::before_dom_children): ... here, avoiding the need
2775 to build scop_info->loop_nest.
2776 (record_loop_in_sese): Remove.
2777 * sese.h (sese_info_t::loop_nest): Remove.
2778 * sese.c (new_sese_info): Do not allocate loop_nest.
2779 (free_sese_info): Do not free loop_nest.
2781 2017-09-27 Jakub Jelinek <jakub@redhat.com>
2784 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
2785 lhs from calls if the lhs has addressable type.
2787 2017-09-27 Richard Biener <rguenther@suse.de>
2789 * graphite.h (scop::max_alias_set): New member.
2790 * graphite-scop-detection.c: Remove references to non-existing
2791 --param in comments.
2792 (build_alias_sets): Record the maximum alias set used for drs.
2793 (build_scops): Support zero as unlimited for
2794 --param graphite-max-arrays-per-scop.
2795 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
2797 (build_poly_sr_1): ... here. Compute alias set based on the
2798 maximum alias set used for drs rather than
2799 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
2801 2017-09-27 Richard Biener <rguenther@suse.de>
2803 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
2804 --param loop-block-tile-size=0 to disable tiling.
2806 2017-09-27 Richard Biener <rguenther@suse.de>
2808 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
2809 (graphite-max-nb-scop-params): Document special value zero.
2810 * domwalk.h (dom_walker::STOP): New symbolical constant.
2811 (dom_walker::dom_walker): Add optional parameter for bb to
2813 (dom_walker::~dom_walker): Declare.
2814 (dom_walker::before_dom_children): Document STOP return value.
2815 (dom_walker::m_user_bb_to_rpo): New member.
2816 (dom_walker::m_bb_to_rpo): Likewise.
2817 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
2818 mapping here if not provided by the user.
2819 (dom_walker::~dom_walker): Free bb to RPO mapping if not
2820 provided by the user.
2821 (dom_walker::STOP): Define.
2822 (dom_walker::walk): Do not compute bb to RPO mapping here.
2823 Support STOP return value from before_dom_children to stop
2825 * graphite-optimize-isl.c (optimize_isl): If the schedule
2826 is the same still generate code if -fgraphite-identity
2827 or -floop-parallelize-all are given.
2828 * graphite-scop-detection.c: Include cfganal.h.
2829 (gather_bbs::gather_bbs): Get and pass through bb to RPO
2831 (gather_bbs::before_dom_children): Return STOP for BBs
2833 (build_scops): Compute bb to RPO mapping and pass it to
2834 the domwalk. Treat --param graphite-max-nb-scop-params=0
2835 as not limiting the number of params.
2836 * graphite.c (graphite_initialize): Remove limit on the
2837 number of basic-blocks in a function.
2838 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
2839 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
2840 default value of 10.
2842 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2844 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
2845 Adjust code to eliminate needing to do the shift right 32-bits
2846 operation after XSCVDPSPN.
2848 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2850 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
2851 ((X / Y) != 0 -> X >= Y): Likewise.
2853 2017-09-26 Carl Love <cel@us.ibm.com>
2855 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
2856 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
2857 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
2858 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
2859 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
2860 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
2861 definitions and overloading.
2862 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
2863 statement for P9V_BUILTIN_XST_LEN_R.
2864 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
2865 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
2866 define_expand and define_insn for the instructions and builtins.
2867 * doc/extend.texi: Update the built-in documentation file for the new
2869 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
2870 define_insn for the instructions
2872 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
2875 * gcc/config/netbsd-protos.h: New file.
2876 * gcc/config/netbsd.c: New file.
2877 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
2878 * gcc/config/t-netbsd: New file.
2879 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
2880 (tmake_file) Add t-netbsd.
2881 (extra_objs) Add netbsd.o.
2883 2017-09-26 Janus Weil <janus@gcc.gnu.org>
2887 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
2889 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2891 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
2892 sign extension from a vector register to a GPR by doing a 32-bit
2893 direct move and then an EXTSW.
2894 (extendsi<mode>2 splitter): Likewise.
2895 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
2896 right or vector extract after doing XSCVDPSPN. Use
2897 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
2899 (movdi_from_sf_zero_ext): Likewise.
2900 (reload_gpr_from_vsxsf): Likewise.
2901 (p8_mfvsrd_4_disf): Delete, no longer used.
2902 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
2903 then needing to move the SFmode to a GPR to use the XSCVDPSP
2904 instruction instead of FRSP and XSCVDPSPN.
2905 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
2906 it is adjacent to the other XSCVSPDP insns.
2907 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
2908 SFmode to be in traditional Altivec registers.
2909 (vsx_xscvdpspn): Eliminate useless alternative constraint.
2910 (vsx_xscvspdpn): Likewise.
2911 (vsx_xscvspdpn_scalar): Likewise.
2913 2017-09-26 Martin Jambor <mjambor@suse.cz>
2915 * tree-sra.c (compare_access_positions): Put integral types first,
2916 stabilize sorting of integral types, remove conditions putting
2917 non-full-precision integers last.
2918 (sort_and_splice_var_accesses): Disable scalarization if a
2919 non-integert would be represented by a non-full-precision integer.
2921 2017-09-26 Joseph Myers <joseph@codesourcery.com>
2923 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
2924 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
2925 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
2926 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
2927 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
2928 conditionals inside the function instead of around it. Call
2929 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
2930 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
2932 2017-09-26 Richard Biener <rguenther@suse.de>
2934 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
2936 (scop_detection::build_scop_breadth): ... this. Removed.
2937 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
2938 (scop_detection::harmful_stmt_in_bb): Likewise.
2939 (scop_detection::graphite_can_represent_stmt): Likewise.
2940 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
2941 (scop_detection::can_represent_loop): Remove recursion, fold in ...
2942 (scop_detection::can_represent_loop_1): ... this. Removed.
2943 (scop_detection::harmful_loop_in_region): Simplify after inlining
2944 the above and remove more quadraticness.
2945 (build_scops): Adjust.
2946 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
2949 2017-09-26 Jakub Jelinek <jakub@redhat.com>
2952 * config/i386/i386.c (ix86_print_operand_address_as): Only test
2953 REGNO (base) == SP_REG if base is a REG.
2956 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
2957 if it is different SSA_NAME.
2958 (optimize_range_tests_cmp_bitwise): New function.
2959 (optimize_range_tests): Call it.
2961 2017-09-26 Richard Biener <rguenther@suse.de>
2963 PR tree-optimization/82321
2964 * graphite.c (canonicalize_loop_closed_ssa): Properly check
2965 for the def being inside the loop.
2967 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2969 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
2971 * config/s390/s390-builtins.def: Fix constraint on op4.
2973 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2975 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
2976 independent expanders.
2977 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
2978 ("vec_ordered", "vec_unordered"): New expanders.
2980 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2982 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
2985 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2987 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
2988 vec_unpacks_lo_v16qi.
2989 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
2991 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2993 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
2994 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
2995 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
2997 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2999 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
3001 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
3003 ("*vec_slb<mode>"): New insn pattern.
3004 ("vec_shr_<mode>"): New expander.
3005 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
3006 and force the shift count operand to V16QImode.
3007 ("vec_srb<mode>"): Set shift count mode to V16QI.
3009 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3011 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
3012 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
3013 ("vec_widen_smult_hi_<mode>"): New expander definitions.
3015 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
3018 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
3020 2017-09-26 Richard Biener <rguenther@suse.de>
3022 PR tree-optimization/82320
3023 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
3026 2017-09-25 Jeff Law <law@redhat.com>
3028 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
3029 prototype for new argument.
3030 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
3031 mostly extracted from rs6000_emit_allocate_stack.
3032 (rs6000_emit_probe_stack_range_stack_clash): New function.
3033 (rs6000_emit_allocate_stack): Call
3034 rs6000_emit_probe_stack_range_stack_clash as needed.
3035 (rs6000_emit_probe_stack_range): Add additional argument
3036 to call to gen_probe_stack_range{si,di}.
3037 (output_probe_stack_range): New.
3038 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
3039 (output_probe_stack_range_stack_clash): New.
3040 (rs6000_emit_prologue): Emit notes into dump file as requested.
3041 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
3042 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
3043 Add additional operand and pass it to output_probe_stack_range.
3045 2017-09-25 Bin Cheng <bin.cheng@arm.com>
3047 PR tree-optimization/82163
3048 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
3049 (checking_verify_loop_closed_ssa): New parameter.
3050 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
3051 (check_loop_closed_ssa_stmt): Delete.
3052 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
3053 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
3054 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
3057 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
3059 * brig-builtins.def: Treat HSAIL barrier builtins as
3060 setjmp/longjump style functions.
3062 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
3064 * target.def (constant_alignment): New hook.
3065 * defaults.h (CONSTANT_ALIGNMENT): Delete.
3066 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
3067 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
3068 * doc/tm.texi: Regenerate.
3069 * targhooks.h (default_constant_alignment): Declare.
3070 (constant_alignment_word_strings): Likewise.
3071 * targhooks.c (default_constant_alignment): New function.
3072 (constant_alignment_word_strings): Likewise.
3073 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
3074 instead of CONSTANT_ALIGNMENT.
3075 * varasm.c (align_variable, get_variable_align, build_constant_desc)
3076 (force_const_mem): Likewise.
3077 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
3078 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
3079 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
3080 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3081 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
3083 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
3084 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3085 constant_alignment_word_strings.
3086 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
3087 (CONSTANT_ALIGNMENT): Likewise.
3088 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3089 (arm_constant_alignment): New function.
3090 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
3091 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3092 constant_alignment_word_strings.
3093 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
3094 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3095 constant_alignment_word_strings.
3096 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
3097 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3098 (cris_constant_alignment): New function.
3099 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
3100 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3101 (epiphany_constant_alignment): New function.
3102 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
3103 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3104 constant_alignment_word_strings.
3105 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
3106 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3107 constant_alignment_word_strings.
3108 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
3109 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3110 constant_alignment_word_strings.
3111 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
3112 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
3113 * config/i386/i386.c (ix86_constant_alignment): Make static.
3114 Use the same interface as the target hook.
3115 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3116 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
3117 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3118 constant_alignment_word_strings.
3119 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
3120 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
3121 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3122 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
3123 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3124 constant_alignment_word_strings.
3125 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
3126 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3127 constant_alignment_word_strings.
3128 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
3129 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3130 constant_alignment_word_strings.
3131 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
3132 * config/microblaze/microblaze.c (microblaze_constant_alignment):
3134 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3135 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
3136 * config/mips/mips.c (mips_constant_alignment): New function.
3137 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3138 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
3139 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
3140 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3141 (mmix_constant_alignment): Make static. Use the same interface
3143 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
3144 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3145 constant_alignment_word_strings.
3146 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
3147 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3148 constant_alignment_word_strings.
3149 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
3150 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3151 constant_alignment_word_strings.
3152 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
3153 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3154 (rs6000_constant_alignment): New function.
3155 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
3156 * config/riscv/riscv.c (riscv_constant_alignment): New function.
3157 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3158 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
3159 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3160 (rs6000_constant_alignment): New function.
3161 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
3162 * config/s390/s390.c (s390_constant_alignment): New function.
3163 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3164 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
3165 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3166 constant_alignment_word_strings.
3167 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
3168 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3169 (sparc_constant_alignment): New function.
3170 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
3171 * config/spu/spu.c (spu_constant_alignment): New function.
3172 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3173 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
3174 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3175 constant_alignment_word_strings.
3176 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
3177 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3178 constant_alignment_word_strings.
3179 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
3180 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3181 constant_alignment_word_strings.
3182 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
3183 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3184 (visium_constant_alignment): New function.
3185 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
3186 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3187 (xtensa_constant_alignment): New function.
3188 * system.h (CONSTANT_ALIGNMENT): Poison.
3190 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
3192 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
3193 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
3194 (rs6000_builtin_valid_without_lhs): New helper function.
3195 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3196 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
3198 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
3200 * target.h (vec_perm_indices): Use unsigned short rather than
3202 (auto_vec_perm_indices): Likewise.
3203 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
3204 Use unsigned int rather than unsigned char.
3205 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
3207 2017-09-25 Richard Biener <rguenther@suse.de>
3209 * cfgloop.h (sort_sibling_loops): Declare.
3210 * cfgloop.c (sort_sibling_loops_cmp): New helper.
3211 (sort_sibling_loops): New function sorting the sibling loop list
3213 * graphite.c (graphite_transform_loops): Sort sibling loops.
3215 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
3217 * target.def (vec_perm_const_ok): Change sel parameter to
3219 * optabs-query.c (can_vec_perm_p): Update accordingly.
3220 * doc/tm.texi: Regenerate.
3221 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
3222 auto_vec_perm_indices and remove separate nelt field.
3223 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
3224 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
3225 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
3226 (aarch64_expand_vec_perm_const): Update accordingly.
3227 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
3228 to vec_perm_indices.
3229 * config/arm/arm.c (expand_vec_perm_d): Change perm to
3230 auto_vec_perm_indices and remove separate nelt field.
3231 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
3232 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
3233 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
3235 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
3236 to vec_perm_indices.
3237 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
3238 sel to vec_perm_indices.
3239 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
3240 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
3241 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
3243 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
3246 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
3249 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
3250 on the FUNCTION_DECL function context if it has a DIE that is a
3253 2017-09-25 Richard Biener <rguenther@suse.de>
3255 PR tree-optimization/82285
3256 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
3259 2017-09-25 Tom de Vries <tom@codesourcery.com>
3263 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
3266 2017-09-25 Richard Biener <rguenther@suse.de>
3268 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
3269 ISL errors other than isl_error_quota happen. Dump if the
3270 schedule is the same.
3271 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
3272 errors instead of aborting inside ISL.
3274 2017-09-25 Iain Sandoe <iain@codesourcery.com>
3277 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
3278 of libgcc_eh for m64.
3279 * config/i386/darwin64.h: Likewise.
3281 2017-09-25 Richard Biener <rguenther@suse.de>
3284 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
3285 attribute for incomplete types nor twice for complete ones.
3287 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
3290 * config/i386/i386.c (ix86_print_operand_address_as): Encode
3291 %esp as %rsp to avoid 0x67 prefix if there is no index or base
3294 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
3297 * config/i386/i386.opt (mprefer-avx256): Use
3298 ix86_target_flags variable.
3299 * config/i386/i386.c (ix86_target_string): Move
3300 -mprefer-avx256 to flag2_opts.
3302 2017-09-22 Jakub Jelinek <jakub@redhat.com>
3305 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
3306 and x != -1 | y != -1 into (x & y) != -1.
3308 2017-09-22 Steve Ellcey <sellcey@cavium.com>
3310 * config.gcc: Add new case statement to set
3311 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
3312 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
3313 s390x-*-linux* case statements. Added aarch64 to the list of
3314 supported architectures.
3316 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3318 PR tree-optimization/82289
3319 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
3320 STMT_VINFO_RELEVANT_P.
3322 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3323 Alan Hayward <alan.hayward@arm.com>
3324 David Sherwood <david.sherwood@arm.com>
3326 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
3327 for VR_RANGE only; don't allow VR_ANTI_RANGE.
3328 (extract_range_from_binary_expr_1): Don't call
3329 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
3331 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3332 Alan Hayward <alan.hayward@arm.com>
3333 David Sherwood <david.sherwood@arm.com>
3335 * target.def (preferred_vector_alignment): New hook.
3336 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
3338 * doc/tm.texi: Regenerate.
3339 * targhooks.h (default_preferred_vector_alignment): Declare.
3340 * targhooks.c (default_preferred_vector_alignment): New function.
3341 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
3343 (DR_TARGET_ALIGNMENT): New macro.
3344 (aligned_access_p): Update commentary.
3345 (vect_known_alignment_in_bytes): New function.
3346 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
3348 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
3349 Calculate the misalignment based on the target alignment rather than
3351 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
3352 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
3353 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
3354 the target alignment, rather than masking the element misalignment
3355 with the number of elements in a vector. Also use the target
3356 alignment when calculating the maximum number of peels.
3357 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
3358 instead of TYPE_ALIGN_UNIT.
3359 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
3360 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
3361 (vect_create_addr_base_for_vector_ref): Update call accordingly.
3362 (vect_create_data_ref_ptr): Likewise.
3363 (vect_setup_realignment): Realign by ANDing with
3364 -DR_TARGET_MISALIGNMENT.
3365 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
3366 the number of peels based on DR_TARGET_ALIGNMENT.
3367 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
3368 with the guaranteed alignment boundary when deciding whether
3370 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
3371 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
3372 (ensure_base_align): Remove stmt_info parameter. Get the
3373 target base alignment from DR_TARGET_ALIGNMENT.
3374 (vectorizable_store): Update call accordingly. Interpret
3375 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
3377 (vectorizable_load): Likewise.
3379 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3380 Alan Hayward <alan.hayward@arm.com>
3381 David Sherwood <david.sherwood@arm.com>
3383 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
3384 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
3385 (vect_enhance_data_refs_alignment): Likewise.
3387 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
3389 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
3390 error. Only quit immediately if parsing is complete.
3391 (BEGIN): Initialize fatal_err and parse_done.
3392 (begin fpu, end fpu): Check number of arguments.
3393 (begin arch, end arch): Likewise.
3394 (begin cpu, end cpu): Likewise.
3395 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
3396 (optalias): Likewise.
3398 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
3400 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
3401 * config/arm/arm-isa.h: Delete. Move definitions to ...
3402 * arm-cpus.in: ... here. Use new feature and fgroup values.
3403 * config/arm/arm.c (arm_option_override): Use lower case for feature
3405 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
3406 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
3407 * config/arm/parsecpu.awk (END): Add new command 'isa'.
3409 (print_isa_bits_for): New function.
3410 (gen_isa): New function.
3411 (gen_comm_data): Use print_isa_bits_for.
3412 (define feature): New keyword.
3413 (define fgroup): New keyword.
3414 * config/arm/t-arm (TM_H): Remove.
3415 (GTM_H): Add arm-isa.h.
3416 (arm-isa.h): Add rule to generate file.
3417 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
3418 case for feature bit names.
3420 2017-09-22 Richard Biener <rguenther@suse.de>
3422 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
3424 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
3425 print of no dependency loops ...
3426 * graphite.c (graphite_transform_loops): ... here.
3427 (canonicalize_loop_closed_ssa_form): Work from inner to outer
3429 (same_close_phi_node, remove_duplicate_close_phi,
3430 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
3431 (canonicalize_loop_closed_ssa): ... here and simplify.
3432 * graphite-optimize-isl.c: Include tree-vectorizer.h.
3433 (optimize_isl): Use dump_printf_loc to tell when we stopped
3434 optimizing because of an ISL timeout.
3436 2017-09-22 Richard Biener <rguenther@suse.de>
3438 PR tree-optimization/82291
3439 * tree-if-conv.c (predicate_mem_writes): Make sure to
3440 remove writes in blocks predicated with false.
3442 2017-09-22 Richard Biener <rguenther@suse.de>
3444 * sese.c: Include cfganal.h.
3445 (if_region_set_false_region): Remove.
3446 (create_if_region_on_edge): Likewise.
3447 (move_sese_in_condition): Re-implement without destroying
3450 2017-09-22 Richard Biener <rguenther@suse.de>
3452 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
3453 Verify both BBs contain loop PHI nodes before dispatching to
3455 (graphite_regenerate_ast_isl): Do not recompute dominators,
3456 do not verify three times. Restructure for clarity.
3457 * graphite-scop-detection.c (same_close_phi_node,
3458 remove_duplicate_close_phi, make_close_phi_nodes_unique,
3459 defined_in_loop_p, canonicalize_loop_closed_ssa,
3460 canonicalize_loop_closed_ssa_form): Simplify, remove excess
3461 checking and SSA rewrite, move to ...
3462 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
3463 (graphite_initialize): Do not pass in ctx, do not reset the
3464 SCEV cache, compute only dominators.
3465 (graphite_transform_loops): Allocate ISL ctx after
3466 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
3467 Maintain post-dominators only around build_scops.
3468 * sese.c (if_region_set_false_region): Make static. Free
3469 and recompute dominators.
3470 (move_sese_in_condition): Assert we don't get called with
3471 post-dominators computed.
3472 * sese.h (if_region_set_false_region): Remove.
3474 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
3476 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
3477 mode attribute for TARGET_AVX512VL.
3479 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
3481 * config/i386/i386.opt (mprefer-avx256): New option.
3482 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
3484 (ix86_preferred_simd_mode): Return 256-bit AVX modes
3485 for TARGET_PREFER_AVX256.
3486 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
3488 2017-09-21 Jeff Law <law@redhat.com>
3490 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
3491 Fix dump output if the only stack space is for pushed registers.
3493 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3495 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
3498 2017-09-21 Martin Sebor <msebor@redhat.com>
3501 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
3502 code (in C++) or code that triggers warnings.
3504 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
3506 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
3508 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
3510 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
3511 * sched-rgn.c: Ditto.
3512 * sel-sched-ir.c: Ditto.
3514 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
3516 * toplev.h (set_random_seed): Adjust return type.
3517 * toplev.c (init_local_tick): Move eager initialization of random_seed
3518 to get_random_seed. Adjust comment.
3519 (init_random_seed): Inline to get_random_seed, delete.
3520 (get_random_seed): Initialize random_seed lazily.
3521 (set_random_seed): Do not return previous value.
3522 (print_switch_value): Do not call get_random_seed.
3524 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
3526 * cgraph.c (delete_function_version): New, broken out from...
3527 (cgraph_node::delete_function_version): ...here. Rename to
3528 cgraph_node::delete_function_version_by_decl. Update all uses.
3529 (cgraph_node::remove): Call delete_function_version.
3531 2017-09-21 Jakub Jelinek <jakub@redhat.com>
3534 * tree-inline.c (expand_call_inline): Emit clobber stmts for
3535 VAR_DECLs to which addressable non-volatile parameters are mapped
3536 and for id->retvar after the return value assignment. Clear
3537 id->retval and id->retbnd after inlining.
3539 2017-09-21 Richard Biener <rguenther@suse.de>
3541 PR tree-optimization/82276
3542 PR tree-optimization/82244
3543 * tree-vrp.c (build_assert_expr_for): Set
3544 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
3546 (remove_range_assertions): Revert earlier change.
3548 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
3551 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
3553 2017-09-21 Richard Biener <rguenther@suse.de>
3555 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3556 Restore valid IL after code generation errors.
3557 * graphite.c (graphite_transform_loops): Diagnose code
3558 generation issues as MSG_MISSED_OPTIMIZATION and continue
3559 with processing SCOPs.
3561 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3562 Alan Hayward <alan.hayward@arm.com>
3563 David Sherwood <david.sherwood@arm.com>
3565 * calls.c (compute_argument_addresses): Use simplify_gen_binary
3566 rather than choosing between plus_constant and gen_rtx_<CODE>.
3567 * expr.c (emit_push_insn): Likewise.
3568 (expand_expr_real_2): Likewise.
3570 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3571 Alan Hayward <alan.hayward@arm.com>
3572 David Sherwood <david.sherwood@arm.com>
3574 * loop-unroll.c (split_iv): Call copy_rtx on the step.
3576 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3577 Alan Hayward <alan.hayward@arm.com>
3578 David Sherwood <david.sherwood@arm.com>
3580 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
3581 calling tree_to_uhwi.
3583 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3584 Alan Hayward <alan.hayward@arm.com>
3585 David Sherwood <david.sherwood@arm.com>
3587 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
3588 INTEGER_CST rather than a negative test for ADDR_EXPR.
3590 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3591 Alan Hayward <alan.hayward@arm.com>
3592 David Sherwood <david.sherwood@arm.com>
3594 * tree-vrp.c (extract_range_from_binary_expr_1): Check
3595 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
3597 2017-09-21 Richard Biener <rguenther@suse.de>
3599 PR tree-optimization/71351
3600 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
3601 graphite_create_new_loop_guard): Remove, fold remaining parts
3603 (translate_isl_ast_node_for): ... here and simplify.
3605 2017-09-21 Jakub Jelinek <jakub@redhat.com>
3608 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
3609 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
3610 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
3611 alternative always use QI mode, for -Os imov (=R,R) alternative
3612 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
3615 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3616 Jeff Law <law@redhat.com>
3618 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
3619 (allocate_stack_space): New function, partially extracted from
3621 (s390_emit_prologue): Track offset to most recent stack probe.
3622 Code to allocate space moved into allocate_stack_space.
3623 Dump actions when no stack is allocated.
3624 (s390_prologue_plus_offset): New function.
3625 (s390_emit_stack_probe): Likewise.
3627 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
3629 * common.opt (Wa, Wl, Wp, g, gz=): Add
3631 (gno-column-info): Remove.
3632 (gcolumn-info): Drop RejectNegative.
3634 (gno-record-gcc-switches): Remove.
3635 (grecord-gcc-switches): Drop RejectNegative.
3636 (gno-split-dwarf): Remove.
3637 (gsplit-dwarf): Drop RejectNegative.
3638 (gno-strict-dwarf): Remove.
3639 (gstrict-dwarf): Drop RejectNegative.
3640 * config/darwin.opt (gfull, gused): Add RejectNegative.
3641 * dwarf2out.c (gen_producer_string): Drop
3642 gno-record-gcc-switches handler.
3643 * optc-gen.awk: Add g to prefixes with negative forms.
3644 * opts-common.c (remapping_prefix_p): New.
3645 (find_opt): Check it.
3646 (generate_canonical_option): Test g prefix.
3647 (option_map): Add -gno- mapping.
3648 (add_misspelling_candidates): Check remapping_prefix_p.
3650 2017-09-20 Jeff Law <law@redhat.com>
3652 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
3653 thinko in stack clash protection support.
3655 * explow.c (compute_stack_clash_protection_loop_data): Use
3656 CONST_INT_P instead of explicit test. Verify object is a
3657 CONST_INT_P before looking at INTVAL.
3658 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
3659 instead of explicit test.
3661 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
3664 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
3665 address instead of to r1 and r11.
3667 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
3669 * config.gcc: Support "knm".
3670 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
3671 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3673 * config/i386/i386.c (m_KNM): Define.
3674 (processor_target_table): Add "knm".
3676 (ix86_option_override_internal): Add "knm".
3677 (ix86_issue_rate): Add PROCESSOR_KNM.
3678 (ix86_adjust_cost): Ditto.
3679 (ia32_multipass_dfa_lookahead): Ditto.
3680 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
3681 (fold_builtin_cpu): Add M_INTEL_KNM.
3682 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
3683 (processor_type): Add PROCESSOR_KNM.
3684 * config/i386/x86-tune.def: Add m_KNM.
3685 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
3687 2017-09-20 Richard Biener <rguenther@suse.de>
3689 PR tree-optimization/80213
3690 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
3691 are allowed in empty BBs as well.
3692 (canonicalize_loop_closed_ssa): Also look for other complex
3694 (scop_detection::get_sese): Include the loop-closed PHI block
3696 (scop_detection::merge_sese): Remove code adding extra blocks.
3697 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
3698 (build_scops): Assert the final returned scop is invalid.
3700 2017-09-20 Richard Biener <rguenther@suse.de>
3702 PR tree-optimization/82264
3703 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
3705 (vn_phi_lookup): Likewise.
3706 (vn_phi_insert): Likewise.
3708 2017-09-20 Jakub Jelinek <jakub@redhat.com>
3710 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
3711 that fits into uhwi or shwi, add DW_AT_const_value regardless
3712 of early_dwarf without going through RTL, using add_AT_unsigned
3715 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
3716 (DEBUG_LTO_ABBREV_SECTION): Likewise.
3717 (DEBUG_LTO_MACINFO_SECTION): Likewise.
3718 (DEBUG_MACRO_SECTION): Likewise.
3719 (DEBUG_LTO_MACRO_SECTION): Likewise.
3720 (DEBUG_STR_DWO_SECTION): Likewise.
3721 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
3722 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
3723 (DEBUG_LTO_DWO_LINE_SECTION): Define.
3724 (DEBUG_LTO_LINE_STR_SECTION): Define.
3725 (init_sections_and_labels): Initialize debug_line_str_section
3726 variable. Initialize debug_loc_section for -gdwarf-5 to
3727 DEBUG_LOCLISTS_SECTION. Formatting fixes.
3729 2017-09-20 Richard Biener <rguenther@suse.de>
3731 * graphite-sese-to-poly.c (extract_affine): Properly handle
3732 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
3734 2017-09-20 Richard Biener <rguenther@suse.de>
3736 PR tree-optimization/81373
3737 * graphite-scop-detection.c (build_cross_bb_scalars_def):
3738 Force SESE live-out defs to be handled even if they are
3741 2017-09-19 Jeff Law <law@redhat.com>
3743 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
3744 nothing for stack adjustments with REG_STACK_CHECK.
3745 * sched-deps.c (parse_add_or_inc): Reject insns with
3746 REG_STACK_CHECK from dependency breaking.
3747 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
3748 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
3749 * reg-notes.def (STACK_CHECK): New note.
3751 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
3752 (ix86_expand_prologue): Dump stack clash info as needed.
3753 Call ix86_adjust_stack_and_probe_stack_clash as needed.
3755 * function.c (dump_stack_clash_frame_info): New function.
3756 * function.h (dump_stack_clash_frame_info): Prototype.
3757 (enum stack_clash_probes): New enum.
3759 * config/alpha/alpha.c (alpha_expand_prologue): Also check
3760 flag_stack_clash_protection.
3761 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
3762 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
3763 (arm_frame_pointer_required): Likewise.
3764 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
3765 (ia64_expand_prologue): Likewise.
3766 * config/mips/mips.c (mips_expand_prologue): Likewise.
3767 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
3768 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
3769 (sparc_flat_expand_prologue): Likewise.
3770 * config/spu/spu.c (spu_expand_prologue): Likewise.
3772 * explow.c: Include "params.h".
3773 (anti_adjust_stack_and_probe_stack_clash): New function.
3774 (get_stack_check_protect): Likewise.
3775 (compute_stack_clash_protection_loop_data): Likewise.
3776 (emit_stack_clash_protection_loop_start): Likewise.
3777 (emit_stack_clash_protection_loop_end): Likewise.
3778 (allocate_dynamic_stack_space): Use get_stack_check_protect.
3779 Use anti_adjust_stack_and_probe_stack_clash.
3780 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
3781 (emit_stack_clash_protection_loop_start): Likewise.
3782 (emit_stack_clash_protection_loop_end): Likewise.
3783 * rtl.h (get_stack_check_protect): Prototype.
3784 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
3785 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
3786 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
3788 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
3790 * doc/tm.texi: Rebuilt.
3791 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
3792 get_stack_check_protect.
3793 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
3794 * config/arm/arm.c (arm_expand_prologue): Likewise.
3795 (arm_frame_pointer_required): Likewise.
3796 * config/i386/i386.c (ix86_expand_prologue): Likewise.
3797 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
3798 * config/mips/mips.c (mips_expand_prologue): Likewise.
3799 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
3800 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
3801 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
3802 (sparc_flat_expand_prologue): Likewise.
3804 * common.opt (-fstack-clash-protection): New option.
3805 * flag-types.h (enum stack_check_type): Note difference between
3806 -fstack-check= and -fstack-clash-protection.
3807 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
3808 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
3809 * toplev.c (process_options): Issue warnings/errors for cases
3810 not handled with -fstack-clash-protection.
3811 * doc/invoke.texi (-fstack-clash-protection): Document new option.
3812 (-fstack-check): Note additional problem with -fstack-check=generic.
3813 Note that -fstack-check is primarily for Ada and refer users
3814 to -fstack-clash-protection for stack-clash-protection.
3815 Document new params for stack clash protection.
3817 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
3819 * config/i386/i386.c (ix86_split_long_move): Do not handle
3820 address used for LEA in a special way.
3822 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
3824 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
3826 2017-09-19 Martin Sebor <msebor@redhat.com>
3829 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
3830 of incompatible types.
3832 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
3834 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
3835 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
3836 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3837 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
3839 2017-09-19 Richard Biener <rguenther@suse.de>
3841 PR tree-optimization/82244
3842 * tree-vrp.c (remove_range_assertions): Do not propagate
3843 a constant to abnormals but replace the assert with a copy.
3845 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
3847 PR rtl-optimization/57878
3848 PR rtl-optimization/68988
3849 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
3850 avoidance test involving non_reload_pseudos. Move frequency test
3851 below the general fragmentation avoidance test.
3853 2017-09-19 Richard Biener <rguenther@suse.de>
3855 PR tree-optimization/69728
3856 * graphite-sese-to-poly.c (schedule_error): New global.
3857 (add_loop_schedule): Handle empty domain by failing the
3859 (build_original_schedule): Handle schedule_error.
3861 2017-09-19 Richard Biener <rguenther@suse.de>
3863 * graphite-scop-detection.c (scop_detection::can_represent_loop):
3864 Do not iterate to sibling loops but only to siblings of inner
3867 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
3870 * config/m68k/m68k.md (moveq feeding equality comparison): Check
3871 that the registers are different.
3873 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
3875 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
3876 to processor_model and "amdfam17h" to arch_names_table.
3877 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
3879 2017-09-18 Jakub Jelinek <jakub@redhat.com>
3882 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
3884 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3885 Alan Hayward <alan.hayward@arm.com>
3886 David Sherwood <david.sherwood@arm.com>
3888 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
3890 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
3892 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
3893 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
3894 vect_schedule_slp_instance.
3895 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
3896 Update call to vect_slp_analyze_node_operations. Simplify return
3898 (vect_slp_analyze_bb_1): Update call accordingly.
3899 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
3900 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
3901 (vect_schedule_slp): Update call accordingly.
3903 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3904 Alan Hayward <alan.hayward@arm.com>
3905 David Sherwood <david.sherwood@arm.com>
3907 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
3908 with types that aren't in fact scalar.
3910 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3912 * tree-vect-slp.c (vect_record_max_nunits): New function,
3914 (vect_build_slp_tree_1): ...here.
3915 (vect_build_slp_tree_2): Call it for phis too.
3917 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3919 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
3920 to vect_get_vec_def_for_operand when getting the mask operand.
3922 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3923 Alan Hayward <alan.hayward@arm.com>
3924 David Sherwood <david.sherwood@arm.com>
3926 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
3929 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3930 Alan Hayward <alan.hayward@arm.com>
3931 David Sherwood <david.sherwood@arm.com>
3933 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
3934 calculation for vector booleans.
3936 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3937 Alan Hayward <alan.hayward@arm.com>
3938 David Sherwood <david.sherwood@arm.com>
3940 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
3942 (vect_transform_stmt): ...here.
3943 (vect_analyze_stmt): Use it instead of calling
3944 vectorizable_live_operation directly.
3946 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
3948 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
3949 non-SIMT targets in acc vector loops.
3951 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
3953 * configure.ac: Add arc and check if assembler supports gdwarf2.
3954 * configure: Regenerate.
3956 2017-09-18 Richard Biener <rguenther@suse.de>
3958 PR tree-optimization/82220
3959 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
3960 epilogue niters from the min_profitable_iters compute.
3962 2017-09-18 Jakub Jelinek <jakub@redhat.com>
3965 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
3966 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
3967 (ix86_init_pic_reg): Revert 2017-09-01 changes.
3969 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
3972 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
3973 switching to a new text section.
3975 2017-09-18 Richard Biener <rguenther@suse.de>
3977 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
3979 (build_alias_set): Reject aliases with no access function.
3981 2017-09-18 Richard Biener <rguenther@suse.de>
3983 PR tree-optimization/79622
3984 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
3986 (build_cross_bb_scalars_use): Likewise.
3988 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
3990 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
3992 2017-09-18 Alan Modra <amodra@gmail.com>
3995 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
3996 stack_pointer_rtx for count 0. Update comments. Break up
3997 large rtl expression.
3999 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
4001 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
4002 Increase to 20 bytes.
4003 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
4004 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
4005 or avx version of the stub.
4007 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
4010 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
4011 compute the minimum stack alignment. Also update preferred stack
4012 boundary for leaf functions.
4014 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
4016 PR tree-optimization/82228
4017 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
4020 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
4022 * common/config/nds32/nds32-common.c
4023 (nds32_option_optimization_table): Refine formatting.
4024 (nds32_option_optimization_table): Use -fsched-pressure and
4025 -fomit-frame-pointer for specific optimization level.
4027 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
4029 * config/nds32/nds32.c: Refine formatting and comments.
4030 * config/nds32/nds32.h: Likewise.
4031 * config/nds32/nds32.md: Likewise.
4032 * config/nds32/nds32-cost.c: Likewise.
4033 * config/nds32/nds32-isr.c: Likewise.
4034 * config/nds32/nds32-md-auxiliary.c: Likewise.
4035 * config/nds32/nds32-multiple.md: Likewise.
4036 * config/nds32/nds32-predicates.c: Likewise.
4038 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
4039 Jakub Jelinek <jakub@redhat.com>
4041 Add support for -std=c++2a.
4042 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
4044 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
4046 2017-09-15 Steve Ellcey <sellcey@cavium.com>
4049 * doc/extend.texi (Common Function Attributes): Add
4050 references to ARM, AArch64, and S/390 specific attributes.
4051 (Function Specific Option Pragmas): Add AArch64 and S/390
4052 to list of back ends that support the target pragma.
4054 2017-09-15 Nathan Sidwell <nathan@acm.org>
4056 * doc/standards.texi: Fix C++17 description. Update URLs for
4059 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
4061 * common.opt (Wcast-align=strict): New warning option.
4062 * doc/invoke.texi: Document -Wcast-align=strict.
4064 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
4066 * cgraph.h (cgraph_thunk_info): Add comments.
4067 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
4068 assert for VIRTUAL_* arguments stricter.
4070 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
4072 PR tree-optimization/71026
4073 * match.pd: Move RDIV patterns from fold-const.c
4074 * fold-const.c (distribute_real_division): Removed.
4075 (fold_binary_loc): Remove calls to distribute_real_divison.
4077 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4079 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
4080 c++1z and gnu++1z as deprecated. Change other references to
4081 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
4082 Change -Wc++1z-compat to -Wc++17-compat.
4083 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
4084 * dwarf2out.c (highest_c_language): Handle C++17.
4085 (gen_compile_unit_die): Likewise.
4087 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4089 PR rtl-optimization/82192
4090 * combine.c (make_extraction): Don't look through non-paradoxical
4091 SUBREGs or TRUNCATE if pos + len is or might be bigger than
4094 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4095 Alan Hayward <alan.hayward@arm.com>
4096 David Sherwood <david.sherwood@arm.com>
4098 * target.def (function_arg_offset): New hook.
4099 * targhooks.h (default_function_arg_offset): Declare.
4100 * targhooks.c (default_function_arg_offset): New function.
4101 * function.c (locate_and_pad_parm): Use
4102 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
4103 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
4104 (TARGET_FUNCTION_ARG_OFFSET): ...this.
4105 * doc/tm.texi: Regenerate.
4106 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
4107 * config/spu/spu.c (spu_function_arg_offset): New function.
4108 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
4109 * system.h (FUNCTION_ARG_OFFSET): Poison.
4111 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4112 Alan Hayard <alan.hayward@arm.com>
4113 David Sherwood <david.sherwood@arm.com>
4115 * target.def (truly_noop_truncation): New hook.
4116 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
4117 than TRULY_NOOP_TRUNCATION.
4118 * hooks.h (hook_bool_uint_uint_true): Declare.
4119 * hooks.c (hook_bool_uint_uint_true): New function.
4120 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
4121 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
4122 * doc/tm.texi: Regenerate.
4123 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
4124 rather than TRULY_NOOP_TRUNCATION in comments.
4125 (simplify_comparison): Likewise.
4126 (record_truncated_value): Likewise.
4127 * expmed.c (extract_bit_field_1): Likewise.
4128 (extract_split_bit_field): Likewise.
4129 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
4130 instead of TRULY_NOOP_TRUNCATION.
4131 * function.c (assign_parm_setup_block): Likewise.
4132 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
4133 * rtlhooks.c: Include target.h.
4134 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
4135 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
4136 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
4137 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
4138 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
4139 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
4140 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
4141 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
4142 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
4143 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
4144 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
4145 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
4146 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
4147 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
4148 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
4149 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
4150 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
4151 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
4152 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
4153 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
4154 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
4155 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
4156 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
4157 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
4158 * config/mips/mips.c (mips_truly_noop_truncation): New function.
4159 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4160 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4161 rather than TRULY_NOOP_TRUNCATION in comments.
4162 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
4163 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
4164 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
4165 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
4166 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
4167 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
4168 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
4169 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
4170 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
4171 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
4172 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
4173 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4174 rather than TRULY_NOOP_TRUNCATION in comments.
4175 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
4176 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
4177 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
4178 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
4179 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
4180 TRULY_NOOP_TRUNCATION condition.
4181 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
4182 (TRULY_NOOP_TRUNCATION): Delete.
4183 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
4184 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
4185 * config/spu/spu.c (spu_truly_noop_truncation): New function.
4186 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4187 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
4188 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
4189 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
4190 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4191 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4192 rather than TRULY_NOOP_TRUNCATION in comments.
4193 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
4194 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
4195 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
4196 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
4197 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
4198 * system.h (TRULY_NOOP_TRUNCATION): Poison.
4200 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
4203 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
4204 (*cmp_ior): Likewise.
4205 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
4206 (*ior_scc_scc_cmp): Likewise.
4207 (*and_scc_scc): Likewise.
4208 (*and_scc_scc_cmp): Likewise.
4210 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4211 Alan Hayard <alan.hayward@arm.com>
4212 David Sherwood <david.sherwood@arm.com>
4214 * target.def (can_change_mode_class): New hook.
4215 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
4216 (hard_regno_nregs): Likewise.
4217 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
4218 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
4219 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
4220 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
4221 (LOAD_EXTEND_OP): Update accordingly.
4222 * doc/tm.texi: Regenerate.
4223 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
4224 CANNOT_CHANGE_MODE_CLASS.
4225 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
4226 (REG_CAN_CHANGE_MODE_P): ...this new macro.
4227 * combine.c (simplify_set): Update accordingly.
4228 * emit-rtl.c (validate_subreg): Likewise.
4229 * recog.c (general_operand): Likewise.
4230 * regcprop.c (mode_change_ok): Likewise.
4231 * reload1.c (choose_reload_regs): Likewise.
4232 (inherit_piecemeal_p): Likewise.
4233 * rtlanal.c (simplify_subreg_regno): Likewise.
4234 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
4235 instead of CANNOT_CHANGE_MODE_CLASS.
4236 (reload_cse_simplify_operands): Likewise.
4237 * reload.c (push_reload): Use targetm.can_change_mode_class
4238 instead of CANNOT_CHANGE_MODE_CLASS.
4239 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
4240 REG_CANNOT_CHANGE_MODE_P.
4241 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4242 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
4243 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4244 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4245 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4246 (arm_can_change_mode_class): New function.
4247 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
4248 than CANNOT_CHANGE_MODE_CLASS in comments.
4249 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4250 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
4251 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
4252 (ix86_can_change_mode_class): ...this new function, inverting the
4253 sense of the return value.
4254 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4255 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4256 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4257 (ia64_can_change_mode_class): New function.
4258 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4259 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
4260 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
4261 (m32c_can_change_mode_class): ...this new function, inverting the
4262 sense of the return value.
4263 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4264 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4265 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
4266 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
4267 (mips_can_change_mode_class): ...this new function, inverting the
4268 sense of the return value.
4269 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4270 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4271 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4272 (msp430_can_change_mode_class): New function.
4273 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4274 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
4275 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4276 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4277 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4278 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
4279 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4280 (pa_cannot_change_mode_class): Replace with...
4281 (pa_can_change_mode_class): ...this new function, inverting the
4282 sense of the return value.
4283 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
4284 than CANNOT_CHANGE_MODE_CLASS in comments.
4285 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4286 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
4287 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4288 (pdp11_cannot_change_mode_class): Replace with...
4289 (pdp11_can_change_mode_class): ...this new function, inverting the
4290 sense of the return value.
4291 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4292 * config/powerpcspe/powerpcspe-protos.h
4293 (rs6000_cannot_change_mode_class_ptr): Delete.
4294 * config/powerpcspe/powerpcspe.c
4295 (rs6000_cannot_change_mode_class_ptr): Delete.
4296 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4297 (rs6000_option_override_internal): Assign to
4298 targetm.can_change_mode_class instead of
4299 rs6000_cannot_change_mode_class_ptr.
4300 (rs6000_cannot_change_mode_class): Replace with...
4301 (rs6000_can_change_mode_class): ...this new function, inverting the
4302 sense of the return value.
4303 (rs6000_debug_cannot_change_mode_class): Replace with...
4304 (rs6000_debug_can_change_mode_class): ...this new function.
4305 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4306 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
4307 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4308 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4309 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
4311 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
4312 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4313 (rs6000_option_override_internal): Assign to
4314 targetm.can_change_mode_class instead of
4315 rs6000_cannot_change_mode_class_ptr.
4316 (rs6000_cannot_change_mode_class): Replace with...
4317 (rs6000_can_change_mode_class): ...this new function, inverting the
4318 sense of the return value.
4319 (rs6000_debug_cannot_change_mode_class): Replace with...
4320 (rs6000_debug_can_change_mode_class): ...this new function.
4321 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4322 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
4323 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
4324 (s390_can_change_mode_class): ...this new function, inverting the
4325 sense of the return value.
4326 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4327 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4328 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
4329 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4330 (sh_cannot_change_mode_class): Replace with...
4331 (sh_can_change_mode_class): ...this new function, inverting the
4332 sense of the return value.
4333 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4334 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4335 (sparc_can_change_mode_class): New function.
4336 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4337 * config/spu/spu.c (spu_can_change_mode_class): New function.
4338 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4339 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4340 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4341 (visium_can_change_mode_class): New function.
4342 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
4344 2017-09-15 Richard Biener <rguenther@suse.de>
4346 PR tree-optimization/82217
4347 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
4348 but not undefined case.
4350 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4353 * postreload.c (reload_cse_simplify_operands): Skip
4354 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
4356 2017-09-15 Richard Biener <rguenther@suse.de>
4358 PR tree-optimization/68823
4359 * graphite-scop-detection.c (build_alias_set): If we have a
4360 possible dependence check whether we can handle them by just
4361 looking at the DRs DR_ACCESS_FNs.
4362 (build_scops): If build_alias_set fails, fail the SCOP.
4364 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
4366 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
4367 to support float128 built-in functions that require the ISA 3.0
4369 (BU_FLOAT128_3_HW): Likewise.
4370 (SQRTF128): Add support for the IEEE 128-bit square root and fma
4372 (FMAF128): Likewise.
4374 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
4375 support for built-in functions that need the ISA 3.0 IEEE 128-bit
4376 floating point instructions.
4377 (rs6000_invalid_builtin): Likewise.
4378 (rs6000_builtin_mask_names): Likewise.
4379 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
4380 (RS6000_BTM_FLOAT128_HW): Likewise.
4381 (RS6000_BTM_COMMON): Likewise.
4382 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
4384 * doc/extend.texi (RS/6000 built-in functions): Document the
4385 IEEE 128-bit floating point square root and fused multiply-add
4388 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
4390 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
4391 reg (r2) isn't in the set of registers defined in the prologue.
4393 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4394 Alan Hayward <alan.hayward@arm.com>
4395 David Sherwood <david.sherwood@arm.com>
4397 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
4398 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
4399 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
4400 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
4401 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
4403 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
4404 max_vectorization_factor.
4405 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
4407 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4408 Alan Hayward <alan.hayward@arm.com>
4409 David Sherwood <david.sherwood@arm.com>
4411 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
4412 (vect_worthwhile_without_simd_p): Declare.
4413 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
4414 (vectorizable_reduction): Use it.
4415 * tree-vect-stmts.c (vectorizable_shift): Likewise.
4416 (vectorizable_operation): Likewise.
4418 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4419 Alan Hayward <alan.hayward@arm.com>
4420 David Sherwood <david.sherwood@arm.com>
4422 * tree-vectorizer.h (vect_get_num_copies): New function.
4423 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
4424 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4425 (vectorizable_induction): Likewise.
4426 (vectorizable_live_operation): Likewise.
4427 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
4428 (vectorizable_bswap): Likewise.
4429 (vectorizable_call): Likewise.
4430 (vectorizable_conversion): Likewise.
4431 (vectorizable_assignment): Likewise.
4432 (vectorizable_shift): Likewise.
4433 (vectorizable_operation): Likewise.
4434 (vectorizable_store): Likewise.
4435 (vectorizable_load): Likewise.
4436 (vectorizable_condition): Likewise.
4437 (vectorizable_comparison): Likewise.
4438 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
4440 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4441 Alan Hayward <alan.hayward@arm.com>
4442 David Sherwood <david.sherwood@arm.com>
4444 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
4445 of vect_init_vector.
4447 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4448 Alan Hayward <alan.hayward@arm.com>
4449 David Sherwood <david.sherwood@arm.com>
4451 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
4452 an inline wrapper that provides a location.
4453 (gimple_build_vector): Likewise.
4454 * gimple-fold.c (gimple_build_vector_from_val): New function.
4455 (gimple_build_vector): Likewise.
4456 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
4457 functions to build the initial value. Always return a gimple value.
4458 (get_initial_defs_for_reduction): Likewise. Only compute
4460 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
4461 vect_init_vector on the results from get_initial_def(s)_for_reduction.
4462 (vectorizable_induction): Use gimple_build_vector rather than
4465 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4466 Alan Hayward <alan.hayward@arm.com>
4467 David Sherwood <david.sherwood@arm.com>
4469 * target.h (vec_perm_indices): New typedef.
4470 (auto_vec_perm_indices): Likewise.
4471 * optabs-query.h: Include target.h
4472 (can_vec_perm_p): Take a vec_perm_indices *.
4473 * optabs-query.c (can_vec_perm_p): Likewise.
4474 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
4475 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
4476 * tree-vect-generic.c (lower_vec_perm): Likewise.
4477 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
4478 (vect_grouped_load_supported): Likewise.
4479 (vect_shift_permute_load_chain): Likewise.
4480 (vect_permute_store_chain): Use auto_vec_perm_indices.
4481 (vect_permute_load_chain): Likewise.
4482 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
4483 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
4484 Update uses of can_vec_perm_p.
4485 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
4486 mode with a number of elements. Take a vec_perm_indices *.
4487 (vect_create_epilog_for_reduction): Update accordingly.
4488 Use auto_vec_perm_indices.
4489 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
4490 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
4491 (vect_transform_slp_perm_load): Likewise.
4492 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
4493 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
4494 (vect_gen_perm_mask_checked): Likewise.
4495 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
4496 (vect_gen_perm_mask_checked): Likewise.
4497 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
4498 (vectorizable_store): Likewise.
4499 (vectorizable_load): Likewise.
4500 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
4501 (vectorizable_bswap): Likewise.
4503 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4504 Alan Hayward <alan.hayward@arm.com>
4505 David Sherwood <david.sherwood@arm.com>
4507 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
4508 * tree.c (build_vector): Likewise.
4509 (build_vector_from_ctor): Update accordingly.
4510 (build_vector_from_val): Likewise.
4511 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
4512 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
4513 * tree-vect-generic.c (add_rshift): Likewise.
4514 (expand_vector_divmod): Likewise.
4515 (optimize_vector_constructor): Likewise.
4516 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
4517 (vect_transform_slp_perm_load): Likewise.
4518 (vect_schedule_slp_instance): Likewise.
4519 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
4520 (vectorizable_call): Likewise.
4521 (vect_gen_perm_mask_any): Likewise. Add elements in order.
4522 * expmed.c (make_tree): Likewise.
4523 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
4524 a vector passed to build_vector.
4525 (fold_convert_const): Likewise.
4526 (exact_inverse): Likewise.
4527 (fold_ternary_loc): Likewise.
4528 (fold_relational_const): Likewise.
4529 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
4530 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
4531 (const_unop): Likewise. Store the reduction accumulator in a
4532 variable rather than an array.
4533 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
4534 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
4535 the new vector, rather than constructing it after the input arrays.
4536 (native_interpret_vector): Use auto_vec<tree> when building
4537 a vector passed to build_vector. Add elements in order.
4538 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
4539 auto_vec<tree> when building a vector passed to build_vector.
4540 (vect_create_epilog_for_reduction): Likewise.
4541 (vectorizable_induction): Likewise.
4542 (get_initial_def_for_reduction): Likewise. Fix indentation of
4544 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
4546 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
4547 passed to build_vector.
4549 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4550 Alan Hayward <alan.hayward@arm.com>
4551 David Sherwood <david.sherwood@arm.com>
4553 * tree-core.h (tree_base::u): Add an "nelts" field.
4554 (tree_vector): Use VECTOR_CST_NELTS as the length.
4555 * tree.c (tree_size): Likewise.
4556 (make_vector): Initialize VECTOR_CST_NELTS.
4557 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
4558 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
4559 TYPE_VECTOR_SUBPARTS.
4560 * expr.c (const_vector_mask_from_tree): Consistently use "units"
4561 as the number of units, setting it from VECTOR_CST_NELTS.
4562 (const_vector_from_tree): Likewise.
4563 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
4564 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
4565 (fold_negate_expr_1): Likewise.
4566 (fold_convert_const): Likewise.
4567 (const_binop): Likewise. Differentiate the number of output and
4569 (const_unop): Likewise.
4570 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
4571 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
4572 in cases that did the opposite.
4574 2017-09-14 Richard Biener <rguenther@suse.de>
4576 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
4579 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
4581 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
4583 2017-09-14 Jakub Jelinek <jakub@redhat.com>
4586 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
4587 if and where to split a bb, except for splitting before debug insn
4588 sequences followed by non-label real insn. Delete debug insns
4589 in between basic blocks.
4591 * combine.c (make_compound_operation_int): Formatting fixes.
4593 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
4594 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4595 * config/netbsd.h (LINK_EH_SPEC): Likewise.
4596 * config/sol2.h (LINK_EH_SPEC): Likewise.
4597 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4598 * config/s390/linux.h (LINK_SPEC): Likewise.
4599 * config/freebsd.h (LINK_EH_SPEC): Likewise.
4600 * config/openbsd.h (LINK_EH_SPEC): Likewise.
4601 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4602 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
4603 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
4604 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4605 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
4606 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
4608 2017-09-13 Jakub Jelinek <jakub@redhat.com>
4610 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
4612 (ENDFILE_LINUX_SPEC): Likewise.
4613 (LINK_EH_SPEC): Likewise.
4614 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
4615 (LINK_OS_LINUX_SPEC32): Likewise.
4616 (LINK_OS_LINUX_SPEC64): Likewise.
4617 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
4618 (LINK_OS_LINUX_SPEC): Likewise.
4620 2017-09-13 Martin Liska <mliska@suse.cz>
4623 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
4624 CASE_HIGH is NULL_TREE.
4626 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
4627 Alan Hayward <alan.hayward@arm.com>
4628 David Sherwood <david.sherwood@arm.com>
4630 * target.def (secondary_memory_needed): New hook.
4631 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
4632 instead of SECONDARY_MEMORY_NEEDED.
4633 (secondary_memory_needed_mode): Likewise.
4634 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
4635 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
4636 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
4637 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
4638 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
4639 * doc/tm.texi: Regenerate.
4640 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
4641 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
4642 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4643 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
4644 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
4645 * config/i386/i386.c (inline_secondary_memory_needed): Put the
4646 mode argument first and change the reg_class arguments to reg_class_t.
4647 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
4648 Make static. Update the call to inline_secondary_memory_needed.
4649 (ix86_register_move_cost): Update the call to
4650 inline_secondary_memory_needed.
4651 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4652 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
4654 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
4655 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4657 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
4658 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
4659 * config/mips/mips.c (mips_secondary_memory_needed): Make static
4660 and match hook interface. Add comment from mips.h.
4661 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4662 * config/mmix/mmix.md (truncdfsf2): Refer to
4663 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4665 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
4666 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
4667 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4668 (pa_secondary_memory_needed): New function.
4669 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
4670 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
4671 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4672 (pdp11_secondary_memory_needed): Make static and match hook interface.
4673 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
4674 * config/powerpcspe/powerpcspe-protos.h
4675 (rs6000_secondary_memory_needed_ptr): Delete.
4676 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
4678 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4679 (rs6000_option_override_internal): Assign to
4680 targetm.secondary_memory_needed rather than
4681 rs6000_secondary_memory_needed_ptr.
4682 (rs6000_secondary_memory_needed): Match hook interface.
4683 (rs6000_debug_secondary_memory_needed): Likewise.
4684 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
4685 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
4686 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
4687 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4688 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
4689 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
4691 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
4692 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4693 (rs6000_option_override_internal): Assign to
4694 targetm.secondary_memory_needed rather than
4695 rs6000_secondary_memory_needed_ptr.
4696 (rs6000_secondary_memory_needed): Match hook interface.
4697 (rs6000_debug_secondary_memory_needed): Likewise.
4698 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
4699 * config/s390/s390.c (s390_secondary_memory_needed): New function.
4700 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4701 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
4702 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4703 (sparc_secondary_memory_needed): New function.
4704 * lra-constraints.c (check_and_process_move): Refer to
4705 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4707 (curr_insn_transform): Likewise.
4708 (process_alt_operands): Use targetm.secondary_memory_needed
4709 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4710 (check_secondary_memory_needed_p): Likewise.
4711 (choose_split_class): Likewise.
4712 * reload.c: Unconditionally include code that was previously
4713 conditional on SECONDARY_MEMORY_NEEDED.
4714 (push_secondary_reload): Use targetm.secondary_memory_needed
4715 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4716 (push_reload): Likewise.
4717 * reload1.c: Unconditionally include code that was previously
4718 conditional on SECONDARY_MEMORY_NEEDED.
4719 (choose_reload_regs): Use targetm.secondary_memory_needed
4720 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4721 (gen_reload): Likewise.
4722 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
4724 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
4725 Alan Hayward <alan.hayward@arm.com>
4726 David Sherwood <david.sherwood@arm.com>
4728 * target.def (secondary_memory_needed_mode): New hook:
4729 * targhooks.c (default_secondary_memory_needed_mode): Declare.
4730 * targhooks.h (default_secondary_memory_needed_mode): New function.
4731 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
4732 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
4733 * doc/tm.texi: Regenerate.
4734 * lra-constraints.c (check_and_process_move): Use
4735 targetm.secondary_memory_needed_mode instead of
4736 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
4737 (curr_insn_transform): Likewise.
4738 * reload.c (get_secondary_mem): Likewise.
4739 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4740 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
4742 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4743 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4744 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
4745 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4746 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
4748 * config/powerpcspe/powerpcspe-protos.h
4749 (rs6000_secondary_memory_needed_mode): Delete.
4750 * config/powerpcspe/powerpcspe.c
4751 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4752 (rs6000_secondary_memory_needed_mode): Make static.
4753 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4754 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
4756 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
4758 (rs6000_secondary_memory_needed_mode): Make static.
4759 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4760 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
4761 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4762 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4763 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
4765 (sparc_secondary_memory_needed_mode): New function.
4766 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
4768 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
4770 * config/aarch64/constraints.md (Umq): New constraint.
4771 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
4773 (mov<mode>): Update condition.
4775 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4777 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
4778 when bitposition is the same.
4780 2017-09-13 Richard Biener <rguenther@suse.de>
4782 * dwarf2out.c (output_die_symbol): Remove.
4783 (output_die): Do not output a DIEs symbol.
4785 2017-09-13 Richard Biener <rguenther@suse.de>
4788 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
4789 default-def to avoid breaking iterator update with the weird
4790 interaction with cgraph_update_edges_for_call_stmt_node.
4792 2017-09-13 Richard Biener <rguenther@suse.de>
4794 * tree-cfg.c (verify_gimple_assign_binary): Add verification
4795 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
4796 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
4797 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
4799 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
4801 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
4802 Disable pc relative literal load irrespective of
4803 TARGET_FIX_ERR_A53_84341 for default.
4805 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
4807 * config/sparc/sparc.c (output_return): Output the source location of
4808 the insn in the delay slot, if any.
4809 (output_sibcall): Likewise.
4811 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
4814 * common.opt (-static-pie): New alias.
4815 (shared): Negate static-pie.
4816 (-no-pie): Update help text.
4818 (static-pie): New option.
4819 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
4820 -static-pie support.
4821 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
4822 (LINK_EH_SPEC): Likewise.
4823 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4824 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
4825 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
4826 * gcc.c (LINK_COMMAND_SPEC): Likewise.
4827 (init_gcc_specs): Likewise.
4828 (init_spec): Likewise.
4829 (display_help): Update help message for -pie.
4830 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
4833 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
4835 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
4836 (movdi_aarch64): Likewise.
4837 (movti_aarch64): Likewise.
4839 2017-09-12 Simon Wright <simon@pushface.org>
4842 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
4843 calculation of the minor version, always output as 0.
4845 2017-09-12 Jakub Jelinek <jakub@redhat.com>
4848 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
4849 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
4850 on it early, rather than manual conversion late. For
4851 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
4852 instead of performing manual conversion.
4854 2017-09-12 Carl Love <cel@us.ibm.com>
4856 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
4857 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
4859 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4860 VMULOSW): Add definitions.
4861 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4862 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4863 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4864 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4865 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4867 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
4869 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
4871 (movti_aarch64): Likewise.
4872 (movdf_aarch64): Likewise.
4873 (movtf_aarch64): Likewise.
4874 (load_pairdi): Likewise.
4875 (store_pairdi): Likewise.
4876 (load_pairdf): Likewise.
4877 (store_pairdf): Likewise.
4878 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
4879 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
4880 (ldr_got_small_<mode>): Likewise.
4881 (ldr_got_small_28k_<mode>): Likewise.
4882 (ldr_got_tiny): Likewise.
4883 * config/aarch64/iterators.md (ldst_sz): New.
4884 (ldpstp_sz): Likewise.
4885 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
4887 (thunderx_load): Split load_8 to load_16.
4888 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
4890 (thunderx2t99_storepair_basic): Split store_8 to store_16.
4891 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
4892 (xgene1_store_pair): Split store_8 to store_16.
4893 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
4894 (falkor_st_0_st_sd): Split store_8 to store_16.
4896 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
4898 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
4899 and store1/2/3/4 to store_4/8/12/16.
4900 * config/aarch64/aarch64.md: Update for rename.
4901 * config/arm/arm.md: Likewise.: Likewise.
4902 * config/arm/arm.c: Likewise.
4903 * config/arm/thumb1.md: Likewise.
4904 * config/arm/thumb2.md: Likewise.
4905 * config/arm/vfp.md: Likewise.
4906 * config/arm/arm-generic.md: Likewise.
4907 * config/arm/arm1020e.md: Likewise.
4908 * config/arm/arm1026ejs.md: Likewise.
4909 * config/arm/arm1136jfs.md: Likewise.
4910 * config/arm/arm926ejs.md: Likewise.
4911 * config/arm/cortex-a15.md: Likewise.
4912 * config/arm/cortex-a17.md: Likewise.
4913 * config/arm/cortex-a5.md: Likewise.
4914 * config/arm/cortex-a53.md: Likewise.
4915 * config/arm/cortex-a57.md: Likewise.
4916 * config/arm/cortex-a7.md: Likewise.
4917 * config/arm/cortex-a8.md: Likewise.
4918 * config/arm/cortex-a9.md: Likewise.
4919 * config/arm/cortex-m4.md: Likewise.
4920 * config/arm/cortex-m7.md: Likewise.
4921 * config/arm/cortex-r4.md: Likewise.
4922 * config/arm/exynos-m1.md: Likewise.
4923 * config/arm/fa526.md: Likewise.
4924 * config/arm/fa606te.md: Likewise.
4925 * config/arm/fa626te.md: Likewise.
4926 * config/arm/fa726te.md: Likewise.
4927 * config/arm/fmp626.md: Likewise.
4928 * config/arm/iwmmxt.md: Likewise.
4929 * config/arm/ldmstm.md: Likewise.
4930 * config/arm/marvell-pj4.md: Likewise.
4931 * config/arm/xgene1.md: Likewise.
4932 * config/aarch64/thunderx.md: Likewise.
4933 * config/aarch64/thunderx2t99.md: Likewise.
4934 * config/aarch64/falkor.md: Likewise.
4936 2017-09-12 Martin Liska <mliska@suse.cz>
4938 * attribs.c (private_lookup_attribute): New function.
4939 * attribs.h (private_lookup_attribute): Declared here.
4940 (lookup_attribute): Called from this place.
4942 2017-09-12 Richard Biener <rguenther@suse.de>
4944 PR tree-optimization/82157
4945 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
4946 stmts with side-effects.
4948 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4949 Alan Hayward <alan.hayward@arm.com>
4950 David Sherwood <david.sherwood@arm.com>
4952 * target.def (hard_regno_nregs): New hook.
4953 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
4954 * targhooks.h (default_hard_regno_nregs): Declare.
4955 * targhooks.c (default_hard_regno_nregs): New function.
4956 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
4957 (TARGET_HARD_REGNO_NREGS): ...this hook.
4958 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
4959 (CLASS_MAX_NREGS): Likewise.
4960 * doc/tm.texi: Regenerate.
4961 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
4962 instead of HARD_REGNO_NREGS.
4963 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
4964 HARD_REGNO_NREGS in the comment.
4965 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
4966 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
4967 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
4968 Return an unsigned int.
4969 (TARGET_HARD_REGNO_NREGS): Redefine.
4970 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
4971 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
4972 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
4973 (arc_hard_regno_nregs): New function.
4974 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
4975 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
4976 (arm_hard_regno_nregs): New function.
4977 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
4978 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
4979 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
4980 (TARGET_HARD_REGNO_NREGS): Redefine.
4981 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
4982 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
4983 (HARD_REGNO_NREGS): Delete.
4984 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
4985 (cr16_hard_regno_nregs): New function.
4986 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
4987 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
4988 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
4989 (cris_hard_regno_nregs): New function.
4990 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
4991 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
4992 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
4993 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
4994 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
4995 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
4996 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
4997 (frv_hard_regno_nregs): Make static. Take and return an
4999 (frv_class_max_nregs): Remove outdated copy of documentation.
5000 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
5001 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
5002 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
5003 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
5004 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
5005 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
5006 (TARGET_HARD_REGNO_NREGS): Redefine.
5007 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
5008 (CLASS_MAX_NREGS): Update comment.
5009 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
5010 (ia64_hard_regno_nregs): New function.
5011 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
5012 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
5013 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
5014 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
5015 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
5017 (m32c_hard_regno_nregs): Likewise. Make static.
5018 (TARGET_HARD_REGNO_NREGS): Redefine.
5019 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
5020 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
5021 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
5022 (m68k_hard_regno_nregs): New function.
5023 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
5024 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
5025 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
5026 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
5027 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
5028 Take and return an unsigned int.
5029 (TARGET_HARD_REGNO_NREGS): Redefine.
5030 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
5031 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
5032 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
5033 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
5034 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
5035 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
5036 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
5037 (msp430_hard_regno_nregs): Make static. Take and return an
5039 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
5040 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
5041 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
5042 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
5043 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
5044 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
5045 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
5046 (TARGET_HARD_REGNO_NREGS): Redefine.
5047 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
5048 (PA_HARD_REGNO_NREGS): ...this.
5049 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
5050 (PA_HARD_REGNO_NREGS): ...this.
5051 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
5052 (pa_hard_regno_nregs): New function.
5053 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
5054 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
5055 (pdp11_hard_regno_nregs): New function.
5056 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
5057 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
5058 (rs6000_hard_regno_nregs_hook): New function.
5059 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
5060 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
5061 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
5062 Take and return an unsigned int. Move earlier in file.
5063 (TARGET_HARD_REGNO_NREGS): Redefine.
5064 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
5065 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
5066 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
5067 (rl78_hard_regno_nregs): Make static. Take and return an
5069 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
5070 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
5071 (rs6000_hard_regno_nregs_hook): New function.
5072 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
5073 * config/rx/rx.c (rx_hard_regno_nregs): New function.
5074 (TARGET_HARD_REGNO_NREGS): Redefine.
5075 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
5076 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
5077 instead of HARD_REGNO_NREGS.
5078 (s390_hard_regno_nregs): New function.
5079 (s390_hard_regno_mode_ok): Add comment from s390.h.
5080 (TARGET_HARD_REGNO_NREGS): Redefine.
5081 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
5082 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
5083 (sh_hard_regno_nregs): New function.
5084 (sh_pass_in_reg_p): Use it.
5085 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
5086 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
5087 (sparc_hard_regno_nregs): New function.
5088 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
5089 * config/spu/spu.c (spu_hard_regno_nregs): New function.
5090 (spu_function_arg_advance): Use it, supplying a valid register number.
5091 (TARGET_HARD_REGNO_NREGS): Redefine.
5092 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
5093 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
5094 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
5095 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
5096 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
5097 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
5098 (CLASS_MAX_NREGS): Remove copy of old documentation.
5099 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
5100 (visium_hard_regno_nregs): New function.
5101 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
5102 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
5103 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
5104 xtensa_hard_regno_nregs): New function.
5105 * system.h (HARD_REGNO_NREGS): Poison.
5107 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5109 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
5110 hard_regno_nregs instead of HARD_REGNO_NREGS.
5111 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
5112 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
5113 (c6x_expand_epilogue): Likewise.
5114 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
5115 (frv_read_iacc_argument): Likewise.
5116 * config/sh/sh.c: Include regs.h.
5117 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
5118 (regs_used): Likewise.
5119 (output_stack_adjust): Likewise.
5120 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
5121 * expmed.c: Include regs.h.
5122 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
5123 * ree.c: Include regs.h.
5124 (combine_reaching_defs): Use hard_regno_nregs instead of
5126 (add_removable_extension): Likewise.
5128 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5130 * regs.h (hard_regno_nregs): Turn into a function.
5131 (end_hard_regno): Update accordingly.
5132 * caller-save.c (setup_save_areas): Likewise.
5133 (save_call_clobbered_regs): Likewise.
5134 (replace_reg_with_saved_mem): Likewise.
5135 (insert_restore): Likewise.
5136 (insert_save): Likewise.
5137 * combine.c (can_change_dest_mode): Likewise.
5138 (move_deaths): Likewise.
5139 (distribute_notes): Likewise.
5140 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
5141 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
5142 (rs6000_split_multireg_move): Likewise.
5143 (rs6000_register_move_cost): Likewise.
5144 (rs6000_memory_move_cost): Likewise.
5145 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
5146 (rs6000_split_multireg_move): Likewise.
5147 (rs6000_register_move_cost): Likewise.
5148 (rs6000_memory_move_cost): Likewise.
5149 * cselib.c (cselib_reset_table): Likewise.
5150 (cselib_lookup_1): Likewise.
5151 * emit-rtl.c (set_mode_and_regno): Likewise.
5152 * function.c (aggregate_value_p): Likewise.
5153 * ira-color.c (setup_profitable_hard_regs): Likewise.
5154 (check_hard_reg_p): Likewise.
5155 (calculate_saved_nregs): Likewise.
5156 (assign_hard_reg): Likewise.
5157 (improve_allocation): Likewise.
5158 (calculate_spill_cost): Likewise.
5159 * ira-emit.c (modify_move_list): Likewise.
5160 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
5161 (ira_hard_reg_in_set_p): Likewise.
5162 * ira.c (setup_reg_mode_hard_regset): Likewise.
5163 (clarify_prohibited_class_mode_regs): Likewise.
5164 (check_allocation): Likewise.
5165 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5166 (lra_setup_reg_renumber): Likewise.
5167 (setup_try_hard_regno_pseudos): Likewise.
5168 (spill_for): Likewise.
5169 (assign_hard_regno): Likewise.
5170 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
5171 * lra-constraints.c (in_class_p): Likewise.
5172 (lra_constraint_offset): Likewise.
5173 (simplify_operand_subreg): Likewise.
5174 (lra_constraints): Likewise.
5175 (split_reg): Likewise.
5176 (split_if_necessary): Likewise.
5177 (invariant_p): Likewise.
5178 (inherit_in_ebb): Likewise.
5179 * lra-lives.c (process_bb_lives): Likewise.
5180 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
5181 (get_hard_regs): Likewise.
5182 (do_remat): Likewise.
5183 * lra-spills.c (assign_spill_hard_regs): Likewise.
5184 * mode-switching.c (create_pre_exit): Likewise.
5185 * postreload.c (reload_combine_recognize_pattern): Likewise.
5186 * recog.c (peep2_find_free_register): Likewise.
5187 * regcprop.c (kill_value_regno): Likewise.
5188 (set_value_regno): Likewise.
5189 (copy_value): Likewise.
5190 (maybe_mode_change): Likewise.
5191 (find_oldest_value_reg): Likewise.
5192 (copyprop_hardreg_forward_1): Likewise.
5193 * regrename.c (check_new_reg_p): Likewise.
5194 (regrename_do_replace): Likewise.
5195 * reload.c (push_reload): Likewise.
5196 (combine_reloads): Likewise.
5197 (find_dummy_reload): Likewise.
5198 (operands_match_p): Likewise.
5199 (find_reloads): Likewise.
5200 (find_equiv_reg): Likewise.
5201 (reload_adjust_reg_for_mode): Likewise.
5202 * reload1.c (count_pseudo): Likewise.
5203 (count_spilled_pseudo): Likewise.
5204 (find_reg): Likewise.
5205 (clear_reload_reg_in_use): Likewise.
5206 (free_for_value_p): Likewise.
5207 (allocate_reload_reg): Likewise.
5208 (choose_reload_regs): Likewise.
5209 (reload_adjust_reg_for_temp): Likewise.
5210 (emit_reload_insns): Likewise.
5211 (delete_output_reload): Likewise.
5212 * rtlanal.c (subreg_get_info): Likewise.
5213 * sched-deps.c (sched_analyze_reg): Likewise.
5214 * sel-sched.c (init_regs_for_mode): Likewise.
5215 (mark_unavailable_hard_regs): Likewise.
5216 (choose_best_reg_1): Likewise.
5217 (verify_target_availability): Likewise.
5218 * valtrack.c (dead_debug_insert_temp): Likewise.
5219 * var-tracking.c (track_loc_p): Likewise.
5220 (emit_note_insn_var_location): Likewise.
5221 * varasm.c (make_decl_rtl): Likewise.
5222 * reginfo.c (choose_hard_reg_mode): Likewise.
5223 (init_reg_modes_target): Refer directly to
5224 this_target_regs->x_hard_regno_nregs.
5226 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5228 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
5229 instead of hard_regno_nregs.
5231 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5233 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
5234 end_hard_regno instead of hard_regno_nregs.
5235 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
5236 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
5237 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
5238 * ira-color.c (improve_allocation): Likewise.
5239 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5240 * lra-lives.c (mark_regno_live): Likewise.
5241 (mark_regno_dead): Likewise.
5242 * lra-remat.c (operand_to_remat): Likewise.
5243 * lra.c (collect_non_operand_hard_regs): Likewise.
5244 * postreload.c (reload_combine_note_store): Likewise.
5245 (move2add_valid_value_p): Likewise.
5246 * reload.c (regno_clobbered_p): Likewise.
5248 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5250 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
5252 * config/v850/v850.c (v850_reorg): Likewise.
5253 * reload.c (refers_to_regno_for_reload_p): Likewise.
5254 (find_equiv_reg): Likewise.
5255 * reload1.c (reload_reg_reaches_end_p): Likewise.
5257 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5259 * caller-save.c (add_used_regs): Use REG_NREGS instead of
5261 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
5262 * config/arm/arm.c (output_move_neon): Likewise.
5263 (arm_attr_length_move_neon): Likewise.
5264 (neon_split_vcombine): Likewise.
5265 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
5266 (c6x_mark_reg_written): Likewise.
5267 (c6x_dwarf_register_span): Likewise.
5268 * config/i386/i386.c (ix86_save_reg): Likewise.
5269 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
5270 (rws_access_reg): Likewise.
5271 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
5272 * mode-switching.c (create_pre_exit): Likewise.
5273 * ree.c (combine_reaching_defs): Likewise.
5274 (add_removable_extension): Likewise.
5275 * regcprop.c (find_oldest_value_reg): Likewise.
5276 (copyprop_hardreg_forward_1): Likewise.
5277 * reload.c (reload_inner_reg_of_subreg): Likewise.
5278 (push_reload): Likewise.
5279 (combine_reloads): Likewise.
5280 (find_dummy_reload): Likewise.
5281 (reload_adjust_reg_for_mode): Likewise.
5282 * reload1.c (find_reload_regs): Likewise.
5283 (forget_old_reloads_1): Likewise.
5284 (reload_reg_free_for_value_p): Likewise.
5285 (reload_adjust_reg_for_temp): Likewise.
5286 (emit_reload_insns): Likewise.
5287 (delete_output_reload): Likewise.
5288 * sel-sched.c (choose_best_reg_1): Likewise.
5289 (choose_best_pseudo_reg): Likewise.
5291 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5292 Alan Hayward <alan.hayward@arm.com>
5293 David Sherwood <david.sherwood@arm.com>
5295 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
5296 * target.def (slow_unaligned_access): New hook.
5297 * targhooks.h (default_slow_unaligned_access): Declare.
5298 * targhooks.c (default_slow_unaligned_access): New function.
5299 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
5300 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
5301 * doc/tm.texi: Regenerate.
5302 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
5303 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
5304 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
5306 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
5307 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
5309 (rs6000_slow_unaligned_access): New function.
5310 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
5311 (expand_block_compare): Likewise.
5312 (expand_strn_compare): Likewise.
5313 (rs6000_rtx_costs): Likewise.
5314 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
5315 (riscv_slow_unaligned_access): Likewise.
5316 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
5317 (riscv_slow_unaligned_access_p): ...this and make static.
5318 (riscv_option_override): Update accordingly.
5319 (riscv_slow_unaligned_access): New function.
5320 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
5321 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
5322 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
5323 (rs6000_slow_unaligned_access): New function.
5324 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
5325 (rs6000_rtx_costs): Likewise.
5326 * config/rs6000/rs6000-string.c (expand_block_compare)
5327 (expand_strn_compare): Use targetm.slow_unaligned_access instead
5328 of SLOW_UNALIGNED_ACCESS.
5329 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
5330 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
5331 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
5332 of SLOW_UNALIGNED_ACCESS.
5333 * expmed.c (simple_mem_bitfield_p): Likewise.
5334 * expr.c (alignment_for_piecewise_move): Likewise.
5335 (emit_group_load_1): Likewise.
5336 (emit_group_store): Likewise.
5337 (copy_blkmode_from_reg): Likewise.
5338 (emit_push_insn): Likewise.
5339 (expand_assignment): Likewise.
5340 (store_field): Likewise.
5341 (expand_expr_real_1): Likewise.
5342 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
5343 * lra-constraints.c (simplify_operand_subreg): Likewise.
5344 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
5345 * gimple-ssa-store-merging.c: Likewise in block comment at start
5347 * tree-ssa-strlen.c: Include target.h.
5348 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
5349 of SLOW_UNALIGNED_ACCESS.
5350 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
5352 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5354 PR rtl-optimization/82185
5355 * expmed.c (emit_store_flag_int): Only test tem if it has been
5358 2017-09-12 Richard Biener <rguenther@suse.de>
5361 * match.pd ((FTYPE) N CMP CST): Fix typo.
5363 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
5365 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
5367 (mips_near_type_p): Add 'short_call' attribute as a synonym
5369 * doc/extend.texi (short_call): Document new function attribute.
5371 2017-09-12 Jakub Jelinek <jakub@redhat.com>
5374 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
5375 assertion check that in the condition.
5376 (get_atomic_generic_size): Likewise. Before testing if parameter
5377 has pointer type, if it has array type, call for C++
5378 default_conversion to perform array-to-pointer conversion.
5380 2017-09-12 Richard Biener <rguenther@suse.de>
5382 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
5383 for operations we cannot scalarize.
5385 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
5387 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
5388 vectors heap vectors. Clean up comments.
5389 Make visited_bbs a reference.
5390 (profitable_jump_thread_path): Make GC
5391 vectors heap vectors. Clean up comments.
5393 (convert_and_register_jump_thread_path): Make GC vectors heap
5395 (check_subpath_and_update_thread_path): Same. Clean up comments.
5396 Make visited_bbs a reference.
5397 (handle_phi): Abstract common code to to
5398 register_jump_thread_path_if_profitable.
5399 Rename VAR_BB to DEF_BB.
5401 Make GC vectors heap vectors.
5402 Make visited_bbs a reference.
5403 (handle_assignment): Same.
5404 (register_jump_thread_path_if_profitable): New.
5405 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
5407 Make GC vectors heap vectors. Clean up comments.
5408 Make visited_bbs a reference.
5409 (find_jump_threads_backwards): Make visited_bbs live in the stack.
5410 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
5413 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
5416 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
5417 words of E_DImode object are reachable by xtensa_uimm8x4 access.
5419 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
5421 Revert r251800 and r251799.
5423 2017-09-11 Martin Jambor <mjambor@suse.cz>
5426 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
5427 arguments in advance.
5428 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
5429 use it to find predecessor edges.
5430 (naive_outof_ssa): Collect vector of predecessors.
5432 2017-09-08 Jason Merrill <jason@redhat.com>
5434 PR c++/70029 - ICE with ref-qualifier and -flto
5435 * langhooks.h (struct lang_hooks_for_types): Add
5436 copy_lang_qualifiers.
5437 * attribs.c (build_type_attribute_qual_variant): Use it.
5438 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
5440 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
5441 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
5443 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
5446 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
5447 (*mulsi3_sp64): New instruction.
5448 (mulsi3): New expander.
5450 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
5452 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
5454 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5456 * sancov.c: Include memmodel.h.
5458 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
5461 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
5464 2017-09-07 Carl Love <cel@us.ibm.com>
5466 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
5467 the sldi instruction.
5469 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
5471 * sancov.c: Include tm_p.h.
5473 2017-09-07 Jakub Jelinek <jakub@redhat.com>
5476 * output.h (switch_to_other_text_partition): New declaration.
5477 * varasm.c (switch_to_other_text_partition): New function.
5478 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
5479 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
5480 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
5481 to the other text partition before emitting LCL label and switch back
5482 after emitting the word after it.
5484 2017-09-07 Richard Biener <rguenther@suse.de>
5486 * passes.def (pass_split_crit_edges): Remove instance before PRE.
5487 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
5488 critical edges here, after loop init.
5489 (pass_data_pre): Remove PROP_no_crit_edges flags.
5490 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
5491 for valueization of call args to avoid leaking VN_TOP.
5492 (visit_use): Assert we do not visit default defs.
5493 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
5494 Use error_mark_node to more easily detect leaking VN_TOP.
5495 All default-defs are varying, not VN_TOP. Mark them visited.
5496 (run_scc_vn): Make code match comment.
5498 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5500 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
5501 OPTION_MASK_FLOAT128_KEYWORD.
5502 (POWERPC_MASKS): Likewise.
5503 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
5504 support for the -mfloat128-type option, and make -mfloat128
5505 default on PowerPC Linux systems. Define or undefine
5506 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
5507 Define __float128 to be __ieee128 if IEEE 128-bit support is
5508 enabled, or undefine it.
5509 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
5510 Delete defining __FLOAT128_TYPE__.
5511 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
5512 -mfloat128-type option and make -mfloat128 default on PowerPC
5514 (TARGET_FLOAT128_TYPE): Likewise.
5515 (-mfloat128-type): Likewise.
5516 * config/rs6000/rs6000.c (rs6000_option_override_internal):
5517 Delete the -mfloat128-type option and make -mfloat128 default on
5518 PowerPC Linux systems. Always use __ieee128 to be the keyword for
5519 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
5520 128-bit floating point is enabled. Change tests from using
5521 -mfloat128-type to -mfloat128.
5522 (rs6000_mangle_type): Use the correct mangling for the __float128
5523 type even if normal long double is restricted to 64-bits.
5524 (floatn_mode): Enable the _Float128 type by default on VSX Linux
5526 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
5527 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
5528 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
5530 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
5531 documentation for -mfloat128.
5533 2017-09-06 Olivier Hainque <hainque@adacore.com>
5535 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
5537 2017-09-06 Wish Wu <wishwu007@gmail.com>
5538 Jakub Jelinek <jakub@redhat.com>
5540 * asan.c (initialize_sanitizer_builtins): Add
5541 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
5542 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
5543 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
5544 BT_FN_VOID_UINT64_PTR variables.
5545 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
5546 (BT_FN_VOID_UINT16_UINT16): Likewise.
5547 (BT_FN_VOID_UINT32_UINT32): Likewise.
5548 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
5549 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
5550 (BT_FN_VOID_UINT64_PTR): Likewise.
5551 * common.opt (flag_sanitize_coverage): New variable.
5552 (fsanitize-coverage=trace-pc): Remove.
5553 (fsanitize-coverage=): Add.
5554 * flag-types.h (enum sanitize_coverage_code): New enum.
5555 * fold-const.c (fold_range_test): Disable non-short-circuit
5556 optimization if flag_sanitize_coverage.
5557 (fold_truth_andor): Likewise.
5558 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
5559 * opts.c (COVERAGE_SANITIZER_OPT): Define.
5560 (coverage_sanitizer_opts): New array.
5561 (get_closest_sanitizer_option): Add OPTS argument, handle also
5562 OPT_fsanitize_coverage_.
5563 (parse_sanitizer_options): Adjusted to also handle
5564 OPT_fsanitize_coverage_.
5565 (common_handle_option): Add OPT_fsanitize_coverage_.
5566 * sancov.c (instrument_comparison, instrument_switch): New function.
5567 (sancov_pass): Add trace-cmp support.
5568 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
5569 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
5570 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
5571 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
5572 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
5573 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
5574 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
5575 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
5576 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
5577 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
5579 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
5581 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
5582 error. Only quit immediately if parsing is complete.
5583 (BEGIN): Initialize fatal_err and parse_done.
5584 (begin fpu, end fpu): Check number of arguments.
5585 (begin arch, end arch): Likewise.
5586 (begin cpu, end cpu): Likewise.
5587 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
5588 (optalias): Likewise.
5590 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
5592 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
5593 * config/arm/arm-isa.h: Delete. Move definitions to ...
5594 * arm-cpus.in: ... here. Use new feature and fgroup values.
5595 * config/arm/arm.c (arm_option_override): Use lower case for feature
5597 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
5598 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
5599 * config/arm/parsecpu.awk (END): Add new command 'isa'.
5601 (print_isa_bits_for): New function.
5602 (gen_isa): New function.
5603 (gen_comm_data): Use print_isa_bits_for.
5604 (define feature): New keyword.
5605 (define fgroup): New keyword.
5606 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
5607 (arm-isa.h): Add rule to generate file.
5608 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
5609 case for feature bit names.
5611 2017-09-06 Richard Biener <rguenther@suse.de>
5613 * tree-ssa-pre.c (NECESSARY): Remove.
5614 (create_expression_by_pieces): Do not touch pass-local flags.
5615 (insert_into_preds_of_block): Likewise.
5616 (do_pre_regular_insertion): Likewise.
5617 (eliminate_insert): Likewise.
5618 (eliminate_dom_walker::before_dom_children): Likewise.
5619 (fini_eliminate): Do not look at inserted_exprs.
5620 (mark_operand_necessary): Remove.
5621 (remove_dead_inserted_code): Replace with simple work-list
5622 algorithm based on inserted_exprs and SSA uses.
5623 (pass_pre::execute): Re-order fini_eliminate and
5624 remove_dead_inserted_code.
5626 2017-09-06 Olivier Hainque <hainque@adacore.com>
5628 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
5629 for VxWorks 7. Adjust surrounding comments.
5631 2017-09-06 Richard Biener <rguenther@suse.de>
5633 * gimple-ssa-strength-reduction.c
5634 (find_candidates_dom_walker::before_dom_children): Also allow
5637 2017-09-06 Richard Biener <rguenther@suse.de>
5639 PR tree-optimization/82108
5640 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
5641 for gap in the non-permutation SLP case.
5643 2017-09-06 Martin Jambor <mjambor@suse.cz>
5645 PR tree-optimization/82078
5646 * tree-sra.c (sort_and_splice_var_accesses): Move call to
5647 add_access_to_work_queue...
5648 (build_accesses_from_assign): ...here.
5649 (propagate_all_subaccesses): Make sure racc is the group
5650 representative, if there is one.
5652 2017-09-06 Jakub Jelinek <jakub@redhat.com>
5655 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
5658 2017-09-06 Richard Biener <rguenther@suse.de>
5660 * gimple-ssa-strength-reduction.c
5661 (find_candidates_dom_walker::before_doom_children): Use a
5662 type and not a mode check.
5664 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
5667 * config/arm/predicates.md (arm_general_adddi_operand): Create new
5669 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
5671 2017-09-05 Jeff Law <law@redhat.com>
5673 PR tree-optimization/64910
5674 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
5675 cases where we have 3 or more operands.
5677 2017-09-05 Jakub Jelinek <jakub@redhat.com>
5680 * omp-low.c (lower_omp_for): Recompute tree invariant if
5681 gimple_omp_for_initial/final is ADDR_EXPR.
5684 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
5685 into gimple val before gimplification fo the COND_EXPR.
5687 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
5689 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
5690 REGION_COPY argument.
5691 (thread_through_all_blocks): Remove unused argument to
5692 duplicate_thread_path.
5694 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5695 Alan Hayward <alan.hayward@arm.com>
5696 David Sherwood <david.sherwood@arm.com>
5698 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
5699 Take a scalar_mode rather than a machine_mode.
5700 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
5701 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
5702 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
5703 (aarch64_gen_adjusted_ldpstp): Likewise.
5704 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
5706 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5707 Alan Hayward <alan.hayward@arm.com>
5708 David Sherwood <david.sherwood@arm.com>
5710 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
5711 Take a scalar_int_mode instead of a machine_mode.
5712 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
5713 (aarch64_output_scalar_simd_mov_immediate): Likewise.
5714 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
5715 (aarch64_simd_attr_length_rglist): Delete.
5716 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
5717 a scalar_int_mode instead of a machine_mode.
5718 (aarch64_add_offset): Likewise.
5719 (aarch64_internal_mov_immediate): Likewise
5720 (aarch64_add_constant_internal): Likewise.
5721 (aarch64_add_constant): Likewise.
5722 (aarch64_movw_imm): Likewise.
5723 (aarch64_rtx_arith_op_extract_p): Likewise.
5724 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
5725 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
5726 Remove assert that the mode isn't a vector.
5727 (aarch64_output_scalar_simd_mov_immediate): Likewise.
5728 (aarch64_expand_mov_immediate): Update calls after above changes.
5729 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
5730 (aarch64_and_bitmask_imm): Check for scalar integer modes.
5731 (aarch64_move_imm): Likewise.
5732 (aarch64_can_const_movi_rtx_p): Likewise.
5733 (aarch64_strip_extend): Likewise.
5734 (aarch64_extr_rtx_p): Likewise.
5735 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
5736 a CONST_INT when the mode parameter is VOIDmode.
5737 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
5739 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5741 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
5742 * stor-layout.c (bitwise_mode_for_mode): Likewise.
5743 (bitwise_type_for_mode): Update accordingly.
5745 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5747 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
5748 * stor-layout.c (mode_for_size_tree): Likewise.
5749 (mode_for_array): Update accordingly.
5750 (layout_decl): Likewise.
5751 (compute_record_mode): Likewise. Only set the mode once.
5753 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5755 * target.def (get_mask_mode): Change return type to opt_mode.
5757 * doc/tm.texi: Regenerate.
5758 * targhooks.h (default_get_mask_mode): Return an opt_mode.
5759 * targhooks.c (default_get_mask_mode): Likewise.
5760 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
5761 * optabs-query.c (can_vec_mask_load_store_p): Update use of
5762 targetm.get_mask_mode.
5763 * tree.c (build_truth_vector_type): Likewise.
5765 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5767 * machmode.h (mode_for_vector): Return an opt_mode.
5768 * stor-layout.c (mode_for_vector): Likewise.
5769 (mode_for_int_vector): Update accordingly.
5770 (layout_type): Likewise.
5771 * config/i386/i386.c (emit_memmov): Likewise.
5772 (ix86_expand_set_or_movmem): Likewise.
5773 (ix86_expand_vector_init): Likewise.
5774 (ix86_get_mask_mode): Likewise.
5775 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
5777 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
5778 * expmed.c (extract_bit_field_1): Likewise.
5779 * expr.c (expand_expr_real_2): Likewise.
5780 * optabs-query.c (can_vec_perm_p): Likewise.
5781 (can_vec_mask_load_store_p): Likewise.
5782 * optabs.c (expand_vec_perm): Likewise.
5783 * targhooks.c (default_get_mask_mode): Likewise.
5784 * tree-vect-stmts.c (vectorizable_store): Likewise.
5785 (vectorizable_load): Likewise.
5786 (get_vectype_for_scalar_type_and_size): Likewise.
5788 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5790 * machmode.h (mode_for_int_vector): New function.
5791 * stor-layout.c (mode_for_int_vector): Likewise.
5792 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
5793 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
5794 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
5795 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
5796 (s390_expand_vcond): Likewise.
5798 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5800 * machmode.h (opt_machine_mode): New type.
5801 (opt_mode<T>): Allow construction from anything that can be
5803 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
5804 (mode_for_size): Return an opt_machine_mode.
5805 * stor-layout.c (mode_for_size): Likewise.
5806 (mode_for_size_tree): Update call accordingly.
5807 (bitwise_mode_for_mode): Likewise.
5808 (make_fract_type): Likewise.
5809 (make_accum_type): Likewise.
5810 * caller-save.c (replace_reg_with_saved_mem): Update call
5812 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5813 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5814 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5815 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5816 * expmed.c (extract_bit_field_1): Likewise.
5817 * reload.c (get_secondary_mem): Likewise.
5818 * varasm.c (assemble_integer): Likewise.
5819 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
5822 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5824 * machmode.h (decimal_float_mode_for_size): New function.
5825 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
5826 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
5827 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
5828 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
5829 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
5830 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
5832 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5834 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
5835 (get_builtin_sync_mode): Likewise.
5836 (expand_ifn_atomic_compare_exchange): Likewise.
5837 (expand_builtin_atomic_clear): Likewise.
5838 (expand_builtin_atomic_test_and_set): Likewise.
5839 (fold_builtin_atomic_always_lock_free): Likewise.
5840 * calls.c (compute_argument_addresses): Likewise.
5841 (emit_library_call_value_1): Likewise.
5842 (store_one_arg): Likewise.
5843 * combine.c (combine_instructions): Likewise.
5844 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
5845 * config/arm/arm.c (arm_function_value): Likewise.
5846 (aapcs_allocate_return_reg): Likewise.
5847 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
5848 * config/i386/i386.c (construct_container): Likewise.
5849 (ix86_gimplify_va_arg): Likewise.
5850 (ix86_expand_sse_cmp): Likewise.
5851 (emit_memmov): Likewise.
5852 (emit_memset): Likewise.
5853 (expand_small_movmem_or_setmem): Likewise.
5854 (ix86_expand_pextr): Likewise.
5855 (ix86_expand_pinsr): Likewise.
5856 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
5857 * config/microblaze/microblaze.c (microblaze_block_move_straight):
5859 * config/mips/mips.c (mips_function_value_1) Likewise.
5860 (mips_block_move_straight): Likewise.
5861 (mips_expand_ins_as_unaligned_store): Likewise.
5862 * config/powerpcspe/powerpcspe.c
5863 (rs6000_darwin64_record_arg_advance_flush): Likewise.
5864 (rs6000_darwin64_record_arg_flush): Likewise.
5865 * config/rs6000/rs6000.c
5866 (rs6000_darwin64_record_arg_advance_flush): Likewise.
5867 (rs6000_darwin64_record_arg_flush): Likewise.
5868 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
5869 (sparc_function_value_1): Likewise.
5870 * config/spu/spu.c (adjust_operand): Likewise.
5871 (spu_emit_branch_or_set): Likewise.
5872 (arith_immediate_p): Likewise.
5873 * emit-rtl.c (gen_lowpart_common): Likewise.
5874 * expr.c (expand_expr_real_1): Likewise.
5875 * function.c (assign_parm_setup_block): Likewise.
5876 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
5877 * reload1.c (alter_reg): Likewise.
5878 * stor-layout.c (mode_for_vector): Likewise.
5879 (layout_type): Likewise.
5881 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5883 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
5884 (spu_convert_move): Likewise.
5885 * lower-subreg.c (resolve_simple_move): Likewise.
5887 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5890 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
5891 define_insn to a define_expand.
5892 (altivec_vsum2sws_direct): New define_insn.
5893 (altivec_vsumsws): Convert from a define_insn to a define_expand.
5895 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
5897 * config/arm/arm.c (arm_option_params_internal): Improve setting of
5900 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
5905 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
5906 realign stack if stack alignment needed is less than incoming
5909 2017-09-05 Marek Polacek <polacek@redhat.com>
5912 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
5915 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
5917 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
5919 2017-09-05 Richard Biener <rguenther@suse.de>
5921 PR tree-optimization/82084
5922 * fold-const.c (can_native_encode_string_p): Handle wide characters.
5924 2017-09-05 Richard Biener <rguenther@suse.de>
5926 PR tree-optimization/82102
5927 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
5929 2017-09-05 Martin Liska <mliska@suse.cz>
5931 PR tree-optimization/82032
5932 * tree-cfg.c (generate_range_test): New function.
5933 * tree-cfg.h (generate_range_test): Declared here.
5934 * tree-cfgcleanup.c (convert_single_case_switch): New function.
5935 (cleanup_control_expr_graph): Use it.
5936 * tree-switch-conversion.c (try_switch_expansion): Remove
5938 (emit_case_nodes): Use generate_range_test.
5940 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
5943 * config/i386/i386.md (*<btsc><mode>_mask): Add
5944 TARGET_USE_BT to insn constraint.
5945 (*btr<mode>_mask): Ditto.
5947 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
5949 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
5950 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
5952 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
5955 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
5956 TARGET_NEON and TARGET_IWMMXT.
5957 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
5958 TARGET_NEON and TARGET_IWMMXT.
5959 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
5961 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
5963 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
5964 (ix86_rewrite_tls_address): Ditto.
5965 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
5966 (ix86_rewrite_tls_address_1): Ditto.
5967 (ix86_rewrite_tls_address): Ditto.
5968 * config/i386/predicates.md (tls_address_pattern): New predicate.
5969 * config/i386/i386.md (TLS address splitter): New splitter.
5971 2017-09-04 Richard Biener <rguenther@suse.de>
5973 PR tree-optimization/82084
5974 * fold-const.h (can_native_encode_string_p): Declare.
5975 * fold-const.c (can_native_encode_string_p): Factor out from ...
5976 (native_encode_string): ... here.
5977 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
5978 vectorizing stores from constants we later cannot handle.
5980 2017-09-04 Marek Polacek <polacek@redhat.com>
5983 * doc/invoke.texi: Update -Wtautological-compare documentation.
5985 2017-09-04 Jeff Law <law@redhat.com>
5987 PR tree-optimization/64910
5988 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
5989 swap the first and last operand if the last is a constant.
5991 2017-09-04 Marek Polacek <polacek@redhat.com>
5994 * convert.c (do_narrow): When sanitizing signed integer overflows,
5995 bail out for signed types.
5996 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
5998 2017-09-04 Richard Biener <rguenther@suse.de>
6000 PR tree-optimization/82060
6001 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
6002 Move devirtualization after stmt folding and before EH/AB/noreturn
6003 cleanup to get the stmt refs canonicalized. Use a bool instead
6004 of gimple_modified_p since that doesn't work for NOPs. Schedule
6005 NOPs generated by folding for removal.
6007 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6008 Alan Hayward <alan.hayward@arm.com>
6009 David Sherwood <david.sherwood@arm.com>
6011 * coretypes.h (pad_direction): New enum.
6012 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
6013 (FUNCTION_ARG_PADDING): Likewise.
6014 * target.def (function_arg_padding): New hook.
6015 * targhooks.h (default_function_arg_padding): Declare.
6016 * targhooks.c (default_function_arg_padding): New function.
6017 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
6018 (TARGET_FUNCTION_ARG_PADDING): ...this.
6019 * doc/tm.texi: Regenerate.
6020 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
6021 instead of direction.
6022 (compute_argument_addresses): Likewise.
6023 (load_register_parameters): Likewise.
6024 (emit_library_call_value_1): Likewise.
6025 (store_one_arg): Use targetm.calls.function_arg_padding instead
6026 of FUNCTION_ARG_PADDING.
6027 (must_pass_in_stack_var_size_or_pad): Likewise.
6028 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
6029 (emit_group_store): Likewise.
6030 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
6031 instead of FUNCTION_ARG_PADDING.
6032 (emit_push_insn): Likewise, and propagate enum change throughout
6034 * function.h (direction): Delete.
6035 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
6037 * function.c (assign_parm_find_stack_rtl): Likewise.
6038 (assign_parm_setup_block_p): Likewise.
6039 (assign_parm_setup_block): Likewise.
6040 (gimplify_parameters): Likewise.
6041 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
6042 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
6044 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
6045 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
6046 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
6047 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
6048 (aarch64_function_arg_padding): ...this new function.
6049 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
6050 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6051 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
6052 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
6053 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
6054 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6055 (arm_pad_arg_upward): Replace with...
6056 (arm_function_arg_padding): ...this new function.
6057 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
6059 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
6060 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
6061 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6062 (ia64_hpux_function_arg_padding): Replace with...
6063 (ia64_function_arg_padding): ...this new function. Use pad_direction
6064 instead of direction. Check for TARGET_HPUX.
6065 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
6066 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6067 (iq2000_function_arg_padding): New function.
6068 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
6069 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
6070 (mips_function_arg_padding): ...this new function.
6071 (mips_pad_reg_upward): Update accordingly.
6072 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6073 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
6074 targetm.calls.function_arg_padding.
6075 (FUNCTION_ARG_PADDING): Delete.
6076 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
6077 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
6078 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
6079 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
6080 (nios2_block_reg_padding): Return pad_direction instead of direction.
6081 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
6082 instead of direction.
6083 (nios2_function_arg_padding): Likewise. Make static.
6084 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6085 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
6086 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
6087 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
6088 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6089 (pa_function_arg_padding): Make static. Return pad_direction instead
6091 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
6092 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
6093 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
6094 instead of direction. Use targetm.calls.function_arg_padding.
6095 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
6096 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
6097 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
6098 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
6099 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
6101 (function_arg_padding): Rename to...
6102 (rs6000_function_arg_padding): ...this. Make static. Return
6103 pad_direction instead of direction.
6104 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
6105 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
6106 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
6107 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
6108 instead of direction. Use targetm.calls.function_arg_padding.
6109 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
6110 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
6111 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
6112 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
6113 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6114 (function_arg_padding): Rename to...
6115 (rs6000_function_arg_padding): ...this. Make static. Return
6116 pad_direction instead of direction.
6117 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
6118 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
6119 * config/s390/s390.c (s390_function_arg_padding): New function.
6120 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6121 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
6122 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
6123 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6124 (function_arg_padding): Rename to...
6125 (sparc_function_arg_padding): ...this. Make static. Return
6126 pad_direction instead of direction.
6127 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
6128 * config/spu/spu.c (spu_function_arg_padding): New function.
6129 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6130 * system.h (FUNCTION_ARG_PADDING): Poison.
6132 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6133 Alan Hayward <alan.hayward@arm.com>
6134 David Sherwood <david.sherwood@arm.com>
6136 * target.def (modes_tieable_p): New hook.
6137 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
6138 (TARGET_MODES_TIEABLE_P): ...this.
6139 * doc/tm.texi.in: Regenerate.
6140 * hooks.h (hook_bool_mode_mode_true): Declare.
6141 * hooks.c (hook_bool_mode_mode_true): New function.
6142 * combine.c (subst): Use targetm.modes_tieable_p instead of
6144 * dse.c (find_shift_sequence): Likewise.
6145 * expmed.c (extract_low_bits): Likewise.
6146 * lower-subreg.c: Include target.h.
6147 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
6149 * rtlanal.c (rtx_cost): Likewise.
6150 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
6151 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
6152 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
6153 (TARGET_MODES_TIEABLE_P): Redefine.
6154 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
6155 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
6156 (TARGET_MODES_TIEABLE_P): Redefine.
6157 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
6158 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
6159 (arc_modes_tieable_p): New function.
6160 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
6161 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
6162 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
6163 (arm_modes_tieable_p): Make static.
6164 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
6165 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
6166 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
6167 (TARGET_MODES_TIEABLE_P): Redefine.
6168 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
6169 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
6170 (TARGET_MODES_TIEABLE_P): Redefine.
6171 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
6172 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
6173 (cr16_modes_tieable_p): New function.
6174 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
6175 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
6176 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
6177 (TRULY_NOOP_TRUNCATION): Update comment.
6178 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
6179 (TRULY_NOOP_TRUNCATION): Update comment.
6180 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
6181 (frv_modes_tieable_p): New function.
6182 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
6183 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
6184 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
6185 (TARGET_MODES_TIEABLE_P): Redefine.
6186 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
6187 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
6188 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
6189 (TARGET_MODES_TIEABLE_P): Redefine.
6190 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
6191 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
6192 (ia64_modes_tieable_p): New function.
6193 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
6194 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
6195 (iq2000_modes_tieable_p): New function.
6196 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
6197 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
6198 (lm32_modes_tieable_p): New function.
6199 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
6200 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
6201 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
6202 (TARGET_MODES_TIEABLE_P): Redefine.
6203 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
6204 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
6205 (m32r_modes_tieable_p): New function.
6206 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
6207 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
6208 (m68k_modes_tieable_p): New function.
6209 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
6210 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
6211 (mcore_modes_tieable_p): New function.
6212 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
6213 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
6215 (TARGET_MODES_TIEABLE_P): Redefine.
6216 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
6217 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
6218 * config/mips/mips.c (mips_modes_tieable_p): Make static.
6219 (TARGET_MODES_TIEABLE_P): Redefine.
6220 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
6221 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
6222 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
6223 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
6224 (mn10300_modes_tieable_p): ...this and make static.
6225 (TARGET_MODES_TIEABLE_P): Redefine.
6226 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
6227 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
6228 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
6229 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
6230 (msp430_modes_tieable_p): Make static.
6231 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
6232 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
6233 (TARGET_MODES_TIEABLE_P): Redefine.
6234 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
6235 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
6236 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
6237 (TARGET_MODES_TIEABLE_P): Redefine.
6238 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
6239 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
6240 * config/pa/pa.c (pa_modes_tieable_p): Make static.
6241 (TARGET_MODES_TIEABLE_P): Redefine.
6242 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
6243 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
6244 (pdp11_modes_tieable_p): New function.
6245 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
6246 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
6247 (rs6000_modes_tieable_p): New function.
6248 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
6249 * config/powerpcspe/powerpcspe.md: Update comment.
6250 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
6251 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
6252 (TARGET_MODES_TIEABLE_P): Redefine.
6253 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
6254 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
6255 (rl78_modes_tieable_p): New function.
6256 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
6257 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
6258 (rs6000_modes_tieable_p): New function.
6259 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
6260 * config/rs6000/rs6000.md: Update comment.
6261 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
6262 * config/rx/rx.c (rx_modes_tieable_p): New function.
6263 (TARGET_MODES_TIEABLE_P): Redefine.
6264 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
6265 * config/s390/s390.c (s390_modes_tieable_p): New function.
6266 (TARGET_MODES_TIEABLE_P): Redefine.
6267 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
6268 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
6269 (sh_modes_tieable_p): New function.
6270 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
6271 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
6272 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
6273 (sparc_modes_tieable_p): Make static.
6274 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
6275 * config/spu/spu.c (spu_modes_tieable_p): New function.
6276 (TARGET_MODES_TIEABLE_P): Redefine.
6277 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
6278 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
6279 (TARGET_MODES_TIEABLE_P): Redefine.
6280 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
6281 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
6282 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
6283 * config/v850/v850.c (v850_modes_tieable_p): New function.
6284 (TARGET_MODES_TIEABLE_P): Redefine.
6285 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
6286 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
6287 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
6288 (visium_modes_tieable_p): New function.
6289 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
6290 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
6291 (xtensa_modes_tieable_p): New function.
6292 * system.h (MODES_TIEABLE_P): Poison.
6294 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6295 Alan Hayward <alan.hayward@arm.com>
6296 David Sherwood <david.sherwood@arm.com>
6298 * target.def (hard_regno_mode_ok): New hook.
6299 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
6300 (TARGET_HARD_REGNO_MODE_OK): ...this.
6301 * doc/tm.texi.in: Regenerate.
6302 * hooks.h (hook_bool_uint_mode_true): Declare.
6303 * hooks.c (hook_bool_uint_mode_true): New function.
6304 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
6306 * genpreds.c (write_insn_preds_c): Add an include of target.h.
6307 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
6308 instead of HARD_REGNO_MODE_OK.
6309 * caller-save.c: Include target.h.
6310 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
6312 * combine.c (can_combine_p): Likewise.
6313 (combinable_i3pat): Likewise.
6314 (can_change_dest_mode): Likewise.
6315 * expr.c (init_expr_target): Likewise.
6316 (convert_move): Likewise.
6317 (convert_modes): Likewise.
6318 * ira.c (setup_prohibited_class_mode_regs): Likewise.
6319 (setup_prohibited_mode_move_regs): Likewise.
6320 * ira.h (target_ira): Likewise.
6321 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6322 * lra-constraints.c (process_alt_operands): Likewise.
6323 (split_reg): Likewise.
6324 * recog.c (peep2_find_free_register): Likewise.
6325 * ree.c (combine_reaching_defs): Likewise.
6326 * regcprop.c (maybe_mode_change): Likewise.
6327 * reginfo.c (init_reg_sets_1): Likewise.
6328 (choose_hard_reg_mode): Likewise.
6329 (simplifiable_subregs): Likewise.
6330 * regrename.c (check_new_reg_p): Likewise.
6331 * reload.c (find_valid_class): Likewise.
6332 (find_valid_class_1): Likewise.
6333 (reload_inner_reg_of_subreg): Likewise.
6334 (push_reload): Likewise.
6335 (combine_reloads): Likewise.
6336 (find_dummy_reload): Likewise.
6337 (find_reloads): Likewise.
6338 * reload1.c (find_reg): Likewise.
6339 (set_reload_reg): Likewise.
6340 (allocate_reload_reg): Likewise.
6341 (choose_reload_regs): Likewise.
6342 (reload_adjust_reg_for_temp): Likewise.
6343 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
6344 (simplify_subreg_regno): Likewise.
6345 * sel-sched.c (init_regs_for_mode): Likewise.
6346 * varasm.c (make_decl_rtl): Likewise.
6347 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
6348 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
6350 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
6351 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
6352 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6353 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
6354 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
6355 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6356 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
6357 (arc_mode_class): Delete.
6358 (HARD_REGNO_MODE_OK): Delete.
6359 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6360 (arc_hard_regno_mode_ok): Rename old array to...
6361 (arc_hard_regno_mode_ok_modes): ...this.
6362 (arc_conditional_register_usage): Update accordingly.
6363 (arc_mode_class): Make static.
6364 (arc_hard_regno_mode_ok): New function.
6365 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
6366 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
6367 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6368 (arm_hard_regno_mode_ok): Make static.
6369 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
6371 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
6372 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
6373 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
6375 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6376 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
6377 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
6378 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
6379 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
6380 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6381 * config/bfin/predicates.md (valid_reg_operand): Use
6382 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
6383 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
6384 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
6385 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6386 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
6387 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
6388 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6389 (cr16_hard_regno_mode_ok): Make static and return a bool.
6390 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
6391 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6392 (cris_hard_regno_mode_ok): New function.
6393 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
6394 (epiphany_mode_class): Delete.
6395 (HARD_REGNO_MODE_OK): Delete.
6396 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
6397 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6398 (hard_regno_mode_ok): Rename to...
6399 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
6400 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
6401 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
6403 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
6404 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
6405 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6406 (frv_hard_regno_mode_ok): Make static and return a bool.
6407 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
6409 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
6410 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
6411 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
6412 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
6414 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6415 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
6416 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
6417 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
6419 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6420 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
6421 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6422 (ia64_hard_regno_mode_ok): New function.
6423 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
6424 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6425 (iq2000_hard_regno_mode_ok): New function.
6426 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
6427 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6428 (lm32_hard_regno_mode_ok): New function.
6429 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
6430 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
6431 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
6432 instead of HARD_REGNO_MODE_OK.
6433 (m32c_hard_regno_ok): Rename to...
6434 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
6435 (m32c_cannot_change_mode_class): Update accordingly.
6436 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6437 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
6438 (m32r_mode_class): Delete.
6439 (HARD_REGNO_MODE_OK): Delete.
6440 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6441 (m32r_hard_regno_mode_ok): Rename to...
6442 (m32r_hard_regno_modes): ...this.
6443 (m32r_mode_class): Make static.
6444 (m32r_hard_regno_mode_ok): New function.
6445 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
6446 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
6447 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6448 (m68k_hard_regno_mode_ok): Make static.
6449 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
6450 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6451 (mcore_hard_regno_mode_ok): New function.
6452 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
6453 (HARD_REGNO_MODE_OK): Delete.
6454 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
6456 (microblaze_hard_regno_mode_ok_p): ...this and make static.
6457 (microblaze_hard_regno_mode_ok): New function.
6458 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6459 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
6460 (mips_hard_regno_mode_ok): Delete.
6461 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
6462 (mips_hard_regno_mode_ok_p): ...this and make static.
6463 (mips_hard_regno_mode_ok_p): Rename to...
6464 (mips_hard_regno_mode_ok_uncached): ...this.
6465 (mips_hard_regno_mode_ok): New function.
6466 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
6467 of HARD_REGNO_MODE_OK.
6468 (mips_option_override): Update after above name changes.
6469 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6470 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
6471 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
6472 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
6473 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
6474 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6475 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
6476 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
6477 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
6478 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6479 (msp430_hard_regno_mode_ok): Make static and return a bool.
6480 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
6481 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
6482 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
6484 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6485 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
6486 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
6487 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
6488 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
6489 (PA_HARD_REGNO_MODE_OK): ...this
6490 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
6491 (PA_HARD_REGNO_MODE_OK): ...this.
6492 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6493 (pa_hard_regno_mode_ok): New function.
6494 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
6495 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6496 (pdp11_hard_regno_mode_ok): New function.
6497 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
6498 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
6500 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
6502 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6503 (rs6000_hard_regno_mode_ok): Rename to...
6504 (rs6000_hard_regno_mode_ok_uncached): ...this.
6505 (rs6000_init_hard_regno_mode_ok): Update accordingly.
6506 (rs6000_hard_regno_mode_ok): New function.
6507 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
6508 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
6509 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
6510 (riscv_hard_regno_mode_ok): ...this and make static.
6511 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6512 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
6513 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
6514 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6515 (rl78_hard_regno_mode_ok): Make static and return bool.
6516 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
6517 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
6519 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
6520 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6521 (rs6000_hard_regno_mode_ok): Rename to...
6522 (rs6000_hard_regno_mode_ok_uncached): ...this.
6523 (rs6000_init_hard_regno_mode_ok): Update accordingly.
6524 (rs6000_hard_regno_mode_ok): New function.
6525 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
6526 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
6527 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6528 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
6529 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
6530 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
6531 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6532 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
6533 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
6534 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6535 (sh_hard_regno_mode_ok): Make static.
6536 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
6537 instead of HARD_REGNO_MODE_OK.
6538 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
6539 (sparc_mode_class): Delete.
6540 (HARD_REGNO_MODE_OK): Delete.
6541 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6542 (hard_regno_mode_classes): Make static.
6543 (sparc_mode_class): Likewise.
6544 (sparc_hard_regno_mode_ok): New function.
6545 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
6546 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
6547 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
6549 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6550 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
6551 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
6552 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
6553 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
6554 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6555 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
6556 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
6557 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6558 (visium_hard_regno_mode_ok): New function.
6559 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
6560 instead of HARD_REGNO_MODE_OK.
6561 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
6562 (HARD_REGNO_MODE_OK): Delete.
6563 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
6564 (xtensa_hard_regno_mode_ok_p): ...this and make static.
6565 (xtensa_option_override): Update accordingly.
6566 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6567 (xtensa_hard_regno_mode_ok): New function.
6568 * system.h (HARD_REGNO_MODE_OK): Poison.
6570 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6571 Alan Hayward <alan.hayward@arm.com>
6572 David Sherwood <david.sherwood@arm.com>
6574 * target.def (hard_regno_call_part_clobbered): New hook.
6575 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
6576 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
6577 * doc/tm.texi: Regenerate.
6578 * hooks.h (hook_bool_uint_mode_false): Declare.
6579 * hooks.c (hook_bool_uint_mode_false): New function.
6580 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6581 * cselib.c (cselib_process_insn): Use
6582 targetm.hard_regno_call_part_clobbered instead of
6583 HARD_REGNO_CALL_PART_CLOBBERED.
6584 * ira-conflicts.c (ira_build_conflicts): Likewise.
6585 * ira-costs.c (ira_tune_allocno_costs): Likewise.
6586 * lra-constraints.c (need_for_call_save_p): Likewise.
6587 * lra-lives.c: Include target.h.
6588 (check_pseudos_live_through_calls): Use
6589 targetm.hard_regno_call_part_clobbered instead of
6590 HARD_REGNO_CALL_PART_CLOBBERED.
6591 * regcprop.c: Include target.h.
6592 (copyprop_hardreg_forward_1): Use
6593 targetm.hard_regno_call_part_clobbered instead of
6594 HARD_REGNO_CALL_PART_CLOBBERED.
6595 * reginfo.c (choose_hard_reg_mode): Likewise.
6596 * regrename.c (check_new_reg_p): Likewise.
6597 * reload.c (find_equiv_reg): Likewise.
6598 * reload1.c (emit_reload_insns): Likewise.
6599 * sched-deps.c (deps_analyze_insn): Likewise.
6600 * sel-sched.c (init_regs_for_mode): Likewise.
6601 (mark_unavailable_hard_regs): Likewise.
6602 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
6603 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6604 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
6606 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6607 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6608 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
6610 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
6612 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6613 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6614 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
6616 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6617 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6618 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
6620 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6621 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
6623 * config/powerpcspe/powerpcspe.c
6624 (rs6000_hard_regno_call_part_clobbered): New function.
6625 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6626 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6627 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
6629 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6630 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6631 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
6633 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6634 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6635 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
6637 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6638 Alan Hayward <alan.hayward@arm.com>
6639 David Sherwood <david.sherwood@arm.com>
6641 * rtl.h (subreg_memory_offset): Declare.
6642 * emit-rtl.c (subreg_memory_offset): New function.
6643 * expmed.c (store_bit_field_1): Use it.
6644 * expr.c (undefined_operand_subword_p): Likewise.
6645 * simplify-rtx.c (simplify_subreg): Likewise.
6647 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
6649 PR rtl-optimization/57448
6652 * optabs.c (expand_atomic_load): Place compiler memory barriers if
6653 using atomic_load pattern.
6654 (expand_atomic_store): Likewise.
6656 2017-09-04 Jakub Jelinek <jakub@redhat.com>
6659 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
6660 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
6661 handling. Use replace_call_with_value with NULL instead of
6662 gsi_replace, unlink_stmt_vdef and release_defs.
6664 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
6667 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
6669 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6672 * rtl.h (emit_library_call_value_1): Declare.
6673 (emit_library_call): Replace declaration with a series of overloads.
6674 Remove the parameter count argument.
6675 (emit_library_call_value): Likewise.
6676 * calls.c (emit_library_call_value_1): Make global. Replace varargs
6677 with an "rtx_mode_t *".
6678 (emit_library_call_value): Delete.
6679 (emit_library_call): Likewise.
6680 * asan.c (asan_emit_stack_protection): Update calls accordingly.
6681 (asan_emit_allocas_unpoison): Likewise.
6682 * builtins.c (expand_builtin_powi): Likewise.
6683 (expand_asan_emit_allocas_unpoison): Likewise.
6684 * cfgexpand.c (expand_main_function): Likewise.
6685 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
6686 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
6687 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
6688 * config/arm/arm.c (arm_trampoline_init): Likewise.
6689 (arm_call_tls_get_addr): Likewise.
6690 (arm_expand_divmod_libfunc): Likewise.
6691 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
6692 (smulsi3_highpart): Likewise.
6693 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
6694 (c6x_expand_compare): Likewise.
6695 (c6x_expand_movmem): Likewise.
6696 * config/frv/frv.c (frv_trampoline_init): Likewise.
6697 * config/i386/i386.c (ix86_trampoline_init): Likewise.
6698 (ix86_expand_divmod_libfunc): Likewise.
6699 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
6700 (ia64_expand_compare): Likewise.
6701 (ia64_profile_hook): Likewise.
6702 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
6703 (nonlocal_goto): Likewise.
6704 (restore_stack_nonlocal): Likewise.
6705 * config/m32r/m32r.c (block_move_call): Likewise.
6706 (m32r_trampoline_init): Likewise.
6707 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
6708 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
6709 (m68k_call_m68k_read_tp): Likewise.
6710 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
6711 (microblaze_expand_divide): Likewise.
6712 * config/mips/mips.h (mips_args): Likewise.
6713 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
6714 (MIPS_ICACHE_SYNC): Likewise.
6715 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
6716 (nios2_trampoline_init): Likewise.
6717 * config/pa/pa.c (hppa_tls_call): Likewise.
6718 (pa_trampoline_init): Likewise.
6719 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
6720 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
6721 (expand_strn_compare): Likewise.
6722 (rs6000_generate_compare): Likewise.
6723 (rs6000_expand_float128_convert): Likewise.
6724 (output_profile_hook): Likewise.
6725 (rs6000_trampoline_init): Likewise.
6726 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
6727 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
6728 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
6729 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
6730 (rs6000_generate_compare): Likewise.
6731 (rs6000_expand_float128_convert): Likewise.
6732 (output_profile_hook): Likewise.
6733 (rs6000_trampoline_init): Likewise.
6734 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
6735 * config/sh/sh.c (sh_trampoline_init): Likewise.
6736 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
6737 (sparc_emit_float_lib_cmp): Likewise.
6738 (sparc32_initialize_trampoline): Likewise.
6739 (sparc64_initialize_trampoline): Likewise.
6740 (sparc_profile_hook): Likewise.
6741 * config/spu/spu.c (ea_load_store): Likewise.
6742 * config/spu/spu.md (floatunssidf2): Likewise.
6743 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
6744 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
6745 * config/visium/visium.c (expand_block_move_4): Likewise.
6746 (expand_block_move_2): Likewise.
6747 (expand_block_move_1): Likewise.
6748 (expand_block_set_4): Likewise.
6749 (expand_block_set_2): Likewise.
6750 (expand_block_set_1): Likewise.
6751 (visium_trampoline_init): Likewise.
6752 (visium_profile_hook): Likewise.
6753 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
6754 (xtensa_setup_frame_addresses): Likewise.
6755 (xtensa_trampoline_init): Likewise.
6756 * except.c (sjlj_emit_function_enter): Likewise.
6757 (sjlj_emit_function_exit): Likewise.
6758 * explow.c (allocate_dynamic_stack_space): Likewise.
6759 (probe_stack_range): Likewise.
6760 * expr.c (convert_mode_scalar): Likewise.
6761 * optabs.c (expand_binop): Likewise.
6762 (expand_twoval_binop_libfunc): Likewise.
6763 (expand_unop): Likewise.
6764 (prepare_cmp_insn): Likewise.
6765 (prepare_float_lib_cmp): Likewise.
6766 (expand_float): Likewise.
6767 (expand_fix): Likewise.
6768 (expand_fixed_convert): Likewise.
6769 (maybe_emit_sync_lock_test_and_set): Likewise.
6770 (expand_atomic_compare_and_swap): Likewise.
6771 (expand_mem_thread_fence): Likewise.
6772 (expand_atomic_fetch_op): Likewise.
6774 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
6776 * doc/generic.texi (OpenACC): Adjust URL.
6777 * doc/invoke.texi (C Dialect Options): Ditto.
6779 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
6781 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
6782 predicate for operand 1. Add (m,<S>) constraint.
6783 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
6784 Prevent memory operand 1 with register operand 2.
6786 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
6788 PR rtl-optimization/82024
6789 * combine.c (try_combine): If the combination result is a PARALLEL,
6790 and we only need to retain the SET in there that would be placed
6791 at I2, check that we can place that at I3 instead, before doing so.
6793 2017-09-01 Jakub Jelinek <jakub@redhat.com>
6796 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
6798 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
6799 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
6802 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
6803 Jeff Law <law@redhat.com>
6805 * varasm.c (bss_initializer_p): Do not put constants into .bss
6806 (categorize_decl_for_section): Handle bss_initializer_p returning
6807 false when DECL_INITIAL is NULL.
6809 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6812 * config/s390/s390.c (s390_can_inline_p): New function.
6814 2017-09-01 Jeff Law <law@redhat.com>
6816 PR tree-optimization/82052
6817 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
6818 Always initialize the returned slot after a hash table miss
6819 when INSERT is true.
6821 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
6823 * config/s390/s390.md (mem_signal_fence): Remove.
6824 * doc/md.texi (mem_signal_fence): Remove.
6825 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
6827 * target-insns.def (mem_signal_fence): Remove.
6829 2017-09-01 Jakub Jelinek <jakub@redhat.com>
6832 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
6835 * asan.c (create_odr_indicator): Strip name encoding from assembler
6836 name before appending it after __odr_asan_.
6838 2017-09-01 Martin Liska <mliska@suse.cz>
6840 PR tree-optimization/82059
6841 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
6842 frequency only when an edge is redirected.
6844 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6846 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
6847 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
6848 (arc_conditional_register_usage): Remove ARC600 lp_count
6850 (arc_file_start): Emit Tag_ARC_CPU_variation.
6851 (arc_can_use_doloop_p): New conditions to use ZOLs.
6852 (hwloop_fail): New function.
6853 (hwloop_optimize): Likewise.
6854 (hwloop_pattern_reg): Likewise.
6855 (arc_doloop_hooks): New struct, to be used with reorg_loops.
6856 (arc_reorg_loops): New function, calls reorg_loops.
6857 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
6858 (arc600_corereg_hazard): Remove ZOL checking, case handled by
6860 (arc_loop_hazard): Remove function, functionality moved into
6862 (arc_hazard): Remove arc_loop_hazard call.
6863 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
6864 into hwloop_optimize.
6865 (arc_label_align): Remove ZOL handling.
6866 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
6867 * config/arc/arc.md (doloop_begin): Remove pattern.
6868 (doloop_begin_i): Likewise.
6869 (doloop_end_i): Likewise.
6870 (doloop_fallback): Likewise.
6871 (doloop_fallback_m): Likewise.
6872 (doloop_end): Reimplement expand.
6873 (arc_lp): New pattern for LP instruction.
6874 (loop_end): New pattern.
6875 (loop_fail): Likewise.
6876 (decrement_and_branch_until_zero): Likewise.
6877 * config/arc/arc.opt (mlpc-width): New option.
6878 * doc/invoke.texi (mlpc-width): Document option.
6880 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6882 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
6883 (arc_ccfsm_advance): Fix checking for delay slots.
6884 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
6886 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6888 * config/arc/arc.md (movqi_insn): Add stores to save constant long
6890 (movhi_insn): Update store instruction constraint which are saving
6891 6-bit short immediates.
6892 (movsi_insn): Consider also short scaled load operations.
6893 (zero_extendhisi2_i): Use Usd constraint instead of T.
6894 (extendhisi2_i): Add q constraint.
6895 (arc_clzsi2): Add type and length attributes.
6896 (arc_ctzsi2): Likewise.
6897 * config/arc/constraints.md (Usc): Update constraint, the
6898 assembler can parse two relocations for a single instruction.
6900 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6902 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
6903 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
6905 2017-08-31 Olivier Hainque <hainque@adacore.com>
6907 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
6908 match as powerpc-wrs-vxworks*.
6910 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
6912 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
6913 register constraint for by-element operand.
6914 (aarch64_mls_elt_merge<mode>): Likewise.
6916 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6918 * config/arc/arc.c (arc_can_follow_jump): Check for short
6921 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6923 * config.gcc: Use g.opt for arc.
6924 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
6925 functionality moved to ...
6926 (legitimate_scaled_address_p): New function, ...here.
6927 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
6928 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
6929 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
6931 (arc_override_options): Handle G option.
6932 (arc_output_pic_addr_const): Correct function definition.
6933 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
6934 (arc_decl_anon_ns_mem_p): Delete.
6935 (arc_in_small_data_p): Overhaul this function to take into
6936 consideration the value given via G option.
6937 (arc_rewrite_small_data_1): Renamed and corrected old
6938 arc_rewrite_small_data function.
6939 (arc_rewrite_small_data): New function.
6940 (small_data_pattern): Don't use pic_offset_table_rtx.
6941 * config/arc/arc.h (CC1_SPEC): Recognize G option.
6942 * config/arc/simdext.md (movmisalignv2hi): Use
6943 prepare_move_operands function.
6945 (movmisalign*): Likewise.
6946 * doc/invoke.texi (ARC options): Document -G option.
6948 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6950 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
6952 * config/arc/arc.c (arc_print_operand): Output scalled address for
6953 sdata whenever is possible.
6954 (arc_in_small_data_p): Allow sdata for 64bit datum when double
6955 load/stores are available.
6956 (compact_sda_memory_operand): Check for the alignment required by
6957 code density instructions.
6958 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
6960 * config/arc/constraints.md (Usd): Update constraint.
6961 (Us0): New constraint.
6962 (Usc): Update constraint.
6964 2017-08-31 Richard Biener <rguenther@suse.de>
6967 * dwarf2out.c (dwarf2out_early_global_decl): Process each
6970 2017-08-31 Tamar Christina <tamar.christina@arm.com>
6972 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
6973 Resize type_signature.
6975 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6976 Alan Hayward <alan.hayward@arm.com>
6977 David Sherwood <david.sherwood@arm.com>
6979 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
6980 subregs whose inner modes can be stored in GPRs.
6981 (aarch64_classify_index): Likewise.
6983 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6984 Alan Hayward <alan.hayward@arm.com>
6985 David Sherwood <david.sherwood@arm.com>
6987 * config/aarch64/iterators.md (V_cmp_result): Rename to...
6988 (V_INT_EQUIV): ...this.
6989 (v_cmp_result): Rename to...
6990 (v_int_equiv): ...this.
6991 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
6992 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
6993 (copysign<mode>3): Likewise.
6994 (aarch64_simd_bsl<mode>_internal): Likewise.
6995 (aarch64_simd_bsl<mode>): Likewise.
6996 (vec_cmp<mode><mode>): Likewise.
6997 (vcond<mode><mode>): Likewise.
6998 (vcond<v_cmp_mixed><mode>): Likewise.
6999 (vcondu<mode><v_cmp_mixed>): Likewise.
7000 (aarch64_cm<optab><mode>): Likewise.
7001 (aarch64_cmtst<mode>): Likewise.
7002 (aarch64_fac<optab><mode>): Likewise.
7003 (vec_perm_const<mode>): Likewise.
7004 (vcond_mask_<mode><v_cmp_result>): Rename to...
7005 (vcond_mask_<mode><v_int_equiv>): ...this.
7006 (vec_cmp<mode><v_cmp_result>): Rename to...
7007 (vec_cmp<mode><v_int_equiv>): ...this.
7009 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
7010 Alan Hayward <alan.hayward@arm.com>
7011 David Sherwood <david.sherwood@arm.com>
7013 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
7015 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
7016 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
7017 (UNSPEC_LD4_DREG): New unspecs.
7018 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
7019 (aarch64_ld2<mode>_dreg_be): Replace with...
7020 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
7022 (aarch64_ld3<mode>_dreg_le)
7023 (aarch64_ld3<mode>_dreg_be): Replace with...
7024 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
7026 (aarch64_ld4<mode>_dreg_le)
7027 (aarch64_ld4<mode>_dreg_be): Replace with...
7028 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
7031 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7033 PR tree-optimization/81987
7034 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
7035 insert an initializer in a location not dominated by the stride
7038 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
7040 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
7041 on the entire header of the finally block in the fallthru case.
7043 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
7045 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
7047 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
7049 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
7050 rs6000_emit_move_from_cr and call renamed function.
7051 (rs6000_emit_prologue): Call renamed functions.
7052 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
7053 movesi_from_cr, remove volatile CRs.
7055 2017-08-30 Jon Beniston <jon@beniston.com>
7056 Richard Biener <rguenther@suse.de>
7058 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
7059 of VECTOR_MODE_P check.
7060 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
7061 element vector types.
7063 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7065 * df.h (df_read_modify_subreg_p): Remove in favor of...
7066 * rtl.h (read_modify_subreg_p): ...this new function. Take a
7067 const_rtx instead of an rtx.
7068 * cprop.c (local_cprop_find_used_regs): Update accordingly.
7069 * df-problems.c (df_word_lr_mark_ref): Likewise.
7070 * ira-lives.c (mark_pseudo_reg_live): Likewise.
7071 (mark_pseudo_reg_dead): Likewise.
7072 (mark_ref_dead): Likewise.
7073 * reginfo.c (init_subregs_of_mode): Likewise.
7074 * sched-deps.c (sched_analyze_1): Likewise.
7075 * df-scan.c (df_def_record_1): Likewise.
7076 (df_uses_record): Likewise.
7077 (df_read_modify_subreg_p): Remove in favor of...
7078 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
7079 const_rtx instead of an rtx.
7081 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7082 Alan Hayward <alan.hayward@arm.com>
7083 David Sherwood <david.sherwood@arm.com>
7085 * rtl.h (partial_subreg_p): New function.
7086 * caller-save.c (save_call_clobbered_regs): Use it.
7087 * calls.c (expand_call): Likewise.
7088 * combine.c (combinable_i3pat): Likewise.
7089 (simplify_set): Likewise.
7090 (make_extraction): Likewise.
7091 (make_compound_operation_int): Likewise.
7092 (gen_lowpart_or_truncate): Likewise.
7093 (force_to_mode): Likewise.
7094 (make_field_assignment): Likewise.
7095 (reg_truncated_to_mode): Likewise.
7096 (record_truncated_value): Likewise.
7097 (move_deaths): Likewise.
7098 * cse.c (record_jump_cond): Likewise.
7099 (cse_insn): Likewise.
7100 * cselib.c (cselib_lookup_1): Likewise.
7101 * expmed.c (extract_bit_field_using_extv): Likewise.
7102 * function.c (assign_parm_setup_reg): Likewise.
7103 * ifcvt.c (noce_convert_multiple_sets): Likewise.
7104 * ira-build.c (create_insn_allocnos): Likewise.
7105 * lra-coalesce.c (merge_pseudos): Likewise.
7106 * lra-constraints.c (match_reload): Likewise.
7107 (simplify_operand_subreg): Likewise.
7108 (curr_insn_transform): Likewise.
7109 * lra-lives.c (process_bb_lives): Likewise.
7110 * lra.c (new_insn_reg): Likewise.
7111 (lra_substitute_pseudo): Likewise.
7112 * regcprop.c (mode_change_ok): Likewise.
7113 (maybe_mode_change): Likewise.
7114 (copyprop_hardreg_forward_1): Likewise.
7115 * reload.c (push_reload): Likewise.
7116 (find_reloads): Likewise.
7117 (find_reloads_subreg_address): Likewise.
7118 * reload1.c (alter_reg): Likewise.
7119 (eliminate_regs_1): Likewise.
7120 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7122 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
7124 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
7125 back to if statements, including unpack.
7127 2017-08-30 Martin Liska <mliska@suse.cz>
7130 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
7132 (func_checker::compare_asm_inputs_outputs): ... this function.
7133 (func_checker::compare_gimple_asm): Use the function to compare
7134 also ASM constrains.
7135 * ipa-icf-gimple.h: Rename the function.
7137 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7138 Alan Hayward <alan.hayward@arm.com>
7139 David Sherwood <david.sherwood@arm.com>
7141 * coretypes.h (complex_mode): New type.
7142 * gdbhooks.py (build_pretty_printer): Handle it.
7143 * machmode.h (complex_mode): New class.
7144 (complex_mode::includes_p): New function.
7145 (is_complex_int_mode): Likewise.
7146 (is_complex_float_mode): Likewise.
7147 * genmodes.c (get_mode_class): Handle complex mode classes.
7148 * function.c (expand_function_end): Use is_complex_int_mode.
7150 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7151 Alan Hayward <alan.hayward@arm.com>
7152 David Sherwood <david.sherwood@arm.com>
7154 * coretypes.h (scalar_mode_pod): New typedef.
7155 * gdbhooks.py (build_pretty_printer): Handle it.
7156 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
7157 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
7158 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
7159 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
7162 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7163 Alan Hayward <alan.hayward@arm.com>
7164 David Sherwood <david.sherwood@arm.com>
7166 * machmode.h (mode_for_vector): Take a scalar_mode instead
7168 * stor-layout.c (mode_for_vector): Likewise.
7169 * explow.c (promote_mode): Use as_a <scalar_mode>.
7170 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
7172 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7173 Alan Hayward <alan.hayward@arm.com>
7174 David Sherwood <david.sherwood@arm.com>
7176 * target.def (preferred_simd_mode): Take a scalar_mode
7177 instead of a machine_mode.
7178 * targhooks.h (default_preferred_simd_mode): Likewise.
7179 * targhooks.c (default_preferred_simd_mode): Likewise.
7180 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
7181 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
7182 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
7183 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
7184 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
7185 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
7186 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
7187 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
7189 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
7190 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
7191 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
7192 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
7193 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
7194 * doc/tm.texi: Regenerate.
7195 * optabs-query.c (can_vec_mask_load_store_p): Return false for
7198 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7199 Alan Hayward <alan.hayward@arm.com>
7200 David Sherwood <david.sherwood@arm.com>
7202 * target.def (scalar_mode_supported_p): Take a scalar_mode
7203 instead of a machine_mode.
7204 * targhooks.h (default_scalar_mode_supported_p): Likewise.
7205 * targhooks.c (default_scalar_mode_supported_p): Likewise.
7206 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
7207 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
7208 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
7209 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
7210 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
7211 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
7212 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
7213 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
7214 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
7215 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
7216 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
7217 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
7219 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
7220 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
7221 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
7222 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
7223 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
7225 * doc/tm.texi: Regenerate.
7227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7228 Alan Hayward <alan.hayward@arm.com>
7229 David Sherwood <david.sherwood@arm.com>
7231 * coretypes.h (opt_scalar_mode): New typedef.
7232 * gdbhooks.py (build_pretty_printers): Handle it.
7233 * machmode.h (mode_iterator::get_2xwider): Add overload for
7235 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
7237 * expr.c (convert_mode_scalar): Likewise.
7238 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7239 * optabs.c (expand_float): Likewise.
7240 (expand_fix): Likewise.
7241 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7243 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7244 Alan Hayward <alan.hayward@arm.com>
7245 David Sherwood <david.sherwood@arm.com>
7247 * optabs.c (expand_float): Explicitly check for scalars before
7248 using a branching expansion.
7249 (expand_fix): Likewise.
7251 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7252 Alan Hayward <alan.hayward@arm.com>
7253 David Sherwood <david.sherwood@arm.com>
7255 * expr.c (convert_mode): Split scalar handling out into...
7256 (convert_mode_scalar): ...this new function. Treat the modes
7259 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7260 Alan Hayward <alan.hayward@arm.com>
7261 David Sherwood <david.sherwood@arm.com>
7263 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
7265 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
7267 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7268 Alan Hayward <alan.hayward@arm.com>
7269 David Sherwood <david.sherwood@arm.com>
7271 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
7272 rather than a machine_mode.
7273 (fixed_from_string): Likewise.
7274 (fixed_convert): Likewise.
7275 (fixed_convert_from_int): Likewise.
7276 (fixed_convert_from_real): Likewise.
7277 (real_convert_from_fixed): Likewise.
7278 * fixed-value.c (fixed_from_double_int): Likewise.
7279 (fixed_from_string): Likewise.
7280 (fixed_convert): Likewise.
7281 (fixed_convert_from_int): Likewise.
7282 (fixed_convert_from_real): Likewise.
7283 (real_convert_from_fixed): Likewise.
7284 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
7286 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7287 Alan Hayward <alan.hayward@arm.com>
7288 David Sherwood <david.sherwood@arm.com>
7290 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
7291 of separate mode class checks. Do not allow vector modes here.
7292 (immed_wide_int_const): Use as_a <scalar_mode>.
7293 * explow.c (trunc_int_for_mode): Likewise.
7294 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
7295 (wi::shwi): Likewise.
7296 (wi::min_value): Likewise.
7297 (wi::max_value): Likewise.
7298 * dwarf2out.c (loc_descriptor): Likewise.
7299 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
7302 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7303 Alan Hayward <alan.hayward@arm.com>
7304 David Sherwood <david.sherwood@arm.com>
7306 * tree.h (SCALAR_TYPE_MODE): New macro.
7307 * expr.c (expand_expr_addr_expr_1): Use it.
7308 (expand_expr_real_2): Likewise.
7309 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
7310 (fold_convert_const_fixed_from_int): Likewise.
7311 (fold_convert_const_fixed_from_real): Likewise.
7312 (native_encode_fixed): Likewise
7313 (native_encode_complex): Likewise
7314 (native_encode_vector): Likewise.
7315 (native_interpret_fixed): Likewise.
7316 (native_interpret_real): Likewise.
7317 (native_interpret_complex): Likewise.
7318 (native_interpret_vector): Likewise.
7319 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
7320 (simd_clone_adjust_argument_types): Likewise.
7321 (simd_clone_init_simd_arrays): Likewise.
7322 (simd_clone_adjust): Likewise.
7323 * stor-layout.c (layout_type): Likewise.
7324 * tree.c (build_minus_one_cst): Likewise.
7325 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
7326 * tree-inline.c (estimate_move_cost): Likewise.
7327 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
7328 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
7329 (vectorizable_reduction): Likewise.
7330 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
7331 (vect_recog_mixed_size_cond_pattern): Likewise.
7332 (check_bool_pattern): Likewise.
7333 (adjust_bool_pattern): Likewise.
7334 (search_type_for_mask_1): Likewise.
7335 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
7336 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7337 (vectorizable_load): Likewise.
7338 (vectorizable_store): Likewise.
7339 * ubsan.c (ubsan_encode_value): Likewise.
7340 * varasm.c (output_constant): Likewise.
7342 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7343 Alan Hayward <alan.hayward@arm.com>
7344 David Sherwood <david.sherwood@arm.com>
7346 * coretypes.h (scalar_mode): New class.
7347 * machmode.h (scalar_mode): Likewise.
7348 (scalar_mode::includes_p): New function.
7349 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
7350 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
7351 * genmodes.c (get_mode_class): Handle remaining scalar modes.
7352 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
7353 * expmed.c (store_bit_field_1): Likewise.
7354 (extract_bit_field_1): Likewise.
7355 * expr.c (write_complex_part): Likewise.
7356 (read_complex_part): Likewise.
7357 (emit_move_complex_push): Likewise.
7358 (expand_expr_real_2): Likewise.
7359 * function.c (assign_parm_setup_reg): Likewise.
7360 (assign_parms_unsplit_complex): Likewise.
7361 * optabs.c (expand_binop): Likewise.
7362 * rtlanal.c (subreg_get_info): Likewise.
7363 * simplify-rtx.c (simplify_immed_subreg): Likewise.
7364 * varasm.c (output_constant_pool_2): Likewise.
7366 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7367 Alan Hayward <alan.hayward@arm.com>
7368 David Sherwood <david.sherwood@arm.com>
7370 * expmed.c (extract_high_half): Use scalar_int_mode and remove
7372 (expmed_mult_highpart_optab): Likewise.
7373 (expmed_mult_highpart): Likewise.
7375 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7376 Alan Hayward <alan.hayward@arm.com>
7377 David Sherwood <david.sherwood@arm.com>
7379 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
7380 instead of a machine_mode.
7381 (builtin_memset_read_str): Likewise.
7382 * builtins.c (c_readstr): Likewise.
7383 (builtin_memcpy_read_str): Likewise.
7384 (builtin_strncpy_read_str): Likewise.
7385 (builtin_memset_read_str): Likewise.
7386 (builtin_memset_gen_str): Likewise.
7387 (expand_builtin_signbit): Use scalar_int_mode for local variables.
7388 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
7389 instead of a machine_mode.
7390 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
7392 (make_extraction): Likewise.
7393 (try_widen_shift_mode): Take and return scalar_int_modes instead
7395 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
7396 a scalar_int_mode instead of a machine_mode.
7397 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
7398 (avr_addr_space_pointer_mode): Likewise.
7399 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
7400 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
7401 (msp430_unwind_word_mode): Likewise.
7402 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
7403 (spu_addr_space_pointer_mode): Likewise.
7404 (spu_addr_space_address_mode): Likewise.
7405 (spu_libgcc_cmp_return_mode): Likewise.
7406 (spu_libgcc_shift_count_mode): Likewise.
7407 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
7408 (rl78_addr_space_pointer_mode): Likewise.
7409 (fl78_unwind_word_mode): Likewise.
7410 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
7412 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
7413 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
7414 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
7415 (mips_valid_pointer_mode): Likewise.
7416 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
7417 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
7418 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
7420 (ft32_addr_space_address_mode): Likewise.
7421 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
7422 scalar_int_mode instead of a machine_mode.
7423 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
7425 (m32c_addr_space_address_mode): Likewise.
7426 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
7427 (rs6000_eh_return_filter_mode): Likewise.
7428 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
7429 (rs6000_eh_return_filter_mode): Likewise.
7430 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
7431 (s390_libgcc_shift_count_mode): Likewise.
7432 (s390_unwind_word_mode): Likewise.
7433 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
7435 * target.def (mode_rep_extended): Likewise.
7436 (valid_pointer_mode): Likewise.
7437 (addr_space.valid_pointer_mode): Likewise.
7438 (eh_return_filter_mode): Return a scalar_int_mode rather than
7440 (libgcc_cmp_return_mode): Likewise.
7441 (libgcc_shift_count_mode): Likewise.
7442 (unwind_word_mode): Likewise.
7443 (addr_space.pointer_mode): Likewise.
7444 (addr_space.address_mode): Likewise.
7445 * doc/tm.texi: Regenerate.
7446 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
7448 (do_jump): Use scalar_int_mode for local variables.
7449 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
7450 rather than a machine_mode.
7451 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
7452 (scompare_loc_descriptor_wide): Likewise.
7453 (scompare_loc_descriptor_narrow): Likewise.
7454 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
7456 * except.c (sjlj_emit_dispatch_table): Likewise.
7457 (expand_builtin_eh_copy_values): Likewise.
7458 * explow.c (convert_memory_address_addr_space_1): Likewise.
7459 Take a scalar_int_mode rather than a machine_mode.
7460 (convert_memory_address_addr_space): Take a scalar_int_mode rather
7461 than a machine_mode.
7462 (memory_address_addr_space): Use scalar_int_mode for local variables.
7463 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
7464 rather than a machine_mode.
7465 * expmed.c (mask_rtx): Likewise.
7466 (init_expmed_one_conv): Likewise.
7467 (expand_mult_highpart_adjust): Likewise.
7468 (extract_high_half): Likewise.
7469 (expmed_mult_highpart_optab): Likewise.
7470 (expmed_mult_highpart): Likewise.
7471 (expand_smod_pow2): Likewise.
7472 (expand_sdiv_pow2): Likewise.
7473 (emit_store_flag_int): Likewise.
7474 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
7476 (extract_low_bits): Likewise.
7477 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
7479 * expr.c (pieces_addr::adjust): Likewise.
7480 (can_store_by_pieces): Likewise.
7481 (store_by_pieces): Likewise.
7482 (clear_by_pieces_1): Likewise.
7483 (expand_expr_addr_expr_1): Likewise.
7484 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
7485 (expand_expr_real_1): Likewise.
7486 (try_casesi): Likewise.
7487 * final.c (shorten_branches): Likewise.
7488 * fold-const.c (fold_convert_const_int_from_fixed): Change the
7489 type of "mode" to machine_mode.
7490 * internal-fn.c (expand_arith_overflow_result_store): Take a
7491 scalar_int_mode rather than a machine_mode.
7492 (expand_mul_overflow): Use scalar_int_mode for local variables.
7493 * loop-doloop.c (doloop_modify): Likewise.
7494 (doloop_optimize): Likewise.
7495 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
7496 than a machine_mode.
7497 (expand_doubleword_shift_condmove): Likewise.
7498 (expand_doubleword_shift): Likewise.
7499 (expand_doubleword_clz): Likewise.
7500 (expand_doubleword_popcount): Likewise.
7501 (expand_doubleword_parity): Likewise.
7502 (expand_absneg_bit): Use scalar_int_mode for local variables.
7503 (prepare_float_lib_cmp): Likewise.
7504 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
7505 rather than a machine_mode.
7506 (convert_memory_address_addr_space): Likewise.
7507 (get_mode_bounds): Likewise.
7508 (get_address_mode): Return a scalar_int_mode rather than a
7510 * rtlanal.c (get_address_mode): Likewise.
7511 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
7512 than a machine_mode.
7513 * targhooks.c (default_mode_rep_extended): Likewise.
7514 (default_valid_pointer_mode): Likewise.
7515 (default_addr_space_valid_pointer_mode): Likewise.
7516 (default_eh_return_filter_mode): Return a scalar_int_mode rather
7517 than a machine_mode.
7518 (default_libgcc_cmp_return_mode): Likewise.
7519 (default_libgcc_shift_count_mode): Likewise.
7520 (default_unwind_word_mode): Likewise.
7521 (default_addr_space_pointer_mode): Likewise.
7522 (default_addr_space_address_mode): Likewise.
7523 * targhooks.h (default_eh_return_filter_mode): Likewise.
7524 (default_libgcc_cmp_return_mode): Likewise.
7525 (default_libgcc_shift_count_mode): Likewise.
7526 (default_unwind_word_mode): Likewise.
7527 (default_addr_space_pointer_mode): Likewise.
7528 (default_addr_space_address_mode): Likewise.
7529 (default_mode_rep_extended): Take a scalar_int_mode rather than
7531 (default_valid_pointer_mode): Likewise.
7532 (default_addr_space_valid_pointer_mode): Likewise.
7533 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
7535 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
7536 rather than a machine_mode.
7537 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
7538 for local variables.
7539 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7540 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
7541 than a machine_mode.
7543 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7544 Alan Hayward <alan.hayward@arm.com>
7545 David Sherwood <david.sherwood@arm.com>
7547 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
7548 the mode argument to scalar_int_mode.
7549 (do_jump_by_parts_zero_rtx): Likewise.
7550 (do_jump_by_parts_equality_rtx): Likewise.
7551 (do_jump_by_parts_greater): Take a mode argument.
7552 (do_jump_by_parts_equality): Likewise.
7553 (do_jump_1): Update calls accordingly.
7555 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7556 Alan Hayward <alan.hayward@arm.com>
7557 David Sherwood <david.sherwood@arm.com>
7559 * is-a.h (safe_dyn_cast): New function.
7560 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
7561 (jump_table_for_label): Likewise.
7562 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
7563 instead of an rtx_insn *.
7564 (shorten_branches): Use dyn_cast instead of LABEL_P and
7565 JUMP_TABLE_DATA_P. Use jump_table_for_label and
7566 rtx_jump_table_data::get_data_mode.
7567 (final_scan_insn): Likewise.
7569 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7570 Alan Hayward <alan.hayward@arm.com>
7571 David Sherwood <david.sherwood@arm.com>
7573 * combine.c (try_combine): Use is_a <scalar_int_mode> when
7574 trying to combine a full-register integer set with a subreg
7577 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7578 Alan Hayward <alan.hayward@arm.com>
7579 David Sherwood <david.sherwood@arm.com>
7581 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
7582 that is always either address_mode or pointer_mode.
7584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7585 Alan Hayward <alan.hayward@arm.com>
7586 David Sherwood <david.sherwood@arm.com>
7588 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
7589 when the two are known to be equal.
7591 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7592 Alan Hayward <alan.hayward@arm.com>
7593 David Sherwood <david.sherwood@arm.com>
7595 * simplify-rtx.c (simplify_const_unary_operation): Use
7596 is_a <scalar_int_mode> instead of checking for a nonzero
7597 precision. Forcibly convert op_mode to a scalar_int_mode
7598 in that case. More clearly differentiate the operand and
7599 result modes and use the former when deciding what the value
7600 of a count-bits operation should be. Use is_int_mode instead
7601 of checking for a MODE_INT. Remove redundant check for whether
7602 this mode has a zero precision.
7604 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7605 Alan Hayward <alan.hayward@arm.com>
7606 David Sherwood <david.sherwood@arm.com>
7608 * optabs.c (widen_leading): Change the type of the mode argument
7609 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
7610 (widen_bswap): Likewise.
7611 (expand_parity): Likewise.
7612 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
7613 (expand_ffs): Likewise.
7614 (epand_unop): Check for scalar integer modes before calling the
7617 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7618 Alan Hayward <alan.hayward@arm.com>
7619 David Sherwood <david.sherwood@arm.com>
7621 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
7623 (expand_expr_real_1): Update call accordingly.
7625 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7626 Alan Hayward <alan.hayward@arm.com>
7627 David Sherwood <david.sherwood@arm.com>
7629 * expmed.c (store_bit_field_using_insv): Add op0_mode and
7630 value_mode arguments. Use scalar_int_mode internally.
7631 (store_bit_field_1): Rename the new integer mode from imode
7632 to op0_mode and use it instead of GET_MODE (op0). Update calls
7633 to store_split_bit_field, store_bit_field_using_insv and
7634 store_fixed_bit_field.
7635 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
7636 Use scalar_int_mode internally. Use a bit count rather than a mode
7637 when calculating the largest bit size for get_best_mode.
7638 Update calls to store_split_bit_field and store_fixed_bit_field_1.
7639 (store_fixed_bit_field_1): Add mode and value_mode arguments.
7640 Remove assertion that OP0 has a scalar integer mode.
7641 (store_split_bit_field): Add op0_mode and value_mode arguments.
7642 Update calls to extract_fixed_bit_field.
7643 (extract_bit_field_using_extv): Add an op0_mode argument.
7644 Use scalar_int_mode internally.
7645 (extract_bit_field_1): Rename the new integer mode from imode to
7646 op0_mode and use it instead of GET_MODE (op0). Update calls to
7647 extract_split_bit_field, extract_bit_field_using_extv and
7648 extract_fixed_bit_field.
7649 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
7650 to extract_split_bit_field and extract_fixed_bit_field_1.
7651 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
7652 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
7654 (extract_split_bit_field): Add an op0_mode argument. Update call
7655 to extract_fixed_bit_field.
7657 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7658 Alan Hayward <alan.hayward@arm.com>
7659 David Sherwood <david.sherwood@arm.com>
7661 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
7662 * explow.c (hard_function_value): Likewise.
7663 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
7664 convert_to_mode call outside the loop.
7665 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
7666 for the mode iterator. Require the mode specified by max_pieces
7668 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
7670 (copy_blkmode_to_reg): Likewise.
7671 (set_storage_via_setmem): Likewise.
7672 * optabs.c (prepare_cmp_insn): Likewise.
7673 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7674 * stor-layout.c (finish_bitfield_representative): Likewise.
7676 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7677 Alan Hayward <alan.hayward@arm.com>
7678 David Sherwood <david.sherwood@arm.com>
7680 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
7681 * expr.c (convert_move): Use them.
7682 (convert_modes): Likewise.
7683 (store_expr_with_bounds): Likewise.
7685 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7686 Alan Hayward <alan.hayward@arm.com>
7687 David Sherwood <david.sherwood@arm.com>
7689 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
7690 parameter for the mode of "x". Remove the "known_x", "known_mode"
7691 and "known_ret" arguments. Change the type of the mode argument
7693 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
7694 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
7695 (reg_num_sign_bit_copies_for_combine): Likewise.
7696 * rtlanal.c (nonzero_bits1): Likewise.
7697 (num_sign_bit_copies1): Likewise.
7698 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
7699 (reg_num_sign_bit_copies_general): Likewise.
7700 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
7701 (reg_nonzero_bits_general): Likewise.
7703 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7704 Alan Hayward <alan.hayward@arm.com>
7705 David Sherwood <david.sherwood@arm.com>
7707 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
7708 than in subroutines. Return 1 for non-integer modes.
7709 (cached_num_sign_bit_copies): Change the type of the mode parameter
7711 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
7712 classes. Handle CONST_INT_P first and then check whether X also
7713 has a scalar integer mode. Check the same thing for inner registers
7714 of a SUBREG and for values that are being extended or truncated.
7716 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7717 Alan Hayward <alan.hayward@arm.com>
7718 David Sherwood <david.sherwood@arm.com>
7720 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
7721 in subroutines. Return the mode mask for non-integer modes.
7722 (cached_nonzero_bits): Change the type of the mode parameter
7724 (nonzero_bits1): Likewise. Remove early exit for other mode
7725 classes. Handle CONST_INT_P first and then check whether X
7726 also has a scalar integer mode.
7728 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7729 Alan Hayward <alan.hayward@arm.com>
7730 David Sherwood <david.sherwood@arm.com>
7732 * expr.c (widest_int_mode_for_size): Make the comment match the code.
7733 Return a scalar_int_mode and assert that the size is greater than
7735 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
7736 (op_by_pieces_d::op_by_pieces_d): Likewise.
7737 (op_by_pieces_d::run): Likewise.
7738 (can_store_by_pieces): Likewise.
7740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7741 Alan Hayward <alan.hayward@arm.com>
7742 David Sherwood <david.sherwood@arm.com>
7744 * combine.c (extract_left_shift): Add a mode argument and update
7746 (make_compound_operation_int): Change the type of the mode parameter
7747 to scalar_int_mode and update the call to extract_left_shift.
7749 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7750 Alan Hayward <alan.hayward@arm.com>
7751 David Sherwood <david.sherwood@arm.com>
7753 * combine.c (simplify_and_const_int): Change the type of the mode
7754 parameter to scalar_int_mode.
7755 (simplify_and_const_int_1): Likewise. Update recursive call.
7757 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7758 Alan Hayward <alan.hayward@arm.com>
7759 David Sherwood <david.sherwood@arm.com>
7761 * combine.c (simplify_compare_const): Check that the mode is a
7762 scalar_int_mode (rather than VOIDmode) before testing its
7764 (simplify_comparison): Move COMPARISON_P handling out of the
7765 loop and restrict the latter part of the loop to scalar_int_modes.
7766 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
7767 and when considering SUBREG_REGs. Use is_int_mode instead of
7768 checking GET_MODE_CLASS against MODE_INT.
7770 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7771 Alan Hayward <alan.hayward@arm.com>
7772 David Sherwood <david.sherwood@arm.com>
7774 * combine.c (try_widen_shift_mode): Move check for equal modes to...
7775 (simplify_shift_const_1): ...here. Use scalar_int_mode for
7776 shift_unit_mode and for modes involved in scalar shifts.
7778 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7779 Alan Hayward <alan.hayward@arm.com>
7780 David Sherwood <david.sherwood@arm.com>
7782 * combine.c (force_int_to_mode): New function, split out from...
7783 (force_to_mode): ...here. Keep xmode up-to-date and use it
7784 instead of GET_MODE (x).
7786 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7787 Alan Hayward <alan.hayward@arm.com>
7788 David Sherwood <david.sherwood@arm.com>
7790 * optabs-query.h (extraction_insn::struct_mode): Change type to
7791 opt_scalar_int_mode and update comment.
7792 (extraction_insn::field_mode): Change type to scalar_int_mode.
7793 (extraction_insn::pos_mode): Likewise.
7794 * combine.c (make_extraction): Update accordingly.
7795 * optabs-query.c (get_traditional_extraction_insn): Likewise.
7796 (get_optab_extraction_insn): Likewise.
7797 * recog.c (simplify_while_replacing): Likewise.
7798 * expmed.c (narrow_bit_field_mem): Change the type of the mode
7799 parameter to opt_scalar_int_mode.
7801 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7802 Alan Hayward <alan.hayward@arm.com>
7803 David Sherwood <david.sherwood@arm.com>
7805 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
7806 to a scalar_int_mode instead of a machine_mode.
7807 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
7808 (get_best_mode): Return a boolean and use a pointer argument to store
7809 the selected mode. Replace the limit mode parameter with a bit limit.
7810 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
7811 for the values returned by bit_field_mode_iterator::next_mode.
7812 (store_bit_field): Update call to get_best_mode.
7813 (store_fixed_bit_field): Likewise.
7814 (extract_fixed_bit_field): Likewise.
7815 * expr.c (optimize_bitfield_assignment_op): Likewise.
7816 * fold-const.c (optimize_bit_field_compare): Likewise.
7817 (fold_truth_andor_1): Likewise.
7818 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
7819 Update for new type of m_mode.
7820 (get_best_mode): As above.
7822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7823 Alan Hayward <alan.hayward@arm.com>
7824 David Sherwood <david.sherwood@arm.com>
7826 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
7827 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
7828 (store_bit_field): Check is_a <scalar_int_mode> before calling
7829 strict_volatile_bitfield_p.
7830 (extract_bit_field): Likewise.
7832 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7833 Alan Hayward <alan.hayward@arm.com>
7834 David Sherwood <david.sherwood@arm.com>
7836 * target.def (cstore_mode): Return a scalar_int_mode.
7837 * doc/tm.texi: Regenerate.
7838 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
7839 * targhooks.h (default_cstore_mode): Likewise.
7840 * targhooks.c (default_cstore_mode): Likewise, using a forced
7842 * expmed.c (emit_cstore): Expect the target of the cstore to be
7845 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7846 Alan Hayward <alan.hayward@arm.com>
7847 David Sherwood <david.sherwood@arm.com>
7849 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
7851 (niter_desc): Likewise mode.
7852 (iv_analyze): Add a mode parameter.
7854 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
7855 and change its type to scalar_int_mode.
7856 * loop-iv.c: Update commentary at head of file.
7857 (iv_constant): Pass the mode paraeter before the rtx it describes
7858 and change its type to scalar_int_mode. Remove VOIDmode handling.
7859 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
7860 (iv_extend): Likewise.
7861 (shorten_into_mode): Likewise.
7862 (iv_add): Use scalar_int_mode.
7863 (iv_mult): Likewise.
7864 (iv_shift): Likewise.
7865 (canonicalize_iv_subregs): Likewise.
7866 (get_biv_step_1): Pass the outer_mode parameter before the rtx
7867 it describes and change its mode to scalar_int_mode. Also change
7868 the type of the returned inner_mode to scalar_int_mode.
7869 (get_biv_step): Likewise, turning outer_mode from a pointer
7870 into a direct parameter. Update call to get_biv_step_1.
7871 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
7872 iv_constant and get_biv_step.
7873 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
7874 and change its type to scalar_int_mode. Don't initialise iv->mode
7875 to VOIDmode and remove later checks for its still being VOIDmode.
7876 Update calls to iv_analyze_op and iv_analyze_expr. Check
7877 is_a <scalar_int_mode> when changing the mode under consideration.
7878 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
7879 Update call to iv_analyze_expr.
7880 (iv_analyze_op): Add a mode parameter. Reject subregs whose
7881 inner register is not also a scalar_int_mode. Update call to
7883 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
7884 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
7885 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
7886 separate mode class checks. Update calls to iv_analyze. Remove
7887 fix-up of VOIDmodes after iv_analyze_biv.
7888 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
7889 don't have a scalar_int_mode. Update call to biv_p.
7891 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7892 Alan Hayward <alan.hayward@arm.com>
7893 David Sherwood <david.sherwood@arm.com>
7895 * cfgexpand.c (convert_debug_memory_address): Use
7896 as_a <scalar_int_mode>.
7897 * combine.c (expand_compound_operation): Likewise.
7898 (make_extraction): Likewise.
7899 (change_zero_ext): Likewise.
7900 (simplify_comparison): Likewise.
7901 * cse.c (cse_insn): Likewise.
7902 * dwarf2out.c (minmax_loc_descriptor): Likewise.
7903 (mem_loc_descriptor): Likewise.
7904 (loc_descriptor): Likewise.
7905 * expmed.c (init_expmed_one_mode): Likewise.
7906 (synth_mult): Likewise.
7907 (emit_store_flag_1): Likewise.
7908 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
7909 of a comparison with size.
7910 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
7911 (reduce_to_bit_field_precision): Likewise.
7912 * function.c (expand_function_end): Likewise.
7913 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
7914 * loop-doloop.c (doloop_modify): Likewise.
7915 * optabs.c (expand_binop): Likewise.
7916 (expand_unop): Likewise.
7917 (expand_copysign_absneg): Likewise.
7918 (prepare_cmp_insn): Likewise.
7919 (maybe_legitimize_operand): Likewise.
7920 * recog.c (const_scalar_int_operand): Likewise.
7921 * rtlanal.c (get_address_mode): Likewise.
7922 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7923 (simplify_cond_clz_ctz): Likewise.
7924 * tree-nested.c (get_nl_goto_field): Likewise.
7925 * tree.c (build_vector_type_for_mode): Likewise.
7926 * var-tracking.c (use_narrower_mode): Likewise.
7928 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7929 Alan Hayward <alan.hayward@arm.com>
7930 David Sherwood <david.sherwood@arm.com>
7932 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
7933 * builtins.c (expand_builtin_signbit): Use it.
7934 * cfgexpand.c (expand_debug_expr): Likewise.
7935 * dojump.c (do_jump): Likewise.
7936 (do_compare_and_jump): Likewise.
7937 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
7938 * expmed.c (make_tree): Likewise.
7939 * expr.c (expand_expr_real_2): Likewise.
7940 (expand_expr_real_1): Likewise.
7941 (try_casesi): Likewise.
7942 * fold-const-call.c (fold_const_call_ss): Likewise.
7943 * fold-const.c (unextend): Likewise.
7944 (extract_muldiv_1): Likewise.
7945 (fold_single_bit_test): Likewise.
7946 (native_encode_int): Likewise.
7947 (native_encode_string): Likewise.
7948 (native_interpret_int): Likewise.
7949 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
7950 * internal-fn.c (expand_addsub_overflow): Likewise.
7951 (expand_neg_overflow): Likewise.
7952 (expand_mul_overflow): Likewise.
7953 (expand_arith_overflow): Likewise.
7954 * match.pd: Likewise.
7955 * stor-layout.c (layout_type): Likewise.
7956 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
7957 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
7958 * tree-ssanames.c (get_range_info): Likewise.
7959 * tree-switch-conversion.c (array_value_type) Likewise.
7960 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
7961 (vect_recog_divmod_pattern): Likewise.
7962 (vect_recog_mixed_size_cond_pattern): Likewise.
7963 * tree-vrp.c (extract_range_basic): Likewise.
7964 (simplify_float_conversion_using_ranges): Likewise.
7965 * tree.c (int_fits_type_p): Likewise.
7966 * ubsan.c (instrument_bool_enum_load): Likewise.
7967 * varasm.c (mergeable_string_section): Likewise.
7968 (narrowing_initializer_constant_valid_p): Likewise.
7969 (output_constant): Likewise.
7971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7972 Alan Hayward <alan.hayward@arm.com>
7973 David Sherwood <david.sherwood@arm.com>
7975 * machmode.h (NARROWEST_INT_MODE): New macro.
7976 * expr.c (alignment_for_piecewise_move): Use it instead of
7977 GET_CLASS_NARROWEST_MODE (MODE_INT).
7978 (push_block): Likewise.
7979 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
7981 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7983 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7984 Alan Hayward <alan.hayward@arm.com>
7985 David Sherwood <david.sherwood@arm.com>
7987 * postreload.c (move2add_valid_value_p): Change the type of the
7988 mode parameter to scalar_int_mode.
7989 (move2add_use_add2_insn): Add a mode parameter and use it instead
7991 (move2add_use_add3_insn): Likewise.
7992 (reload_cse_move2add): Update accordingly.
7994 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7995 Alan Hayward <alan.hayward@arm.com>
7996 David Sherwood <david.sherwood@arm.com>
7998 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
8000 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
8001 * optabs.c (expand_unop): Likewise.
8003 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8004 Alan Hayward <alan.hayward@arm.com>
8005 David Sherwood <david.sherwood@arm.com>
8007 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
8008 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
8009 (popcount_loc_descriptor): Likewise.
8010 (bswap_loc_descriptor): Likewise.
8011 (rotate_loc_descriptor): Likewise.
8012 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
8013 calling the functions above.
8015 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8016 Alan Hayward <alan.hayward@arm.com>
8017 David Sherwood <david.sherwood@arm.com>
8019 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
8021 (try_combine): Likewise.
8022 (simplify_if_then_else): Likewise.
8023 * cse.c (cse_insn): Likewise.
8024 * dwarf2out.c (mem_loc_descriptor): Likewise.
8025 * emit-rtl.c (gen_lowpart_common): Likewise.
8026 * simplify-rtx.c (simplify_truncation): Likewise.
8027 (simplify_binary_operation_1): Likewise.
8028 (simplify_const_relational_operation): Likewise.
8029 (simplify_ternary_operation): Likewise.
8030 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
8032 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8033 Alan Hayward <alan.hayward@arm.com>
8034 David Sherwood <david.sherwood@arm.com>
8036 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
8037 * reload.c (push_reload): Likewise.
8038 (find_reloads): Likewise.
8040 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8041 Alan Hayward <alan.hayward@arm.com>
8042 David Sherwood <david.sherwood@arm.com>
8044 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
8045 (make_compound_operation_int): Likewise.
8046 (change_zero_ext): Likewise.
8047 * expr.c (convert_move): Likewise.
8048 (convert_modes): Likewise.
8049 * fwprop.c (forward_propagate_subreg): Likewise.
8050 * loop-iv.c (get_biv_step_1): Likewise.
8051 * optabs.c (widen_operand): Likewise.
8052 * postreload.c (move2add_valid_value_p): Likewise.
8053 * recog.c (simplify_while_replacing): Likewise.
8054 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8055 (simplify_binary_operation_1): Likewise. Remove redundant
8056 mode equality check.
8058 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8059 Alan Hayward <alan.hayward@arm.com>
8060 David Sherwood <david.sherwood@arm.com>
8062 * combine.c (combine_simplify_rtx): Add checks for
8063 is_a <scalar_int_mode>.
8064 (simplify_if_then_else): Likewise.
8065 (make_field_assignment): Likewise.
8066 (simplify_comparison): Likewise.
8067 * ifcvt.c (noce_try_bitop): Likewise.
8068 * loop-invariant.c (canonicalize_address_mult): Likewise.
8069 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8071 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8072 Alan Hayward <alan.hayward@arm.com>
8073 David Sherwood <david.sherwood@arm.com>
8075 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
8076 is_a <scalar_int_mode> instead of != BLKmode.
8078 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8079 Alan Hayward <alan.hayward@arm.com>
8080 David Sherwood <david.sherwood@arm.com>
8082 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
8083 instead of != VOIDmode.
8084 * combine.c (if_then_else_cond): Likewise.
8085 (change_zero_ext): Likewise.
8086 * dwarf2out.c (mem_loc_descriptor): Likewise.
8087 (loc_descriptor): Likewise.
8088 * rtlanal.c (canonicalize_condition): Likewise.
8089 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
8091 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8092 Alan Hayward <alan.hayward@arm.com>
8093 David Sherwood <david.sherwood@arm.com>
8095 * simplify-rtx.c (simplify_binary_operation_1): Use
8096 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
8098 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8099 Alan Hayward <alan.hayward@arm.com>
8100 David Sherwood <david.sherwood@arm.com>
8102 * wide-int.h (int_traits<unsigned char>) New class.
8103 (int_traits<unsigned short>) Likewise.
8104 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
8105 Use GET_MODE_UNIT_PRECISION and remove redundant test for
8107 * combine.c (set_nonzero_bits_and_sign_copies): Use
8108 is_a <scalar_int_mode>.
8109 (find_split_point): Likewise.
8110 (combine_simplify_rtx): Likewise.
8111 (simplify_logical): Likewise.
8112 (expand_compound_operation): Likewise.
8113 (expand_field_assignment): Likewise.
8114 (make_compound_operation): Likewise.
8115 (extended_count): Likewise.
8116 (change_zero_ext): Likewise.
8117 (simplify_comparison): Likewise.
8118 * dwarf2out.c (scompare_loc_descriptor): Likewise.
8119 (ucompare_loc_descriptor): Likewise.
8120 (minmax_loc_descriptor): Likewise.
8121 (mem_loc_descriptor): Likewise.
8122 (loc_descriptor): Likewise.
8123 * expmed.c (init_expmed_one_mode): Likewise.
8124 * lra-constraints.c (lra_constraint_offset): Likewise.
8125 * optabs.c (prepare_libcall_arg): Likewise.
8126 * postreload.c (move2add_note_store): Likewise.
8127 * reload.c (operands_match_p): Likewise.
8128 * rtl.h (load_extend_op): Likewise.
8129 * rtlhooks.c (gen_lowpart_general): Likewise.
8130 * simplify-rtx.c (simplify_truncation): Likewise.
8131 (simplify_unary_operation_1): Likewise.
8132 (simplify_binary_operation_1): Likewise.
8133 (simplify_const_binary_operation): Likewise.
8134 (simplify_const_relational_operation): Likewise.
8135 (simplify_subreg): Likewise.
8136 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8137 * var-tracking.c (adjust_mems): Likewise.
8138 (prepare_call_arguments): Likewise.
8140 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8141 Alan Hayward <alan.hayward@arm.com>
8142 David Sherwood <david.sherwood@arm.com>
8144 * machmode.h (is_int_mode): New fuction.
8145 * combine.c (find_split_point): Use it.
8146 (combine_simplify_rtx): Likewise.
8147 (simplify_if_then_else): Likewise.
8148 (simplify_set): Likewise.
8149 (simplify_shift_const_1): Likewise.
8150 (simplify_comparison): Likewise.
8151 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
8152 * cse.c (notreg_cost): Likewise.
8153 (cse_insn): Likewise.
8154 * cselib.c (cselib_lookup_1): Likewise.
8155 * dojump.c (do_jump_1): Likewise.
8156 (do_compare_rtx_and_jump): Likewise.
8157 * dse.c (get_call_args): Likewise.
8158 * dwarf2out.c (rtl_for_decl_init): Likewise.
8159 (native_encode_initializer): Likewise.
8160 * expmed.c (emit_store_flag_1): Likewise.
8161 (emit_store_flag): Likewise.
8162 * expr.c (convert_modes): Likewise.
8163 (store_field): Likewise.
8164 (expand_expr_real_1): Likewise.
8165 * fold-const.c (fold_read_from_constant_string): Likewise.
8166 * gimple-ssa-sprintf.c (get_format_string): Likewise.
8167 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
8168 * optabs.c (expand_binop): Likewise.
8169 (expand_unop): Likewise.
8170 (expand_abs_nojump): Likewise.
8171 (expand_one_cmpl_abs_nojump): Likewise.
8172 * simplify-rtx.c (mode_signbit_p): Likewise.
8173 (val_signbit_p): Likewise.
8174 (val_signbit_known_set_p): Likewise.
8175 (val_signbit_known_clear_p): Likewise.
8176 (simplify_relational_operation_1): Likewise.
8177 * tree.c (vector_type_mode): Likewise.
8179 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8180 Alan Hayward <alan.hayward@arm.com>
8181 David Sherwood <david.sherwood@arm.com>
8183 * machmode.h (smallest_mode_for_size): Fix formatting.
8184 (smallest_int_mode_for_size): New function.
8185 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
8186 instead of smallest_mode_for_size.
8187 * combine.c (make_extraction): Likewise.
8188 * config/arc/arc.c (arc_expand_movmem): Likewise.
8189 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
8190 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8191 * config/s390/s390.c (s390_expand_insv): Likewise.
8192 * config/sparc/sparc.c (assign_int_registers): Likewise.
8193 * config/spu/spu.c (spu_function_value): Likewise.
8194 (spu_function_arg): Likewise.
8195 * coverage.c (get_gcov_type): Likewise.
8196 (get_gcov_unsigned_t): Likewise.
8197 * dse.c (find_shift_sequence): Likewise.
8198 * expmed.c (store_bit_field_1): Likewise.
8199 * expr.c (convert_move): Likewise.
8200 (store_field): Likewise.
8201 * internal-fn.c (expand_arith_overflow): Likewise.
8202 * optabs-query.c (get_best_extraction_insn): Likewise.
8203 * optabs.c (expand_twoval_binop_libfunc): Likewise.
8204 * stor-layout.c (layout_type): Likewise.
8205 (initialize_sizetypes): Likewise.
8206 * targhooks.c (default_get_mask_mode): Likewise.
8207 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
8209 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8210 Alan Hayward <alan.hayward@arm.com>
8211 David Sherwood <david.sherwood@arm.com>
8213 * machmode.h (opt_mode::else_blk): New function.
8214 (int_mode_for_mode): Declare.
8215 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
8216 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
8218 * cfgexpand.c (expand_debug_expr): Likewise.
8219 * combine.c (gen_lowpart_or_truncate): Likewise.
8220 (gen_lowpart_for_combine): Likewise.
8221 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
8222 * config/avr/avr.c (avr_to_int_mode): Likewise.
8223 (avr_out_plus_1): Likewise.
8224 (avr_out_plus): Likewise.
8225 (avr_out_round): Likewise.
8226 * config/i386/i386.c (ix86_split_to_parts): Likewise.
8227 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8228 (s390_expand_vcond): Likewise.
8229 * config/spu/spu.c (spu_split_immediate): Likewise.
8230 (spu_expand_mov): Likewise.
8231 * dse.c (get_stored_val): Likewise.
8232 * expmed.c (store_bit_field_1): Likewise.
8233 (convert_extracted_bit_field): Use int_mode_for_mode instead of
8235 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
8236 (extract_low_bits): Likewise.
8237 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
8238 handling rather than repeating the check.
8239 (emit_group_store): Likewise.
8240 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
8241 * optabs.c (expand_absneg_bit): Likewise.
8242 (expand_copysign_absneg): Likewise.
8243 (expand_copysign_bit): Likewise.
8244 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
8245 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
8246 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
8247 * var-tracking.c (prepare_call_arguments): Likewise.
8248 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
8249 int_mode_for_mode instead of mode_for_size.
8250 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8253 Alan Hayward <alan.hayward@arm.com>
8254 David Sherwood <david.sherwood@arm.com>
8256 * machmode.h (int_mode_for_size): New function.
8257 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
8258 instead of mode_for_size.
8259 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
8261 * combine.c (expand_field_assignment): Use int_mode_for_size
8262 instead of mode_for_size.
8263 (make_extraction): Likewise.
8264 (simplify_shift_const_1): Likewise.
8265 (simplify_comparison): Likewise.
8266 * dojump.c (do_jump): Likewise.
8267 * dwarf2out.c (mem_loc_descriptor): Likewise.
8268 * emit-rtl.c (init_derived_machine_modes): Likewise.
8269 * expmed.c (flip_storage_order): Likewise.
8270 (convert_extracted_bit_field): Likewise.
8271 * expr.c (copy_blkmode_from_reg): Likewise.
8272 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
8273 * internal-fn.c (expand_mul_overflow): Likewise.
8274 * lower-subreg.c (simple_move): Likewise.
8275 * optabs-libfuncs.c (init_optabs): Likewise.
8276 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8277 * tree.c (vector_type_mode): Likewise.
8278 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
8279 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
8280 * tree-vect-generic.c (expand_vector_parallel): Likewise.
8281 * tree-vect-stmts.c (vectorizable_load): Likewise.
8282 (vectorizable_store): Likewise.
8284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8285 Alan Hayward <alan.hayward@arm.com>
8286 David Sherwood <david.sherwood@arm.com>
8288 * coretypes.h (pod_mode): New type.
8289 (scalar_int_mode_pod): New typedef.
8290 * machmode.h (pod_mode): New class.
8291 (int_n_data_t::m): Change type to scalar_int_mode_pod.
8292 * genmodes.c (emit_mode_int_n): Update accordingly.
8293 * lower-subreg.h (target_lower_subreg): Change type to
8294 scalar_int_mode_pod.
8295 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
8296 scalar_int_mode_pod.
8298 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8299 Alan Hayward <alan.hayward@arm.com>
8300 David Sherwood <david.sherwood@arm.com>
8302 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
8303 machine_mode to scalar_int_mode.
8304 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
8305 (rs6000_option_override_internal): Remove cast to int.
8306 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
8307 machine_mode to scalar_int_mode.
8308 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
8309 (rs6000_option_override_internal): Remove cast to int.
8310 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
8311 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
8313 * config/s390/s390.c (s390_expand_builtin): Likewise.
8314 * coretypes.h (scalar_int_mode): New type.
8315 (opt_scalar_int_mode): New typedef.
8316 * machmode.h (scalar_int_mode): New class.
8317 (scalar_int_mode::includes_p): New function.
8318 (byte_mode): Change type to scalar_int_mode.
8319 (word_mode): Likewise.
8320 (ptr_mode): Likewise.
8321 * emit-rtl.c (byte_mode): Likewise.
8322 (word_mode): Likewise.
8323 (ptr_mode): Likewise.
8324 (init_derived_machine_modes): Update accordingly.
8325 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
8326 and MODE_PARTIAL_INT.
8327 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
8328 opt_scalar_int_mode.
8330 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8331 Alan Hayward <alan.hayward@arm.com>
8332 David Sherwood <david.sherwood@arm.com>
8334 * target.def (libgcc_floating_mode_supported_p): Take a
8336 * doc/tm.texi: Regenerate.
8337 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
8339 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
8340 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
8343 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8344 Alan Hayward <alan.hayward@arm.com>
8345 David Sherwood <david.sherwood@arm.com>
8347 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
8348 * doc/tm.texi: Regenerate.
8349 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
8350 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
8351 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
8352 * targhooks.h (default_floatn_mode): Likewise.
8353 * targhooks.c (default_floatn_mode): Likewise.
8354 * tree.c (build_common_tree_nodes): Update accordingly.
8356 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8357 Alan Hayward <alan.hayward@arm.com>
8358 David Sherwood <david.sherwood@arm.com>
8360 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
8361 (mode_iterator::iterate_p): Likewise.
8362 (mode_iterator::get_wider): Likewise.
8363 * expr.c (init_expr_target): Use opt_scalar_float_mode.
8365 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8366 Alan Hayward <alan.hayward@arm.com>
8367 David Sherwood <david.sherwood@arm.com>
8369 * coretypes.h (opt_scalar_float_mode): New typedef.
8370 * machmode.h (float_mode_for_size): New function.
8371 * emit-rtl.c (double_mode): Delete.
8372 (init_emit_once): Use float_mode_for_size.
8373 * stor-layout.c (layout_type): Likewise.
8374 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
8376 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8377 Alan Hayward <alan.hayward@arm.com>
8378 David Sherwood <david.sherwood@arm.com>
8380 * output.h (assemble_real): Take a scalar_float_mode.
8381 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
8382 * config/arm/arm.md (consttable_4): Likewise.
8383 (consttable_8): Likewise.
8384 (consttable_16): Likewise.
8385 * config/mips/mips.md (consttable_float): Likewise.
8386 * config/s390/s390.c (s390_output_pool_entry): Likewise.
8387 * varasm.c (assemble_real): Take a scalar_float_mode.
8388 (output_constant_pool_2): Update accordingly.
8389 (output_constant): Likewise.
8391 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8392 Alan Hayward <alan.hayward@arm.com>
8393 David Sherwood <david.sherwood@arm.com>
8395 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
8396 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
8397 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
8398 (native_encode_real): Likewise.
8399 (native_interpret_real): Likewise.
8400 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
8401 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8403 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8404 Alan Hayward <alan.hayward@arm.com>
8405 David Sherwood <david.sherwood@arm.com>
8407 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
8408 <scalar_float_mode>. Simplify.
8409 (gen_extend_conv_libfunc): Likewise.
8411 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8412 Alan Hayward <alan.hayward@arm.com>
8413 David Sherwood <david.sherwood@arm.com>
8415 * coretypes.h (scalar_float_mode): New type.
8416 * machmode.h (mode_traits::from_int): Use machine_mode if
8417 USE_ENUM_MODES is defined.
8418 (is_a): New function.
8420 (dyn_cast): Likewise.
8421 (scalar_float_mode): New class.
8422 (scalar_float_mode::includes_p): New function.
8423 (is_float_mode): Likewise.
8424 * gdbhooks.py (MachineModePrinter): New class.
8425 (build_pretty_printer): Use it for scalar_float_mode.
8426 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
8427 (format_helper::format_helper): Turn into a template.
8428 * genmodes.c (get_mode_class): New function.
8429 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
8430 or machine_mode if none.
8431 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
8432 as_a <scalar_float_mode>.
8433 * dwarf2out.c (mem_loc_descriptor): Likewise.
8434 (insert_float): Likewise.
8435 (add_const_value_attribute): Likewise.
8436 * simplify-rtx.c (simplify_immed_subreg): Likewise.
8437 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
8438 (expand_unop): Update accordingly.
8439 (expand_abs_nojump): Likewise.
8440 (expand_copysign_absneg): Take a scalar_float_mode.
8441 (expand_copysign_bit): Likewise.
8442 (expand_copysign): Update accordingly.
8444 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8445 Alan Hayward <alan.hayward@arm.com>
8446 David Sherwood <david.sherwood@arm.com>
8448 * coretypes.h (opt_mode): New class.
8449 * machmode.h (opt_mode): Likewise.
8450 (opt_mode::else_void): New function.
8451 (opt_mode::require): Likewise.
8452 (opt_mode::exists): Likewise.
8453 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
8454 (GET_MODE_2XWIDER_MODE): Likewise.
8455 (mode_iterator::get_wider): Update accordingly.
8456 (mode_iterator::get_2xwider): Likewise.
8457 (mode_iterator::get_known_wider): Likewise, turning into a template.
8458 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
8459 forcing a wider mode to exist.
8460 * config/cr16/cr16.h (LONG_REG_P): Likewise.
8461 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8462 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
8463 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
8464 * lower-subreg.c (init_lower_subreg): Likewise.
8465 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
8466 on the final iteration.
8467 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
8468 a wider mode exists before asking for a move pattern.
8469 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
8470 forcing a wider mode to exist.
8471 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
8472 returning false if no such mode exists.
8473 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
8474 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
8475 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
8476 Avoid checking for a MODE_INT if we already know the mode is not a
8478 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
8479 forcing a wider mode to exist.
8480 (expmed_mult_highpart_optab): Likewise.
8481 (expmed_mult_highpart): Likewise.
8482 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
8484 * lto-streamer-in.c (lto_input_mode_table): Likewise.
8485 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
8486 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8487 * internal-fn.c (expand_mul_overflow): Update use of
8488 GET_MODE_2XWIDER_MODE.
8489 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8490 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
8491 GET_MODE_WIDER_MODE.
8492 (convert_plusminus_to_widen): Likewise.
8493 * tree-switch-conversion.c (array_value_type): Likewise.
8494 * var-tracking.c (emit_note_insn_var_location): Likewise.
8495 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8496 Return false inside rather than outside the loop if no wider mode
8498 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
8499 and GET_MODE_2XWIDER_MODE
8500 (can_compare_p): Use else_void.
8501 * gdbhooks.py (OptMachineModePrinter): New class.
8502 (build_pretty_printer): Use it for opt_mode.
8504 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8505 Alan Hayward <alan.hayward@arm.com>
8506 David Sherwood <david.sherwood@arm.com>
8508 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
8509 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
8511 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8512 Alan Hayward <alan.hayward@arm.com>
8513 David Sherwood <david.sherwood@arm.com>
8515 * machmode.h (mode_traits): New structure.
8516 (get_narrowest_mode): New function.
8517 (mode_iterator::start): Likewise.
8518 (mode_iterator::iterate_p): Likewise.
8519 (mode_iterator::get_wider): Likewise.
8520 (mode_iterator::get_known_wider): Likewise.
8521 (mode_iterator::get_2xwider): Likewise.
8522 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
8523 (FOR_EACH_MODE): Likewise.
8524 (FOR_EACH_MODE_FROM): Likewise.
8525 (FOR_EACH_MODE_UNTIL): Likewise.
8526 (FOR_EACH_WIDER_MODE): Likewise.
8527 (FOR_EACH_2XWIDER_MODE): Likewise.
8528 * builtins.c (expand_builtin_strlen): Use new mode iterators.
8529 * combine.c (simplify_comparison): Likewise
8530 * config/i386/i386.c (type_natural_mode): Likewise.
8531 * cse.c (cse_insn): Likewise.
8532 * dse.c (find_shift_sequence): Likewise.
8533 * emit-rtl.c (init_derived_machine_modes): Likewise.
8534 (init_emit_once): Likewise.
8535 * explow.c (hard_function_value): Likewise.
8536 * expmed.c (extract_fixed_bit_field_1): Likewise.
8537 (extract_bit_field_1): Likewise.
8538 (expand_divmod): Likewise.
8539 (emit_store_flag_1): Likewise.
8540 * expr.c (init_expr_target): Likewise.
8541 (convert_move): Likewise.
8542 (alignment_for_piecewise_move): Likewise.
8543 (widest_int_mode_for_size): Likewise.
8544 (emit_block_move_via_movmem): Likewise.
8545 (copy_blkmode_to_reg): Likewise.
8546 (set_storage_via_setmem): Likewise.
8547 (compress_float_constant): Likewise.
8548 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8549 * optabs-query.c (get_best_extraction_insn): Likewise.
8550 * optabs.c (expand_binop): Likewise.
8551 (expand_twoval_unop): Likewise.
8552 (expand_twoval_binop): Likewise.
8553 (widen_leading): Likewise.
8554 (widen_bswap): Likewise.
8555 (expand_parity): Likewise.
8556 (expand_unop): Likewise.
8557 (prepare_cmp_insn): Likewise.
8558 (prepare_float_lib_cmp): Likewise.
8559 (expand_float): Likewise.
8560 (expand_fix): Likewise.
8561 (expand_sfix_optab): Likewise.
8562 * postreload.c (move2add_use_add2_insn): Likewise.
8563 * reg-stack.c (reg_to_stack): Likewise.
8564 * reginfo.c (choose_hard_reg_mode): Likewise.
8565 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8566 * stor-layout.c (mode_for_size): Likewise.
8567 (smallest_mode_for_size): Likewise.
8568 (mode_for_vector): Likewise.
8569 (finish_bitfield_representative): Likewise.
8570 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
8571 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
8572 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8573 * var-tracking.c (prepare_call_arguments): Likewise.
8575 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8576 Alan Hayward <alan.hayward@arm.com>
8577 David Sherwood <david.sherwood@arm.com>
8579 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
8580 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
8581 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
8582 * machmode.h (mode_size): Move earlier in file.
8583 (mode_precision): Likewise.
8584 (mode_inner): Likewise.
8585 (mode_nunits): Likewise.
8586 (mode_unit_size): Likewise.
8587 (unit_unit_precision): Likewise.
8588 (mode_wider): Likewise.
8589 (mode_2xwider): Likewise.
8590 (machine_mode): New class.
8591 (mode_to_bytes): New function.
8592 (mode_to_bits): Likewise.
8593 (mode_to_precision): Likewise.
8594 (mode_to_inner): Likewise.
8595 (mode_to_unit_size): Likewise.
8596 (mode_to_unit_precision): Likewise.
8597 (mode_to_nunits): Likewise.
8598 (GET_MODE_SIZE): Use mode_to_bytes.
8599 (GET_MODE_BITSIZE): Use mode_to_bits.
8600 (GET_MODE_PRECISION): Use mode_to_precision.
8601 (GET_MODE_INNER): Use mode_to_inner.
8602 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
8603 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
8604 (GET_MODE_NUNITS): Use mode_to_nunits.
8605 * system.h (ALWAYS_INLINE): New macro.
8606 * config/powerpcspe/powerpcspe-c.c
8607 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
8608 int for arg1_mode and arg2_mode.
8610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8611 Alan Hayward <alan.hayward@arm.com>
8612 David Sherwood <david.sherwood@arm.com>
8614 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
8615 Prefix mode names with E_ in case statements.
8616 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8617 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
8618 (aarch64_split_simd_move): Likewise.
8619 (aarch64_gen_storewb_pair): Likewise.
8620 (aarch64_gen_loadwb_pair): Likewise.
8621 (aarch64_gen_store_pair): Likewise.
8622 (aarch64_gen_load_pair): Likewise.
8623 (aarch64_get_condition_code_1): Likewise.
8624 (aarch64_constant_pool_reload_icode): Likewise.
8625 (get_rsqrte_type): Likewise.
8626 (get_rsqrts_type): Likewise.
8627 (get_recpe_type): Likewise.
8628 (get_recps_type): Likewise.
8629 (aarch64_gimplify_va_arg_expr): Likewise.
8630 (aarch64_simd_container_mode): Likewise.
8631 (aarch64_emit_load_exclusive): Likewise.
8632 (aarch64_emit_store_exclusive): Likewise.
8633 (aarch64_expand_compare_and_swap): Likewise.
8634 (aarch64_gen_atomic_cas): Likewise.
8635 (aarch64_emit_bic): Likewise.
8636 (aarch64_emit_atomic_swap): Likewise.
8637 (aarch64_emit_atomic_load_op): Likewise.
8638 (aarch64_evpc_trn): Likewise.
8639 (aarch64_evpc_uzp): Likewise.
8640 (aarch64_evpc_zip): Likewise.
8641 (aarch64_evpc_ext): Likewise.
8642 (aarch64_evpc_rev): Likewise.
8643 (aarch64_evpc_dup): Likewise.
8644 (aarch64_gen_ccmp_first): Likewise.
8645 (aarch64_gen_ccmp_next): Likewise.
8646 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
8647 (alpha_emit_xfloating_libcall): Likewise.
8648 (emit_insxl): Likewise.
8649 (alpha_arg_type): Likewise.
8650 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
8651 (arc_preferred_simd_mode): Likewise.
8652 (arc_secondary_reload): Likewise.
8653 (get_arc_condition_code): Likewise.
8654 (arc_print_operand): Likewise.
8655 (arc_legitimate_constant_p): Likewise.
8656 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8657 * config/arc/arc.md (casesi_load): Likewise.
8658 (casesi_compact_jump): Likewise.
8659 * config/arc/predicates.md (proper_comparison_operator): Likewise.
8660 (cc_use_register): Likewise.
8661 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8662 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
8663 (arm_init_iwmmxt_builtins): Likewise.
8664 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
8665 (neon_expand_vector_init): Likewise.
8666 (arm_attr_length_move_neon): Likewise.
8667 (maybe_get_arm_condition_code): Likewise.
8668 (arm_emit_vector_const): Likewise.
8669 (arm_preferred_simd_mode): Likewise.
8670 (arm_output_iwmmxt_tinsr): Likewise.
8671 (thumb1_output_casesi): Likewise.
8672 (thumb2_output_casesi): Likewise.
8673 (arm_emit_load_exclusive): Likewise.
8674 (arm_emit_store_exclusive): Likewise.
8675 (arm_expand_compare_and_swap): Likewise.
8676 (arm_evpc_neon_vuzp): Likewise.
8677 (arm_evpc_neon_vzip): Likewise.
8678 (arm_evpc_neon_vrev): Likewise.
8679 (arm_evpc_neon_vtrn): Likewise.
8680 (arm_evpc_neon_vext): Likewise.
8681 (arm_validize_comparison): Likewise.
8682 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
8683 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
8684 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
8685 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
8686 (c6x_preferred_simd_mode): Likewise.
8687 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
8688 (epiphany_rtx_costs): Likewise.
8689 * config/epiphany/predicates.md (proper_comparison_operator):
8691 * config/frv/frv.c (condexec_memory_operand): Likewise.
8692 (frv_emit_move): Likewise.
8693 (output_move_single): Likewise.
8694 (output_condmove_single): Likewise.
8695 (frv_hard_regno_mode_ok): Likewise.
8696 (frv_matching_accg_mode): Likewise.
8697 * config/h8300/h8300.c (split_adds_subs): Likewise.
8698 (h8300_rtx_costs): Likewise.
8699 (h8300_print_operand): Likewise.
8700 (compute_mov_length): Likewise.
8701 (output_logical_op): Likewise.
8702 (compute_logical_op_length): Likewise.
8703 (compute_logical_op_cc): Likewise.
8704 (h8300_shift_needs_scratch_p): Likewise.
8705 (output_a_shift): Likewise.
8706 (compute_a_shift_length): Likewise.
8707 (compute_a_shift_cc): Likewise.
8708 (expand_a_rotate): Likewise.
8709 (output_a_rotate): Likewise.
8710 * config/i386/i386.c (classify_argument): Likewise.
8711 (function_arg_advance_32): Likewise.
8712 (function_arg_32): Likewise.
8713 (function_arg_64): Likewise.
8714 (function_value_64): Likewise.
8715 (ix86_gimplify_va_arg): Likewise.
8716 (ix86_legitimate_constant_p): Likewise.
8717 (put_condition_code): Likewise.
8718 (split_double_mode): Likewise.
8719 (ix86_avx256_split_vector_move_misalign): Likewise.
8720 (ix86_expand_vector_logical_operator): Likewise.
8721 (ix86_split_idivmod): Likewise.
8722 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
8723 (ix86_build_const_vector): Likewise.
8724 (ix86_build_signbit_mask): Likewise.
8725 (ix86_match_ccmode): Likewise.
8726 (ix86_cc_modes_compatible): Likewise.
8727 (ix86_expand_branch): Likewise.
8728 (ix86_expand_sse_cmp): Likewise.
8729 (ix86_expand_sse_movcc): Likewise.
8730 (ix86_expand_int_sse_cmp): Likewise.
8731 (ix86_expand_vec_perm_vpermi2): Likewise.
8732 (ix86_expand_vec_perm): Likewise.
8733 (ix86_expand_sse_unpack): Likewise.
8734 (ix86_expand_int_addcc): Likewise.
8735 (ix86_split_to_parts): Likewise.
8736 (ix86_vectorize_builtin_gather): Likewise.
8737 (ix86_vectorize_builtin_scatter): Likewise.
8738 (avx_vpermilp_parallel): Likewise.
8739 (inline_memory_move_cost): Likewise.
8740 (ix86_tieable_integer_mode_p): Likewise.
8741 (x86_maybe_negate_const_int): Likewise.
8742 (ix86_expand_vector_init_duplicate): Likewise.
8743 (ix86_expand_vector_init_one_nonzero): Likewise.
8744 (ix86_expand_vector_init_one_var): Likewise.
8745 (ix86_expand_vector_init_concat): Likewise.
8746 (ix86_expand_vector_init_interleave): Likewise.
8747 (ix86_expand_vector_init_general): Likewise.
8748 (ix86_expand_vector_set): Likewise.
8749 (ix86_expand_vector_extract): Likewise.
8750 (emit_reduc_half): Likewise.
8751 (ix86_emit_i387_round): Likewise.
8752 (ix86_mangle_type): Likewise.
8753 (ix86_expand_round_sse4): Likewise.
8754 (expand_vec_perm_blend): Likewise.
8755 (canonicalize_vector_int_perm): Likewise.
8756 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
8757 (expand_vec_perm_1): Likewise.
8758 (expand_vec_perm_interleave3): Likewise.
8759 (expand_vec_perm_even_odd_pack): Likewise.
8760 (expand_vec_perm_even_odd_1): Likewise.
8761 (expand_vec_perm_broadcast_1): Likewise.
8762 (ix86_vectorize_vec_perm_const_ok): Likewise.
8763 (ix86_expand_vecop_qihi): Likewise.
8764 (ix86_expand_mul_widen_hilo): Likewise.
8765 (ix86_expand_sse2_abs): Likewise.
8766 (ix86_expand_pextr): Likewise.
8767 (ix86_expand_pinsr): Likewise.
8768 (ix86_preferred_simd_mode): Likewise.
8769 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
8770 * config/i386/sse.md (*andnot<mode>3): Likewise.
8771 (<mask_codefor><code><mode>3<mask_name>): Likewise.
8772 (*<code><mode>3): Likewise.
8773 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
8774 (ia64_expand_atomic_op): Likewise.
8775 (ia64_arg_type): Likewise.
8776 (ia64_mode_to_int): Likewise.
8777 (ia64_scalar_mode_supported_p): Likewise.
8778 (ia64_vector_mode_supported_p): Likewise.
8779 (expand_vec_perm_broadcast): Likewise.
8780 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
8781 (iq2000_function_arg_advance): Likewise.
8782 (iq2000_function_arg): Likewise.
8783 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
8784 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
8785 (m68k_libcall_value): Likewise.
8786 (m68k_function_value): Likewise.
8787 (sched_attr_op_type): Likewise.
8788 * config/mcore/mcore.c (mcore_output_move): Likewise.
8789 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
8791 (microblaze_function_arg): Likewise.
8792 * config/mips/mips.c (mips16_build_call_stub): Likewise.
8793 (mips_print_operand): Likewise.
8794 (mips_mode_ok_for_mov_fmt_p): Likewise.
8795 (mips_vector_mode_supported_p): Likewise.
8796 (mips_preferred_simd_mode): Likewise.
8797 (mips_expand_vpc_loongson_even_odd): Likewise.
8798 (mips_expand_vec_unpack): Likewise.
8799 (mips_expand_vi_broadcast): Likewise.
8800 (mips_expand_vector_init): Likewise.
8801 (mips_expand_vec_reduc): Likewise.
8802 (mips_expand_msa_cmp): Likewise.
8803 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
8804 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
8805 (cc_flags_for_mode): Likewise.
8806 * config/msp430/msp430.c (msp430_print_operand): Likewise.
8807 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
8808 (nds32_output_casesi_pc_relative): Likewise.
8809 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8810 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
8811 (nvptx_gen_unpack): Likewise.
8812 (nvptx_gen_pack): Likewise.
8813 (nvptx_gen_shuffle): Likewise.
8814 (nvptx_gen_wcast): Likewise.
8815 (nvptx_preferred_simd_mode): Likewise.
8816 * config/pa/pa.c (pa_secondary_reload): Likewise.
8817 * config/pa/predicates.md (base14_operand): Likewise.
8818 * config/powerpcspe/powerpcspe-c.c
8819 (altivec_resolve_overloaded_builtin): Likewise.
8820 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
8822 (rs6000_preferred_simd_mode): Likewise.
8823 (output_vec_const_move): Likewise.
8824 (rs6000_expand_vector_extract): Likewise.
8825 (rs6000_split_vec_extract_var): Likewise.
8826 (reg_offset_addressing_ok_p): Likewise.
8827 (rs6000_legitimate_offset_address_p): Likewise.
8828 (rs6000_legitimize_address): Likewise.
8829 (rs6000_emit_set_const): Likewise.
8830 (rs6000_const_vec): Likewise.
8831 (rs6000_emit_move): Likewise.
8832 (spe_build_register_parallel): Likewise.
8833 (rs6000_darwin64_record_arg_recurse): Likewise.
8834 (swap_selector_for_mode): Likewise.
8835 (spe_init_builtins): Likewise.
8836 (paired_init_builtins): Likewise.
8837 (altivec_init_builtins): Likewise.
8838 (do_load_for_compare): Likewise.
8839 (rs6000_generate_compare): Likewise.
8840 (rs6000_expand_float128_convert): Likewise.
8841 (emit_load_locked): Likewise.
8842 (emit_store_conditional): Likewise.
8843 (rs6000_output_function_epilogue): Likewise.
8844 (rs6000_handle_altivec_attribute): Likewise.
8845 (rs6000_function_value): Likewise.
8846 (emit_fusion_gpr_load): Likewise.
8847 (emit_fusion_p9_load): Likewise.
8848 (emit_fusion_p9_store): Likewise.
8849 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
8850 (fusion_gpr_mem_load): Likewise.
8851 (fusion_addis_mem_combo_load): Likewise.
8852 (fusion_addis_mem_combo_store): Likewise.
8853 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
8854 (fusion_gpr_mem_load): Likewise.
8855 (fusion_addis_mem_combo_load): Likewise.
8856 (fusion_addis_mem_combo_store): Likewise.
8857 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8859 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
8860 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
8861 (rs6000_preferred_simd_mode): Likewise.
8862 (output_vec_const_move): Likewise.
8863 (rs6000_expand_vector_extract): Likewise.
8864 (rs6000_split_vec_extract_var): Likewise.
8865 (reg_offset_addressing_ok_p): Likewise.
8866 (rs6000_legitimate_offset_address_p): Likewise.
8867 (rs6000_legitimize_address): Likewise.
8868 (rs6000_emit_set_const): Likewise.
8869 (rs6000_const_vec): Likewise.
8870 (rs6000_emit_move): Likewise.
8871 (rs6000_darwin64_record_arg_recurse): Likewise.
8872 (swap_selector_for_mode): Likewise.
8873 (paired_init_builtins): Likewise.
8874 (altivec_init_builtins): Likewise.
8875 (rs6000_expand_float128_convert): Likewise.
8876 (emit_load_locked): Likewise.
8877 (emit_store_conditional): Likewise.
8878 (rs6000_output_function_epilogue): Likewise.
8879 (rs6000_handle_altivec_attribute): Likewise.
8880 (rs6000_function_value): Likewise.
8881 (emit_fusion_gpr_load): Likewise.
8882 (emit_fusion_p9_load): Likewise.
8883 (emit_fusion_p9_store): Likewise.
8884 * config/rx/rx.c (rx_gen_move_template): Likewise.
8885 (flags_from_mode): Likewise.
8886 * config/s390/predicates.md (s390_alc_comparison): Likewise.
8887 (s390_slb_comparison): Likewise.
8888 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
8889 (s390_vector_mode_supported_p): Likewise.
8890 (s390_cc_modes_compatible): Likewise.
8891 (s390_match_ccmode_set): Likewise.
8892 (s390_canonicalize_comparison): Likewise.
8893 (s390_emit_compare_and_swap): Likewise.
8894 (s390_branch_condition_mask): Likewise.
8895 (s390_rtx_costs): Likewise.
8896 (s390_secondary_reload): Likewise.
8897 (__SECONDARY_RELOAD_CASE): Likewise.
8898 (s390_expand_cs): Likewise.
8899 (s390_preferred_simd_mode): Likewise.
8900 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
8901 * config/sh/sh.c (sh_print_operand): Likewise.
8902 (dump_table): Likewise.
8903 (sh_secondary_reload): Likewise.
8904 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8905 * config/sh/sh.md (casesi_worker_1): Likewise.
8906 (casesi_worker_2): Likewise.
8907 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
8908 (fcc_comparison_operator): Likewise.
8909 * config/sparc/sparc.c (sparc_expand_move): Likewise.
8910 (emit_soft_tfmode_cvt): Likewise.
8911 (sparc_preferred_simd_mode): Likewise.
8912 (output_cbranch): Likewise.
8913 (sparc_print_operand): Likewise.
8914 (sparc_expand_vec_perm_bmask): Likewise.
8915 (vector_init_bshuffle): Likewise.
8916 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
8917 (spu_vector_mode_supported_p): Likewise.
8918 (spu_expand_insv): Likewise.
8919 (spu_emit_branch_or_set): Likewise.
8920 (spu_handle_vector_attribute): Likewise.
8921 (spu_builtin_splats): Likewise.
8922 (spu_builtin_extract): Likewise.
8923 (spu_builtin_promote): Likewise.
8924 (spu_expand_sign_extend): Likewise.
8925 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
8926 (tilegx_simd_int): Likewise.
8927 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
8928 (tilepro_simd_int): Likewise.
8929 * config/v850/v850.c (const_double_split): Likewise.
8930 (v850_print_operand): Likewise.
8931 (ep_memory_offset): Likewise.
8932 * config/vax/vax.c (vax_rtx_costs): Likewise.
8933 (vax_output_int_move): Likewise.
8934 (vax_output_int_add): Likewise.
8935 (vax_output_int_subtract): Likewise.
8936 * config/visium/predicates.md (visium_branch_operator): Likewise.
8937 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
8938 (visium_print_operand_address): Likewise.
8939 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8940 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
8941 (xtensa_expand_conditional_branch): Likewise.
8942 (xtensa_copy_incoming_a7): Likewise.
8943 (xtensa_output_literal): Likewise.
8944 * dfp.c (decimal_real_maxval): Likewise.
8945 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
8947 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8948 Alan Hayward <alan.hayward@arm.com>
8949 David Sherwood <david.sherwood@arm.com>
8951 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
8952 (mode_nunits_inline): Likewise.
8953 (mode_inner_inline): Likewise.
8954 (mode_unit_size_inline): Likewise.
8955 (mode_unit_precision_inline): Likewise.
8956 (emit_insn_modes_h): Likewise. Also emit a #define of the
8958 (emit_mode_wider): Add an E_ prefix to mode names.
8959 (emit_mode_complex): Likewise.
8960 (emit_mode_inner): Likewise.
8961 (emit_mode_adjustments): Likewise.
8962 (emit_mode_int_n): Likewise.
8963 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
8964 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
8965 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
8966 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
8967 (CRC32_BUILTIN, ENTRY): Likewise.
8968 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
8969 (aarch64_pop_regs): Likewise.
8970 (aarch64_process_components): Likewise.
8971 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
8972 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
8973 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
8974 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
8975 * config/arm/arm.c (arm_init_libfuncs): Likewise.
8976 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
8978 * config/i386/i386-builtin.def (pcmpestr): Likewise.
8979 (pcmpistr): Likewise.
8980 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
8981 * config/mmix/mmix.c (mmix_output_condition): Likewise.
8982 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
8984 * config/rl78/rl78.c (mduc_regs): Likewise.
8985 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8986 (htm_expand_builtin): Likewise.
8987 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
8988 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
8989 * config/xtensa/xtensa.c (print_operand): Likewise.
8990 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
8991 (NUM_MODE_VECTOR_INT): Likewise.
8992 * genoutput.c (null_operand): Likewise.
8993 (output_operand_data): Likewise.
8994 * genrecog.c (print_parameter_value): Likewise.
8995 * lra.c (debug_operand_data): Likewise.
8997 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8999 * dwarf2out.c (scompare_loc_descriptor_wide)
9000 (scompare_loc_descriptor_narrow): New functions, split out from...
9001 (scompare_loc_descriptor): ...here.
9002 * expmed.c (emit_store_flag_int): New function, split out from...
9003 (emit_store_flag): ...here.
9005 2017-08-30 Richard Biener <rguenther@suse.de>
9007 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
9008 (dwarf2out_early_finish): Move setting of AT_pubnames from
9009 early debug output to early finish.
9011 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9013 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
9014 and -mdata-region to the assembler.
9016 2017-08-30 Richard Biener <rguenther@suse.de>
9018 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
9020 (gen_subprogram_die): Add DW_AT_object_pointer only early.
9021 (dwarf2out_early_global_decl): Only generate a DIE for the
9022 abstract origin if it doesn't already exist or is a declaration DIE.
9023 (resolve_addr): Do not add the linkage name twice when
9024 generating a stub DIE for the DW_TAG_GNU_call_site target.
9026 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9028 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9029 Use machine_mode rather than int for arg1_mode.
9031 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
9034 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
9035 that the second argument of the built-in functions to unpack
9036 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
9037 switch statement instead a lot of if statements.
9038 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
9039 Allow 64-bit values to be in Altivec registers as well as
9040 traditional floating point registers.
9041 (pack<mode>, FMOVE128_VSX iterator): Likewise.
9043 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
9045 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
9046 MAX_REGS_PER_ADDRESS == 1.
9048 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
9050 * config/i386/i386.opt (flag_fentry): Do not init to -1.
9051 * config/i386/i386.c (ix86_option_override_internal): Simplify
9052 setting of opts->x_flag_entry.
9054 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9055 Jakub Jelinek <jakub@redhat.com>
9056 Richard Biener <rguenther@suse.de>
9058 PR tree-optimization/81503
9059 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
9060 folded constant fits in the target type; reorder tests for clarity.
9062 2017-08-29 Martin Liska <mliska@suse.cz>
9064 * passes.def: Include pass_lower_switch.
9065 * stmt.c (dump_case_nodes): Remove and move to
9066 tree-switch-conversion.
9067 (case_values_threshold): Likewise.
9068 (expand_switch_as_decision_tree_p): Likewise.
9069 (emit_case_decision_tree): Likewise.
9070 (expand_case): Likewise.
9071 (balance_case_nodes): Likewise.
9072 (node_has_low_bound): Likewise.
9073 (node_has_high_bound): Likewise.
9074 (node_is_bounded): Likewise.
9075 (emit_case_nodes): Likewise.
9076 (struct simple_case_node): New struct.
9077 (add_case_node): Remove.
9078 (emit_case_dispatch_table): Use vector instead of case_list.
9079 (reset_out_edges_aux): Remove.
9080 (compute_cases_per_edge): Likewise.
9081 (expand_case): Build list of simple_case_node.
9082 (expand_sjlj_dispatch_table): Use it.
9083 * tree-switch-conversion.c (struct case_node): Moved from
9084 stmt.c and adjusted.
9085 (emit_case_nodes): Likewise.
9086 (node_has_low_bound): Likewise.
9087 (node_has_high_bound): Likewise.
9088 (node_is_bounded): Likewise.
9089 (case_values_threshold): Likewise.
9090 (reset_out_edges_aux): Likewise.
9091 (compute_cases_per_edge): Likewise.
9092 (add_case_node): Likewise.
9093 (dump_case_nodes): Likewise.
9094 (balance_case_nodes): Likewise.
9095 (expand_switch_as_decision_tree_p): Likewise.
9096 (emit_jump): Likewise.
9097 (emit_case_decision_tree): Likewise.
9098 (try_switch_expansion): Likewise.
9099 (do_jump_if_equal): Likewise.
9100 (emit_cmp_and_jump_insns): Likewise.
9101 (fix_phi_operands_for_edge): New function.
9102 (record_phi_operand_mapping): Likewise.
9103 (class pass_lower_switch): New pass.
9104 (pass_lower_switch::execute): New function.
9105 (make_pass_lower_switch): Likewise.
9106 (conditional_probability):
9107 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
9108 * tree-pass.h: Add make_pass_lower_switch.
9110 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9113 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
9116 2017-08-29 Richard Biener <rguenther@suse.de>
9118 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
9119 we do not add a DW_AT_inline attribute twice.
9120 (gen_subprogram_die): Remove code setting DW_AT_inline on
9121 DECL_ABSTRACT_P nodes.
9123 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
9125 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
9126 calls to internal functions.
9127 (gimplify_modify_expr): Likewise.
9128 * tree-call-cdce.c (use_internal_fn): Likewise.
9129 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
9130 (convert_to_divmod): Set the nothrow flag.
9131 * tree-if-conv.c (predicate_mem_writes): Likewise.
9132 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
9133 (vectorizable_call): Likewise.
9134 (vectorizable_store): Likewise.
9135 (vectorizable_load): Likewise.
9136 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
9137 (vect_recog_mask_conversion_pattern): Likewise.
9139 2017-08-29 Martin Liska <mliska@suse.cz>
9142 * gcc.c (driver_handle_option): Add new argument.
9143 * opts-common.c (handle_option): Pass
9144 target_option_override_hook.
9145 * opts-global.c (lang_handle_option): Add new option.
9146 (set_default_handlers): Add new argument.
9147 (decode_options): Likewise.
9148 * opts.c (target_handle_option): Likewise.
9149 (common_handle_option): Call target_option_override_hook.
9150 * opts.h (struct cl_option_handler_func): Add hook for
9151 target option override.
9152 (struct cl_option_handlers): Likewise.
9153 (set_default_handlers): Add new argument.
9154 (decode_options): Likewise.
9155 (common_handle_option): Likewise.
9156 (target_handle_option): Likewise.
9157 * toplev.c (toplev::main): Pass targetm.target_option.override
9160 2017-08-29 Richard Biener <rguenther@suse.de>
9161 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9163 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
9164 life to the active subtree.
9166 2017-08-28 Jeff Law <law@redhat.com>
9168 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
9169 derive_equivalences.
9170 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
9172 (edge_info::derive_equivalences): New private member function
9174 * tree-ssa-dom.c (class edge_info): Changed from a struct
9175 to a class. Add ctor/dtor, methods and data members.
9176 (edge_info::edge_info): Renamed from allocate_edge_info.
9177 Initialize additional members.
9178 (edge_info::~edge_info): New.
9179 (free_dom_edge_info): Delete the edge info.
9180 (record_edge_info): Use new class & associated member functions.
9181 Tighten forms for testing for edge equivalences.
9182 (record_temporary_equivalences): Iterate over the simple
9183 equivalences rather than assuming there's only one per edge.
9184 (cprop_into_successor_phis): Iterate over the simple
9185 equivalences rather than assuming there's only one per edge.
9186 (optimize_stmt): Use operand_equal_p rather than pointer
9187 equality for mini-DSE code.
9189 2017-08-28 Nathan Sidwell <nathan@acm.org>
9191 * gcc.c (execute): Fold SIGPIPE handling into switch
9192 statement. Adjust internal error message.
9194 2017-08-28 Richard Biener <rguenther@suse.de>
9197 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
9198 Do nothing for removed DIEs.
9200 2017-08-28 Richard Biener <rguenther@suse.de>
9202 PR tree-optimization/81977
9203 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
9206 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
9209 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
9210 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
9211 using targetm.gen_mem_thread_fence.
9213 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
9216 * config/i386/i386.md (*<btsc><mode>): Change operand 2
9217 predicate to register_operand. Reorder operands.
9218 (*btr<mode>): Ditto.
9219 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
9220 (*btr<mode>_mask): Ditto.
9222 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
9224 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
9225 * config/rs6000/xmmintrin.h: New file.
9226 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
9228 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9231 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
9232 parameter and_insn and return it.
9233 (recombine_lvx_pattern): Insert a copy to ensure availability of
9234 the base register of the copied masking operation at the point of
9235 the instruction replacement.
9236 (recombine_stvx_pattern): Likewise.
9238 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
9240 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
9241 undocumented switches.
9242 (-mpower9-dform-vector): Likewise.
9243 (-mpower9-dform): Likewise.
9244 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
9245 comments to delete references to -mpower9-dform* switches.
9246 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
9247 Delete reference to -mpower9-dform* switches, test for
9248 -mpower9-vector instead.
9249 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
9250 (OTHER_P9_VECTOR_MASKS): Likewise.
9251 (POWERPC_MASKS): Likewise.
9252 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
9253 tests against -mpower9-dform* to -mpower9-vector. Delete code
9254 that checked for -mpower9-dform* consistancy with other options.
9255 Add test for -mpower9-misc to enable other power9 switches.
9256 (rs6000_init_hard_regno_mode_ok): Likewise.
9257 (rs6000_option_override_internal): Likewise.
9258 (rs6000_emit_prologue): Likewise.
9259 (rs6000_emit_epilogue): Likewise.
9260 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
9261 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
9262 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
9264 (emit_fusion_p9_store): Likewise.
9265 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
9266 resetting these macros if the assembler does not support ISA 3.0
9268 (TARGET_P9_DFORM_VECTOR): Likewise.
9269 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
9270 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
9272 2017-08-25 Alan Modra <amodra@gmail.com>
9274 PR rtl-optimization/81747
9275 * cse.c (cse_extended_basic_block): Don't attempt to record
9276 equivalences for degenerate conditional branches that branch
9277 to their fall-through.
9279 2017-08-24 Martin Sebor <msebor@redhat.com>
9282 * gimple-fold.c (size_must_be_zero_p): New function.
9283 (gimple_fold_builtin_memory_op): Call it.
9285 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
9287 * config/rs6000/mm_malloc.h: New file.
9289 2017-08-24 Bin Cheng <bin.cheng@arm.com>
9291 PR tree-optimization/81913
9292 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
9293 analysis when either IVs in condition can wrap.
9295 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
9297 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
9298 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
9300 2017-08-24 Richard Biener <rguenther@suse.de>
9303 * targhooks.c (default_target_can_inline_p): Properly
9304 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
9305 is present and always compare.
9306 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
9307 infer -mfpmath=sse from TARGET_SSE_P.
9308 (ix86_can_inline_p): Properly use target_option_default_node when
9309 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
9311 2017-08-24 Richard Biener <rguenther@suse.de>
9314 * dwarf2out.c (output_die): Handle flag_generate_offload like
9316 (output_comp_unit): Likewise.
9317 (gen_array_type_die): Likewise.
9318 (dwarf2out_early_finish): Likewise.
9319 (note_variable_value_in_expr): Likewise.
9320 (dwarf2out_finish): Likewise. Adjust assert.
9321 * cgraphunit.c (symbol_table::compile): Move setting of
9322 flag_generate_offload earlier ...
9323 (symbol_table::finalize_compilation_unit): ... here, before
9324 early debug finalization.
9326 2017-08-24 Richard Biener <rguenther@suse.de>
9328 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
9329 and ipa-fnsummary.h.
9330 (ix86_can_inline_p): When ix86_fpmath flags do not match
9331 check whether the callee uses FP math at all.
9333 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
9336 * tree-ssanames.c (get_nonzero_bits): Use element_precision
9337 instead of TYPE_PRECISION.
9339 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
9340 Alan Hayward <alan.hayward@arm.com>
9341 David Sherwood <david.sherwood@arm.com>
9343 * combine.c (make_extraction): Use subreg_offset_from_lsb.
9345 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
9347 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
9349 (ix86_frame::stack_realign_allocate): New field.
9350 (struct machine_frame_state): Modify comments.
9351 (machine_frame_state::sp_realigned_fp_end): New field.
9352 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
9354 (sp_valid_at): Add assertion to assure no attempt to access invalid
9355 offset of a realigned stack.
9356 (fp_valid_at): Likewise.
9357 (choose_baseaddr): Modify comments.
9358 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
9359 ix86_expand_prologue.
9360 (ix86_expand_prologue): Modify stack realignment and allocation.
9361 (ix86_expand_epilogue): Modify comments.
9362 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
9363 avx2_runtime, avx512f, and avx512f_runtime.
9365 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
9367 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
9368 (mstackrealign): Do not init to -1.
9369 * config/i386/i386.c (ix86_option_override_internal):
9370 Check opts_set, not opts when setting default value of
9371 opts->x_ix86_force_align_arg_pointer.
9373 2017-08-23 Richard Biener <rguenther@suse.de>
9375 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
9376 lang_hooks.decl_printable_name.
9377 * print-rtl-function.c (print_rtx_function): Likewise.
9378 * tree-pretty-print.c (dump_function_header): Likewise.
9380 2017-08-23 Richard Biener <rguenther@suse.de>
9383 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
9384 -g0 at compile-time.
9386 2017-08-23 Tamar Christina <tamar.christina@arm.com>
9389 * doc/sourcebuild.texi (Other hardware attributes):
9392 2017-08-23 Tamar Christina <tamar.christina@arm.com>
9395 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
9396 Fix single-use check.
9398 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
9400 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
9402 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
9404 * doc/install.texi: Modify to add more details on running selected
9407 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
9409 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
9410 is combined with -mabi=ms.
9411 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
9414 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9416 PR tree-optimization/81488
9417 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
9418 and cached_basis fields.
9419 (MAX_SPREAD): New constant.
9420 (alloc_cand_and_find_basis): Initialize new fields.
9421 (clear_visited): New function.
9422 (create_phi_basis_1): Rename from create_phi_basis, set visited
9423 and cached_basis fields.
9424 (create_phi_basis): New wrapper function.
9425 (phi_add_costs_1): Rename from phi_add_costs, add spread
9426 parameter, set visited field, short-circuit when limits reached.
9427 (phi_add_costs): New wrapper function.
9428 (record_phi_increments_1): Rename from record_phi_increments, set
9430 (record_phi_increments): New wrapper function.
9431 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
9432 (phi_incr_cost): New wrapper function.
9433 (all_phi_incrs_profitable_1): Rename from
9434 all_phi_incrs_profitable, set visited field.
9435 (all_phi_incrs_profitable): New wrapper function.
9437 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
9438 Alan Hayward <alan.hayward@arm.com>
9439 David Sherwood <david.sherwood@arm.com>
9441 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
9442 that takes the outer and inner modes.
9443 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
9444 comparison as the canonical test for a paradoxical subreg.
9445 * combine.c (simplify_set): Use paradoxical_subreg_p.
9446 (make_extraction): Likewise.
9447 (force_to_mode): Likewise.
9448 (rtx_equal_for_field_assignment_p): Likewise.
9449 (gen_lowpart_for_combine): Likewise.
9450 (simplify_comparison): Likewise.
9451 * cse.c (equiv_constant): Likewise.
9452 * expmed.c (store_bit_field_1): Likewise.
9453 * final.c (alter_subreg): Likewise.
9454 * fwprop.c (propagate_rtx): Likewise.
9455 (forward_propagate_subreg): Likewise.
9456 * ira-conflicts.c (ira_build_conflicts): Likewise.
9457 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
9458 * lra-constraints.c (curr_insn_transform): Likewise.
9459 (split_reg): Likewise.
9460 * lra-eliminations.c (move_plus_up): Likewise.
9461 (lra_eliminate_regs_1): Likewise.
9462 * recog.c (general_operand): Likewise.
9463 * ree.c (combine_reaching_defs): Likewise.
9464 * reload.c (push_reload): Likewise.
9465 (find_reloads): Likewise.
9466 * reload1.c (elimination_effects): Likewise.
9467 (compute_reload_subreg_offset): Likewise.
9468 (choose_reload_regs): Likewise.
9469 * rtlanal.c (subreg_lsb_1): Likewise.
9470 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9471 (simplify_subreg): Likewise.
9472 * var-tracking.c (track_loc_p): Likewise.
9473 * emit-rtl.c (byte_lowpart_offset): Likewise.
9474 (paradoxical_subreg_p): Delete out-of-line definition.
9476 2017-08-22 Jeff Law <law@redhat.com>
9478 PR tree-optimization/81741
9479 PR tree-optimization/71947
9480 * tree-ssa-dom.c: Include tree-inline.h.
9481 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
9482 equivalences if one is more expensive to compute than the other.
9483 * tree-ssa-scopedtables.h (class const_or_copies): Make
9484 record_const_or_copy_raw method private.
9485 (class avail_exprs_stack): New method simplify_binary_operation.
9486 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
9487 avail_exprs_stack::simplify_binary_operation as needed.
9488 (avail_exprs_stack::simplify_binary_operation): New function.
9490 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
9492 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
9493 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
9494 (DOT_SYMBOLS): Likewise.
9495 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
9496 (RELOCATABLE_NEEDS_FIXUP): Likewise.
9497 (RS6000_ABI_NAME): Likewise.
9498 (TARGET_CMODEL): Likewise.
9499 (TOC_SECTION_ASM_OP): Likewise.
9500 (SET_CMODEL): New macro.
9501 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
9503 2017-08-22 Richard Biener <rguenther@suse.de>
9505 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
9506 to allow for free-lang-data replacements similar to verify_type_variant.
9508 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
9509 Alan Hayward <alan.hayward@arm.com>
9510 David Sherwood <david.sherwood@arm.com>
9512 * config/aarch64/aarch64.md (casesi): Use DImode rather than
9513 VOIDmode for the LABEL_REF.
9515 2017-08-22 Richard Biener <rguenther@suse.de>
9517 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
9519 2017-08-22 Richard Biener <rguenther@suse.de>
9521 * common.opt (feliminate-dwarf2-dups): Ignore.
9522 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
9523 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
9524 same_die_p_wrap, compute_section_prefix,
9525 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
9526 (comdat_symbol_id, comdat_symbol_number): Likewise.
9527 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
9529 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
9530 (output_die): Mark unreachable path unreachable.
9531 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
9532 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
9533 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
9534 (dwarf2out_early_finish): Likewise.
9536 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
9538 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
9540 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
9543 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
9544 not VAR_P. Filter attribute warnings with OPT_Wattributes.
9545 (avr_attribute_table) <io, io_low, address>: Initialize
9546 .decl_required with true.
9548 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
9550 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
9551 undocumented debugging options.
9552 (-mvsx-scalar-double): Likewise.
9553 (-mallow-df-permute): Likewise.
9554 (-mvectorize-builtins): Likewise.
9555 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9556 (rs6000_builtin_vectorized_function): Likewise.
9557 (rs6000_builtin_md_vectorized_function): Likewise.
9558 (rs6000_opt_vars): Likewise.
9560 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
9563 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
9564 (*btrq_imm): Rename from *btrq.
9565 (*btcq_imm): Rename from *btcq.
9566 (btsc): New code attribute.
9567 (*<btsc><mode>): New insn pattern.
9568 (*btr<mode>): Ditto.
9569 (*<btsc><mode>_mask): New insn_and_split pattern.
9570 (*btr<mode>_mask): Ditto.
9572 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9573 Alan Hayward <alan.hayward@arm.com>
9574 David Sherwood <david.sherwood@arm.com>
9576 * function.c (pad_below): Simplify padding calculation.
9578 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9579 Alan Hayward <alan.hayward@arm.com>
9580 David Sherwood <david.sherwood@arm.com>
9582 * target.def (function_prologue): Remove frame size argument.
9583 (function_epilogue): Likewise.
9584 * doc/tm.texi: Regenerate.
9585 * final.c (final_start_function): Update call to function_prologue.
9586 (final_end_function): Update call to function_epilogue.
9587 (default_function_pro_epilogue): Remove frame size argument.
9588 * output.h (default_function_pro_epilogue): Likewise.
9589 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
9590 (arm_output_function_prologue): Likewise.
9591 * config/frv/frv.c (frv_function_prologue): Likewise.
9592 (frv_function_epilogue): Likewise.
9593 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
9594 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
9595 (ia64_output_function_epilogue): Likewise.
9596 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
9597 (m32r_output_function_epilogue): Likewise.
9598 * config/microblaze/microblaze.c (microblaze_function_prologue)
9599 (microblaze_function_epilogue): Likewise.
9600 * config/mips/mips.c (mips_output_function_prologue): Likewise.
9601 (mips_output_function_epilogue): Likewise.
9602 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
9603 (mmix_target_asm_function_epilogue): Likewise.
9604 * config/msp430/msp430.c (msp430_start_function): Likewise.
9605 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
9606 (nds32_asm_function_epilogue): Likewise.
9607 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
9608 * config/pa/pa.c (pa_output_function_prologue): Likewise.
9609 (pa_output_function_epilogue): Likewise.
9610 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
9611 (rs6000_output_function_epilogue): Likewise.
9612 * config/rl78/rl78.c (rl78_start_function): Likewise.
9613 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
9614 (rs6000_output_function_epilogue): Likewise.
9615 * config/rx/rx.c (rx_output_function_prologue): Likewise.
9616 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
9617 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
9618 (sparc_asm_function_epilogue): Likewise.
9620 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9622 * tree.h (type_has_mode_precision_p): New function.
9623 * convert.c (convert_to_integer_1): Use it.
9624 * expr.c (expand_expr_real_2): Likewise.
9625 (expand_expr_real_1): Likewise.
9626 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
9627 * match.pd: Likewise.
9628 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
9629 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
9630 * tree-tailcall.c (process_assignment): Likewise.
9631 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9632 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
9633 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
9634 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9635 (vectorizable_assignment): Likewise.
9636 (vectorizable_shift): Likewise.
9637 (vectorizable_operation): Likewise.
9638 * tree-vrp.c (register_edge_assert_for_2): Likewise.
9640 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
9642 * match.pd: Add pow (C, x) simplification.
9644 2017-08-21 Richard Biener <rguenther@suse.de>
9646 PR tree-optimization/81900
9647 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
9648 for blocks with abnormal predecessors.
9649 (compute_antic): Do not set visited flag prematurely.
9651 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
9654 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
9656 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9658 * stor-layout.h (vector_type_mode): Move to...
9659 * tree.h (vector_type_mode): ...here.
9660 * stor-layout.c (vector_type_mode): Move to...
9661 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
9663 2017-08-21 Richard Biener <rguenther@suse.de>
9665 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
9666 register_external_die hooks.
9667 (debug_false_tree_charstarstar_uhwistar): Declare.
9668 (debug_nothing_tree_charstar_uhwi): Likewise.
9669 * debug.c (do_nothing_debug_hooks): Adjust.
9670 (debug_false_tree_charstarstar_uhwistar): New do nothing.
9671 (debug_nothing_tree_charstar_uhwi): Likewise.
9672 * dbxout.c (dbx_debug_hooks): Adjust.
9673 (xcoff_debug_hooks): Likewise.
9674 * sdbout.c (sdb_debug_hooks): Likewise.
9675 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
9676 * dwarf2out.c (macinfo_label_base): New global.
9677 (dwarf2out_register_external_die): New function for the
9678 register_external_die hook.
9679 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
9680 (dwarf2_debug_hooks): Use them.
9681 (dwarf2_lineno_debug_hooks): Adjust.
9682 (struct die_struct): Add with_offset flag.
9683 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
9684 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
9685 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
9686 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
9687 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
9688 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
9689 defining section names for the early LTO debug variants.
9690 (reset_indirect_string): New helper.
9691 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
9692 (print_dw_val): Add support for offsetted symbol references.
9693 (get_ultimate_context): Split out from is_cxx.
9694 (is_cxx): Use get_ultimate_context.
9695 (is_fortran): Add decl overload.
9696 (compute_comp_unit_symbol): Split out worker from
9697 compute_section_prefix.
9698 (compute_section_prefix): Call compute_comp_unit_symbol and
9699 set comdat_type_p here.
9700 (output_die): Skip DIE symbol output for the LTO added one.
9701 Handle DIE symbol references with offset.
9702 (output_comp_unit): Guard section name mangling properly.
9703 For LTO debug sections emit a symbol at the section beginning
9704 which we use to refer to its DIEs.
9705 (add_abstract_origin_attribute): For DIEs registered via
9706 dwarf2out_register_external_die directly refer to the early
9707 DIE rather than indirectly through the shadow one we created.
9708 Remove obsolete call to dwarf2out_abstract_function for
9709 non-function/block origins.
9710 (gen_array_type_die): When generating early LTO debug do
9711 not emit DW_AT_string_length.
9712 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
9713 late when in LTO. As suggested place a gcc_unreachable for
9714 the DECL_ABSTRACT_P case.
9715 (gen_subprogram_die): Avoid another specification DIE
9716 for early built declarations/definitions for the late LTO case.
9717 (gen_variable_die): Add type references for late duplicated VLA dies
9719 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
9720 we have the abstract instance already.
9721 (process_scope_var): Adjust decl DIE contexts in LTO which
9722 first puts them in limbo.
9723 (gen_decl_die): Do not generate type DIEs late apart from
9724 types for VLAs or for decls we do not yet have a DIE. Do not
9725 call dwarf2out_abstract_function late.
9726 (dwarf2out_early_global_decl): Make sure to create DIEs
9727 for abstract instances of a decl first.
9728 (dwarf2out_late_global_decl): Adjust comment.
9729 (output_macinfo_op): With multiple macro sections use
9730 macinfo_label_base to distinguish labels.
9731 (output_macinfo): Likewise. Update macinfo_label_base.
9732 Pass in the line info label.
9733 (note_variable_value_in_expr): When generating LTO resolve
9734 all variable values here by generating DIEs as needed.
9735 (init_sections_and_labels): Add early LTO debug flag parameter
9736 and generate different sections and names if set. Add generation
9737 counter for the labels so we can have multiple of them.
9738 (reset_dies): Helper to allow DIEs to be output multiple times.
9739 (dwarf2out_finish): When outputting DIEs to the fat part of an
9740 LTO object first reset DIEs.
9741 (dwarf2out_early_finish): Output early DIEs when generating LTO.
9742 (modified_type_die): Check for decl_ultimate_origin being self
9744 (gen_type_die_with_usage): Likewise.
9745 (gen_typedef_die): Allow decl_ultimate_origin being self.
9746 (set_decl_abstract_flags): Remove.
9747 (set_block_abstract_flags): Likewise.
9748 (dwarf2out_abstract_function): Treat the early generated DIEs
9749 as the abstract copy and only add DW_AT_inline and
9750 DW_AT_artificial here and call set_decl_origin_self.
9751 If the DIE has an abstract origin don't do anything.
9752 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
9753 if we have none yet (Go fails to build one, PR78628).
9754 (variably_modified_type_p): Prevent endless recursion for Ada
9755 cyclic pointer types.
9756 * lto-streamer-in.c: Include debug.h.
9757 (dref_queue): New global.
9758 (lto_read_tree_1): Stream in DIE references.
9759 (lto_input_tree): Register DIE references.
9760 (input_function): Stream DECL_DEBUG_ARGS.
9761 * lto-streamer-out.c: Include debug.h.
9762 (lto_write_tree_1): Output DIE references.
9763 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
9764 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
9765 (output_function): Stream DECL_DEBUG_ARGS.
9766 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
9767 Stream DECL_ABSTRACT_ORIGIN.
9768 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
9769 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
9770 DECL_CONTEXT for file-scope decls.
9771 * lto-streamer.h (struct dref_entry): Declare.
9772 (dref_queue): Likewise.
9773 * cfgexpand.c (pass_expand::execute): Do not call the
9774 outlining_inline_function hook here.
9775 * lto-wrapper.c (debug_obj): New global.
9776 (tool_cleanup): Unlink it if required.
9777 (debug_objcopy): New function.
9778 (run_gcc): Handle early debug sections in the IL files by
9779 extracting them to separate files, partially linkin them and
9780 feeding the result back as result to the linker.
9781 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
9782 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
9783 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
9784 sections into a separate segment.
9785 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
9787 (darwin_asm_dwarf_section): Likewise.
9788 (darwin_asm_output_dwarf_offset): Likewise.
9789 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
9791 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9792 Alan Hayward <alan.hayward@arm.com>
9793 David Sherwood <david.sherwood@arm.com>
9795 * read-md.h (md_reader::record_potential_iterator_use): Replace
9796 pointer argument with an rtx and an index.
9797 * read-rtl.c (iterator_group::apply_iterator): Likewise.
9798 (apply_mode_iterator): Likewise.
9799 (apply_code_iterator): Likewise.
9800 (apply_int_iterator): Likewise.
9801 (apply_subst_iterator): Likewise.
9802 (record_iterator_use): Likewise.
9803 (record_attribute_use): Likewise.
9804 (md_reader::record_potential_iterator_use): Likewise. Update calls
9805 to record_iterator_use and apply_iterator.
9806 (iterator_use): Replace ptr with x and index.
9807 (attribute_use): Likewise.
9808 (apply_attribute_uses): Update calls to apply_iterator.
9809 (apply_iterators): Likewise. Update initialization of iterator_use.
9810 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
9811 and record_potential_iterator_use.
9812 (rtx_reader::read_rtx_operand): Likewise.
9814 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9815 Alan Hayward <alan.hayward@arm.com>
9816 David Sherwood <david.sherwood@arm.com>
9818 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
9821 2017-08-21 Richard Biener <rguenther@suse.de>
9824 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
9825 at struct end conservatively when comparing common bases.
9827 2017-08-21 Richard Biener <rguenther@suse.de>
9829 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
9830 (mem_ref_in_stmt): Remove.
9831 (determine_max_movement): Use ref index to get at the reference.
9832 (invariantness_dom_walker::before_dom_children): Deal with
9833 lim data already initialized.
9834 (gather_mem_refs_stmt): Initialize lim data and record ref index.
9836 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
9838 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
9839 (TARGET_ISA_ROUND): Ditto.
9840 (TARGET_ROUND): Ditto.
9841 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
9842 * config/i386/i386.md: Ditto.
9843 * config/i386/sse.md: Ditto.
9844 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
9845 with OPTION_MASK_ISA_SSE4_1.
9847 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
9850 * doc/extend.texi (x86 Built-in Functions): Correct the name of
9851 __builtin_ia32_lzcnt_u16.
9853 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
9856 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
9857 (rs6000_set_current_function): Rewrite function to use it.
9859 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
9862 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
9863 and TYPE_WARN_IF_NOT_ALIGN.
9864 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
9865 (handle_warn_if_not_align): New.
9866 (place_union_field): Call handle_warn_if_not_align.
9867 (place_field): Call handle_warn_if_not_align.
9868 Copy TYPE_WARN_IF_NOT_ALIGN.
9869 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
9870 (layout_type): Likewise.
9871 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
9873 (tree_decl_common): Add warn_if_not_align.
9874 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
9875 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
9876 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
9877 (DECL_WARN_IF_NOT_ALIGN): Likewise.
9878 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
9879 * doc/extend.texi: Document warn_if_not_aligned attribute.
9880 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
9882 2017-08-17 Martin Liska <mliska@suse.cz>
9885 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
9886 (get_data_dependence): Use it as pointer type.
9887 (distribute_loop): Likewise.
9889 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9891 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
9892 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
9893 (p8_vmrgow_<mode>_direct): New define_insn.
9894 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
9895 handle endianness for vmrgew and vmrgow permute patterns.
9897 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
9899 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
9900 * config/rs6000/rs6000-cpus.def: Remove comment.
9901 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
9902 (POWERPC_MASKS): Likewise.
9903 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
9904 use of TARGET_VSX_TIMODE.
9905 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
9906 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
9907 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
9908 (rs6000_option_override_internal): Remove dead code.
9909 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
9910 (rs6000_legitimize_reload_address): Likewise.
9911 (rs6000_legitimate_address_p): Likewise.
9912 (rs6000_opt_masks): Delete "vsx-timode".
9913 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
9914 from function comment.
9915 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
9916 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
9917 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
9919 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
9920 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
9921 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
9924 (define_peephole2): Likewise.
9926 2017-08-17 Martin Sebor <msebor@redhat.com>
9929 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
9930 past the end of an array.
9931 (test_pp_format): Add test cases.
9933 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
9935 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
9936 missing ECF_NOTHROW flags.
9938 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
9941 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
9942 operands residing in integer registers.
9943 (*vsx_le_perm_load_<mode>): Likewise.
9944 (*vsx_le_perm_store_<mode>): Likewise.
9945 (define_peephole2): Add peepholes to optimize the above.
9947 2017-08-17 Marek Polacek <polacek@redhat.com>
9950 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
9951 to mimic what shorten_compare did. Change the return type to bool.
9952 (fold_cond_expr_with_comparison): Update call to
9953 operand_equal_for_comparison_p.
9954 (fold_ternary_loc): Likewise.
9956 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
9958 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
9960 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
9962 2017-08-17 Richard Biener <rguenther@suse.de>
9964 * tree-ssa-structalias.c (solve_graph): When propagating
9965 to successors update the graphs succ edges and avoid duplicate work.
9967 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
9970 * config/i386/i386.c (ix86_option_override_internal): Save target
9971 specific options after ix86_stack_protector_guard_reg was changed.
9973 2017-08-17 Richard Biener <rguenther@suse.de>
9975 PR tree-optimization/81827
9976 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
9977 (new_var_info): Initialize it conservatively.
9978 (get_call_vi): Mark register vars.
9979 (new_scalar_tmp_constraint_exp): Likewise.
9980 (handle_rhs_call): Likewise.
9981 (handle_const_call): Likewise.
9982 (create_function_info_for): Likewise.
9983 (solve_constraints): Sort varinfos to separate register from
9984 non-register vars to pack points-to solution bitmaps during
9987 2017-08-17 Marek Polacek <polacek@redhat.com>
9989 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
9991 2017-08-17 Richard Biener <rguenther@suse.de>
9993 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
9994 to true when overflow is undefined and we saturated the result.
9996 2017-08-17 Alan Modra <amodra@gmail.com>
9999 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
10000 Don't use store multiple if only one reg needs saving.
10001 (interesting_frame_related_regno): New function.
10002 (rs6000_frame_related): Don't emit frame info for regs that
10004 (rs6000_emit_epilogue): Likewise.
10006 2017-08-16 Nathan Sidwell <nathan@acm.org>
10008 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
10009 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
10010 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
10011 * tree.c (free_lang_data_in_type): Use else-if chain. Always
10012 clear TYPE_LANG_1. Remove obsolete member-function stripping.
10013 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
10014 (verify_type): Adjust for TYPE_BINFO move.
10015 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
10016 process TYPE_BINFO directly.
10017 (hash_tree): Likewise.
10018 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
10020 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
10023 2017-08-16 David Malcolm <dmalcolm@redhat.com>
10025 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
10027 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
10030 * config/i386/i386.md (*anddi_1_btr): Change predicates of
10031 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
10032 Add ix86_binary_operator_ok to insn constraint.
10033 (*iordi_1_bts): Ditto.
10034 (*xordi_1_btc): Ditto.
10035 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
10036 Update corresponding peephole2 pattern.
10040 2017-08-16 Bin Cheng <bin.cheng@arm.com>
10042 PR tree-optimization/81832
10043 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
10044 copy loop header which has IFN_LOOP_DIST_ALIAS call.
10046 2017-08-16 Marek Polacek <polacek@redhat.com>
10049 * fold-const.c (fold_indirect_ref_1): Restore original behavior
10050 regarding size_zero_node.
10052 2017-08-16 Martin Liska <mliska@suse.cz>
10055 * config.gcc: Respect previously set extra_objs in case
10058 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
10060 PR tree-optimization/81835
10061 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
10062 the phi SSA_NAME. Check that the condition in a COND_EXPR does
10063 not depend on the phi.
10065 2017-08-16 Alan Modra <amodra@gmail.com>
10067 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
10070 2017-08-16 Alan Modra <amodra@gmail.com>
10072 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
10073 (save_reg_p): ..into this. Update all callers.
10074 (first_reg_to_save): Simplify.
10076 2017-08-16 Alan Modra <amodra@gmail.com>
10078 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
10081 2017-08-15 Joseph Myers <joseph@codesourcery.com>
10085 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
10086 constant count if that count is less than 32.
10088 2017-08-15 Nathan Sidwell <nathan@acm.org>
10090 * gcc.c (execute): Emit friendlier message if inferior is killed
10091 by an external cause.
10093 2017-08-15 Richard Biener <rguenther@suse.de>
10095 PR tree-optimization/81790
10096 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
10097 CONSTRUCTORs from simplifying and VN.
10099 2017-08-14 Martin Sebor <msebor@redhat.com>
10101 * builtin-attrs.def: Add comments.
10103 2017-08-14 Martin Sebor <msebor@redhat.com>
10106 * doc/extend.texi (attribute nonstring): Document new attribute.
10108 2017-08-14 Martin Sebor <msebor@redhat.com>
10111 * tree-diagnostic.c (default_tree_printer): Handle %G.
10112 * gimple-pretty-print.h (percent_G_format): Declare new function.
10113 * gimple-pretty-print.c (percent_G_format): Define.
10114 * tree-pretty-print.c (percent_K_format): Add argument.
10116 2017-08-14 Martin Sebor <msebor@redhat.com>
10118 PR translation/79998
10119 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
10120 Remove a stray space.
10122 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
10125 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
10126 (*iordi_1_bts): Ditto.
10127 (*xordi_1_btc): Ditto.
10129 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10132 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
10133 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10135 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
10136 quoted strings, and make more translator-friendly.
10137 (darwin_rs6000_override_options): Likewise.
10138 (rs6000_option_override_internal): Likewise.
10139 (rs6000_return_in_memory): Fix overlong line.
10140 (init_cmulative_args): Use quoted strings, and make more
10141 translator-friendly.
10142 (rs6000_pass_by_reference): Fix overlong line.
10143 (def_builtin): Use quoted strings.
10144 (altivec_expand_predicate_builtin): Use quoted strings, and make
10145 more translator-friendly.
10146 (htm_expand_builtin): Use quoted strings.
10147 (cpu_expand_builtin): Use quoted strings, and make more
10148 translator-friendly.
10149 (altivec_expand_builtin): Likewise.
10150 (paired_expand_predicate_builtin): Likewise.
10151 (rs6000_invalid_builtin): Likewise.
10152 (builtin_function_type): Use quoted strings.
10153 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
10154 more translator-friendly.
10155 (rs6000_trampoline_init): Likewise.
10156 (rs6000_handle_altivec_attribute): Likewise.
10157 (rs6000_inner_target_options): Use quoted strings.
10158 (rs6000_disable_incompatible_switches): Likewise.
10159 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
10160 strings, and make more translator-friendly.
10161 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
10163 2017-08-14 Bin Cheng <bin.cheng@arm.com>
10165 PR tree-optimization/81799
10166 * tree-loop-distribution.c (version_loop_by_alias_check): Force
10167 cond_expr to simple gimple operand.
10169 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
10171 PR middle-end/46932
10172 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
10174 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
10178 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
10179 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
10180 TARGET_GASISR_PROLOGUES.
10181 * config/avr/avr.c (avr_option_override): Same.
10182 (avr_pass_pre_proep::execute): Same.
10184 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
10187 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
10188 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
10190 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
10192 * config/i386/i386.md (*load_tp_<mode>): Redefine as
10193 define_insn_and_split. Split to a memory load from 0 in
10194 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
10195 using PTR mode iterator.
10196 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
10197 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
10198 (*add_tp_<mode>): Redefine as define_insn_and_split.
10199 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
10200 address space. Merge with *add_tp_x32 using PTR mode iterator.
10201 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
10202 Split to an add with a memory load from 0 in
10203 DEFAULT_TLS_SEG_REG address space.
10205 2017-08-12 Andrew Pinski <apinski@cavium.com>
10207 * config/aarch64/aarch64-option-extensions.def (rdma):
10208 Fix feature string to what Linux prints out in /proc/cpuinfo.
10210 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
10213 * dwarf2out.c (modified_type_die): For C typedef types that have
10214 an ultimate origin, process the ultimate origin instead of the
10216 (gen_typedef_die): Assert that input DECLs have no ultimate
10218 (gen_type_die_with_usage): For typedef variants that have an
10219 ultimate origin, just call gen_decl_die on the original DECL.
10220 (process_scope_var): Avoid creating DIEs for local typedefs and
10221 concrete static variables.
10223 2017-08-12 Alan Modra <amodra@gmail.com>
10227 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
10228 match gnu-user.h startfile.
10229 (ENDFILE_LINUX_SPEC): Similarly.
10231 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
10234 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
10236 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
10238 2017-08-11 Tamar Christina <tamar.christina@arm.com>
10239 * config/aarch64/aarch64.md (mov<mode>): Change.
10240 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
10241 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
10242 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
10244 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
10246 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
10247 for storage order barriers.
10249 2017-08-11 Martin Liska <mliska@suse.cz>
10252 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
10253 variables of void type.
10255 2017-08-11 Martin Liska <mliska@suse.cz>
10257 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
10258 TARGET_SUPPORTS_ALIASES.
10259 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
10260 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
10261 (optimize_weakref): Likewise.
10262 * symtab.c (symtab_node::noninterposable_alias): Likewise.
10263 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
10264 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
10266 2017-08-11 Martin Liska <mliska@suse.cz>
10269 * config/i386/i386.c (make_resolver_func): Do complete
10270 refactoring of the function.
10272 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
10275 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
10276 * config/i386/i386.c (ix86_stack_protect_guard): Use
10277 ix86_stack_protect_guard_symbol_str to generate varible declaration.
10278 * doc/invoke.texi (x86 Options): Document
10279 -mstack-protector-guard-symbol= option.
10281 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
10283 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
10284 * config/i386/i386.c (ix86_split_stack_guard): New function.
10285 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
10286 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
10287 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
10288 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
10289 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
10290 (split_stack_space_check): Call ix86_split_stack_guard.
10292 2017-08-10 Martin Sebor <msebor@redhat.com>
10294 * print-tree.c (print_node): Print location using the established
10296 Replace spaces with colons.
10297 (debug_raw, debug): Ditto.
10299 2017-08-10 Martin Sebor <msebor@redhat.com>
10302 * pretty-print.c (pp_format): Correct the handling of %s precision.
10304 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
10307 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
10309 (ix86_finalize_stack_frame_flags): This. Also clear
10310 frame_pointer_needed if -fno-omit-frame-pointer is used without
10312 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
10313 with ix86_finalize_stack_frame_flags.
10314 (ix86_expand_epilogue): Likewise.
10315 (ix86_expand_split_stack_prologue): Likewise.
10316 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
10318 2017-08-10 Martin Liska <mliska@suse.cz>
10321 * c-attribs.c (handle_target_attribute):
10322 Report warning for an empty string argument of target attribute.
10324 2017-08-09 Jakub Jelinek <jakub@redhat.com>
10327 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
10329 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
10330 or DECL_NONLOCAL labels.
10331 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
10332 or DECL_NONLOCAL labels here.
10334 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
10336 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
10337 to indicate when early gimple folding has been disabled.
10338 (rs6000_gimple_fold_builtin): Add debug content.
10339 (rs6000_invalid_builtin): Fix whitespace.
10340 (rs6000_expand_builtin): Fix whitespace.
10341 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
10343 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
10346 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
10347 SAVE_MULTIPLE if not all the registers that saves, should be saved.
10349 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
10351 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
10352 (qdf24xx): Likewise.
10353 * config/aarch64/aarch64.md: Include falkor.md.
10354 * config/aarch64/falkor.md: New.
10356 2017-08-09 Marek Polacek <polacek@redhat.com>
10359 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
10360 * diagnostic.c (emit_diagnostic): Add a comment.
10361 (emit_diagnostic_valist): New function.
10363 2017-08-09 Marek Polacek <polacek@redhat.com>
10366 * input.c (make_location): New overload.
10367 * input.h (make_location): Declare.
10369 2017-08-08 Alan Modra <amodra@gmail.com>
10370 H.J. Lu <hongjiu.lu@intel.com>
10373 * gcc.c (NO_PIE_SPEC): Delete.
10374 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
10376 (LINK_PIE_SPEC): ..to here.
10377 (LINK_COMMAND_SPEC): Support -no-pie.
10378 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
10379 chain of crtbegin*.o selection, update for PIE_SPEC changes and
10381 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
10382 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
10383 (ENDFILE_CRTEND_SPEC): Similarly.
10385 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
10388 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
10389 (mstack-protector-guard-offset=): Ditto.
10390 * config/i386/i386.c (ix86_option_override): Handle
10391 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
10393 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
10394 ix86_stack_protect_guard_offset variables.
10395 (TARGET_STACK_PROTECT_GUARD): Always define.
10396 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
10397 and -mstack-protector-guard-offset= options.
10399 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10401 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
10402 boundary case for the last candidate.
10404 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10406 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
10407 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
10409 2017-08-08 Tamar Christina <tamar.christina@arm.com>
10411 PR middle-end/19706
10412 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
10413 * config/aarch64/aarch64-builtins.c
10414 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
10415 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
10416 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
10418 2017-08-08 Tamar Christina <tamar.christina@arm.com>
10419 Andrew Pinski <pinskia@gmail.com>
10421 PR middle-end/19706
10422 * internal-fn.def (XORSIGN): New.
10423 * optabs.def (xorsign_optab): New.
10424 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
10425 (convert_expand_mult_copysign): New.
10426 (pass_optimize_widening_mul::execute): Call
10427 convert_expand_mult_copysign.
10429 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10431 PR tree-optimization/81354
10432 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
10433 Insert on edges rather than explicitly creating landing pads.
10434 (analyze_candidates_and_replace): Commit edge inserts.
10436 2017-08-08 Richard Biener <rguenther@suse.de>
10438 PR middle-end/81719
10439 * tree-ssa-loop-niter.c: Include tree-dfa.h.
10440 (expand_simple_operations): Also look through ADDR_EXPRs with
10441 MEM_REF bases treating them as POINTER_PLUS_EXPR.
10443 2017-08-08 Richard Biener <rguenther@suse.de>
10445 PR tree-optimization/81723
10446 * tree-vect-slp.c (struct bst_traits): New hash traits.
10447 (bst_fail): New global.
10448 (vect_build_slp_tree_2): New worker, split out from ...
10449 (vect_build_slp_tree): ... this now wrapping it with using
10450 bst_fail set to cache SLP tree build fails. Properly handle
10452 (vect_analyze_slp_instance): Allocate and free bst_fail.
10454 2017-08-08 Martin Liska <mliska@suse.cz>
10457 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
10458 LABEL_DECLs that can be from a different function.
10460 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10462 PR tree-optimization/81744
10463 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
10464 loop's number of iterations.
10466 2017-08-08 Martin Liska <mliska@suse.cz>
10468 * asan.c: Include header files.
10469 * attribs.c (build_decl_attribute_variant): New function moved
10471 (build_type_attribute_qual_variant): Likewise.
10472 (cmp_attrib_identifiers): Likewise.
10473 (simple_cst_list_equal): Likewise.
10474 (omp_declare_simd_clauses_equal): Likewise.
10475 (attribute_value_equal): Likewise.
10476 (comp_type_attributes): Likewise.
10477 (build_type_attribute_variant): Likewise.
10478 (lookup_ident_attribute): Likewise.
10479 (remove_attribute): Likewise.
10480 (merge_attributes): Likewise.
10481 (merge_type_attributes): Likewise.
10482 (merge_decl_attributes): Likewise.
10483 (merge_dllimport_decl_attributes): Likewise.
10484 (handle_dll_attribute): Likewise.
10485 (attribute_list_equal): Likewise.
10486 (attribute_list_contained): Likewise.
10487 * attribs.h (lookup_attribute): New function moved from tree.[ch].
10488 (lookup_attribute_by_prefix): Likewise.
10489 * bb-reorder.c: Include header files.
10490 * builtins.c: Likewise.
10491 * calls.c: Likewise.
10492 * cfgexpand.c: Likewise.
10493 * cgraph.c: Likewise.
10494 * cgraphunit.c: Likewise.
10495 * convert.c: Likewise.
10496 * dwarf2out.c: Likewise.
10497 * final.c: Likewise.
10498 * fold-const.c: Likewise.
10499 * function.c: Likewise.
10500 * gimple-expr.c: Likewise.
10501 * gimple-fold.c: Likewise.
10502 * gimple-pretty-print.c: Likewise.
10503 * gimple.c: Likewise.
10504 * gimplify.c: Likewise.
10505 * hsa-common.c: Likewise.
10506 * hsa-gen.c: Likewise.
10507 * internal-fn.c: Likewise.
10508 * ipa-chkp.c: Likewise.
10509 * ipa-cp.c: Likewise.
10510 * ipa-devirt.c: Likewise.
10511 * ipa-fnsummary.c: Likewise.
10512 * ipa-inline.c: Likewise.
10513 * ipa-visibility.c: Likewise.
10515 * lto-cgraph.c: Likewise.
10516 * omp-expand.c: Likewise.
10517 * omp-general.c: Likewise.
10518 * omp-low.c: Likewise.
10519 * omp-offload.c: Likewise.
10520 * omp-simd-clone.c: Likewise.
10521 * opts-global.c: Likewise.
10522 * passes.c: Likewise.
10523 * predict.c: Likewise.
10524 * sancov.c: Likewise.
10525 * sanopt.c: Likewise.
10526 * symtab.c: Likewise.
10527 * toplev.c: Likewise.
10528 * trans-mem.c: Likewise.
10529 * tree-chkp.c: Likewise.
10530 * tree-eh.c: Likewise.
10531 * tree-into-ssa.c: Likewise.
10532 * tree-object-size.c: Likewise.
10533 * tree-parloops.c: Likewise.
10534 * tree-profile.c: Likewise.
10535 * tree-ssa-ccp.c: Likewise.
10536 * tree-ssa-live.c: Likewise.
10537 * tree-ssa-loop.c: Likewise.
10538 * tree-ssa-sccvn.c: Likewise.
10539 * tree-ssa-structalias.c: Likewise.
10540 * tree-ssa.c: Likewise.
10541 * tree-streamer-in.c: Likewise.
10542 * tree-vectorizer.c: Likewise.
10543 * tree-vrp.c: Likewise.
10544 * tsan.c: Likewise.
10545 * ubsan.c: Likewise.
10546 * varasm.c: Likewise.
10547 * varpool.c: Likewise.
10548 * tree.c: Remove functions moved to attribs.[ch].
10549 * tree.h: Likewise.
10550 * config/aarch64/aarch64.c: Add attrs.h header file.
10551 * config/alpha/alpha.c: Likewise.
10552 * config/arc/arc.c: Likewise.
10553 * config/arm/arm.c: Likewise.
10554 * config/avr/avr.c: Likewise.
10555 * config/bfin/bfin.c: Likewise.
10556 * config/c6x/c6x.c: Likewise.
10557 * config/cr16/cr16.c: Likewise.
10558 * config/cris/cris.c: Likewise.
10559 * config/darwin.c: Likewise.
10560 * config/epiphany/epiphany.c: Likewise.
10561 * config/fr30/fr30.c: Likewise.
10562 * config/frv/frv.c: Likewise.
10563 * config/ft32/ft32.c: Likewise.
10564 * config/h8300/h8300.c: Likewise.
10565 * config/i386/winnt.c: Likewise.
10566 * config/ia64/ia64.c: Likewise.
10567 * config/iq2000/iq2000.c: Likewise.
10568 * config/lm32/lm32.c: Likewise.
10569 * config/m32c/m32c.c: Likewise.
10570 * config/m32r/m32r.c: Likewise.
10571 * config/m68k/m68k.c: Likewise.
10572 * config/mcore/mcore.c: Likewise.
10573 * config/microblaze/microblaze.c: Likewise.
10574 * config/mips/mips.c: Likewise.
10575 * config/mmix/mmix.c: Likewise.
10576 * config/mn10300/mn10300.c: Likewise.
10577 * config/moxie/moxie.c: Likewise.
10578 * config/msp430/msp430.c: Likewise.
10579 * config/nds32/nds32-isr.c: Likewise.
10580 * config/nds32/nds32.c: Likewise.
10581 * config/nios2/nios2.c: Likewise.
10582 * config/nvptx/nvptx.c: Likewise.
10583 * config/pa/pa.c: Likewise.
10584 * config/pdp11/pdp11.c: Likewise.
10585 * config/powerpcspe/powerpcspe.c: Likewise.
10586 * config/riscv/riscv.c: Likewise.
10587 * config/rl78/rl78.c: Likewise.
10588 * config/rx/rx.c: Likewise.
10589 * config/s390/s390.c: Likewise.
10590 * config/sh/sh.c: Likewise.
10591 * config/sol2.c: Likewise.
10592 * config/sparc/sparc.c: Likewise.
10593 * config/spu/spu.c: Likewise.
10594 * config/stormy16/stormy16.c: Likewise.
10595 * config/tilegx/tilegx.c: Likewise.
10596 * config/tilepro/tilepro.c: Likewise.
10597 * config/v850/v850.c: Likewise.
10598 * config/vax/vax.c: Likewise.
10599 * config/visium/visium.c: Likewise.
10600 * config/xtensa/xtensa.c: Likewise.
10602 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10605 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
10606 constraints since the -mupper-regs-* switches have been
10608 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
10609 into a vector from a double word element that was extracted from
10610 another vector, and eliminate extra XXPERMDI instructions.
10611 (vsx_concat_<mode>_2): Likewise.
10612 (vsx_concat_<mode>_3): Likewise.
10613 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
10614 concat to allow optimizing inserts from previous extracts.
10616 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
10618 * config/i386/i386.c (ix86_stack_protect_guard): Generate
10619 memory reference to a SSP offset in TLS address space.
10620 (ix86_print_operand) <case '@'>: Remove.
10621 (ix86_print_operand_punct_valid_p): Remove '@' code.
10622 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
10623 UNSPEC_SP_TLS_TEST.
10624 (stack_tls_protect_set_<mode>): Remove.
10625 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
10626 (stack_tls_protect_test_<mode>): Remove.
10627 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
10629 2017-08-07 Olivier Hainque <hainque@adacore.com>
10632 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
10634 2017-08-07 Douglas Rupp <rupp@adacore.com>
10636 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
10637 variable was referenced as multidir in command.
10639 2017-08-07 Jakub Jelinek <jakub@redhat.com>
10642 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
10645 2017-08-07 Martin Liska <mliska@suse.cz>
10647 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
10648 * config/rl78/rl78.c: Add include of attribs.h.
10649 * config/sh/sh.c: Likewise.
10650 * config/v850/v850.c: Likewise.
10652 2017-08-07 Tom de Vries <tom@codesourcery.com>
10654 PR middle-end/78266
10655 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
10657 2017-08-07 Martin Liska <mliska@suse.cz>
10659 * config/mips/mips.c: Include attribs.h.
10661 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
10664 * doc/invoke.texi: Document change in behvaior for -Ofast for
10667 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
10669 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
10671 (aarch64_pop_regs): Likewise.
10672 (aarch64_gen_load_pair): Likewise.
10673 (aarch64_save_callee_saves): Likewise.
10674 (aarch64_restore_callee_saves): Likewise.
10676 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10678 * config/i386/i386.c: Revert the last change.
10680 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10683 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
10685 (ix86_finalize_stack_frame_flags): This. Also clear
10686 frame_pointer_needed if -fno-omit-frame-pointer is used without
10688 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
10689 with ix86_finalize_stack_frame_flags.
10690 (ix86_expand_epilogue): Likewise.
10691 (ix86_expand_split_stack_prologue): Likewise.
10693 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10696 * config/i386/i386.c (get_builtin_code_for_version): Set priority
10697 to P_AES for Westmere.
10699 2017-08-07 Jonathan Yong <10walls@gmail.com>
10701 * config/i386/mingw.opt (fset-stack-executable): Removed.
10702 * config/i386/cygming.opt (fset-stack-executable): Moved
10704 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
10706 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
10708 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
10710 2017-08-07 Marek Polacek <polacek@redhat.com>
10712 PR middle-end/81737
10713 * fold-const.c (fold_indirect_ref_1): Check type_domain.
10715 2017-08-07 Martin Liska <mliska@suse.cz>
10717 * attribs.h (canonicalize_attr_name): New function.
10718 (cmp_attribs): Move from c-format.c and adjusted.
10719 (is_attribute_p): Moved from tree.h.
10720 * tree-inline.c: Add new includes.
10721 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
10722 (private_is_attribute_p): Remove.
10723 (private_lookup_attribute): Likewise.
10724 (private_lookup_attribute_by_prefix): Simplify.
10725 (remove_attribute): Use is_attribute_p.
10726 * tree.h: Remove removed declarations.
10728 2017-08-07 Jakub Jelinek <jakub@redhat.com>
10730 PR middle-end/81698
10731 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
10732 instead of computing it in the function. Formatting fix.
10733 (expand_case): Don't rely on default_edge being the first edge,
10734 clear it if removing it, pass default_edge to
10735 emit_case_dispatch_table.
10736 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
10739 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
10741 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
10742 insn in the function, emit NOP after the insn.
10744 2017-08-06 Tom de Vries <tom@codesourcery.com>
10746 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
10749 2017-08-06 Tom de Vries <tom@codesourcery.com>
10751 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
10754 2017-08-04 Yury Gribov <tetra2005@gmail.com>
10756 PR tree-optimization/57371
10757 * match.pd: New pattern.
10759 2017-08-04 Marek Polacek <polacek@redhat.com>
10761 PR middle-end/81695
10762 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
10763 perform the computation in offset_int.
10765 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10767 PR tree-optimization/81136
10768 * tree-vectorizer.h: Include tree-hash-traits.h.
10769 (vec_base_alignments): New typedef.
10770 (vec_info): Add a base_alignments field.
10771 (vect_record_base_alignments): Declare.
10772 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
10774 (DR_IS_CONDITIONAL_IN_STMT): New macro.
10775 (create_data_ref): Add an is_conditional_in_stmt argument.
10776 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
10777 the is_conditional_in_stmt field.
10778 (data_ref_loc): Add an is_conditional_in_stmt field.
10779 (get_references_in_stmt): Set the is_conditional_in_stmt field.
10780 (find_data_references_in_stmt): Update call to create_data_ref.
10781 (graphite_find_data_references_in_stmt): Likewise.
10782 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
10783 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
10784 (vect_record_base_alignment): New function.
10785 (vect_record_base_alignments): Likewise.
10786 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
10787 for nested statements even if we fail to compute a misalignment.
10788 Use pooled base alignments for unconditional references.
10789 (vect_find_same_alignment_drs): Compare base addresses instead
10791 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
10792 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
10794 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10796 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
10797 Add an explicit name for the enum. Use auto_vec for slp_instances
10798 and grouped_stores.
10799 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
10801 (_bb_vec_info): Add a constructor and destructor.
10802 (vinfo_for_stmt): Return NULL for uids of -1 as well.
10803 (destroy_loop_vec_info): Delete.
10804 (vect_destroy_datarefs): Likewise.
10805 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
10806 (vec_info::vec_info): New function.
10807 (vec_info::~vec_info): Likewise.
10808 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
10809 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
10810 destroy_loop_vec_info.
10811 * tree-vect-loop.c (new_loop_vec_info): Replace with...
10812 (_loop_vec_info::_loop_vec_info): ...this.
10813 (destroy_loop_vec_info): Replace with...
10814 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
10815 the stmt_vec_infos. Leave handling of vec_info information to its
10816 destructor. Remove explicit vector releases.
10817 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
10818 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
10819 * tree-vect-slp.c (new_bb_vec_info): Replace with...
10820 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
10821 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
10822 (destroy_bb_vec_info): Replace with...
10823 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
10824 information to its destructor.
10825 (vect_slp_analyze_bb_1): Use new and delete instead of
10826 new_bb_vec_info and destroy_bb_vec_info.
10827 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
10830 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10832 * tree-data-ref.h (subscript): Add access_fn field.
10833 (data_dependence_relation): Add could_be_independent_p.
10834 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
10835 (same_access_functions): Move to tree-data-ref.c.
10836 * tree-data-ref.c (ref_contains_union_access_p): New function.
10837 (access_fn_component_p): Likewise.
10838 (access_fn_components_comparable_p): Likewise.
10839 (dr_analyze_indices): Add a reference to access_fn_component_p.
10840 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
10842 (constant_access_functions): Likewise.
10843 (add_other_self_distances): Likewise.
10844 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
10845 (initialize_data_dependence_relation): Use XCNEW and remove
10846 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
10847 of access functions that have the same type. Allow the
10848 subsequence to end with different bases in some circumstances.
10849 Record the chosen access functions in SUB_ACCESS_FN.
10850 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
10851 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
10852 (subscript_dependence_tester_1): Likewise dra and drb.
10853 (build_classic_dist_vector): Update calls accordingly.
10854 (subscript_dependence_tester): Likewise.
10855 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
10856 DDR_COULD_BE_INDEPENDENT_P.
10857 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
10858 comp_alias_ddrs instead of may_alias_ddrs.
10859 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
10861 (vect_analyze_data_ref_dependence): Use it if
10862 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
10863 distance vectors if that fails.
10864 (dependence_distance_ge_vf): New function.
10865 (vect_prune_runtime_alias_test_list): Use it. Don't clear
10866 LOOP_VINFO_MAY_ALIAS_DDRS.
10868 2017-08-04 Richard Biener <rguenther@suse.de>
10870 PR middle-end/81705
10871 * fold-const.c (fold_binary_loc): Properly restrict
10872 minus_var0 && minus_var1 case when associating undefined overflow
10875 2017-08-04 Tom de Vries <tom@codesourcery.com>
10877 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
10879 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10881 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10882 Don't start diagnostic messages with a capital letter.
10883 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10885 (rs6000_invalid_builtin): Likewise.
10886 (rs6000_trampoline_init): Likewise.
10888 2017-08-03 Jakub Jelinek <jakub@redhat.com>
10891 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
10892 after setting changeable df flags.
10894 2017-08-03 Richard Biener <rguenther@suse.de>
10896 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
10897 up if the use is in USE - X.
10899 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
10901 * toplev.c (dumpfile.h): New include.
10902 (internal_error_reentered): New static function. Use it...
10903 (internal_error_function): ...here to handle reentered internal_error.
10905 2017-08-03 Richard Biener <rguenther@suse.de>
10907 PR middle-end/81148
10908 * fold-const.c (split_tree): Add minus_var and minus_con
10909 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
10910 here but always use minus_*.
10911 (associate_trees): Assert we never associate with MINUS_EXPR
10912 and NULL first operand. Do not recurse for PLUS_EXPR operands
10913 when associating as MINUS_EXPR either.
10914 (fold_binary_loc): Track minus_var and minus_con.
10916 2017-08-03 Tom de Vries <tom@codesourcery.com>
10919 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
10920 ACCEL_COMPILER, apply finish_options on
10921 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
10923 2017-08-03 Tom de Vries <tom@codesourcery.com>
10926 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
10927 function_entry_patch_area_size > 0.
10929 2017-08-03 Jakub Jelinek <jakub@redhat.com>
10932 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
10933 instead of 10??LU, perform unit multiplication in wide_int,
10934 don't change alloc_object_size_limit if the limit is larger
10937 PR tree-optimization/81655
10938 PR tree-optimization/81588
10939 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
10940 the case when ranges[i].low and high are 1 for unsigned type with
10943 PR middle-end/81052
10944 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
10945 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
10947 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10949 * tree-vrp.h: Add include guard.
10951 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
10954 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
10955 (ud2): New insn pattern.
10956 * config/i386/i386.c (ix86_expand_epilogue):
10957 For naked functions, generate ud2 instead of trap insn.
10959 2017-08-02 Marek Polacek <polacek@redhat.com>
10962 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
10964 2017-08-02 Tom de Vries <tom@codesourcery.com>
10965 Cesar Philippidis <cesar@codesourcery.com>
10967 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
10968 Add missing edge probabilities.
10970 2017-08-02 Tamar Christina <tamar.christina@arm.com>
10972 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
10973 Correct endianness.
10975 2017-08-02 Jakub Jelinek <jakub@redhat.com>
10977 PR middle-end/79499
10978 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
10979 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
10980 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
10981 prologue_seq sequences - if any.
10983 2017-08-02 Richard Biener <rguenther@suse.de>
10985 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
10986 via vectors if supported, integer extracts via punning if supported
10987 or otherwise vector extracts.
10989 2017-08-02 Richard Biener <rguenther@suse.de>
10991 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
10993 (bitmap_insert_into_set): ... this.
10995 2017-08-02 Richard Biener <rguenther@suse.de>
10997 PR tree-optimization/81633
10999 2015-08-17 Alan Hayward <alan.hayward@arm.com>
11001 PR tree-optimization/71752
11002 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
11004 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
11006 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
11007 (machine_function::call_ms2sysv_pad_out): Remove field.
11008 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
11009 (ix86_compute_frame_layout): Likewise.
11011 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
11014 * config/i386/i386.c (ix86_set_func_type): Disallow naked
11015 attribute with interrupt attribute.
11017 2017-08-01 Andrew Pinski <apinski@cavium.com>
11019 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
11020 BIT_INSERT_EXPR's operand 1
11021 to see if the types precision matches.
11023 2017-08-01 Martin Liska <mliska@suse.cz>
11025 PR middle-end/70140
11026 * builtins.c (expand_builtin_memcpy_args): Remove.
11027 (expand_builtin_memcpy): Call newly added function
11028 expand_builtin_memory_copy_args.
11029 (expand_builtin_memcpy_with_bounds): Likewise.
11030 (expand_builtin_mempcpy): Remove last argument.
11031 (expand_builtin_mempcpy_with_bounds): Likewise.
11032 (expand_builtin_memory_copy_args): New function created from
11033 expand_builtin_mempcpy_args with small modifications.
11034 (expand_builtin_mempcpy_args): Remove.
11035 (expand_builtin_stpcpy): Remove unused argument.
11036 (expand_builtin): Likewise.
11037 (expand_builtin_with_bounds): Likewise.
11039 2017-08-01 Martin Liska <mliska@suse.cz>
11042 Make mempcpy more optimal (PR middle-end/70140).
11044 2017-08-01 Jakub Jelinek <jakub@redhat.com>
11047 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
11048 __builtin_vec_cmpne verify both arguments are compatible vectors
11049 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
11050 verify arg1_type is a pointer or array type. For __builtin_vec_st,
11051 move computation of aligned to after checking the argument types.
11055 * config/rs6000/vsx.md (vextract_fp_from_shorth,
11056 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
11059 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
11060 Doug Rupp <rupp@adacore.com>
11061 Olivier Hainque <hainque@adacore.com>
11063 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
11064 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
11066 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
11067 for TARGET_OS_CPP_BUILTIN.
11068 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
11069 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
11071 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
11072 passing required abi options to the assembler for EABI configurations.
11073 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
11074 of .text.hot and .text.unlikely sections for kernel modules when
11075 using ARM style exceptions.
11076 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
11077 options. Add EXTRA_CC1_SPEC.
11078 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
11080 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
11082 (ARM_TARGET2_DWARF_FORMAT): Define.
11083 * config/arm/t-vxworks: Adjust multilib control to removal of the
11084 Diab command line options.
11086 2017-08-01 Martin Liska <mliska@suse.cz>
11088 PR gcov-profile/81561
11089 * gcov.c (unblock): Make unblocking safe as we need to preserve
11090 index correspondence of blocks and block_lists.
11092 2017-08-01 Richard Biener <rguenther@suse.de>
11094 PR tree-optimization/81181
11095 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
11096 (compute_antic): ... end of iteration here.
11098 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
11100 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
11101 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
11102 (ftree-slp-vectorize): Likewise.
11103 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
11104 can no longer be set independent of flag_tree_loop_vectorize.
11105 * omp-general.c (emp_max_vf): Likewise.
11106 * opts.c (enable_fdo_optimizations): Remove references to
11107 flag_tree_vectorize, these are now implicit.
11108 (common_handle_option): Remove handling for OPT_ftree_vectorize,
11109 and leave it for the options machinery.
11111 2017-08-01 Martin Liska <mliska@suse.cz>
11113 PR middle-end/70140
11114 * builtins.c (expand_builtin_memcpy_args): Remove.
11115 (expand_builtin_memcpy): Call newly added function
11116 expand_builtin_memory_copy_args.
11117 (expand_builtin_memcpy_with_bounds): Likewise.
11118 (expand_builtin_mempcpy): Remove last argument.
11119 (expand_builtin_mempcpy_with_bounds): Likewise.
11120 (expand_builtin_memory_copy_args): New function created from
11121 expand_builtin_mempcpy_args with small modifications.
11122 (expand_builtin_mempcpy_args): Remove.
11123 (expand_builtin_stpcpy): Remove unused argument.
11124 (expand_builtin): Likewise.
11125 (expand_builtin_with_bounds): Likewise.
11127 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
11130 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
11131 print "ds:" only for immediates in generic address space.
11133 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
11136 * config/i386/i386.c (ix86_funciton_naked): New prototype.
11137 (ix86_function_ok_for_sibcall): Return false for naked functions.
11139 2017-08-01 Richard Biener <rguenther@suse.de>
11141 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
11142 (compute_antic): Seed worklist with exit block predecessors.
11143 * cfganal.c (dfs_find_deadend): For a cycle return the source
11144 of the edge closing it.
11146 2017-08-01 Tamar Christina <tamar.christina@arm.com>
11148 * config/aarch64/aarch64.c
11149 (aarch64_can_const_movi_rtx_p): Move 0 check.
11151 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11153 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
11154 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
11156 * match.pd: Ditto in address comparison pattern.
11158 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11160 PR tree-optimization/81627
11161 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
11162 closed ssa form for store-store chain.
11164 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11166 PR tree-optimization/81620
11167 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
11168 for store-store chain.
11170 2017-08-01 Jakub Jelinek <jakub@redhat.com>
11172 PR tree-optimization/81588
11173 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
11174 ranges[i].in_p, invert comparison code ccode. For >/>=,
11175 swap rhs1 and rhs2 and comparison code unconditionally,
11176 for </<= don't do that. Don't swap rhs1/rhs2 again if
11177 ranges[i].in_p, instead invert comparison code ccode if
11178 opcode or oe->rank is BIT_IOR_EXPR.
11181 * optabs.def (vec_extract_optab, vec_init_optab): Change from
11182 a direct optab to conversion optab.
11183 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
11184 with GET_MODE_INNER as last argument instead of optab_handler.
11185 * expmed.c (extract_bit_field_1): Likewise. Use vector from
11186 vector extraction if possible and optab is available.
11187 * expr.c (store_constructor): Use convert_optab_handler instead
11188 of optab_handler. Use vector initialization from smaller
11189 vectors if possible and optab is available.
11190 * tree-vect-stmts.c (vectorizable_load): Likewise.
11191 * doc/md.texi (vec_extract, vec_init): Document that the optabs
11192 now have two modes.
11193 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
11194 of vec_init from half-sized vectors with the same element mode.
11195 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
11196 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
11197 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
11198 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
11199 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
11200 after mode in gen_vec_extract* calls.
11201 (vec_extract<mode>): Renamed to ...
11202 (vec_extract<mode><ssescalarmodelower>): ... this.
11203 (vec_extract<mode><ssehalfvecmodelower>): New expander.
11204 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
11205 element mode after mode in gen_vec_init* calls.
11206 (VEC_INIT_HALF_MODE): New mode iterator.
11207 (vec_init<mode>): Renamed to ...
11208 (vec_init<mode><ssescalarmodelower>): ... this.
11209 (vec_init<mode><ssehalfvecmodelower>): New expander.
11210 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
11211 (vec_extractv2sfsf): ... this.
11212 (vec_initv2sf): Renamed to ...
11213 (vec_initv2sfsf): ... this.
11214 (vec_extractv2si): Renamed to ...
11215 (vec_extractv2sisi): ... this.
11216 (vec_initv2si): Renamed to ...
11217 (vec_initv2sisi): ... this.
11218 (vec_extractv4hi): Renamed to ...
11219 (vec_extractv4hihi): ... this.
11220 (vec_initv4hi): Renamed to ...
11221 (vec_initv4hihi): ... this.
11222 (vec_extractv8qi): Renamed to ...
11223 (vec_extractv8qiqi): ... this.
11224 (vec_initv8qi): Renamed to ...
11225 (vec_initv8qiqi): ... this.
11226 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
11227 (vec_init<mode>): Renamed to ...
11228 (vec_init<mode><VEC_base_l>): ... this.
11229 (vec_extract<mode>): Renamed to ...
11230 (vec_extract<mode><VEC_base_l>): ... this.
11231 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
11232 (vec_initv2sfsf): ... this.
11233 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
11234 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
11235 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
11236 element mode after mode in gen_vec_init* calls.
11237 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
11238 (vec_init<mode><Vel>): ... this.
11239 (vec_extract<mode>): Renamed to ...
11240 (vec_extract<mode><Vel>): ... this.
11241 * config/aarch64/iterators.md (Vel): New mode attribute.
11242 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
11243 Add element mode after mode in gen_vec_extract* calls.
11244 * config/s390/vector.md (non_vec_l): New mode attribute.
11245 (vec_extract<mode>): Renamed to ...
11246 (vec_extract<mode><non_vec_l>): ... this.
11247 (vec_init<mode>): Renamed to ...
11248 (vec_init<mode><non_vec_l>): ... this.
11249 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
11250 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
11252 * config/arm/iterators.md (V_elem_l): New mode attribute.
11253 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
11254 (vec_extract<mode><V_elem_l>): ... this.
11255 (vec_extractv2di): Renamed to ...
11256 (vec_extractv2didi): ... this.
11257 (vec_init<mode>): Renamed to ...
11258 (vec_init<mode><V_elem_l>): ... this.
11259 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
11260 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
11261 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
11262 Add element mode after gen_vec_extract* calls.
11263 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
11264 (vec_init<mode><unitmode>): ... this.
11265 (vec_extract<mode>): Renamed to ...
11266 (vec_extract<mode><unitmode>): ... this.
11267 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
11268 (vec_init<mode><unitmode>): ... this.
11269 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
11270 (vec_initv2sfsf): ... this.
11271 (vec_extractv2sf): Renamed to ...
11272 (vec_extractv2sfsf): ... this.
11273 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
11274 Add element mode after gen_vec_extract* calls.
11275 * config/mips/mips.md (unitmode): New mode iterator.
11276 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
11277 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
11278 * config/spu/spu.md (inner_l): New mode attribute.
11279 (vec_init<mode>): Renamed to ...
11280 (vec_init<mode><inner_l>): ... this.
11281 (vec_extract<mode>): Renamed to ...
11282 (vec_extract<mode><inner_l>): ... this.
11283 * config/sparc/sparc.md (veltmode): New mode iterator.
11284 (vec_init<VMALL:mode>): Renamed to ...
11285 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
11286 * config/ia64/vect.md (vec_initv2si): Renamed to ...
11287 (vec_initv2sisi): ... this.
11288 (vec_initv2sf): Renamed to ...
11289 (vec_initv2sfsf): ... this.
11290 (vec_extractv2sf): Renamed to ...
11291 (vec_extractv2sfsf): ... this.
11292 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
11293 (vec_init<mode>): Renamed to ...
11294 (vec_init<mode><VEC_base_l>): ... this.
11295 (vec_extract<mode>): Renamed to ...
11296 (vec_extract<mode><VEC_base_l>): ... this.
11297 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
11298 (vec_initv2sfsf): ... this.
11299 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
11300 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
11301 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
11302 gen_vec_init* calls.
11304 2017-08-01 Richard Biener <rguenther@suse.de>
11306 PR tree-optimization/81297
11307 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
11308 TREE_OVERFLOW from INTEGER_CSTs.
11310 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
11312 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
11314 2017-07-31 Carl Love <cel@us.ibm.com>
11316 * config/rs6000/rs6000-c: Add support for built-in functions
11317 vector signed char vec_xl_be (signed long long, signed char *);
11318 vector unsigned char vec_xl_be (signed long long, unsigned char *);
11319 vector signed int vec_xl_be (signed long long, signed int *);
11320 vector unsigned int vec_xl_be (signed long long, unsigned int *);
11321 vector signed long long vec_xl_be (signed long long, signed long long *);
11322 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
11323 vector signed short vec_xl_be (signed long long, signed short *);
11324 vector unsigned short vec_xl_be (signed long long, unsigned short *);
11325 vector double vec_xl_be (signed long long, double *);
11326 vector float vec_xl_be (signed long long, float *);
11327 * config/rs6000/altivec.h (vec_xl_be): Add #define.
11328 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
11329 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
11331 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
11332 (altivec_expand_builtin): Add switch statement to call
11333 altivec_expand_xl_be for each builtin.
11334 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
11335 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
11336 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
11337 __builtin_vsx_le_be_v16qi.
11338 * doc/extend.texi: Update the built-in documentation file for the
11339 new built-in functions.
11341 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
11344 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
11346 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
11348 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11350 * config.gcc: Add z14.
11351 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
11352 CPU model numbers for z13s and z14.
11353 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
11355 * config/s390/s390-opts.h (enum processor_type): Rename
11356 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
11357 * config/s390/s390.c (processor_table): Add field for CPU name to
11358 be passed to Binutils.
11359 (s390_asm_output_machine_for_arch): Use the new field in
11360 processor_table for Binutils.
11361 (s390_expand_builtin): Replace arch12 with z14.
11362 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
11363 (s390_get_sched_attrmask): Likewise.
11364 (s390_get_unit_mask): Likewise.
11365 * config/s390/s390.opt: Add z14 to processor_type enum.
11367 2017-07-31 Martin Jambor <mjambor@suse.cz>
11370 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
11371 regardless of optimization level.
11373 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
11374 Martin Liska <mliska@suse.cz>
11376 * predict.def: Remove old comment and adjust probability.
11377 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
11378 PREDICT statements.
11380 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
11383 * config/i386/i386.c (ix86_function_naked): New function.
11384 (ix86_can_use_return_insn_p): Return false for naked functions.
11385 (ix86_expand_prologue): Skip prologue for naked functions.
11386 (ix86_expand_epilogue): Skip epilogue for naked functions
11387 and emit trap instruction.
11388 (ix86_warn_func_return): New function.
11389 (ix86_attribute_table): Add "naked" attribute specification.
11390 (TARGET_WARN_FUNC_RETURN): Define.
11391 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
11393 2017-07-31 Martin Liska <mliska@suse.cz>
11395 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
11396 (dump_gimple_bb_header): Always dump BB info.
11397 (pp_cfg_jump): Do not append info about BB when dumping a jump.
11399 2017-07-31 Martin Liska <mliska@suse.cz>
11402 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
11403 also with current_function_decl non-null equality.
11405 2017-07-31 Jakub Jelinek <jakub@redhat.com>
11408 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
11409 change type to the element type, instead add eltype variable and
11410 use it where we are interested in the element type.
11412 PR tree-optimization/81603
11413 * ipa-polymorphic-call.c
11414 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
11415 offset arithmetic in offset_int, bail out if the resulting bit offset
11416 doesn't fit into shwi.
11418 2017-07-31 Martin Liska <mliska@suse.cz>
11420 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
11421 (gimplify_save_expr): Fix comment.
11423 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
11426 * config/i386/i386.c (ix86_function_arg): Update arguments for
11428 (ix86_compute_frame_layout): Set the initial stack offset to
11429 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
11430 INCOMING_FRAME_SP_OFFSET.
11431 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
11432 stack before exception handler returns.
11433 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
11434 the 'ERROR_CODE' for exception handler.
11436 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
11438 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
11439 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
11440 (ASM_OUTPUT_REG_POP): Ditto.
11441 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
11442 instead of asm_fprintf to output pure string.
11444 2017-07-29 Jakub Jelinek <jakub@redhat.com>
11446 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
11447 to imported_module_or_decl hook.
11448 (debug_nothing_tree_tree_tree_bool): Remove.
11449 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
11450 * debug.c (do_nothing_debug_hooks): Use
11451 debug_nothing_tree_tree_tree_bool_bool instead of
11452 debug_nothing_tree_tree_tree_bool.
11453 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
11454 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
11455 * sdbout.c (sdb_debug_hooks): Likewise.
11456 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
11457 (gen_namespace_die): Add DW_AT_export_symbols attribute if
11459 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
11460 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
11461 attribute, don't add anything.
11463 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11465 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
11466 (fold_build2_stat_loc): Likewise.
11467 (fold_build3_stat_loc): Likewise.
11468 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
11469 (fold_build1_loc): Remove macro.
11470 (fold_build2_loc): Likewise.
11471 (fold_build3_loc): Likewise.
11473 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11475 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
11476 (gimple_build_debug_bind_source_stat): Likewise.
11477 * gimple.h (gimple_build_debug_bind): Remove macro.
11478 (gimple_build_debug_bind_source): Likewise.
11480 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11482 * bitmap.c (bitmap_alloc): Adjust.
11483 (bitmap_gc_alloc): Likewise.
11484 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
11486 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11488 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
11489 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
11490 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
11491 (bitmap_gc_alloc_stat): Likewise.
11492 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
11494 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11496 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
11497 * rtl.h (shallow_copy_rtx): Remove macro.
11499 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11501 * emit-rtl.c (gen_raw_REG): Adjust.
11502 * gengenrtl.c (gendef): Likewise.
11503 * rtl.c (rtx_alloc_stat): Remove _stat from name.
11504 * rtl.h (rtx_alloc): Remove macro.
11506 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11508 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
11509 (build_tree_list_stat): Likewise.
11510 * tree.h (build_tree_list): Remove macro.
11511 (build_tree_list_vec): Likewise.
11513 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11515 * tree.c (make_vector_stat): Remove _stat from name.
11516 (build_vector_stat): Likewise.
11517 * tree.h (make_vector_stat): Remove macro.
11518 (build_vector_stat): Likewise.
11520 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11522 * tree.h (build_var_debug_value): Remove prototype.
11524 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11526 * tree.c (tree_cons_stat): Remove _stat from name.
11527 * tree.h (tree_cons): Remove macro.
11529 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11531 * tree.c (build_vl_exp_stat): Remove _stat from name.
11532 * tree.h (build_vl_exp): Remove macro.
11534 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11536 * tree.c (build_decl_stat): Remove _stat from name.
11537 * tree.h (build_decl): Remove macro.
11539 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11541 * gimple.c (gimple_build_with_ops_stat): Adjust.
11542 (gimple_alloc_stat): Remove _stat from name.
11543 * gimple.h (gimple_alloc): Remove macro.
11545 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11547 * tree.c (make_tree_vec_stat): Remove _stat from name.
11548 (grow_tree_vec_stat): Likewise.
11549 * tree.h (make_tree_vec_stat): Adjust prototype.
11550 (grow_tree_vec_stat): Likewise.
11551 (make_tree_vec): Remove macro.
11552 (grow_tree_vec): Likewise.
11554 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11556 * fold-const.c (fold_build1_stat_loc): Adjust.
11557 (fold_build2_stat_loc): Likewise.
11558 (fold_build3_stat_loc): Likewise.
11559 * tree.c (build0_stat): Remove _stat from name.
11560 (build1_stat): Likewise.
11561 (build2_stat): Likewise.
11562 (build3_stat): Likewise.
11563 (build4_stat): Likewise.
11564 (build5_stat): Likewise.
11565 * tree.h (build1_loc): Remove macro, and rename _stat function
11567 (build2_loc): Likewise.
11568 (build3_loc): Likewise.
11569 (build4_loc): Likewise.
11570 (build5_loc): Likewise.
11572 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11574 * tree.c (make_int_cst_stat): Remove _stat from name.
11575 * tree.h (make_int_cst_stat): Adjust prototype.
11576 (make_int_cst): Remove macro.
11578 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11580 * tree.c (make_tre_binfo_stat): Remove _stat from name.
11581 * tree.h (make_tree_binfo_stat): Adjust prototype.
11582 (make_tree_binfo): Remove.
11584 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11586 * tree.c (copy_node_stat): Rename to copy_node.
11587 (build_distinct_type_copy): Adjust.
11588 * tree.h (copy_node_stat): Adjust prototype.
11589 (copy_node): Remove macro.
11591 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11593 * tree.c (make_node_stat): rename to make_node.
11594 (build_tree_list_stat): Adjust.
11595 (build0_stat): Likewise.
11596 (build2_stat): Likewise.
11597 (build3_stat): Likewise.
11598 (build4_stat): Likewise.
11599 (build5_stat): Likewise.
11600 (build_decl_stat): Likewise.
11601 * tree.h (make_node_stat): Adjust prototype.
11602 (make_node): remove macro.
11604 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
11606 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
11607 (PPC_FEATURE2_SCV): Likewise.
11608 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
11610 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11612 * config/aarch64/aarch64.c
11613 (aarch64_internal_mov_immediate): Add new special pattern.
11614 * config/aarch64/aarch64.md (*movdi_aarch64):
11615 Add reg/32bit const mov case.
11617 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11618 Richard Sandiford <richard.sandiford@linaro.org>
11620 * config/aarch64/aarch64.md (mov<mode>): Generalize.
11621 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
11622 Add integer and movi cases.
11623 (movi-split-hf-df-sf split, fp16): New.
11624 (enabled): Added TARGET_FP_F16INST.
11625 * config/aarch64/iterators.md (GPF_HF): New.
11626 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
11628 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11630 * config/aarch64/aarch64.c
11631 (aarch64_simd_container_mode): Add prototype.
11632 (aarch64_expand_mov_immediate): Add HI support.
11633 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
11634 (aarch64_can_const_movi_rtx_p): New.
11635 (aarch64_preferred_reload_class):
11636 Remove restrictions of using FP registers for certain SIMD operations.
11637 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
11638 (aarch64_valid_floating_const): Add integer move validation.
11639 (aarch64_simd_imm_scalar_p): Remove.
11640 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
11641 (aarch64_legitimate_constant_p): Expand list of supported cases.
11642 * config/aarch64/aarch64-protos.h
11643 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
11644 (aarch64_reinterpret_float_as_int): New.
11645 (aarch64_simd_imm_scalar_p): Remove.
11646 * config/aarch64/constraints.md (Uvi): New.
11647 (Dd): Split into Ds and new Dd.
11648 * config/aarch64/aarch64.md (*movsi_aarch64):
11650 (*movdi_aarch64): Add SIMD mov case.
11652 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11654 * tree-predcom.c: (struct chain): Handle store-store chain in which
11655 stores for elimination only store loop invariant values.
11656 (execute_pred_commoning_chain): Ditto.
11657 (prepare_initializers_chain_store_elim): Ditto.
11658 (prepare_finalizers): Ditto.
11659 (is_inv_store_elimination_chain): New function.
11660 (initialize_root_vars_store_elim_1): New function.
11662 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11664 * tree-predcom.c: Revise general description of the pass.
11665 (enum chain_type): New enum type for store elimination.
11666 (struct chain): New field supporting store elimination.
11667 (struct component): Ditto.
11668 (dump_chain): Dump store-stores chain.
11669 (release_chain): Release resources.
11670 (split_data_refs_to_components): Compute and create component
11671 contains only stores for elimination.
11672 (get_chain_last_ref_at): New function.
11673 (make_invariant_chain): Initialization.
11674 (make_rooted_chain): Specify chain type in parameter and record it.
11675 (add_looparound_copies): Skip for store-stores chain.
11676 (determine_roots_comp): Compute type of chain and pass it to
11678 (initialize_root_vars_store_elim_2): New function.
11679 (finalize_eliminated_stores): New function.
11680 (remove_stmt): Handle store for elimination.
11681 (execute_pred_commoning_chain): Execute predictive commoning on
11682 store-store chains.
11683 (determine_unroll_factor): Skip unroll for store-stores chain.
11684 (prepare_initializers_chain_store_elim): New function.
11685 (prepare_initializers_chain): Hanlde store-store chain.
11686 (prepare_finalizers_chain, prepare_finalizers): New function.
11687 (tree_predictive_commoning_loop): Return integer value indicating
11688 if loop is unrolled or lcssa form is corrupted.
11689 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
11691 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11693 * tree-predcom.c (initialize_root): Delete.
11694 (execute_pred_commoning_chain): Initialize root vars and replace
11695 reference of non-combined chain directly, rather than call above
11698 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11700 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
11701 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
11703 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11705 * tree-predcom.c (struct chain): New field init_seq.
11706 (release_chain): Release init_seq.
11707 (prepare_initializers_chain): Record intialization stmts in above
11709 (insert_init_seqs): New function.
11710 (tree_predictive_commoning_loop): Call insert_init_seqs.
11712 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11714 * tree-predcom.c (determine_roots_comp): Skip trivial components.
11716 2017-07-28 Richard Biener <rguenther@suse.de>
11718 * match.pd: Remove superfluous :c.
11719 * genmatch.c (simplify::id): Add member.
11720 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
11722 (current_id): New global.
11723 (dt_node::parent): Move from ...
11724 (dt_operand::parent): ... here. Add for_id member.
11725 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
11726 (decision_tree::find_node): Relax order requirement when
11727 merging DT_TRUE nodes to ones inbetween the current simplify
11728 and the one we try to merge with. Add diagnostic whenever
11729 we need to enforce pattern order by not merging.
11730 (decision_tree::insert): Set current_id.
11731 (decision_tree::print_node): Dump parent node and for_id.
11732 (parser::last_id): Add member.
11733 (parser::push_simplify): Assign unique id.
11734 (parser::parser): Initialize last_id.
11736 2017-07-28 Martin Liska <mliska@suse.cz>
11739 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
11740 gimple_build_debug_bind.
11742 2017-07-28 Richard Biener <rguenther@suse.de>
11744 PR tree-optimization/81502
11745 * match.pd: Add pattern combining BIT_INSERT_EXPR with
11747 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
11749 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
11750 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
11751 for BIT_FIELD_REF args.
11752 * fold-const.c (make_bit_field_ref): Likewise.
11753 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
11755 2017-07-28 Jakub Jelinek <jakub@redhat.com>
11758 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
11759 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
11760 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
11761 Or it into SANITIZER_UNDEFINED.
11762 * ubsan.c: Include gimple-fold.h and varasm.h.
11763 (ubsan_expand_ptr_ifn): New function.
11764 (instrument_pointer_overflow): New function.
11765 (maybe_instrument_pointer_overflow): New function.
11766 (instrument_object_size): Formatting fix.
11767 (pass_ubsan::execute): Call instrument_pointer_overflow
11768 and maybe_instrument_pointer_overflow.
11769 * internal-fn.c (expand_UBSAN_PTR): New function.
11770 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
11771 * sanitizer.def (__ubsan_handle_pointer_overflow,
11772 __ubsan_handle_pointer_overflow_abort): New builtins.
11773 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
11774 * internal-fn.def (UBSAN_PTR): New internal function.
11775 * opts.c (sanitizer_opts): Add pointer-overflow.
11776 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
11777 * fold-const.c (build_range_check): Compute pointer range check in
11778 integral type if pointer arithmetics would be needed. Formatting
11781 2017-07-28 Martin Liska <mliska@suse.cz>
11784 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
11785 parameters that are of a variable-length.
11787 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
11789 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
11791 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
11792 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
11793 (CRT_CALL_STATIC_FUNCTION): Likewise.
11794 (ASM_DEFAULT_SPEC): New define.
11795 (ASM_SPEC32): Likewise.
11796 (ASM_SPEC64): Likewise.
11797 (ASM_SPEC_COMMON): Likewise.
11798 (ASM_SPEC): Likewise.
11799 (INVALID_64BIT): Likewise.
11800 (LINK_OS_DEFAULT_SPEC): Likewise.
11801 (LINK_OS_SPEC32): Likewise.
11802 (LINK_OS_SPEC64): Likewise.
11803 (POWERPC_LINUX): Likewise.
11804 (PTRDIFF_TYPE): Likewise.
11805 (RESTORE_FP_PREFIX): Likewise.
11806 (RESTORE_FP_SUFFIX): Likewise.
11807 (SAVE_FP_PREFIX): Likewise.
11808 (SAVE_FP_SUFFIX): Likewise.
11809 (SIZE_TYPE): Likewise.
11810 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
11811 (TARGET_64BIT): Likewise.
11812 (TARGET_64BIT): Likewise.
11813 (TARGET_AIX): Likewise.
11814 (WCHAR_TYPE_SIZE): Likewise.
11815 (WCHAR_TYPE): Undefine.
11816 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
11817 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
11818 (CPP_OS_RTEMS_SPEC): Delete.
11819 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
11820 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
11822 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
11824 2017-07-28 Jakub Jelinek <jakub@redhat.com>
11826 PR tree-optimization/81578
11827 * tree-parloops.c (build_new_reduction): Bail out if
11828 reduction_code isn't one of the standard OpenMP reductions.
11829 Move the details printing after that decision.
11831 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11833 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
11834 related to reload_in_progress.
11835 (splat_input_operand): Likewise.
11836 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
11838 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
11840 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
11841 (TARGET_INSTANTIATE_DECLS): Likewise.
11842 (legitimate_indexed_address_p): Delete reload_in_progress code.
11843 (rs6000_debug_legitimate_address_p): Likewise.
11844 (rs6000_eliminate_indexed_memrefs): Likewise.
11845 (rs6000_emit_le_vsx_store): Likewise.
11846 (rs6000_emit_move_si_sf_subreg): Likewise.
11847 (rs6000_emit_move): Likewise.
11848 (register_to_reg_type): Likewise.
11849 (rs6000_pre_atomic_barrier): Likewise.
11850 (rs6000_machopic_legitimize_pic_address): Likewise.
11851 (rs6000_allocate_stack_temp): Likewise.
11852 (rs6000_address_for_fpconvert): Likewise.
11853 (rs6000_address_for_altivec): Likewise.
11854 (rs6000_secondary_memory_needed_rtx): Delete function.
11855 (rs6000_check_sdmode): Likewise.
11856 (rs6000_alloc_sdmode_stack_slot): Likewise.
11857 (rs6000_instantiate_decls): Likewise.
11858 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
11859 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
11860 Delete reload_in_progress.
11861 (*vec_reload_and_plus_<mptrsize>): Likewise.
11862 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
11863 (vsx_div_v2di): Likewise.
11864 (vsx_udiv_v2di): Likewise.
11866 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11868 * config/rs6000/rs6000.opt (mlra): Replace with stub.
11869 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
11870 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
11871 (rs6000_debug_reg_global): Delete print of LRA status.
11872 (rs6000_option_override_internal): Delete dead LRA related code.
11873 (rs6000_lra_p): Delete function.
11874 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
11876 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
11878 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
11879 * config/riscv/rtems.h: New file.
11881 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11882 Sudakshina Das <sudi.das@arm.com>
11884 * config/aarch64/aarch64.md
11885 (define_split for and<mode>3nr_compare): Move
11886 non aarch64_logical_operand to a register.
11887 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
11888 register immediate operand to a register.
11889 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
11891 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11893 PR middle-end/81564
11894 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
11896 2017-07-27 Richard Biener <rguenther@suse.de>
11898 PR tree-optimization/81573
11899 PR tree-optimization/81494
11900 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
11901 multi defuse cycle case.
11903 2017-07-27 Richard Biener <rguenther@suse.de>
11905 PR tree-optimization/81571
11906 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
11909 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
11911 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
11912 earlier and only if MASK_FPU is set. Adjust formatting.
11914 2017-07-27 Martin Liska <mliska@suse.cz>
11916 * opt-functions.awk: Add validation of value of Init.
11917 * optc-gen.awk: Pass new argument.
11919 2017-07-27 Martin Liska <mliska@suse.cz>
11921 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
11922 Fix wrong condition.
11924 2017-07-27 Martin Liska <mliska@suse.cz>
11926 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
11927 BBs and edges seen by autoFDO.
11929 2017-07-27 Richard Biener <rguenther@suse.de>
11931 PR tree-optimization/81502
11932 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
11933 with incompatible but same sized type.
11934 (execute_update_addresses_taken): Likewise.
11936 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
11938 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
11939 flag_tree_loop_vectorize rather than flag_tree_vectorize.
11941 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11944 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
11945 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
11946 Change s_operand to memory_operand.
11948 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
11950 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
11951 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
11952 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
11953 Emit instructions rather than returning an expression. Handle TFmode
11954 and KFmode by casting to TImode.
11955 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
11956 (rs6000_emit_le_vsx_store): Likewise.
11957 * config/rs6000/vsx.md (VSX_TI): New iterator.
11958 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
11959 (*vsx_le_undo_permute_<mode>): Likewise.
11960 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
11961 emit the split sequence.
11962 (*vsx_le_perm_store_<mode>): Likewise.
11964 2017-07-27 Jakub Jelinek <jakub@redhat.com>
11966 PR tree-optimization/81555
11967 PR tree-optimization/81556
11968 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
11969 if true, force CHANGED for the recursive invocation.
11970 (reassociate_bb): Remember original length of ops array, pass
11971 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
11973 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
11974 attributes for noipa attribute. For naked attribute use
11975 lookup_attribute first before lookup_attribute_spec.
11976 * final.c (rest_of_handle_final): Disable IPA RA for functions with
11978 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
11979 for functions with noipa attribute.
11980 (cgraph_externally_visible_p): Return true for functions with noipa
11982 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
11983 for functions with noipa attribute.
11984 * doc/extend.texi: Document noipa function attribute.
11985 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
11986 also for functions with noipa attribute.
11987 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
11989 2017-07-26 Andrew Pinski <apinski@cavium.com>
11991 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
11992 vec_unalign_load_cost and vec_unalign_store_cost.
11994 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11996 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
11997 -mvsx-small-integer option.
11998 (ISA_3_0_MASKS_IEEE): Likewise.
11999 (OTHER_VSX_VECTOR_MASKS): Likewise.
12000 (POWERPC_MASKS): Likewise.
12001 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
12002 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
12003 code, only testing for DImode being allowed in non-VSX floating
12005 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
12006 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
12008 (rs6000_option_override_internal): Delete -mvsx-small-integer.
12009 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
12010 TARGET_P8_VECTOR test.
12011 (rs6000_secondary_reload_simple_move): Likewise.
12012 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
12013 since TARGET_P9_VECTOR was already tested.
12014 (rs6000_opt_masks): Remove -mvsx-small-integer.
12015 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
12016 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
12018 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
12019 test for TARGET_VEXTRACTUB was used, and that uses
12021 (p9 extract splitter): Likewise.
12022 (vsx_extract_<mode>_di_p9): Likewise.
12023 (vsx_extract_<mode>_store_p9): Likewise.
12024 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
12025 for TARGET_P9_VECTOR was used. Delete code that is now dead with
12026 the elimination of TARGET_VSX_SMALL_INTEGER.
12027 (vsx_extract_<mode>_p8): Likewise.
12028 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
12029 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
12030 (vsx_set_<mode>_p9): Likewise.
12031 (vsx_set_v4sf_p9): Likewise.
12032 (vsx_set_v4sf_p9_zero): Likewise.
12033 (vsx_insert_extract_v4sf_p9): Likewise.
12034 (vsx_insert_extract_v4sf_p9_2): Likewise.
12035 * config/rs6000/rs6000.md (sign extend splitter): Change
12036 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
12037 (floatsi<mode>2_lfiwax_mem): Likewise.
12038 (floatunssi<mode>2_lfiwzx_mem): Likewise.
12039 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
12040 since a test for TARGET_P9_VECTOR was used.
12041 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12042 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
12043 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12044 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
12045 TARGET_P8_VECTOR test.
12046 (fix_trunc<mode>si2_stfiwx): Likewise.
12047 (fix_trunc<mode>si2_internal): Likewise.
12048 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
12049 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
12051 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
12052 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
12053 TARGET_P8_VECTOR test.
12054 (fixuns_trunc<mode>si2_stfiwx): Likewise.
12055 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
12056 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
12058 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
12059 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
12060 since a test for TARGET_P9_VECTOR was used.
12061 (splitter for loading small constants): Likewise.
12063 2017-07-26 Andrew Pinski <apinski@cavium.com>
12065 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
12068 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
12071 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
12072 (fp_valid_at): Likewise.
12074 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
12076 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
12077 (qdf24xx_addrcost_table): Likewise.
12078 (cortexa57_tunings): Update to use generic_branch_cost.
12079 (cortexa72_tunings): Likewise.
12080 (cortexa73_tunings): Likewise.
12081 (qdf24xx_tunings): Likewise.
12083 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
12085 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
12086 (thunderx2t99_branch_cost): Likewise.
12087 (cortexa35_tunings): Update to use generic_branch_cost.
12088 (cortexa53_tunings): Likewise.
12089 (cortexa57_tunings): Likewise.
12090 (cortexa72_tunings): Likewise.
12091 (cortexa73_tunings): Likewise.
12092 (thunderx2t99_tunings): Likewise.
12094 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
12096 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
12097 (sparc_option_override): Honour MASK_FSMULD.
12098 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
12099 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
12100 * config/sparc/sparc.opt (mfsmuld): New option.
12101 * doc/invoke.texi (mfsmuld): Document option.
12103 2017-07-26 Marek Polacek <polacek@redhat.com>
12105 PR middle-end/70992
12106 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
12108 2017-07-26 Richard Biener <rguenther@suse.de>
12110 * gimple-match-head.c (do_valueize): Return OP if valueize
12112 (get_def): New helper to get at the def stmt of a SSA name
12113 if valueize allows.
12114 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
12115 do_valueize to get at the def stmt.
12116 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
12118 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
12120 PR middle-end/46932
12121 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
12123 2017-07-26 Martin Liska <mliska@suse.cz>
12126 * function.c (expand_function_start): Make expansion of
12127 nonlocal_goto_save_area after parm_birth_insn.
12129 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
12131 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
12132 from all CPU target flags enable members.
12134 2017-07-26 Richard Biener <rguenther@suse.de>
12136 * genmatch.c (dt_simplify::gen): Make iterator vars const.
12137 (decision_tree::gen): Make 'type' const.
12138 (write_predicate): Likewise.
12140 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
12142 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
12143 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
12144 (rs6000_option_override_internal): Likewise.
12145 (rs6000_expand_vector_set): Likewise.
12146 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
12147 (TARGET_UPPER_REGS_SF): Likewise.
12148 (TARGET_UPPER_REGS_DI): Likewise.
12149 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
12150 (TARGET_DIRECT_MOVE_64BIT): Likewise.
12151 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
12152 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12153 (Splitters for DI constants in Altivec registers): Likewise.
12154 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
12155 (vsx_set_v4sf_p9): Likewise.
12156 (vsx_set_v4sf_p9_zero): Likewise.
12157 (vsx_insert_extract_v4sf_p9): Likewise.
12158 (vsx_insert_extract_v4sf_p9_2): Likewise.
12160 2017-07-25 Carl Love <cel@us.ibm.com>
12162 * doc/extend.texi: Update the built-in documentation file for the
12163 existing built-in functions
12164 vector signed char vec_cnttz (vector signed char);
12165 vector unsigned char vec_cnttz (vector unsigned char);
12166 vector signed short vec_cnttz (vector signed short);
12167 vector unsigned short vec_cnttz (vector unsigned short);
12168 vector signed int vec_cnttz (vector signed int);
12169 vector unsigned int vec_cnttz (vector unsigned int);
12170 vector signed long long vec_cnttz (vector signed long long);
12171 vector unsigned long long vec_cnttz (vector unsigned long long);
12173 2017-07-25 Andrew Pinski <apinski@cavium.com>
12175 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
12176 accesses where the use is for the first operand of a BIT_INSERT.
12178 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
12181 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
12182 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
12184 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
12186 * config/i386/gstabs.h: Delete.
12187 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
12189 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
12191 * config/i386/i386.c (ix86_decompose_address): Do not check for
12192 register RTX when looking at index_reg or base_reg.
12193 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
12195 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
12197 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
12198 to update EH info here.
12200 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12202 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
12204 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12206 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
12208 2017-07-25 Torsten Duwe <duwe@suse.de>
12210 * common.opt: Introduce -fpatchable-function-entry
12211 command line option, and its variables function_entry_patch_area_size
12212 and function_entry_patch_area_start.
12213 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
12214 including a two-value parser.
12215 * target.def (print_patchable_function_entry): New target hook.
12216 * targhooks.h (default_print_patchable_function_entry): New function.
12217 * targhooks.c (default_print_patchable_function_entry): Likewise.
12218 * toplev.c (process_options): Switch off IPA-RA if
12219 patchable function entries are being generated.
12220 * varasm.c (assemble_start_function): Look at the
12221 patchable-function-entry command line switch and current
12222 function attributes and maybe generate NOP instructions by
12223 calling the print_patchable_function_entry hook.
12224 * doc/extend.texi: Document patchable_function_entry attribute.
12225 * doc/invoke.texi: Document -fpatchable_function_entry
12226 command line option.
12227 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
12229 * doc/tm.texi: Re-generate.
12231 2017-07-25 Jakub Jelinek <jakub@redhat.com>
12234 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
12235 TARGET_AVX512DQ rather than TARGET_AVX512BW.
12237 2017-07-25 Tamar Christina <tamar.christina@arm.com>
12239 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
12241 2017-07-25 Richard Biener <rguenther@suse.de>
12243 PR tree-optimization/81455
12244 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
12245 not walk in cycles when looking for guards.
12247 2017-07-25 Richard Biener <rguenther@suse.de>
12249 PR tree-optimization/81529
12250 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
12251 when optimizing backedge uses.
12253 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
12255 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
12257 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
12258 to dl_section_ref. On AIX, append an expression to subtract
12259 the size of the section length to dl_section_ref.
12261 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
12263 * configure.ac: If any of the config.* scripts fail, exit 1.
12264 * configure: Regenerate.
12266 2017-07-25 Richard Biener <rguenther@suse.de>
12268 PR middle-end/81546
12269 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
12270 of immediate uses, be more verbose on errors.
12272 2017-07-25 Richard Biener <rguenther@suse.de>
12274 PR tree-optimization/81510
12275 * tree-vect-loop.c (vect_is_simple_reduction): When the
12276 reduction stmt is not inside the loop bail out.
12278 2017-07-25 Richard Biener <rguenther@suse.de>
12280 PR tree-optimization/81303
12281 * tree-vect-loop-manip.c (vect_loop_versioning): Build
12282 profitability check against LOOP_VINFO_NITERSM1.
12284 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12286 * domwalk.c (cmp_bb_postorder): Simplify.
12287 (sort_bbs_postorder): New function. Use it...
12288 (dom_walker::walk): ...here to optimize common cases.
12290 2017-07-25 Martin Liska <mliska@suse.cz>
12293 * ipa-visibility.c (function_and_variable_visibility): Make the
12294 redirection just on target that supports aliasing.
12295 Fix GNU coding style.
12297 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
12300 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
12302 * config/arm/rtems.h: Likewise.
12303 * config/bfin/rtems.h: Likewise.
12304 * config/i386/rtemself.h: Likewise.
12305 * config/lm32/rtems.h: Likewise.
12306 * config/m32c/rtems.h: Likewise.
12307 * config/m68k/rtemself.h: Likewise.
12308 * config/microblaze/rtems.h: Likewise.
12309 * config/mips/rtems.h: Likewise.
12310 * config/moxie/rtems.h: Likewise.
12311 * config/nios2/rtems.h: Likewise.
12312 * config/powerpcspe/rtems.h: Likewise.
12313 * config/rs6000/rtems.h: Likewise.
12314 * config/rtems.h: Likewise.
12315 * config/sh/rtems.h: Likewise.
12316 * config/sh/rtemself.h: Likewise.
12317 * config/sparc/rtemself.h: Likewise.
12319 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
12322 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
12323 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
12324 * tree-ssa-structalias.c (alias_get_name): Same.
12326 2017-07-25 Bin Cheng <bin.cheng@arm.com>
12329 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
12330 instructions if no du chain is found.
12332 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
12334 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
12336 2017-07-25 Richard Biener <rguenther@suse.de>
12338 PR middle-end/81505
12339 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
12342 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
12344 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
12345 upper-regs options.
12346 (ISA_2_7_MASKS_SERVER): Likewise.
12347 (ISA_3_0_MASKS_IEEE): Likewise.
12348 (OTHER_P8_VECTOR_MASKS): Likewise.
12349 (OTHER_VSX_VECTOR_MASKS): Likewise.
12350 (POWERPC_MASKS): Likewise.
12351 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
12352 duplicate list of options.
12353 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
12354 explicit -mupper-regs options.
12355 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
12356 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
12357 alias for -mupper-regs-df.
12358 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12359 (rs6000_init_hard_regno_mode_ok): Likewise.
12360 (rs6000_option_override_internal): Likewise.
12361 (rs6000_opt_masks): Likewise.
12362 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
12363 options in terms of whether -mvsx or -mpower8-vector was used.
12364 (TARGET_UPPER_REGS_DI): Likewise.
12365 (TARGET_UPPER_REGS_SF): Likewise.
12366 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
12367 -mupper-regs-* options.
12369 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
12371 * passes.c (emergency_dump_function): Print some empty lines and a
12372 header before the RTL dump.
12374 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
12376 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
12378 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
12381 * config/aarch64/aarch64.c (aarch64_classify_symbol):
12382 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
12384 2017-07-24 Carl Love <cel@us.ibm.com>
12386 * config/rs6000/rs6000-c.c: Add support for built-in functions
12387 vector float vec_extract_fp32_from_shorth (vector unsigned short);
12388 vector float vec_extract_fp32_from_shortl (vector unsigned short);
12389 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
12390 vec_extract_fp_from_shortl): Add defines for the two builtins.
12391 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
12392 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
12394 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
12395 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
12396 * doc/extend.texi: Update the built-in documentation file for the
12397 new built-in function.
12399 2017-07-24 Jakub Jelinek <jakub@redhat.com>
12402 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
12404 * doc/generic.texi: Likewise.
12405 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
12406 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
12408 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
12410 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
12411 (aarch64_mls_elt_merge<mode>): Likewise.
12413 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
12415 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
12416 having __cxa_atexit.
12418 2017-07-23 Michael Collison <michael.collison@arm.com>
12420 * config/arm/arm.c (arm_option_override): Deprecate
12421 use of -mstructure-size-boundary.
12422 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
12423 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
12425 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12428 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
12429 Reduce cost estimate for direct moves.
12431 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
12434 * config/i386/i386.c (ix86_option_override_internal): Disable
12435 BMI, BMI2 and TBM instructions for -m16.
12437 2017-07-21 Carl Love <cel@us.ibm.com>
12439 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12440 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12441 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12442 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12443 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12444 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12445 VMULOSW): New enum "unspec" values.
12446 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12447 altivec_vmulosw): New patterns.
12448 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12449 VMULOSW): Add definitions.
12451 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
12453 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
12454 (qdf24xx): Likewise.
12455 * config/aarch64/aarch64-options-extensions.def (rdma); New.
12456 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
12457 (AARCH64_FL_V8_1): Renumber.
12458 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
12459 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
12460 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
12461 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
12462 rdma to feature modifiers list.
12464 2017-07-21 Yury Gribov <tetra2005@gmail.com>
12466 PR middle-end/56727
12467 * ipa-visibility (function_and_variable_visibility): Convert
12468 recursive PLT call to direct call if appropriate.
12470 2017-07-21 Andrew Pinski <apinski@cavium.com>
12472 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
12473 operand 1 to see if the types precision matches.
12474 * fold-const.c (operand_equal_p): Likewise.
12476 2017-07-21 Richard Biener <rguenther@suse.de>
12478 PR tree-optimization/81303
12479 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
12480 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
12481 (vect_peeling_hash_get_lowest_cost): Adjust.
12482 (vect_enhance_data_refs_alignment): Likewise. Use
12483 vect_get_peeling_costs_all_drs to compute the penalty for no
12484 peeling to match up costs.
12486 2017-07-21 Richard Biener <rguenther@suse.de>
12488 PR tree-optimization/81500
12489 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
12490 we didn't identify a reduction path.
12492 2017-07-21 Tom de Vries <tom@codesourcery.com>
12493 Cesar Philippidis <cesar@codesourcery.com>
12495 PR gcov-profile/81442
12496 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
12499 2017-07-21 Tom de Vries <tom@codesourcery.com>
12502 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
12504 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
12505 nvptx_override_options_after_change.
12507 2017-07-21 Ulrich Drepper <drepper@redhat.com>
12509 * dwarf2out.c (output_file_names): Avoid double testing for
12510 dwarf_version >= 5.
12512 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
12514 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
12516 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
12518 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
12520 (try_crossjump_to_edge): Do not punt on partitioned functions.
12522 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
12524 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
12525 Put all BBs reachable only via paths crossing cold region to cold
12527 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
12529 2016-07-21 Richard Biener <rguenther@suse.de>
12531 PR tree-optimization/81303
12532 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
12533 into account prologue and epilogue iterations when raising
12534 min_profitable_iters to sth at least covering one vector iteration.
12536 2017-07-21 Tamar Christina <tamar.christina@arm.com>
12538 * config/arm/arm.c (arm_test_cpu_arch_dat):
12541 2017-07-20 Nathan Sidwell <nathan@acm.org>
12543 Remove TYPE_METHODS.
12544 * tree.h (TYPE_METHODS): Delete.
12545 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
12546 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
12547 (dbxout_type_methods): Scan TYPE_FIELDS.
12548 (dbxout_type): Don't check TYPE_METHODS here.
12549 * function.c (use_register_for_decl): Always ignore register for
12550 class types when not optimizing.
12551 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
12552 * tree.c (free_lang_data_in_type): Stitch out member functions and
12553 templates from TYPE_FIELDS.
12554 (build_distinct_type_copy, verify_type_variant,
12555 verify_type): Member fns are on TYPE_FIELDS.
12556 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
12557 * tree-pretty-print.c (dump_generic_node): Likewise.
12559 2017-07-20 Jakub Jelinek <jakub@redhat.com>
12562 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
12563 V2TImode and V4TImode.
12564 (ix86_expand_vector_extract): Likewise.
12565 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
12566 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
12567 (ssescalarmode): Handle V4TImode and V2TImode.
12568 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
12569 (*vec_extractv2ti, *vec_extractv4ti): New insns.
12570 (VEXTRACTI128_MODE): New mode iterator.
12571 (splitter for *vec_extractv?ti first element): New.
12572 (VEC_INIT_MODE): New mode iterator.
12573 (vec_init<mode>): Consolidate 3 expanders into one using
12574 VEC_INIT_MODE mode iterator.
12576 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
12578 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
12579 non_spilled_static_chain_regno_p.
12581 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
12583 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
12585 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
12587 * bb-reorder.c (connect_traces): Allow copying of blocks within
12590 2017-07-20 Richard Biener <rguenther@suse.de>
12592 * gimple.h (gimple_phi_result): Add gphi * overload.
12593 (gimple_phi_result_ptr): Likewise.
12594 (gimple_phi_arg): Likewise. Adjust index assert to only
12595 allow actual argument accesses rather than all slots available
12597 (gimple_phi_arg_def): Add gphi * overload.
12598 * tree-phinodes.c (make_phi_node): Initialize only actual
12600 (resize_phi_node): Clear memory not covered by old node,
12601 do not initialize excess argument slots.
12602 (reserve_phi_args_for_new_edge): Initialize new argument slot
12605 2017-07-20 Bin Cheng <bin.cheng@arm.com>
12607 PR tree-optimization/81388
12609 2016-07-21 Bin Cheng <bin.cheng@arm.com>
12611 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
12612 by removing computation of may_be_zero.
12614 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12615 Tom de Vries <tom@codesourcery.com>
12617 PR middle-end/81030
12618 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
12619 when gimple level profile disagrees with what RTL expander did.
12621 2017-07-20 Richard Biener <rguenther@suse.de>
12623 PR tree-optimization/61171
12624 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
12625 (vect_analyze_stmt): Add slp instance parameter.
12626 (vectorizable_reduction): Likewise.
12627 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
12628 (vect_is_simple_reduction): Deal with chains not detected
12629 as SLP reduction chain, specifically not properly associated
12630 chains containing a mix of plus/minus.
12631 (get_reduction_op): Remove.
12632 (get_initial_defs_for_reduction): Simplify, pass in whether
12633 this is a reduction chain, pass in the SLP node for the PHIs.
12634 (vect_create_epilog_for_reduction): Get the SLP instance as
12636 (vectorizable_reduction): Get the SLP instance as arg.
12637 During analysis remember the SLP node with the PHIs in the
12638 instance. Simplify getting at the vectorized reduction PHIs.
12639 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
12640 through SLP instance.
12641 (vect_slp_analyze_operations): Likewise.
12642 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
12643 (vect_transform_stmt): Likewise.
12645 2017-07-20 Tom de Vries <tom@codesourcery.com>
12647 PR tree-optimization/81489
12648 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
12649 read of phi arg location to before loop that modifies phi.
12651 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12653 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
12656 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12658 PR middle-end/81331
12659 * except.c (execute): Fix ordering issue.
12661 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
12663 PR rtl-optimization/81423
12664 * combine.c (make_compound_operation_int): Don't try to optimize
12665 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
12667 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
12669 PR rtl-optimization/81423
12670 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
12671 with a constant that is -1 in the truncated to mode.
12673 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12675 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
12676 (determine_unlikely_bbs): ... here.
12677 * predict.h (propagate_unlikely_bbs_forward): Declare.
12678 * cfgexpand.c (pass_expand::execute): Use it.
12679 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
12681 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
12682 propagate_unlikely_bbs_forward.
12684 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12686 PR middle-end/81331
12687 * except.c (maybe_add_nop_after_section_switch): New function.
12690 2017-07-19 Tom de Vries <tom@codesourcery.com>
12692 * gimple.h (gimple_phi_set_arg): Make assert more strict.
12694 2017-07-19 Tom de Vries <tom@codesourcery.com>
12696 * gimple.h (gimple_phi_arg): Make assert more strict.
12698 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
12700 * config.gcc (powerpc*-*-*): Add mmintrin.h.
12701 * config/rs6000/mmintrin.h: New file.
12702 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
12704 2017-07-19 Jakub Jelinek <jakub@redhat.com>
12706 PR tree-optimization/81346
12707 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
12709 2017-07-19 Tom de Vries <tom@codesourcery.com>
12711 * config/nvptx/nvptx.md (VECIM): Add V2DI.
12713 2017-07-19 Tom de Vries <tom@codesourcery.com>
12715 * config/nvptx/nvptx-modes.def: Add V2DImode.
12716 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
12717 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
12718 (nvptx_output_mov_insn): Handle lack of mov.b128.
12719 (nvptx_print_operand): Handle 'H' and 'L' codes.
12720 (nvptx_vector_mode_supported): Allow V2DImode.
12721 (nvptx_preferred_simd_mode): New function.
12722 (nvptx_data_alignment): New function.
12723 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
12724 nvptx_preferred_simd_mode.
12725 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
12727 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
12729 2017-07-19 Tom de Vries <tom@codesourcery.com>
12731 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
12732 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
12733 (nvptx_vector_mode_supported): New function. Allow V2SImode.
12734 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
12735 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
12736 (mov<VECIM>_insn): New define_insn.
12737 (define_expand "mov<VECIM>): New define_expand.
12739 2017-07-19 Tom de Vries <tom@codesourcery.com>
12741 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
12743 2017-07-19 Jakub Jelinek <jakub@redhat.com>
12745 PR tree-optimization/81346
12746 * fold-const.h (fold_div_compare, range_check_type): Declare.
12747 * fold-const.c (range_check_type): New function.
12748 (build_range_check): Use range_check_type.
12749 (fold_div_compare): No longer static, rewritten into
12750 a match.pd helper function.
12751 (fold_comparison): Don't call fold_div_compare here.
12752 * match.pd (X / C1 op C2): New optimization using fold_div_compare
12753 as helper function.
12755 2017-07-19 Nathan Sidwell <nathan@acm.org>
12757 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
12758 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
12759 * tree.c (find_decls_types_r, verify_type): Use
12760 TYPE_{MIN,MAX}_VALUE_RAW.
12761 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
12762 (hash_tree): Likewise.
12763 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
12765 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
12768 2017-07-18 Tom de Vries <tom@codesourcery.com>
12770 PR middle-end/81464
12771 * omp-expand.c (expand_omp_for_static_chunk): Handle
12772 equal-argument loop exit phi.
12774 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
12777 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
12778 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
12779 operand 2 predicate.
12780 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
12781 operand 2 predicate.
12782 (ror,rol -> rorx splitters): Use const_int_operand as
12783 operand 2 predicate.
12785 2017-06-18 Richard Biener <rguenther@suse.de>
12787 PR tree-optimization/81410
12788 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
12789 the gap in the ! slp_perm SLP case after each group.
12791 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12793 PR middle-end/81463
12794 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
12797 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12799 PR middle-end/81462
12800 * predict.c (set_even_probabilities): Cleanup; do not affect
12801 probabilities that are already known.
12802 (combine_predictions_for_bb): Call even when count is set.
12804 2017-07-18 Nathan Sidwell <nathan@acm.org>
12806 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
12809 2017-07-18 Bin Cheng <bin.cheng@arm.com>
12812 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
12813 optimization for loop niter analysis.
12815 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
12818 * config/avr/avr.c (avr_optimize_casesi): Don't use
12819 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
12821 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
12823 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
12824 body_cost_vec from _vect_peel_extended_info.
12825 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
12826 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
12829 2017-07-18 Bin Cheng <bin.cheng@arm.com>
12831 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
12833 2017-07-18 Richard Biener <rguenther@suse.de>
12835 PR tree-optimization/80620
12836 PR tree-optimization/81403
12837 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
12838 info when re-using a VN table entry.
12840 2017-07-18 Richard Biener <rguenther@suse.de>
12842 PR tree-optimization/81418
12843 * tree-vect-loop.c (vectorizable_reduction): Properly compute
12844 vectype_in. Verify that with lane-reducing reduction operations
12845 we have a single def-use cycle.
12847 2017-07-17 Carl Love <cel@us.ibm.com>
12849 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
12851 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12852 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12853 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12854 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12855 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12856 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12857 VMULOSW): New enum "unspec" values.
12858 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
12859 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
12860 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12861 altivec_vmulosw): New patterns.
12862 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12863 VMULOSW): Add definitions.
12864 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
12866 * config/alpha/alpha.c: Include predict.h.
12868 2017-07-17 Yury Gribov <tetra2005@gmail.com>
12870 * tree-vrp.c (compare_assert_loc): Fix comparison function
12871 to return predictable results.
12873 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12875 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
12877 (subdi3): Likewise.
12878 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
12879 * doc/invoke.texi (mexpand-adddi): Update text.
12881 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12883 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
12884 that also clobbers the CC register. The old expand code is moved
12886 (*arc_clzsi2): ... here.
12887 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
12888 the CC register. The old expand code is moved to ...
12889 (arc_ctzsi2): ... here.
12891 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12893 * config/arc/arc.opt (mindexed-loads): Use initial value
12894 TARGET_INDEXED_LOADS_DEFAULT.
12895 (mauto-modify-reg): Use initial value
12896 TARGET_AUTO_MODIFY_REG_DEFAULT.
12897 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
12898 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
12899 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
12900 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
12902 2017-07-17 Martin Liska <mliska@suse.cz>
12905 * opts.c (finish_options): Do not allow -fgnu-tm
12906 w/ -fsanitize={kernel-,}address. Say sorry.
12908 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12911 * tree-loop-distribution.c (classify_partition): Only assert on
12912 numer of iterations.
12913 (merge_dep_scc_partitions): Delete prameter. Update function call.
12914 (distribute_loop): Remove code handling loop with unknown niters.
12915 (pass_loop_distribution::execute): Skip loop with unknown niters.
12917 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12920 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
12921 function sort_partitions_by_post_order.
12923 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12925 PR tree-optimization/81374
12926 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
12927 the max index of basic blocks, rather than number of basic blocks.
12929 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12931 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
12933 (arc_legitimate_pic_operand_p): Likewise.
12934 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
12936 (arc_needs_pcl_p): Likewise.
12937 (arc_legitimate_pc_offset_p): Likewise.
12938 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
12939 function is also used in constrains.md.
12940 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
12941 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
12942 PLUS. Only return true/false in known cases, otherwise assert.
12943 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
12944 is already called in arc_legitimate_constant_p.
12945 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
12947 (LEGITIMATE_PIC_OPERAND_P): Use
12948 arc_raw_symbolic_reference_mentioned_p function.
12949 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
12954 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12955 Andrew Burgess <andrew.burgess@embecosm.com>
12957 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
12958 (arc_return_address_register): New function.
12959 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
12960 (arc_handle_fndecl_attribute): Add naked attribute.
12961 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
12962 (TARGET_WARN_FUNC_RETURN): Likewise.
12963 (arc_allocate_stack_slots_for_args): New function.
12964 (arc_warn_func_return): Likewise.
12965 (machine_function): Change type fn_type.
12966 (arc_compute_function_type): Consider new naked function type,
12967 change function return type.
12968 (arc_must_save_register): Adapt to handle new
12969 arc_compute_function_type's return type.
12970 (arc_expand_prologue): Likewise.
12971 (arc_expand_epilogue): Likewise.
12972 (arc_return_address_regs): Delete.
12973 (arc_return_address_register): New function.
12974 (arc_epilogue_uses): Use above function.
12975 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
12976 (arc_function_type): Change encoding, add naked type.
12977 (ARC_INTERRUPT_P): Change to handle the new encoding.
12978 (ARC_FAST_INTERRUPT_P): Likewise.
12979 (ARC_NORMAL_P): Define.
12980 (ARC_NAKED_P): Likewise.
12981 (arc_compute_function_type): Delete prototype.
12982 * config/arc/arc.md (in_ret_delay_slot): Use
12983 arc_return_address_register function.
12984 (simple_return): Likewise.
12985 (p_return_i): Likewise.
12987 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12989 PR tree-optimization/81428
12990 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
12991 can't be built for those types.
12993 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
12995 Remove stuff dead since r239246.
12997 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
12998 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
12999 (avr_inform_devices): Remove dead stuff.
13001 2017-07-17 Tamar Christina <tamar.christina@arm.com>
13003 * config/arm/arm_neon.h: Fix softp typo.
13005 2017-07-17 Jakub Jelinek <jakub@redhat.com>
13007 PR tree-optimization/81365
13008 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
13009 aggregate moves onto bb predecessor edges, make sure there are no
13010 loads that could alias the lhs in between the start of bb and the
13013 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
13016 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
13017 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
13018 [LSHIFTRT, outer_code = TRUNCATE]: Same.
13020 2017-07-17 Jakub Jelinek <jakub@redhat.com>
13022 PR tree-optimization/81396
13023 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
13024 (init_symbolic_number): Initialize it to 1.
13025 (perform_symbolic_merge): Add n_ops from both operands into the new
13027 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
13028 without base_addr as useless if they need more than one operation.
13029 (bswap_replace): Handle !bswap case for NULL base_addr.
13031 2017-07-17 Tom de Vries <tom@codesourcery.com>
13034 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
13037 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
13039 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
13040 conditional builtin define __FIX_LEON3FT_B2BST.
13042 2017-07-17 Daniel Cederman <cederman@gaisler.com>
13044 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
13045 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
13048 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
13050 PR rtl-optimization/81424
13051 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
13052 to remove potential trapping from operands if -fnon-call-exceptions.
13054 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13056 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
13057 profile_proability for scalling.
13058 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
13060 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13062 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
13064 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13066 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
13067 fixpoint arithmetics.
13069 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13071 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
13072 fixpoint arithmetics.
13074 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13076 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
13077 fixpoint arithmetics.
13079 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13081 * profile-count.h (profile_probability::from_reg_br_prob_note,
13082 profile_probability::to_reg_br_prob_note): New functions.
13083 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
13084 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
13085 * predict.c (probability_reliable_p): Update.
13086 (edge_probability_reliable_p): Update.
13087 (br_prob_note_reliable_p): Update.
13088 (invert_br_probabilities): Update.
13089 (add_reg_br_prob_note): New function.
13090 (combine_predictions_for_insn): Update.
13091 * asan.c (asan_clear_shadow): Update.
13092 * cfgbuild.c (compute_outgoing_frequencies): Update.
13093 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
13094 (update_br_prob_note): Update.
13095 (rtl_verify_edges): Update.
13096 (purge_dead_edges): Update.
13097 (fixup_reorder_chain): Update.
13098 * emit-rtl.c (try_split): Update.
13099 * ifcvt.c (cond_exec_process_insns): Update.
13100 (cond_exec_process_if_block): Update.
13101 (dead_or_predicable): Update.
13102 * internal-fn.c (expand_addsub_overflow): Update.
13103 (expand_neg_overflow): Update.
13104 (expand_mul_overflow): Update.
13105 * loop-doloop.c (doloop_modify): Update.
13106 * loop-unroll.c (compare_and_jump_seq): Update.
13107 * optabs.c (emit_cmp_and_jump_insn_1): Update.
13108 * predict.h: Update.
13109 * reorg.c (mostly_true_jump): Update.
13111 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
13112 * config/alpha/alpha.c (emit_unlikely_jump): Update.
13113 * config/arc/arc.c: (emit_unlikely_jump): Update.
13114 * config/arm/arm.c: (emit_unlikely_jump): Update.
13115 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
13116 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
13117 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
13118 (ix86_print_operand): Update.
13119 (ix86_split_fp_branch): Update.
13120 (predict_jump): Update.
13121 * config/ia64/ia64.c (ia64_print_operand): Update.
13122 * config/mmix/mmix.c (mmix_print_operand): Update.
13123 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
13124 (rs6000_expand_split_stack_prologue): Update.
13125 * config/rs6000/rs6000.c: Update.
13126 * config/s390/s390.c (s390_expand_vec_strlen): Update.
13127 (s390_expand_vec_movstr): Update.
13128 (s390_expand_cs_tdsi): Update.
13129 (s390_expand_split_stack_prologue): Update.
13130 * config/sh/sh.c (sh_print_operand): Update.
13131 (expand_cbranchsi4): Update.
13132 (expand_cbranchdi4): Update.
13133 * config/sparc/sparc.c (output_v9branch): Update.
13134 * config/spu/spu.c (get_branch_target): Update.
13135 (ea_load_store_inline): Update.
13136 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
13137 * config/tilepro/tilepro.c: Update.
13139 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
13141 * gimplify.c (mostly_copy_tree_r): Revert latest change.
13142 (gimplify_save_expr): Likewise.
13144 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13146 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
13148 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13150 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
13152 * timevar.def (TV_IPA_FNSUMMARY): Define.
13154 2017-07-16 Daniel Cederman <cederman@gaisler.com>
13156 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
13157 to back store errata sensitive sequence from being generated.
13158 (sqrtdf2_fix): Likewise.
13160 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13162 * tree-ssa-threadupdate.c (compute_path_counts,
13163 update_joiner_offpath_counts): Use profile_probability.
13165 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
13168 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13170 * config/arm/arm-c.c (arm_cpu_builtins): Define
13171 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
13173 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
13175 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13176 array entries to represent __ieee128 versions of the
13177 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
13178 scalar_extract_sig, and scalar_insert_exp built-in functions.
13179 (altivec_resolve_overloaded_builtin): Add special case handling
13180 for the __builtin_scalar_insert_exp function, as represented by
13181 the P9V_BUILTIN_VEC_VSIEDP constant.
13182 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
13183 exponent support for __ieee128 argument.
13184 (VSESQP): Add scalar extract signature support for __ieee128
13186 (VSTDCNQP): Add scalar test negative support for __ieee128
13188 (VSIEQP): Add scalar insert exponent support for __int128 argument
13189 with __ieee128 result.
13190 (VSIEQPF): Add scalar insert exponent support for __ieee128
13191 argument with __ieee128 result.
13192 (VSTDCQP): Add scalar test data class support for __ieee128
13194 (VSTDCNQP): Add overload support for scalar test negative with
13195 __ieee128 argument.
13196 (VSTDCQP): Add overload support for scalar test data class
13197 __ieee128 argument.
13198 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
13199 UNSPEC_VSX_SXSIGDP.
13200 (UNSPEC_VSX_SIEXPQP): New constant.
13201 (xsxexpqp): New insn for VSX scalar extract exponent quad
13203 (xsxsigqp): New insn for VSX scalar extract significand quad
13205 (xsiexpqpf): New insn for VSX scalar insert exponent quad
13206 precision with floating point argument.
13207 (xststdcqp): New expand for VSX scalar test data class quad
13209 (xststdcnegqp): New expand for VSX scalar test negative quad
13211 (xststdcqp): New insn to match expansions for VSX scalar test data
13212 class quad precision and VSX scalar test negative quad precision.
13213 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
13214 special case operand checking to enforce that second operand of
13215 VSX scalar test data class with quad precision argument is a 7-bit
13217 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
13218 prototypes and descriptions of __ieee128 versions of
13219 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
13220 scalar_test_data_class, and scalar_test_neg built-in functions.
13222 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13224 PR tree-optimization/81162
13225 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
13226 replace a negate with an add.
13228 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
13230 * doc/invoke.texi (arm/-mcpu): Document +crypto.
13232 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13234 * config/arm/arm-c.c (arm_cpu_builtins): Define
13235 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
13237 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13239 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
13240 (armv8-r): Set ARM Cortex-R52 as default CPU.
13241 * config/arm/arm-tables.opt: Regenerate.
13242 * config/arm/arm-tune.md: Regenerate.
13243 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
13245 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
13246 extension for -mcpu=cortex-r52.
13248 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13250 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
13251 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
13252 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
13253 (fp-armv8): Define it as FP_ARMv8 only.
13254 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
13255 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
13257 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
13258 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
13259 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
13260 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
13261 than TARGET_FPU_ARMV8.
13262 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
13263 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
13264 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
13266 * config/arm/neon.md (neon_vrint): Likewise.
13267 (neon_vcvt): Likewise.
13268 (neon_<fmaxmin_op><mode>): Likewise.
13269 (<fmaxmin><mode>3): Likewise.
13270 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
13271 * config/arm/predicates.md (arm_cond_move_operator): Check against
13272 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
13274 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
13276 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
13277 to top of function.
13279 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13281 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
13282 loop in comment with memset.
13284 2017-07-14 Martin Liska <mliska@suse.cz>
13286 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
13287 * dwarf2out.c (is_java): Remove the function.
13288 (output_pubname): Remove usage of the function.
13289 (lower_bound_default): Remove usage of DW_LANG_Java.
13290 (gen_compile_unit_die): Likewise.
13291 * gcc.c: Remove compiler defaults for .java and .zip files.
13292 * gimple-expr.c (remove_suffix): Change as there's no longer
13293 extension than 4-letter one.
13294 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
13295 (gimplify_save_expr): Likewise.
13296 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
13297 as it's possible even for other languages than Java.
13298 * langhooks.h (struct lang_hooks): Remove Java from a comment.
13299 * lto-opts.c (lto_write_options): Remove reference to Java.
13300 * opts.c (strip_off_ending): Update file extension handling.
13301 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
13302 * tree-eh.c (lower_resx): Likewise.
13303 * tree.c (free_lang_data_in_type): Remove dead code.
13304 (find_decls_types_r): Likewise.
13305 (build_common_builtin_nodes): Remove Java from a comment.
13306 (verify_type): Remove dead code.
13307 * varasm.c (assemble_external): Remove Java from a comment.
13309 2017-07-14 Martin Liska <mliska@suse.cz>
13311 * opts.c (finish_options): Add quotes.
13312 (common_handle_option): Likewise.
13314 2017-07-14 Martin Liska <mliska@suse.cz>
13316 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
13317 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
13318 Remove N_SO_PASCAL.
13319 * dwarf2out.c (lower_bound_default): Do not handle
13321 (gen_compile_unit_die): Likewise.
13322 * gcc.c: Remove default extension binding for GNU Pascal.
13323 * stmt.c: Remove Pascal language from a comment.
13324 * xcoffout.c: Likewise.
13326 2017-07-13 David Malcolm <dmalcolm@redhat.com>
13329 * diagnostic-show-locus.c (fixit_cmp): New function.
13330 (layout::layout): Sort m_fixit_hints.
13331 (column_range::column_range): Assert that the values are valid.
13332 (struct char_span): New struct.
13333 (correction::overwrite): New method.
13334 (struct source_line): New struct.
13335 (line_corrections::add_hint): Add assertions. Reimplement memcpy
13336 calls in terms of classes source_line and char_span, and
13337 correction::overwrite.
13338 (selftest::test_overlapped_fixit_printing_2): New function.
13339 (selftest::diagnostic_show_locus_c_tests): Call it.
13341 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
13343 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
13344 early if there is no lhs.
13346 2017-07-13 Martin Liska <mliska@suse.cz>
13348 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
13349 (gen_reference_type_die): Likewise.
13350 * stor-layout.c: Remove Pascal-related comment.
13352 2017-07-13 Martin Liska <mliska@suse.cz>
13354 * opts.c (finish_options): Add quotes to error messages.
13355 (parse_sanitizer_options): Likewise.
13357 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
13359 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
13361 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
13363 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
13365 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
13367 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
13369 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
13371 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
13374 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
13375 provides the hardware capability bits, define the macro
13376 __BUILTIN_CPU_SUPPORTS__.
13377 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
13378 if GLIBC does not provide the hardware capability bits. Add a
13379 gcc_unreachable call if the built-in cpu function is neither
13380 __builtin_cpu_is nor __builtin_cpu_supports.
13381 (rs6000_get_function_versions_dispatcher): Change the warning
13382 that an old GLIBC is used which does not export the capability
13383 bits to be an error.
13384 * doc/extend.texi (target_clones attribute): Document the
13385 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
13386 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
13387 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
13388 the macros defined by GCC if the newer GLIBC is available.
13390 2017-07-12 Jeff Law <law@redhat.com>
13392 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
13393 remaining includes slightly.
13394 * config/riscv/riscv-builtins.c: Include profile-count.h.
13396 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
13399 * config/avr/avr.c (avr_set_current_function): In diagnostic
13400 messages: Quote keywords and (parts of) identifiers.
13401 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
13404 2017-07-12 Carl Love <cel@us.ibm.com>
13406 * config/rs6000/rs6000-c.c: Add support for built-in functions
13407 vector bool char vec_revb (vector bool char);
13408 vector bool short vec_revb (vector short char);
13409 vector bool int vec_revb (vector bool int);
13410 vector bool long long vec_revb (vector bool long long);
13411 * doc/extend.texi: Update the built-in documentation file for the
13412 new built-in functions.
13414 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13416 * config/s390/s390.md: Remove movcc splitter.
13418 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13420 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
13421 load/store on condition.
13423 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
13426 * config/avr/avr.c (avr_encode_section_info)
13427 [progmem && !TREE_READONLY]: Error if progmem object needs
13430 2017-07-11 Michael Collison <michael.collison@arm.com>
13432 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
13435 2017-07-11 Carl Love <cel@us.ibm.com>
13437 * config/rs6000/rs6000-c.c: Add support for builtins
13438 vector unsigned int vec_parity_lsbb (vector signed int);
13439 vector unsigned int vec_parity_lsbb (vector unsigned int);
13440 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
13441 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
13442 vector unsigned long long vec_parity_lsbb (vector signed long long);
13443 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
13444 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
13445 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
13446 * doc/extend.texi: Update the built-in documentation file for the
13447 new built-in functions.
13449 2017-07-11 David Malcolm <dmalcolm@redhat.com>
13451 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
13452 (layout::m_primary_loc): New field.
13453 (layout::layout): Initialize new field. Move location filtering
13454 logic from here to...
13455 (layout::maybe_add_location_range): ...this new method. Add
13456 support for filtering to just the lines already specified by other
13458 (layout::will_show_line_p): New method.
13459 (gcc_rich_location::add_location_if_nearby): New method.
13460 (selftest::test_add_location_if_nearby): New test function.
13461 (selftest::diagnostic_show_locus_c_tests): Call it.
13462 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
13465 2017-07-11 Tom de Vries <tom@codesourcery.com>
13467 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
13468 (bb_first_real_insn): New function.
13469 (nvptx_single): Add extra initialization of broadcasted condition
13472 2017-07-11 Nathan Sidwell <nathan@acm.org>
13474 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
13476 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
13478 * doc/extend.texi (AVR Function Attributes): Remove weblink to
13479 Binutils doc as TEXI will mess them up.
13480 * doc/invoke.texi (AVR Options): Same here.
13482 2017-07-11 Daniel Cederman <cederman@gaisler.com>
13484 * config/sparc/sparc.opt (mfix-ut700): New option.
13485 (mfix-gr712rc): Likewise.
13486 (sparc_fix_b2bst): New variable.
13487 * doc/invoke.texi (SPARC options): Document them.
13488 (ARM options): Fix warnings.
13489 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
13490 instructions to prevent sequences that can trigger the store-store
13491 errata for certain LEON3FT processors.
13492 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
13493 (sparc_option_override): Set sparc_fix_b2bst appropriately.
13494 * config/sparc/sparc.md (fix_b2bst): New attribute.
13495 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
13497 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
13500 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
13502 (*rsqrtsf2_sse): Ditto.
13504 (div<mode>3): Macroize insn from divdf3 and divsf3
13505 using MODEF mode iterator.
13507 2017-07-10 Martin Sebor <msebor@redhat.com>
13509 PR tree-optimization/80397
13510 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
13511 instead of testing for equality to INTEGER_TYPE.
13513 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
13515 * config.gcc: Remove uclibc from arc target spec.
13517 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
13519 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
13521 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13524 * lto-wrapper.c (remove_option): New function.
13525 (merge_and_complain): Merge PIC/PIE options more realistically.
13527 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
13529 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
13533 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
13534 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
13535 * config.in: Regenerate.
13536 * configure: Regenerate.
13537 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
13538 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
13539 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
13540 (TARGET_GASISR_PROLOGUES): ...target mask.
13541 * common/config/avr/avr-common.c
13542 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
13543 Set -mgas-isr-prologues.
13544 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
13545 INSERT_PASS_BEFORE for it.
13546 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
13547 * config/avr/avr.c (avr_option_override)
13548 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
13549 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
13550 (avr_attribute_table) <no_gccisr>: Add new function attribute.
13551 (avr_set_current_function) <is_no_gccisr>: Init machine field.
13552 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
13554 (make_avr_pass_pre_proep): New function.
13555 (emit_push_sfr) <treg>: Add argument to function and use it
13556 instead of TMP_REG.
13557 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
13558 and set machine->gasisr.yes.
13559 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
13560 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
13561 __gcc_isr.n_pushed to .L__stack_usage.
13562 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
13563 (avr_asm_final_postscan_insn): ...this new static function.
13564 * config/avr/avr.h (machine_function)
13565 <is_no_gccisr, use_L__stack_usage>: New fields.
13566 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
13567 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
13568 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
13569 (gasisr, *gasisr): New expander and insn.
13570 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
13571 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
13572 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
13574 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
13576 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
13579 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
13581 Move jump-tables out of .text again.
13584 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
13585 (ASM_OUTPUT_ADDR_VEC): New function.
13586 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
13587 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
13588 INSN_ADDRESSes as asm comment.
13589 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
13590 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
13591 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
13592 * config/avr/avr.md (*tablejump): Adjust comment.
13593 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
13594 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
13596 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
13597 (avr_output_addr_vec): New proto.
13598 (avr_log_t) <insn_addresses>: New field.
13600 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
13603 * config/i386/i386.c (ix86_function_arg_advance): Set
13604 outgoing_args_on_stack to true if there are outgoing arguments
13606 (ix86_function_arg): Likewise.
13607 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
13608 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
13609 * config/i386/i386.h (machine_function): Add
13610 outgoing_args_on_stack.
13612 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
13614 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
13615 supporting pthreds.
13616 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
13618 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
13620 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
13621 (REAL_H): Remove $(MACHMODE_H).
13622 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
13624 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
13625 $(MACHMODE_H) and double-int.h.
13626 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
13628 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
13631 2017-07-07 Andrew Pinski <apinski@cavium.com>
13633 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
13634 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
13636 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13638 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
13639 Add warning if GCC was not configured to link against a GLIBC that
13640 exports the hardware capability bits.
13641 (make_resolver_func): Make resolver function private and not a
13642 COMDAT function. Create the name with clone_function_name instead
13643 of make_unique_name.
13646 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
13647 correct operand in doing the split.
13649 2017-07-07 Carl Love <cel@us.ibm.com>
13651 * config/rs6000/rs6000-c: Add support for built-in function
13652 vector unsigned short vec_pack_to_short_fp32 (vector float,
13654 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
13655 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
13656 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
13657 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
13658 (convert_4f32_8i16): Add define_expand.
13659 * doc/extend.texi: Update the built-in documentation file for the
13660 new built-in function.
13662 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13664 * config/sparc/m8.md: New file.
13665 * config/sparc/sparc.md: Include m8.md.
13667 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13669 * config/sparc/sparc.opt: New option -mvis4b.
13670 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
13671 (sparc_option_override): Handle VIS4B.
13672 (enum sparc_builtins): Define
13673 SPARC_BUILTIN_DICTUNPACK{8,16,32},
13674 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
13675 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
13676 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
13677 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
13678 (check_constant_argument): New function.
13679 (sparc_vis_init_builtins): Define builtins
13680 __builtin_vis_dictunpack{8,16,32},
13681 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
13682 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
13683 __builtin_vis_fpcmpde{8,16,32}shl and
13684 __builtin_vis_fpcmpur{8,16,32}shl.
13685 (sparc_expand_builtin): Check that the constant operands to
13686 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
13687 constant and in range.
13688 * config/sparc/sparc-c.c (sparc_target_macros): Handle
13690 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
13691 (SPARC_IMM5_P): Likewise.
13692 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
13693 (enabled): Handle vis4b.
13694 (UNSPEC_DICTUNPACK): New unspec.
13695 (UNSPEC_FPCMPSHL): Likewise.
13696 (UNSPEC_FPUCMPSHL): Likewise.
13697 (UNSPEC_FPCMPDESHL): Likewise.
13698 (UNSPEC_FPCMPURSHL): Likewise.
13699 (cpu_feature): New CPU feature `vis4b'.
13700 (dictunpack{8,16,32}): New insns.
13701 (FPCSMODE): New mode iterator.
13702 (fpcscond): New code iterator.
13703 (fpcsucond): Likewise.
13704 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
13705 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
13706 (fpcmpde{8,16,32}{si,di}shl): Likewise.
13707 (fpcmpur{8,16,32}{si,di}shl): Likewise.
13708 * config/sparc/constraints.md: Define constraints `q' for unsigned
13709 2-bit integer constants and `t' for unsigned 5-bit integer
13711 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
13713 (imm5_operand_dictunpack16): Likewise.
13714 (imm5_operand_dictunpack32): Likewise.
13715 (imm2_operand): Likewise.
13716 * doc/invoke.texi (SPARC Options): Document -mvis4b.
13717 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
13718 ditunpack* and fpcmp*shl builtins.
13720 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13722 * config.gcc: Handle m8 in --with-{cpu,tune} options.
13723 * config.in: Add HAVE_AS_SPARC6 define.
13724 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
13726 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
13728 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
13729 (CPP_CPU_SPEC): Handle m8.
13730 (ASM_CPU_SPEC): Likewise.
13731 * config/sparc/sparc-opts.h (enum processor_type): Add
13733 * config/sparc/sparc.c (m8_costs): New struct.
13734 (sparc_option_override): Handle TARGET_CPU_m8.
13735 (sparc32_initialize_trampoline): Likewise.
13736 (sparc64_initialize_trampoline): Likewise.
13737 (sparc_issue_rate): Likewise.
13738 (sparc_register_move_cost): Likewise.
13739 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
13740 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
13741 (ASM_CPU64_DEFAULT_SPEC): Likewise.
13742 (CPP_CPU_SPEC): Handle M8.
13743 (ASM_CPU_SPEC): Likewise.
13744 (AS_M8_FLAG): Define.
13745 * config/sparc/sparc.md: Add m8 to the cpu attribute.
13746 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
13747 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
13749 * configure: Regenerate.
13750 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
13753 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13755 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
13757 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
13758 ("*movdi_insn_sp32"): Do not set v3pipe.
13759 ("*movsi_insn"): Likewise.
13760 ("*movdi_insn_sp64"): Likewise.
13761 ("*movsf_insn"): Likewise.
13762 ("*movdf_insn_sp32"): Likewise.
13763 ("*movdf_insn_sp64"): Likewise.
13764 ("*zero_extendsidi2_insn_sp64"): Likewise.
13765 ("*sign_extendsidi2_insn"): Likewise.
13766 ("*mov<VM32:mode>_insn"): Likewise.
13767 ("*mov<VM64:mode>_insn_sp64"): Likewise.
13768 ("*mov<VM64:mode>_insn_sp32"): Likewise.
13769 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
13770 ("<vlop:code><VL:mode>3"): Likewise.
13771 ("*not_<vlop:code><VL:mode>3"): Likewise.
13772 ("*nand<VL:mode>_vis"): Likewise.
13773 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
13774 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
13775 ("one_cmpl<VL:mode>2"): Likewise.
13776 ("faligndata<VM64:mode>_vis"): Likewise.
13777 ("alignaddrsi_vis"): Likewise.
13778 ("alignaddrdi_vis"): Likweise.
13779 ("alignaddrlsi_vis"): Likewise.
13780 ("alignaddrldi_vis"): Likewise.
13781 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
13782 ("bmaskdi_vis"): Likewise.
13783 ("bmasksi_vis"): Likewise.
13784 ("bshuffle<VM64:mode>_vis"): Likewise.
13785 ("cmask8<P:mode>_vis"): Likewise.
13786 ("cmask16<P:mode>_vis"): Likewise.
13787 ("cmask32<P:mode>_vis"): Likewise.
13788 ("pdistn<P:mode>_vis"): Likewise.
13789 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
13791 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13793 * config/sparc/sparc.md ("subtype"): New insn attribute.
13794 ("*wrgsr_sp64"): Set insn subtype.
13795 ("*rdgsr_sp64"): Likewise.
13796 ("alignaddrsi_vis"): Likewise.
13797 ("alignaddrdi_vis"): Likewise.
13798 ("alignaddrlsi_vis"): Likewise.
13799 ("alignaddrldi_vis"): Likewise.
13800 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
13801 ("fexpand_vis"): Likewise.
13802 ("fpmerge_vis"): Likewise.
13803 ("faligndata<VM64:mode>_vis"): Likewise.
13804 ("bshuffle<VM64:mode>_vis"): Likewise.
13805 ("cmask8<P:mode>_vis"): Likewise.
13806 ("cmask16<P:mode>_vis"): Likewise.
13807 ("cmask32<P:mode>_vis"): Likewise.
13808 ("fchksm16_vis"): Likewise.
13809 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
13810 ("fmean16_vis"): Likewise.
13811 ("fp<plusminus_insn>64_vis"): Likewise.
13812 ("<plusminus_insn>v8qi3"): Likewise.
13813 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
13814 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
13815 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
13816 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
13817 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
13818 ("*movqi_insn"): Likewise.
13819 ("*movhi_insn"): Likewise.
13820 ("*movsi_insn"): Likewise.
13821 ("movsi_pic_gotdata_op"): Likewise.
13822 ("*movdi_insn_sp32"): Likewise.
13823 ("*movdi_insn_sp64"): Likewise.
13824 ("movdi_pic_gotdata_op"): Likewise.
13825 ("*movsf_insn"): Likewise.
13826 ("*movdf_insn_sp32"): Likewise.
13827 ("*movdf_insn_sp64"): Likewise.
13828 ("*zero_extendhisi2_insn"): Likewise.
13829 ("*zero_extendqihi2_insn"): Likewise.
13830 ("*zero_extendqisi2_insn"): Likewise.
13831 ("*zero_extendqidi2_insn"): Likewise.
13832 ("*zero_extendhidi2_insn"): Likewise.
13833 ("*zero_extendsidi2_insn_sp64"): Likewise.
13834 ("ldfsr"): Likewise.
13835 ("prefetch_64"): Likewise.
13836 ("prefetch_32"): Likewise.
13837 ("tie_ld32"): Likewise.
13838 ("tie_ld64"): Likewise.
13839 ("*tldo_ldub_sp32"): Likewise.
13840 ("*tldo_ldub1_sp32"): Likewise.
13841 ("*tldo_ldub2_sp32"): Likewise.
13842 ("*tldo_ldub_sp64"): Likewise.
13843 ("*tldo_ldub1_sp64"): Likewise.
13844 ("*tldo_ldub2_sp64"): Likewise.
13845 ("*tldo_ldub3_sp64"): Likewise.
13846 ("*tldo_lduh_sp32"): Likewise.
13847 ("*tldo_lduh1_sp32"): Likewise.
13848 ("*tldo_lduh_sp64"): Likewise.
13849 ("*tldo_lduh1_sp64"): Likewise.
13850 ("*tldo_lduh2_sp64"): Likewise.
13851 ("*tldo_lduw_sp32"): Likewise.
13852 ("*tldo_lduw_sp64"): Likewise.
13853 ("*tldo_lduw1_sp64"): Likewise.
13854 ("*tldo_ldx_sp64"): Likewise.
13855 ("*mov<VM32:mode>_insn"): Likewise.
13856 ("*mov<VM64:mode>_insn_sp64"): Likewise.
13857 ("*mov<VM64:mode>_insn_sp32"): Likewise.
13859 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13861 * config/sparc/sparc.md ("type"): New insn type viscmp.
13862 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
13864 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
13865 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
13866 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
13867 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
13869 ("n7_vis_logical_11cycle"): Likewise.
13870 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
13871 * config/sparc/niagara2.md ("niag3_vis": Likewise.
13872 * config/sparc/niagara.md ("niag_vis"): Likewise.
13873 * config/sparc/ultra3.md ("us3_fga"): Likewise.
13874 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
13876 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13878 * config/sparc/sparc.md: New instruction type `bmask'.
13879 (bmaskdi_vis): Use the `bmask' type.
13880 (bmasksi_vis): Likewise.
13881 * config/sparc/ultra3.md (us3_array): Likewise.
13882 * config/sparc/niagara7.md (n7_array): Likewise.
13883 * config/sparc/niagara4.md (n4_array): Likewise.
13884 * config/sparc/niagara2.md (niag2_vis): Likewise.
13885 (niag3_vis): Likewise.
13886 * config/sparc/niagara.md (niag_vis): Likewise.
13888 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13890 * ipa-comdats.c: Remove optimize check from gate.
13891 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
13892 for functions not optimized.
13893 (ipa_fn_summary_read): Skip optimize check.
13894 (ipa_fn_summary_write): Likewise.
13895 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
13897 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
13899 (can_inline_edge_p): Check flag_pcc_struct_return for match.
13900 (check_callers): Give up on caller which is not optimized.
13901 (inline_small_functions): Likewise.
13902 (ipa_inline): Do not give up when not optimizing.
13903 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
13904 away unoptimizes cdtors.
13905 (whole_program_function_and_variable_visibility): Do
13906 ipa_discover_readonly_nonaddressable_vars in LTO mode.
13907 * ipa.c (process_references): Do not check optimize.
13908 (symbol_table::remove_unreachable_nodes): Update optimize check.
13909 (set_writeonly_bit): Update optimize check.
13910 (pass_ipa_cdtor_merge::gate): Do not check optimize.
13911 (pass_ipa_single_use::gate): Remove.
13913 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13915 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
13916 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
13917 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
13918 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
13919 permute_load, permute_store, adjust_extract, adjust_splat,
13920 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
13921 replace_swap_with_copy, dump_swap_insn_table,
13922 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
13923 recombine_lvx_pattern, recombine_stvx_pattern,
13924 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
13925 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
13926 to file rs6000-p8swap.c.
13927 * config/rs6000/rs6000-p8swap.c: New file.
13928 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
13929 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
13930 and rs6000*-*-* targets.
13932 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13934 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
13936 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13938 * lto-wrapper.c (merge_and_complain): Do not merge
13939 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
13940 fsigned_zeros, ftrapping_math, fwrapv.
13941 (append_compiler_options): Do not track these options.
13942 (append_linker_options): Likewie
13944 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13946 * cgraphunit.c (cgraph_node::finalize_function): When
13947 !flag_toplevel_reorde set no_reorder flag.
13948 (varpool_node::finalize_decl): Likewise.
13949 (symbol_table::compile): Drop no toplevel reorder path.
13951 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13953 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
13954 edges; zero probability is not better than uninitialized.
13956 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
13958 * asan.h (asan_sanitize_allocas_p): Declare.
13959 * asan.c (asan_sanitize_allocas_p): New function.
13960 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
13961 (handle_builtin_alloca): Likewise.
13962 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
13963 if !asan_sanitize_allocas_p.
13964 * params.def (asan-instrument-allocas): Add new option.
13965 * params.h (ASAN_PROTECT_ALLOCAS): Define.
13966 * opts.c (common_handle_option): Disable allocas sanitization for
13969 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
13971 * asan.c: Include gimple-fold.h.
13972 (get_last_alloca_addr): New function.
13973 (handle_builtin_stackrestore): Likewise.
13974 (handle_builtin_alloca): Likewise.
13975 (asan_emit_allocas_unpoison): Likewise.
13976 (get_mem_refs_of_builtin_call): Add new parameter, remove const
13977 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
13978 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
13979 (instrument_builtin_call): Pass gimple iterator to
13980 get_mem_refs_of_builtin_call.
13981 (last_alloca_addr): New global.
13982 * asan.h (asan_emit_allocas_unpoison): Declare.
13983 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
13984 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
13985 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
13986 if function calls alloca.
13987 * gimple-fold.c (replace_call_with_value): Remove static keyword.
13988 * gimple-fold.h (replace_call_with_value): Declare.
13989 * internal-fn.c: Include asan.h.
13990 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
13991 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
13993 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13995 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
13996 (C_SELFTEST_FLAGS): New.
13997 (CPP_SELFTEST_FLAGS): New.
13998 (SELFTEST_DEPS): New, from deps of s-selftest.
13999 (C_SELFTEST_DEPS): New, from deps of s-selftest.
14000 (CPP_SELFTEST_DEPS): New.
14001 (selftest): Add dependency on s-selftest-c++.
14002 (s-selftest): Rename to...
14003 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
14004 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
14005 than SELFTEST_FLAGS.
14006 (selftest-gdb): Rename to...
14007 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
14009 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
14010 (selftest-valgrind): Rename to...
14011 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
14013 (selftest-valgrind): Reintroduce as an alias for
14014 selftest-c-valgrind.
14015 (s-selftest-c++): New.
14016 (selftest-c++-gdb): New.
14017 (selftest-c++-valgrind): New.
14019 2017-07-06 Olivier Hainque <hainque@adacore.com>
14021 * gcc.c (process_command): When deciding if undefined variables
14022 should be ignored when processing specs, accept "gcc -v" as well.
14024 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14026 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
14027 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
14029 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
14031 * config/arm/arm-cpus.in (armv8-r): Add new entry.
14032 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
14033 * config/arm/arm-tables.opt: Regenerate.
14034 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
14036 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
14038 2017-07-06 Carl Love <cel@us.ibm.com>
14040 * ChangeLog: Clean up from mid air collision
14042 2017-07-06 Carl Love <cel@us.ibm.com>
14044 * config/rs6000/rs6000-c.c: Add support for built-in functions
14045 vector signed int vec_subc (vector signed int, vector signed int);
14046 vector signed __int128 vec_subc (vector signed __int128,
14047 vector signed __int128);
14048 vector unsigned __int128 vec_subc (vector unsigned __int128,
14049 vector unsigned __int128);
14050 vector signed int vec_sube (vector signed int, vector signed int,
14051 vector signed int);
14052 vector unsigned int vec_sube (vector unsigned int,
14053 vector unsigned int,
14054 vector unsigned int);
14055 vector signed __int128 vec_sube (vector signed __int128,
14056 vector signed __int128,
14057 vector signed__int128);
14058 vector unsigned __int128 vec_sube (vector unsigned __int128,
14059 vector unsigned __int128,
14060 vector unsigned __int128);
14061 vector signed int vec_subec (vector signed int, vector signed int,
14062 vector signed int);
14063 vector unsigned int vec_subec (vector unsigned int,
14064 vector unsigned int,
14065 vector unsigned int);
14066 vector signed __int128 vec_subec (vector signed __int128,
14067 vector signed __int128,
14068 vector signed__int128);
14069 vector unsigned __int128 vec_subec (vector unsigned __int128,
14070 vector unsigned __int128,
14071 vector unsigned __int128);
14072 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
14073 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
14074 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
14075 BU_ALTIVEC_OVERLOAD_X definitions.
14076 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
14077 * doc/extend.texi: Update the built-in documentation file for the new
14078 built-in functions.
14080 2017-07-06 David Malcolm <dmalcolm@redhat.com>
14083 * diagnostic-show-locus.c (layout::layout): Use start and finish
14084 spelling location for the start and finish of each range.
14085 * genmatch.c (linemap_client_expand_location_to_spelling_point):
14086 Add unused aspect param.
14087 * input.c (expand_location_1): Add "aspect" param, and use it
14088 to access the correct part of the location.
14089 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
14091 (expand_location_to_spelling_point): Likewise.
14092 (linemap_client_expand_location_to_spelling_point): Add "aspect"
14093 param, and pass it to expand_location_1.
14095 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
14097 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
14098 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
14099 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
14100 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
14101 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
14102 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
14103 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
14104 _mm_maskz_getmant_ss): New intrinsics.
14105 (__builtin_ia32_getexpss128_mask): Changed to ...
14106 __builtin_ia32_getexpss128_round ... this.
14107 (__builtin_ia32_getexpsd128_mask): Changed to ...
14108 __builtin_ia32_getexpsd128_round ... this.
14109 * config/i386/i386-builtin-types.def
14110 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
14111 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
14112 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
14113 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
14114 __builtin_ia32_getmantss_mask_round): New builtins.
14115 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
14116 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
14117 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
14118 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
14119 * config/i386/sse.md
14120 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
14121 avx512f_sgetexp<mode><mask_scalar_name>
14122 <round_saeonly_scalar_name> ... this.
14123 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
14124 %0, %1, %2<round_saeonly_op3>}): Changed to ...
14125 vgetexp<ssescalarmodesuffix>
14126 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14127 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
14128 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
14129 avx512f_vgetmant<mode><mask_scalar_name>
14130 <round_saeonly_scalar_name> ... this.
14131 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
14132 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
14133 vgetmant<ssescalarmodesuffix>
14134 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
14135 %0<mask_scalar_operand4>, %1, %2
14136 <round_saeonly_scalar_mask_op4>, %3} ... this.
14137 * config/i386/subst.md (mask_scalar_operand4,
14138 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
14139 round_saeonly_scalar_nimm_predicate): New subst attributes.
14141 2017-07-06 Julia Koval <julia.koval@intel.com>
14143 * config/i386/i386.c (ix86_erase_embedded_rounding):
14144 Remove code for old rounding pattern.
14146 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
14148 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
14150 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
14152 * doc/sourcebuild.texi (Test Directives, Variants of
14153 dg-require-support): Add documentation for dg-require-stack-check.
14155 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
14157 * config/i386/subst.md (mask_scalar, round_scalar,
14158 round_saeonly_scalar): New meta-templates.
14159 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
14160 round_scalar_mask_operand3, round_scalar_mask_op3,
14161 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
14162 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
14163 round_saeonly_scalar_constraint,
14164 round_saeonly_scalar_prefix): New subst attribute.
14165 * config/i386/sse.md
14166 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
14167 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
14168 <round_scalar_name> ... this.
14169 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
14170 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
14171 <round_scalar_name> ... this.
14172 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
14173 <sse>_vm<code><mode>3<mask_scalar_name>
14174 <round_saeonly_scalar_name> ... this.
14175 (v<plusminus_mnemonic><ssescalarmodesuffix>
14176 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
14177 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
14178 v<plusminus_mnemonic><ssescalarmodesuffix>
14179 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14180 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
14181 (v<multdiv_mnemonic><ssescalarmodesuffix>
14182 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
14183 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
14184 v<multdiv_mnemonic><ssescalarmodesuffix>
14185 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14186 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
14187 (v<maxmin_float><ssescalarmodesuffix>
14188 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
14189 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
14190 v<maxmin_float><ssescalarmodesuffix>
14191 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14192 %0<mask_scalar_operand3>, %1, %<iptr>2
14193 <round_saeonly_scalar_mask_op3>} ... this.
14195 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
14197 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
14198 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
14200 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
14201 Alan Hayward <alan.hayward@arm.com>
14202 David Sherwood <david.sherwood@arm.com>
14204 * combine.c (simplify_if_then_else): Remove "enum" before
14206 * compare-elim.c (can_eliminate_compare): Likewise.
14207 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
14209 (aarch64_lookup_simd_builtin_type): Likewise.
14210 (aarch64_simd_builtin_type): Likewise.
14211 (aarch64_init_simd_builtin_types): Likewise.
14212 (aarch64_simd_expand_args): Likewise.
14213 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
14215 (aarch64_reverse_mask): Likewise.
14216 (aarch64_simd_emit_reg_reg_move): Likewise.
14217 (aarch64_gen_adjusted_ldpstp): Likewise.
14218 (aarch64_ccmp_mode_to_code): Likewise.
14219 (aarch64_operands_ok_for_ldpstp): Likewise.
14220 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14221 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
14223 (aarch64_min_divisions_for_recip_mul): Likewise.
14224 (aarch64_reassociation_width): Likewise.
14225 (aarch64_get_condition_code_1): Likewise.
14226 (aarch64_simd_emit_reg_reg_move): Likewise.
14227 (aarch64_simd_attr_length_rglist): Likewise.
14228 (aarch64_reverse_mask): Likewise.
14229 (aarch64_operands_ok_for_ldpstp): Likewise.
14230 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14231 (aarch64_gen_adjusted_ldpstp): Likewise.
14232 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
14234 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
14235 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
14236 (arm_lookup_simd_builtin_type): Likewise.
14237 (arm_simd_builtin_type): Likewise.
14238 (arm_init_simd_builtin_types): Likewise.
14239 (arm_expand_builtin_args): Likewise.
14240 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
14241 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
14242 (ft32_setup_incoming_varargs): Likewise.
14243 (ft32_function_arg): Likewise.
14244 (ft32_function_arg_advance): Likewise.
14245 (ft32_pass_by_reference): Likewise.
14246 (ft32_arg_partial_bytes): Likewise.
14247 (ft32_valid_pointer_mode): Likewise.
14248 (ft32_addr_space_pointer_mode): Likewise.
14249 (ft32_addr_space_legitimate_address_p): Likewise.
14250 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
14252 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
14253 (ix86_emit_outlined_ms2sysv_restore): Likewise.
14254 (iamcu_alignment): Likewise.
14255 (canonicalize_vector_int_perm): Likewise.
14256 (ix86_noce_conversion_profitable_p): Likewise.
14257 (ix86_mpx_bound_mode): Likewise.
14258 (ix86_operands_ok_for_move_multiple): Likewise.
14259 * config/microblaze/microblaze-protos.h
14260 (microblaze_expand_conditional_branch_reg): Likewise.
14261 * config/microblaze/microblaze.c
14262 (microblaze_expand_conditional_branch_reg): Likewise.
14263 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
14265 (rs6000_reassociation_width): Likewise.
14266 (rs6000_invalid_binary_op): Likewise.
14267 (fusion_p9_p): Likewise.
14268 (emit_fusion_p9_load): Likewise.
14269 (emit_fusion_p9_store): Likewise.
14270 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
14272 (riscv_hard_regno_mode_ok_p): Likewise.
14273 (riscv_address_insns): Likewise.
14274 (riscv_split_symbol): Likewise.
14275 (riscv_legitimize_move): Likewise.
14276 (riscv_function_value): Likewise.
14277 (riscv_hard_regno_nregs): Likewise.
14278 (riscv_expand_builtin): Likewise.
14279 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
14280 (riscv_build_integer): Likewise.
14281 (riscv_split_integer): Likewise.
14282 (riscv_legitimate_constant_p): Likewise.
14283 (riscv_cannot_force_const_mem): Likewise.
14284 (riscv_regno_mode_ok_for_base_p): Likewise.
14285 (riscv_valid_base_register_p): Likewise.
14286 (riscv_valid_offset_p): Likewise.
14287 (riscv_valid_lo_sum_p): Likewise.
14288 (riscv_classify_address): Likewise.
14289 (riscv_legitimate_address_p): Likewise.
14290 (riscv_address_insns): Likewise.
14291 (riscv_load_store_insns): Likewise.
14292 (riscv_force_binary): Likewise.
14293 (riscv_split_symbol): Likewise.
14294 (riscv_force_address): Likewise.
14295 (riscv_legitimize_address): Likewise.
14296 (riscv_move_integer): Likewise.
14297 (riscv_legitimize_const_move): Likewise.
14298 (riscv_legitimize_move): Likewise.
14299 (riscv_address_cost): Likewise.
14300 (riscv_subword): Likewise.
14301 (riscv_output_move): Likewise.
14302 (riscv_canonicalize_int_order_test): Likewise.
14303 (riscv_emit_int_order_test): Likewise.
14304 (riscv_function_arg_boundary): Likewise.
14305 (riscv_pass_mode_in_fpr_p): Likewise.
14306 (riscv_pass_fpr_single): Likewise.
14307 (riscv_pass_fpr_pair): Likewise.
14308 (riscv_get_arg_info): Likewise.
14309 (riscv_function_arg): Likewise.
14310 (riscv_function_arg_advance): Likewise.
14311 (riscv_arg_partial_bytes): Likewise.
14312 (riscv_function_value): Likewise.
14313 (riscv_pass_by_reference): Likewise.
14314 (riscv_setup_incoming_varargs): Likewise.
14315 (riscv_print_operand): Likewise.
14316 (riscv_elf_select_rtx_section): Likewise.
14317 (riscv_save_restore_reg): Likewise.
14318 (riscv_for_each_saved_reg): Likewise.
14319 (riscv_register_move_cost): Likewise.
14320 (riscv_hard_regno_mode_ok_p): Likewise.
14321 (riscv_hard_regno_nregs): Likewise.
14322 (riscv_class_max_nregs): Likewise.
14323 (riscv_memory_move_cost): Likewise.
14324 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
14325 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
14326 (rl78_addr_space_address_mode): Likewise.
14327 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14329 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14330 (rs6000_reassociation_width): Likewise.
14331 (rs6000_invalid_binary_op): Likewise.
14332 (fusion_p9_p): Likewise.
14333 (emit_fusion_p9_load): Likewise.
14334 (emit_fusion_p9_store): Likewise.
14335 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
14336 (ok_for_simple_move_operands): Likewise.
14337 (ok_for_simple_move_strict_operands): Likewise.
14338 (ok_for_simple_arith_logic_operands): Likewise.
14339 (visium_legitimize_reload_address): Likewise.
14340 (visium_select_cc_mode): Likewise.
14341 (output_cbranch): Likewise.
14342 (visium_split_double_move): Likewise.
14343 (visium_expand_copysign): Likewise.
14344 (visium_expand_int_cstore): Likewise.
14345 (visium_expand_fp_cstore): Likewise.
14346 * config/visium/visium.c (visium_pass_by_reference): Likewise.
14347 (visium_function_arg): Likewise.
14348 (visium_function_arg_advance): Likewise.
14349 (visium_libcall_value): Likewise.
14350 (visium_setup_incoming_varargs): Likewise.
14351 (visium_legitimate_constant_p): Likewise.
14352 (visium_legitimate_address_p): Likewise.
14353 (visium_legitimize_address): Likewise.
14354 (visium_secondary_reload): Likewise.
14355 (visium_register_move_cost): Likewise.
14356 (visium_memory_move_cost): Likewise.
14357 (prepare_move_operands): Likewise.
14358 (ok_for_simple_move_operands): Likewise.
14359 (ok_for_simple_move_strict_operands): Likewise.
14360 (ok_for_simple_arith_logic_operands): Likewise.
14361 (visium_function_value_1): Likewise.
14362 (rtx_ok_for_offset_p): Likewise.
14363 (visium_legitimize_reload_address): Likewise.
14364 (visium_split_double_move): Likewise.
14365 (visium_expand_copysign): Likewise.
14366 (visium_expand_int_cstore): Likewise.
14367 (visium_expand_fp_cstore): Likewise.
14368 (visium_split_cstore): Likewise.
14369 (visium_select_cc_mode): Likewise.
14370 (visium_split_cbranch): Likewise.
14371 (output_cbranch): Likewise.
14372 (visium_print_operand_address): Likewise.
14373 * expmed.c (flip_storage_order): Likewise.
14374 * expmed.h (emit_cstore): Likewise.
14375 (flip_storage_order): Likewise.
14376 * genrecog.c (validate_pattern): Likewise.
14377 * hsa-gen.c (gen_hsa_addr): Likewise.
14378 * internal-fn.c (expand_arith_overflow): Likewise.
14379 * ira-color.c (allocno_copy_cost_saving): Likewise.
14380 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14381 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
14382 (process_invariant_for_inheritance): Likewise.
14383 * lra-eliminations.c (move_plus_up): Likewise.
14384 * omp-low.c (lower_oacc_reductions): Likewise.
14385 * simplify-rtx.c (simplify_subreg): Likewise.
14386 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
14387 (TARGET_CHKP_BOUND_MODE): Likewise..
14388 * targhooks.c (default_chkp_bound_mode): Likewise.
14389 (default_setup_incoming_vararg_bounds): Likewise.
14390 * targhooks.h (default_chkp_bound_mode): Likewise.
14391 (default_setup_incoming_vararg_bounds): Likewise.
14392 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
14393 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
14394 (have_whole_vector_shift): Likewise.
14395 * tree-vect-stmts.c (vectorizable_load): Likewise.
14396 * doc/tm.texi: Regenerate.
14398 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
14400 Graceful degrade if Binutils PR21472 is not available.
14403 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
14404 .rodata in flash test fails.
14405 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
14406 * confgure: Regenerate.
14407 * config.in: Regenerate.
14408 * config/avr/avr.c (avr_asm_named_section)
14409 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
14410 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
14411 (avr_asm_init_sections): Same.
14413 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14415 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
14416 (fma<VH:mode>4_intrinsic): Likewise.
14417 (*fmsub<VCVTF:mode>4): Likewise.
14418 (*fmsub<VH:mode>4_intrinsic): Likewise.
14420 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
14423 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
14424 Don't depend on "optimize > 0".
14425 (out_movhi_r_mr, out_movqi_mr_r): Same.
14426 (out_movhi_mr_r, out_movqi_r_mr): Same.
14427 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
14428 io_address_operand on "optimize > 0".
14430 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14432 * tree-loop-distribution.c: Add general explanantion on the pass.
14433 (generate_loops_for_partition): Mark distributed loop.
14434 (pg_add_dependence_edges): New parameter. Handle alias data
14435 dependence specially and record it in the parameter if asked.
14436 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
14437 (init_partition_graph_vertices, add_partition_graph_edge): New.
14438 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
14439 (free_partition_graph_vdata, build_partition_graph): New.
14440 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
14441 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
14442 (data_ref_segment_size, latch_dominated_by_data_ref): New.
14443 (compute_alias_check_pairs, version_loop_by_alias_check): New.
14444 (version_for_distribution_p, finalize_partitions): New.
14445 (distribute_loop): Handle alias data dependence specially. Factor
14446 out loop fusion code as functions and call these functions.
14448 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14450 * tree-loop-distribution.c (classify_partition): New parameter and
14451 better handle reduction statement.
14452 (rdg_build_partitions): Revise comment.
14453 (distribute_loop): Compute statements in all partitions and pass it
14454 to classify_partition.
14456 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14458 * tree-loop-distribution.c (enum partition_type): New.
14459 (struct partition): New field type.
14460 (partition_merge_into): Add parameter. Update partition type.
14461 (data_dep_in_cycle_p, update_type_for_merge): New functions.
14462 (build_rdg_partition_for_vertex): Compute partition type.
14463 (rdg_build_partitions): Dump partition type.
14464 (distribute_loop): Update calls to partition_merge_into.
14466 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14468 * tree-loop-distribution.c (struct ddr_hasher): New.
14469 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
14471 (classify_partition): Call get_data_dependence.
14472 (pg_add_dependence_edges): Ditto.
14473 (distribute_loop): Release data dependence hash table.
14475 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14477 * tree-loop-distribution.c (ref_base_address): Delete.
14478 (similar_memory_accesses): Rename ...
14479 (share_memory_accesses): ... to this. Check if partitions access
14480 the same memory reference.
14481 (distribute_loop): Call share_memory_accesses.
14483 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14485 * tree-loop-distribution.c (struct partition): New field recording
14486 its data reference.
14487 (partition_alloc, partition_free): Init and release data refs.
14488 (partition_merge_into): Merge data refs.
14489 (build_rdg_partition_for_vertex): Collect data refs for partition.
14490 (pg_add_dependence_edges): Change parameters from vector to bitmap.
14492 (distribute_loop): Remve data refs from vertice data of partition
14495 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14497 * tree-loop-distribution.c (params.h): Include header file.
14498 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
14499 (datarefs_vec): New global var.
14500 (create_rdg_vertices): Use datarefs_vec directly.
14501 (free_rdg): Don't free data references.
14502 (build_rdg): Update use. Don't free data references.
14503 (distribute_loop): Compute global variable for data references.
14504 Bail out if there are too many data references.
14506 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14508 * tree-loop-distribution.c (loop_nest): New global var.
14509 (build_rdg): Use loop directly, rather than loop nest.
14510 (pg_add_dependence_edges): Remove loop nest parameter. Use global
14512 (distribute_loop): Compute global variable loop nest. Update use.
14514 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14516 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
14517 (partition_merge_into): New parameter. Dump reason for fusion.
14518 (distribute_loop): Update use of partition_merge_into.
14520 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14522 * tree-loop-distribution.c (bb_top_order_index): New.
14523 (bb_top_order_index_size, bb_top_order_cmp): New.
14524 (stmts_from_loop): Use topological order.
14525 (pass_loop_distribution::execute): Compute and release topological
14526 order for basic blocks.
14528 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14530 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
14533 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14535 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
14536 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
14537 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
14538 * internal-fn.def (LOOP_DIST_ALIAS): New.
14539 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
14540 (fold_loop_internal_call): ... this.
14541 (vect_loop_dist_alias_call): New function.
14542 (set_uid_loop_bbs): Call fold_loop_internal_call.
14543 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
14546 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
14549 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
14550 Require dead FLAGS_REG at the beginning of a peephole.
14552 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
14555 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
14556 arguments in the call to __builtin_ia32_sbb_u32.
14557 (_subborrow_u64): Swap _X and _Y arguments in the call to
14558 __builtin_ia32_sbb_u64.
14560 2017-07-04 Jakub Jelinek <jakub@redhat.com>
14563 * tree-vrp.c (compare_assert_loc): Turn into a function template
14564 with stable template parameter. Only test if a->e is NULL,
14565 !a->e == !b->e has been verified already. Use e == NULL or
14566 e != NULL instead of e or ! e tests. If stable is true, don't use
14567 iterative_hash_expr, on the other side allow a or b or both NULL
14568 and sort the NULLs last.
14569 (process_assert_insertions): Sort using compare_assert_loc<false>
14570 instead of compare_assert_loc, later sort using
14571 compare_assert_loc<true> before calling process_assert_insertions_for
14572 in a loop. Use break instead of continue once seen NULL pointer.
14574 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
14576 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
14577 Cortex-R7 and Cortex-R8 processors.
14579 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14581 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
14582 uninitialized while src is not.
14584 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
14586 * common/config/arm/arm-common.c: Adjust include path for
14588 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
14589 (arm-cpu.h): Create in build directory. Adjust dependency rules.
14590 (arm-cpu-data.h): Likewise.
14591 (arm-cpu-cdata.h): Likewise.
14592 * config/arm/arm-cpu.h: Delete.
14593 * config/arm/arm-cpu-cdata.h: Delete.
14594 * config/arm/arm-cpu-data.h: Delete.
14596 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
14598 * config/arm/arm-cpus.in (cortex-a55): New.
14599 (cortex-a75): Likewise.
14600 (cortex-a75.cortex-a55): Likewise.
14601 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
14603 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
14604 * config/arm/arm-cpu-cdata.h: Regenerate.
14605 * config/arm/arm-cpu-data.h: Regenerate.
14606 * config/arm/arm-cpu.h: Regenerate.
14607 * config/arm/arm-tables.opt: Regenerate.
14608 * config/arm/arm-tune.md: Regenerate.
14610 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14612 * haifa-sched.c (sched_create_recovery_edges): Update profile.
14614 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14616 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
14619 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
14621 PR tree-optimization/81292
14622 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
14623 full_string_p, also call adjust_related_strinfos if the adjustment
14624 is simple, otherwise invalidate related strinfos.
14626 2017-07-04 Martin Liska <mliska@suse.cz>
14629 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
14630 newly created variable as DECL_IGNORED_P.
14632 2017-07-04 Martin Liska <mliska@suse.cz>
14635 * ipa-inline.c (inline_small_functions):
14636 Use xstrdup_for_dump.
14638 2017-07-04 Tom de Vries <tom@codesourcery.com>
14640 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
14642 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
14645 * config/darwin.c (darwin_function_switched_text_sections):
14648 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
14650 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
14652 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
14654 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
14656 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14658 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
14659 min_profitable_iters, and th as inclusive lower bounds.
14660 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
14661 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
14662 for min_profitable_iters and min_profitable_estimate.
14663 (vect_transform_loop): Treat th as an inclusive lower bound.
14664 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
14666 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
14669 * config/darwin.c (darwin_function_switched_text_sections):
14670 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
14671 in two pieces, and suppress the use of buf.
14673 2017-07-03 Nathan Sidwell <nathan@acm.org>
14675 * hash-table.h (hash_table_mod1): Fix indentation.
14677 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14679 PR middle-end/81290
14680 * predict.c (force_edge_cold): Be more careful about propagation
14682 * profile-count.h (profile_probability::guessed,
14683 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
14685 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
14687 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
14689 * doc/invoke.texi (rcpc architecture extension): Document it.
14691 2017-07-03 Richard Biener <rguenther@suse.de>
14693 PR tree-optimization/60510
14694 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
14695 the scalar reduction PHI and use it.
14696 (vectorizable_reduction): Properly guard the single_defuse_cycle
14697 path for non-SLP reduction chains where we cannot use it.
14698 Rework reduc_def/index and vector type deduction. Rework
14699 vector operand gathering during reduction op code-gen.
14700 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
14701 chains dissolve the chain and leave it to non-SLP reduction
14704 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14706 * tree-data-ref.h (dr_alignment): Declare.
14707 * tree-data-ref.c (dr_alignment): New function.
14708 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
14709 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
14711 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
14713 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14715 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
14716 and base_misalignment fields.
14717 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
14718 * tree-data-ref.c: Include builtins.h.
14719 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
14720 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
14721 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
14722 * tree-vect-data-refs.c: Include tree-cfg.h.
14723 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
14724 fields instead of calculating an alignment here.
14725 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
14726 innermost_loop_behavior fields.
14728 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14730 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
14732 (DR_STEP_ALIGNMENT): New macro.
14733 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
14734 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
14735 (create_data_ref): Print it.
14736 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
14737 to tell whether the step preserves vector (mis)alignment.
14738 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
14739 Move the check for an integer step and generalise to all INTEGER_CST.
14740 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
14741 Print the outer step alignment.
14743 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14745 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
14746 with offset_alignment.
14747 (DR_ALIGNED_TO): Delete.
14748 (DR_OFFSET_ALIGNMENT): New macro.
14749 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
14750 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
14751 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
14752 (create_data_ref): Likewise.
14753 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
14754 (vect_analyze_data_refs): Likewise.
14755 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
14756 creating dummy innermost behavior.
14758 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14760 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
14761 with a "innermost_loop_behavior *" and refeence tree.
14762 * tree-data-ref.c (dr_analyze_innermost): Likewise.
14763 (create_data_ref): Update call accordingly.
14764 * tree-predcom.c (find_looparound_phi): Likewise.
14766 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14768 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
14769 fields with dr_wrt_vec_loop.
14770 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
14771 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
14772 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
14773 (vect_dr_behavior): New function.
14774 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
14775 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
14776 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
14777 track whether the step preserves the misalignment.
14778 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
14779 Use vect_dr_behavior.
14780 (vect_setup_realignment): Update call accordingly.
14781 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
14782 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
14783 call to vect_create_addr_base_for_vector_ref.
14784 (vect_create_cond_for_align_checks): Likewise.
14785 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
14786 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
14787 (vect_recog_mask_conversion_pattern): Likewise.
14788 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
14789 (new_stmt_vec_info): Remove redundant zeroing.
14791 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
14793 * common/config/arm/arm-common.c (arm_be8_option): New function.
14794 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
14795 (ISA_ARMv6): Add isa_bit_be8.
14796 * config/arm/arm.h (arm_be8_option): Add prototype.
14797 (BE8_SPEC_FUNCTION): New define.
14798 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
14799 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
14800 (mlittle-endian): Similarly.
14801 (mbe8, mbe32): New options.
14802 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
14803 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
14805 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14807 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
14809 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14811 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
14812 (cleanup_tree_cfg_bb): Use it.
14813 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
14815 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
14817 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14820 * loop-doloop.c (add_test): Update profile.
14822 2017-07-03 Martin Liska <mliska@suse.cz>
14825 * sanopt.c (rewrite_usage_of_param): New function.
14826 (sanitize_rewrite_addressable_params): Likewise.
14827 (pass_sanopt::execute): Call rewrite_usage_of_param.
14829 2017-07-03 Richard Biener <rguenther@suse.de>
14831 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
14832 back to using VIEW_CONVERT_EXPR.
14834 2017-07-03 Martin Liska <mliska@suse.cz>
14837 * doc/extend.texi: Document when a resolver function is
14838 generated for target_clones.
14840 2017-07-03 Martin Liska <mliska@suse.cz>
14842 * asan.c (asan_emit_stack_protection): Unpoison just red zones
14843 and shadow memory of auto variables which are subject of
14844 use-after-scope sanitization.
14845 (asan_expand_mark_ifn): Add do set only when is_poison.
14847 2016-07-03 Richard Biener <rguenther@suse.de>
14849 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
14851 (vect_force_simple_reduction): Record reduction def -> phi mapping.
14852 (vectorizable_reduction): Perform reduction PHI creation when
14853 visiting a reduction PHI and adjust and simplify code generation
14854 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
14855 (vect_transform_loop): Visit reduction PHIs.
14856 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
14857 defs into the SLP tree.
14858 (vect_build_slp_tree): Reduction defs terminate the recursion.
14859 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
14861 (vect_get_vec_defs_for_stmt_copy): Export.
14862 (vect_get_vec_defs): Likewise.
14863 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
14865 (vect_get_vec_defs_for_stmt_copy): Declare.
14866 (vect_get_vec_defs): Likewise.
14868 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14870 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
14871 parameter with a "loop" parameter and use it instead of the
14872 loop containing DR_STMT. Don't check simple_iv when doing
14873 BB analysis. Describe the two analysis modes in the comment.
14875 2017-07-03 Tom de Vries <tom@codesourcery.com>
14877 PR tree-optimization/69468
14878 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
14879 (find_same_succ_bb): Handle ignore_edge_flags.
14881 2017-07-03 Tom de Vries <tom@codesourcery.com>
14883 PR tree-optimization/81192
14884 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
14886 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
14888 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
14890 2017-07-03 Tom de Vries <tom@codesourcery.com>
14892 PR tree-optimization/81192
14893 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
14894 BB_SAME_SUCC (bb) == NULL.
14896 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14898 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
14901 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14903 * dumpfile.c: Include profile-count.h
14904 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
14906 (insert_cond_bb): Update profile.
14907 * tree-cfg.h (insert_cond_bb): Update prototype.
14908 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
14909 * tree-dump.c: Do not include tree-cfg.
14911 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14913 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
14915 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14917 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
14920 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14922 * tree-complex.c (expand_complex_div_wide): update profile.
14924 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14925 Alan Hayward <alan.hayward@arm.com>
14926 David Sherwood <david.sherwood@arm.com>
14928 * Makefile.in (MACHMODE_H): Remove insn-modes.h
14929 (CORETYPES_H): New define.
14930 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
14931 (insn-modes-inline.h, s-modes-inline-h): New rules.
14932 (generated_files): Add insn-modes-inline.h.
14933 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
14934 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
14935 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
14936 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
14937 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
14938 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
14939 (build/gencodes.o, build/genconditions.o): Likewise.
14940 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
14941 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
14942 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
14943 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
14944 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
14945 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
14946 * coretypes.h: Include everything up to real.h for generators.
14947 Include insn-modes.h first. Include wide-int-print.h after
14948 wide-int.h. Include insn-modes-inline.h and then machmode.h.
14949 * machmode.h: Don't include insn-modes.h here.
14950 * function-tests.c: Remove includes of signop.h, machmode.h,
14951 double-int.h and wide-int.h.
14953 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
14955 * optc-save-gen.awk: Likewise.
14956 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
14957 * godump.c: Remove include of wide-int-print.h.
14958 * pretty-print.h: Likewise.
14959 * wide-int-print.cc: Likewise.
14960 * wide-int.cc: Likewise.
14961 * hash-map-tests.c: Remove include of signop.h.
14962 * hash-set-tests.c: Likewise.
14963 * rtl-tests.c: Likewise.
14964 * mkconfig.sh: Remove include of machmode.h.
14965 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
14967 (emit_insn_modes_inline_h): ...this new function. Emit the code
14968 into an insn-modes-inline.h header file, adding appropriate
14969 include guards and end comments.
14970 (emit_insn_modes_c_header): Remove include of machmode.h.
14971 (emit_min_insn_modes_c_header): Include coretypes.h rather than
14973 (main): Handle -i flag and call emit_insn_modes_inline_h when
14976 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14978 * tree-ssa-strlen.c (strinfo): Rename the length field to
14979 nonzero_chars. Add a full_string_p field.
14980 (compare_nonzero_chars, zero_length_string_p): New functions.
14981 (get_addr_stridx): Add an offset_out parameter.
14982 Use compare_nonzero_chars.
14983 (get_stridx): Update accordingly. Use compare_nonzero_chars.
14984 (new_strinfo): Update after above changes to strinfo.
14985 (set_endptr_and_length): Set full_string_p.
14986 (get_string_length): Update after above changes to strinfo.
14987 (unshare_strinfo): Update call to new_strinfo.
14988 (maybe_invalidate): Likewise.
14989 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
14990 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
14991 as a uhwi instead of an shwi. Update after above changes to
14992 strinfo and new_strinfo.
14993 (zero_length_string): Assert that chainsi contains full strings.
14994 Use zero_length_string_p. Update call to new_strinfo.
14995 (adjust_related_strinfos): Update after above changes to strinfo.
14996 Copy full_string_p from origsi.
14997 (adjust_last_stmt): Use zero_length_string_p.
14998 (handle_builtin_strlen): Update after above changes to strinfo and
14999 new_strinfo. Install the lhs as the string length if the previous
15000 entry didn't describe a full string.
15001 (handle_builtin_strchr): Update after above changes to strinfo
15003 (handle_builtin_strcpy): Likewise.
15004 (handle_builtin_strcat): Likewise.
15005 (handle_builtin_malloc): Likewise.
15006 (handle_pointer_plus): Likewise.
15007 (handle_builtin_memcpy): Likewise. Track nonzero characters
15008 that aren't necessarily followed by a nul terminator.
15009 (handle_char_store): Likewise.
15011 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15013 PR tree-optimization/80769
15014 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
15015 for malloc and calloc. Document the new invariant that all related
15016 strinfos have delayed lengths or none do.
15017 (verify_related_strinfos): Move earlier in file.
15018 (set_endptr_and_length): New function, split out from...
15019 (get_string_length): ...here. Also set the lengths of related
15021 (zero_length_string): Assert that chainsi has known (rather than
15023 (adjust_related_strinfos): Likewise.
15025 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15027 PR tree-optimization/81136
15028 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
15029 assert that two references with the same misalignment have the same
15030 compile-time misalignment if those compile-time misalignments
15033 2017-07-01 Andi Kleen <ak@linux.intel.com>
15035 * print-tree.c (print_node): Print all attributes.
15037 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15039 * cfg.c (scale_bbs_frequencies): New function.
15040 * cfg.h (scale_bbs_frequencies): Declare it.
15041 * cfgloopanal.c (single_likely_exit): Cleanup.
15042 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
15044 (scale_loop_profile): Likewise.
15045 (loop_version): Likewise.
15046 (create_empty_loop_on_edge): Update.
15047 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
15048 scale_loop_frequencies, scale_loop_profile, loopify,
15049 loop_version): Update prototypes.
15050 * modulo-sched.c (sms_schedule): Update.
15051 * predict.c (unlikely_executed_edge_p): Also check probability.
15052 (probably_never_executed_edge_p): Fix typo.
15053 * tree-if-conv.c (version_loop_for_if_conversion): Update.
15054 * tree-parloops.c (gen_parallel_loop): Update.
15055 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
15056 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
15057 * tree-ssa-loop-split.c (split_loop): Update.
15058 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
15059 * tree-vect-loop-manip.c (vect_do_peeling): Update.
15060 (vect_loop_versioning): Update.
15061 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
15063 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15065 * trans-mem.c (split_bb_make_tm_edge): Update profile.
15067 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15069 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
15070 to keep profile consistent.
15072 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15074 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
15075 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
15076 * profile-count.h (max_safe_multiplier): Make unsigned.
15077 (profile_count::guessed_zero): New.
15079 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15081 * bb-reorder.c (fix_up_crossing_landing_pad,
15082 fix_crossing_conditional_branches): Use make_single_succ_edge
15083 to keep profile consistent.
15085 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15087 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
15090 2017-07-01 Jakub Jelinek <jakub@redhat.com>
15093 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
15094 the right scopes, make sure cond_jump isn't preserved between multiple
15095 iterations. Search for fallthru edge whenever there are 3+ edges and
15096 use find_fallthru_edge for it.
15098 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15100 Patch by Alexander Monakov <amonakov@ispras.ru>
15101 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
15102 probabilities consistently.
15104 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15106 * pa.c (pa_expand_compare_and_swap_loop): Update call of
15107 emit_cmp_and_jump_insns.
15109 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15112 * tree-inline.c (expand_call_inline): Combine profile statuses.
15114 2017-06-30 Andrew Pinski <apinski@cavium.com>
15116 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
15117 fold_stmt returned true.
15119 2017-06-30 Nathan Sidwell <nathan@acm.org>
15121 * ggc.h (empty_string): Delete.
15122 * cfgexpand.c (expand_asm_stmt): Use plain "".
15123 * optabs.c (expand_asm_memory_barrier): Likewise.
15124 * stringpool.c (empty_string): Delete.
15125 (digit_vector, digit_string): Delete.
15126 (ggc_alloc_string): Use plain "", don't optimize single digit
15127 strings. Use ggc_alloc_atomic.
15129 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
15131 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
15132 comparison set and one other set, use the cost of the non-comparison
15135 2017-06-30 Nathan Sidwell <nathan@acm.org>
15137 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
15140 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
15142 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
15143 loops. Remove now unneeded calls to gimple_switch_set_label() that
15144 just set removed labels to NULL_TREE.
15146 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
15148 * tree-ssanames.c (set_range_info_raw): Abstract from ...
15149 (set_range_info): ...here. Only call set_range_info_raw if domain
15151 (set_nonzero_bits): Call set_range_info_raw.
15152 * tree-ssanames.h (set_range_info_raw): New.
15154 2017-06-30 Jakub Jelinek <jakub@redhat.com>
15157 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
15158 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
15159 of nonimmediate_operand and <store_mask_constraint> instead of m
15160 for the input operand. For V8FI iterator, always split if input
15161 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
15162 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
15163 <store_mask_predicate> instead of register_operand and
15164 <store_mask_constraint> instead of v for the input operand. Make
15165 sure both operands aren't MEMs for if not <mask_applied>.
15167 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
15169 * lto-wrapper.c (copy_file) Close both file descriptors before
15172 2017-06-30 Martin Liska <mliska@suse.cz>
15175 * multiple_target.c (create_dispatcher_calls): Make ifunc
15176 also for function that don't have calls or are not referenced.
15178 2017-06-30 Richard Biener <rguenther@suse.de>
15180 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
15181 analyze the first scalar stmt. Move vector type computation
15182 for the BB case here from ...
15183 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
15184 live operation processing in the SLP case properly.
15186 2017-06-30 Richard Biener <rguenther@suse.de>
15188 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
15190 2017-06-30 Martin Liska <mliska@suse.cz>
15193 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
15194 before BUILT_IN_UNWIND_RESUME when ASAN is used.
15196 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
15198 * doc/invoke.texi (AArch64): Add missing options and remove redundant
15201 2017-06-30 Richard Biener <rguenther@suse.de>
15203 PR tree-optimization/81249
15204 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
15205 condition reduction result to original scalar type.
15207 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15209 * profile-count.h (enum profile_quality): Fix typos and whitespace
15212 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15214 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
15215 type for branch probabilities.
15217 2017-06-29 Julian Brown <julian@codesourcery.com>
15218 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15220 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
15221 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
15222 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
15223 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
15225 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15227 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
15228 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
15229 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
15230 CC usage from generic code to here.
15231 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
15232 CC usage into the target macros.
15234 2017-06-29 Maya Rashish <coypu@sdf.org>
15236 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
15239 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15241 * arm/arm-builtins.c: Include profile-count.h
15242 * except.c (sjlj_emit_function_enter): Use
15243 profile_probability::unlikely.
15245 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15247 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
15248 and tocrel_offset be pointer args rather than implicitly using
15250 (legitimate_constant_pool_address_p, rs6000_emit_move,
15251 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
15252 tocrel_offset and use in toc_relative_expr_p call.
15253 (print_operand, print_operand_address): Use static tocrel_base_oac
15254 and tocrel_offset_oac.
15255 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
15258 2017-06-29 Maya Rashish <coypu@sdf.org>
15260 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
15262 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
15264 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
15265 objects, take into account only the alignment of 'op0' and 'mode1' if
15268 2017-06-29 Steve Ellcey <sellcey@cavium.com>
15270 * ccmp.c (ccmp_tree_comparison_p): New function.
15271 (ccmp_candidate_p): Update to use above function.
15272 (get_compare_parts): New function.
15273 (expand_ccmp_next): Update to use new functions.
15274 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
15276 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
15277 take mode as argument.
15278 * ccmp.h (expand_ccmp_expr): Add mode as argument.
15279 * expr.c (expand_expr_real_1): Pass mode as argument.
15281 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
15283 * combine.c (combine_instructions): Print insns to dump_file, together
15286 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15288 * asan.c (asan_emit_stack_protection): Update.
15289 (create_cond_insert_point): Update.
15290 * auto-profile.c (afdo_propagate_circuit): Update.
15291 * basic-block.h (struct edge_def): Turn probability to
15292 profile_probability.
15293 (EDGE_FREQUENCY): Update.
15294 * bb-reorder.c (find_traces_1_round): Update.
15295 (better_edge_p): Update.
15296 (sanitize_hot_paths): Update.
15297 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
15298 (make_single_succ_edge): Update.
15299 (check_bb_profile): Update.
15300 (dump_edge_info): Update.
15301 (update_bb_profile_for_threading): Update.
15302 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
15303 probabilitycount to 0.
15304 * cfgbuild.c (compute_outgoing_frequencies): Update.
15305 * cfgcleanup.c (try_forward_edges): Update.
15306 (outgoing_edges_match): Update.
15307 (try_crossjump_to_edge): Update.
15308 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
15309 (expand_gimple_tailcall): Update.
15310 (construct_init_block): Use make_single_succ_edge.
15311 (construct_exit_block): Use make_single_succ_edge.
15312 * cfghooks.c (verify_flow_info): Update.
15313 (redirect_edge_succ_nodup): Update.
15314 (split_edge): Update.
15315 (account_profile_record): Update.
15316 * cfgloopanal.c (single_likely_exit): Update.
15317 * cfgloopmanip.c (scale_loop_profile): Update.
15318 (set_zero_probability): Remove.
15319 (duplicate_loop_to_header_edge): Update.
15320 * cfgloopmanip.h (loop_version): Update prototype.
15321 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
15322 (force_nonfallthru_and_redirect): Update.
15323 (update_br_prob_note): Update.
15324 (rtl_verify_edges): Update.
15325 (purge_dead_edges): Update.
15326 (rtl_lv_add_condition_to_bb): Update.
15327 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
15328 * cgraphunit.c (init_lowered_empty_function): Update.
15329 (cgraph_node::expand_thunk): Update.
15330 * cilk-common.c: Include profile-count.h
15331 * dojump.c (inv): Remove.
15332 (jumpifnot): Update.
15333 (jumpifnot_1): Update.
15334 (do_jump_1): Update.
15336 (do_jump_by_parts_greater_rtx): Update.
15337 (do_compare_rtx_and_jump): Update.
15338 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
15339 do_jump_1. do_compare_rtx_and_jump): Update prototype.
15340 * dwarf2cfi.c: Include profile-count.h
15341 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
15342 (sjlj_emit_dispatch_table): Likewise.
15343 * explow.c: Include profile-count.h
15344 * expmed.c (emit_store_flag_force): Update.
15345 (do_cmp_and_jump): Update.
15346 * expr.c (compare_by_pieces_d::generate): Update.
15347 (compare_by_pieces_d::finish_mode): Update.
15348 (emit_block_move_via_loop): Update.
15349 (store_expr_with_bounds): Update.
15350 (store_constructor): Update.
15351 (expand_expr_real_2): Update.
15352 (expand_expr_real_1): Update.
15353 * expr.h (try_casesi, try_tablejump): Update prototypes.
15354 * gimple-pretty-print.c (dump_probability): Update.
15355 (dump_profile): New.
15356 (dump_gimple_label): Update.
15357 (dump_gimple_bb_header): Update.
15358 * graph.c (draw_cfg_node_succ_edges): Update.
15359 * hsa-gen.c (convert_switch_statements): Update.
15360 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
15361 (find_if_case_1): Update.
15362 (find_if_case_2): Update.
15363 * internal-fn.c (expand_arith_overflow_result_store): Update.
15364 (expand_addsub_overflow): Update.
15365 (expand_neg_overflow): Update.
15366 (expand_mul_overflow): Update.
15367 (expand_vector_ubsan_overflow): Update.
15368 * ipa-cp.c (good_cloning_opportunity_p): Update.
15369 * ipa-split.c (split_function): Use make_single_succ_edge.
15370 * ipa-utils.c (ipa_merge_profiles): Update.
15371 * loop-doloop.c (add_test): Update.
15372 (doloop_modify): Update.
15373 * loop-unroll.c (compare_and_jump_seq): Update.
15374 (unroll_loop_runtime_iterations): Update.
15375 * lra-constraints.c (lra_inheritance): Update.
15376 * lto-streamer-in.c (input_cfg): Update.
15377 * lto-streamer-out.c (output_cfg): Update.
15378 * mcf.c (adjust_cfg_counts): Update.
15379 * modulo-sched.c (sms_schedule): Update.
15380 * omp-expand.c (expand_omp_for_init_counts): Update.
15381 (extract_omp_for_update_vars): Update.
15382 (expand_omp_ordered_sink): Update.
15383 (expand_omp_for_ordered_loops): Update.
15384 (expand_omp_for_generic): Update.
15385 (expand_omp_for_static_nochunk): Update.
15386 (expand_omp_for_static_chunk): Update.
15387 (expand_cilk_for): Update.
15388 (expand_omp_simd): Update.
15389 (expand_omp_taskloop_for_outer): Update.
15390 (expand_omp_taskloop_for_inner): Update.
15391 * omp-simd-clone.c (simd_clone_adjust): Update.
15392 * optabs.c (expand_doubleword_shift): Update.
15393 (expand_abs): Update.
15394 (emit_cmp_and_jump_insn_1): Update.
15395 (expand_compare_and_swap_loop): Update.
15396 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
15397 * predict.c (predictable_edge_p): Update.
15398 (edge_probability_reliable_p): Update.
15399 (set_even_probabilities): Update.
15400 (combine_predictions_for_insn): Update.
15401 (combine_predictions_for_bb): Update.
15402 (propagate_freq): Update.
15403 (estimate_bb_frequencies): Update.
15404 (force_edge_cold): Update.
15405 * profile-count.c (profile_count::dump): Add missing space into dump.
15406 (profile_count::debug): Add newline.
15407 (profile_count::differs_from_p): Explicitly convert to unsigned.
15408 (profile_count::stream_in): Update.
15409 (profile_probability::dump): New member function.
15410 (profile_probability::debug): New member function.
15411 (profile_probability::differs_from_p): New member function.
15412 (profile_probability::differs_lot_from_p): New member function.
15413 (profile_probability::stream_in): New member function.
15414 (profile_probability::stream_out): New member function.
15415 * profile-count.h (profile_count_quality): Rename to ...
15416 (profile_quality): ... this one.
15417 (profile_probability): New.
15418 (profile_count): Update.
15419 * profile.c (compute_branch_probabilities): Update.
15420 * recog.c (peep2_attempt): Update.
15421 * sched-ebb.c (schedule_ebbs): Update.
15422 * sched-rgn.c (find_single_block_region): Update.
15423 (compute_dom_prob_ps): Update.
15424 (schedule_region): Update.
15425 * sel-sched-ir.c (compute_succs_info): Update.
15426 * stmt.c (struct case_node): Update.
15427 (do_jump_if_equal): Update.
15428 (get_outgoing_edge_probs): Update.
15429 (conditional_probability): Update.
15430 (emit_case_dispatch_table): Update.
15431 (expand_case): Update.
15432 (expand_sjlj_dispatch_table): Update.
15433 (emit_case_nodes): Update.
15434 * targhooks.c: Update.
15435 * tracer.c (better_p): Update.
15436 (find_best_successor): Update.
15437 * trans-mem.c (expand_transaction): Update.
15438 * tree-call-cdce.c: Update.
15439 * tree-cfg.c (gimple_split_edge): Upate.
15440 (move_sese_region_to_fn): Upate.
15441 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
15442 * tree-eh.c (lower_resx): Upate.
15443 (cleanup_empty_eh_move_lp): Upate.
15444 * tree-if-conv.c (version_loop_for_if_conversion): Update.
15445 * tree-inline.c (copy_edges_for_bb): Update.
15446 (copy_cfg_body): Update.
15447 * tree-parloops.c (gen_parallel_loop): Update.
15448 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
15449 (gimple_gen_time_profiler): Update.
15450 * tree-ssa-dce.c (remove_dead_stmt): Update.
15451 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
15452 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
15453 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
15454 (unloop_loops): Update.
15455 (try_peel_loop): Update.
15456 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
15457 * tree-ssa-loop-split.c (connect_loops): Update.
15458 (split_loop): Update.
15459 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
15460 (hoist_guard): Update.
15461 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
15462 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
15463 (value_replacement): Update.
15464 * tree-ssa-reassoc.c (branch_fixup): Update.
15465 * tree-ssa-tail-merge.c (replace_block_by): Update.
15466 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
15467 (create_edge_and_update_destination_phis): Update.
15468 (compute_path_counts): Update.
15469 (recompute_probabilities): Update.
15470 (update_joiner_offpath_counts): Update.
15471 (freqs_to_counts_path): Update.
15472 (duplicate_thread_path): Update.
15473 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
15474 (struct switch_conv_info): Update.
15475 (gen_inbound_check): Update.
15476 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
15477 (vect_do_peeling): Update.
15478 (vect_loop_versioning): Update.
15479 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
15480 (optimize_mask_stores): Update.
15481 * ubsan.c (ubsan_expand_null_ifn): Update.
15482 * value-prof.c (gimple_divmod_fixed_value): Update.
15483 (gimple_divmod_fixed_value_transform): Update.
15484 (gimple_mod_pow2): Update.
15485 (gimple_mod_pow2_value_transform): Update.
15486 (gimple_mod_subtract): Update.
15487 (gimple_mod_subtract_transform): Update.
15488 (gimple_ic): Update.
15489 (gimple_stringop_fixed_value): Update.
15490 (gimple_stringops_transform): Update.
15491 * value-prof.h: Update.
15493 2017-06-29 Carl Love <cel@us.ibm.com>
15495 * config/rs6000/rs6000-c.c: Add support for built-in functions
15496 vector signed int vec_signed (vector float);
15497 vector signed long long vec_signed (vector double);
15498 vector signed int vec_signed2 (vector double, vector double);
15499 vector signed int vec_signede (vector double);
15500 vector signed int vec_signedo (vector double);
15501 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
15502 instruction generator.
15503 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
15504 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
15505 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
15507 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
15508 vunsignede_v2df): Add define_expands.
15509 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
15510 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
15511 VEC_UNSIGNEDO): Add definitions.
15512 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
15513 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
15514 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
15515 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
15516 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
15517 * config/rs6000/altivec.h (vec_signed, vec_signed2,
15518 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
15519 vec_unsignede, vec_unsignedo): Add builtin defines.
15520 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
15522 * doc/extend.texi: Update the built-in documentation file for the
15523 new built-in functions.
15525 2017-06-29 Richard Biener <rguenther@suse.de>
15527 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
15528 reduction chains to LOOP_VINFO_REDUCTIONS.
15529 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
15530 SLP reductions after processing reduction chains.
15532 2017-06-29 Nathan Sidwell <nathan@acm.org>
15534 * builtins.c (fold_builtin_FUNCTION): Use
15535 lang_hooks.decl_printable_name.
15537 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
15539 PR middle-end/81194
15540 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
15541 with only one label.
15542 * stmt.c (expand_case): Assert NCASES is greater than one.
15544 2017-06-29 Richard Biener <rguenther@suse.de>
15546 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
15548 (group_case_labels): Likewise.
15549 (find_taken_edge): Push sanity checking on val to workers...
15550 (find_taken_edge_cond_expr): ... here
15551 (find_taken_edge_switch_expr): ... and here, handle cases
15552 with just a default label.
15553 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
15554 (group_case_labels): Likewise.
15555 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
15556 group_case_labels does anything cleanup the CFG again.
15558 2017-06-29 Bin Cheng <bin.cheng@arm.com>
15560 PR tree-optimization/81196
15561 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
15562 exit condition comparing two IVs.
15564 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
15566 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
15567 profile to the dummy entry at the end of the list of architectures.
15568 * config/arm/arm-cpu-cdata.h: Regenerated.
15570 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15571 Michael Collison <michael.collison@arm.com>
15574 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
15576 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
15577 (*aarch64_reg_<mode>3_minus_mask): New pattern.
15578 (*aarch64_<optab>_reg_di3_mask2): New pattern.
15579 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
15580 of shift when the shift amount is masked with constant equal to
15581 the size of the mode.
15582 * config/aarch64/predicates.md (subreg_lowpart_operator): New
15585 2017-06-29 Martin Liska <mliska@suse.cz>
15587 * config/i386/i386.opt: Change range from [1,5] to [0,5].
15589 2017-06-29 Yury Gribov <tetra2005@gmail.com>
15592 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
15595 (edge_growth_cache_entry::edge_growth_cache_entry): New
15597 (reset_edge_growth_cache): Update to use constructor.
15599 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15601 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
15602 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
15603 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
15605 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
15607 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
15608 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
15610 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
15612 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
15613 (*-linux-uclibc*): Add t-uclibc tmake_file.
15614 * config/t-musl: New.
15615 * config/t-uclibc: New.
15617 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
15619 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
15621 (gen_comm_data): Emit architectural setting of arch_prof.
15622 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
15624 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
15625 (armv8-m.base, armv8-m.main): Likewise.
15626 * arm-protos.h (arm_build_target): Add profile field.
15627 (arch_option): Likewise.
15628 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
15630 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
15631 arm_active_target.profile.
15633 2017-06-28 Richard Biener <rguenther@suse.de>
15635 PR middle-end/81227
15636 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
15637 TYPE_OVERFLOW_WRAPS.
15638 * match.pd (negate_expr_p): Likewise.
15639 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
15640 fold_build2, not fold_binary.
15642 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
15644 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
15645 Convert memory address to Pmode.
15646 (aarch64_print_operand): Assert MEM operands are always Pmode.
15648 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
15651 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
15652 Remove redundant if.
15653 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
15654 * config/arm/aarch-common-protos.h
15655 (aarch_forward_to_shift_is_not_shifted_re): Remove.
15656 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
15658 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
15661 * multiple_target.c (create_dispatcher_calls): Set the default
15662 clone to be static, not public.
15664 2017-06-28 Richard Biener <rguenther@suse.de>
15666 * tree-vect-loop.c (vectorizable_reduction): Move special
15667 cond reduction IV var creation ...
15668 (vect_create_epilog_for_reduction): ... here. Remove induction_index
15669 parameter. Use STMT_VINFO_VECTYPE.
15670 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
15673 2017-06-28 Martin Liska <mliska@suse.cz>
15676 * ipa-visibility.c (non_local_p): Handle visibility.
15678 2017-06-28 Martin Liska <mliska@suse.cz>
15681 * common.opt: Add IntegerRange to various options.
15682 * opt-functions.awk (integer_range_info): New function.
15683 * optc-gen.awk: Add integer_range_info to cl_options struct.
15684 * opts-common.c (decode_cmdline_option): Handle
15685 CL_ERR_INT_RANGE_ARG.
15686 (cmdline_handle_error): Likewise.
15687 * opts.c (print_filtered_help): Show valid interval in
15688 when --help is provided.
15689 * opts.h (struct cl_option): Add range_min and range_max fields.
15690 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
15692 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
15694 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
15695 (x * C EQ/NE y * C): New transformation.
15697 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
15699 * genmultilib (combination_space): Accept '+' in option names.
15701 2017-06-28 Martin Liska <mliska@suse.cz>
15704 * asan.c (instrument_derefs): Bail out inner references
15705 that are hard register variables.
15707 2017-06-28 Jakub Jelinek <jakub@redhat.com>
15710 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
15711 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
15713 2017-06-28 Richard Biener <rguenther@suse.de>
15715 * tree-vectorizer.h (vect_get_vec_defs): Remove.
15716 (vect_get_slp_defs): Adjust.
15717 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
15719 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
15721 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
15722 get_initial_defs_for_reduction instead of vect_get_vec_defs.
15723 (vectorizable_reduction): Adjust.
15724 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
15726 (vect_get_slp_defs): Likewise.
15727 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
15728 (vectorizable_bswap): Adjust.
15729 (vectorizable_call): Likewise.
15730 (vectorizable_conversion): Likewise.
15731 (vectorizable_assignment): Likewise.
15732 (vectorizable_shift): Likewise.
15733 (vectorizable_operation): Likewise.
15734 (vectorizable_store): Likewise.
15735 (vectorizable_condition): Likewise.
15736 (vectorizable_comparison): Likewise.
15738 2017-06-28 Michael Collison <michael.collison@arm.com>
15741 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
15743 (arm_gen_movmemqi): Removed unused variable 'i'.
15744 Convert 'for' loop into 'while' loop.
15745 (arm_expand_prologue): Remove last unnecessary set of insn.
15746 (thumb_pop): Remove unused variable 'pushed_words'.
15747 (thumb_exit): Remove last unnecessary set of regs_to_pop.
15749 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15751 * config/s390/predicates.md: Use s390_rel_address_ok_p.
15752 * config/s390/s390-protos.h: Add prototype of
15753 s390_rel_address_ok_p.
15754 * config/s390/s390.c (s390_got_symbol): New function.
15755 (s390_rel_address_ok_p): New function.
15756 (legitimize_pic_address): Use s390_rel_address_ok_p.
15757 (s390_load_got): Use s390_got_symbol.
15758 (s390_option_override): Issue error if
15759 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
15760 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
15762 * config/s390/s390.opt: New option mpic-data-is-text-relative.
15764 2017-06-27 Andrew Pinski <apinski@cavium.com>
15766 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
15767 (X * copysign (1.0, X)): New pattern.
15768 (X * copysign (1.0, -X)): New pattern.
15769 (copysign (-1.0, CST)): New pattern.
15771 2017-06-27 Joseph Myers <joseph@codesourcery.com>
15773 * genmultilib (combination_space): Remove variable.
15774 Validate reuse rules against regular expression for any sequence
15775 of multilib options in any order.
15777 2017-06-27 Michael Collison <michael.collison@arm.com>
15779 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
15780 call aarch64_split_simd_combine.
15781 * (aarch64_combine_internal<mode>): Delete pattern.
15782 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
15783 Allow register and subreg operands.
15785 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15787 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
15788 specific need, just fallback on defaults.
15789 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
15791 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15792 Olivier Hainque <hainque@adacore.com>
15794 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
15796 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
15797 targets. Pick a default if no particular attempt applied.
15798 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
15801 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15803 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
15804 (x86_64-wrs-vxworks7): Likewise.
15806 2017-06-27 Marek Polacek <polacek@redhat.com>
15809 * ubsan.c (instrument_null): Check get_base_address's result for null.
15811 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
15813 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
15815 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
15817 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
15818 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
15819 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
15820 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
15821 New function types.
15822 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
15823 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
15824 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
15825 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
15826 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
15827 BUILT_IN_FEUPDATEENV): New builtins.
15828 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
15829 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
15830 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
15831 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
15833 (builtin_structptr_types): Adjust size.
15834 * tree.c (builtin_structptr_types): Add four entries.
15836 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15837 Olivier Hainque <hainque@adacore.com>
15839 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
15840 (TLS_SYM): New local macro, forcing reference to __tls__ on
15841 link command lines for VxWorks 7 RTPs, triggering initialization
15843 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
15844 OS features TLS support, true for RTPs on VxWorks 7.
15845 * config/vxworks.c (vxworks_override_options): Setup emutls
15848 2017-06-27 Jakub Jelinek <jakub@redhat.com>
15850 * predict.c (test_prediction_value_range): Use -1U instead of -1
15851 to avoid narrowing conversion warning.
15852 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
15853 to avoid narrowing conversion warning.
15854 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
15856 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
15858 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15860 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
15861 64bit configurations.
15862 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
15863 (SIZE_TYPE): Likewise.
15864 * config/vxworks.c (vxworks_emutls_var_fields): Use
15865 long_unsigned_type_node instead of unsigned_type_node as the offset
15866 field type, which is "pointer" mode in emutls.c.
15868 2017-06-27 Jakub Jelinek <jakub@redhat.com>
15871 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
15873 PR middle-end/81207
15874 * gimple-fold.c (replace_call_with_call_and_fold): Handle
15875 gimple_vuse copying separately from gimple_vdef copying.
15877 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15879 * value-prof.c (free_hist): Remove call to memset and the enclosing if
15882 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
15883 Olivier Hainque <hainque@adacore.com>
15885 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
15886 for all vxworks7 targets.
15887 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
15888 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
15889 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
15890 variations for VX6/VX7 and 32/64bits later on in ...
15891 (VXWORKS_LIB_SPEC): Leverage new macros.
15892 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
15893 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
15895 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
15897 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
15898 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
15900 2017-06-26 Carl Love <cel@us.ibm.com>
15902 * config/rs6000/rs6000-c.c: Add support for built-in functions
15903 vector bool char vec_reve (vector bool char);
15904 vector signed char vec_reve (vector signed char);
15905 vector unsigned char vec_reve (vector unsigned char);
15906 vector bool int vec_reve (vector bool int);
15907 vector signed int vec_reve (vector signed int);
15908 vector unsigned int vec_reve (vector unsigned int);
15909 vector bool long long vec_reve (vector bool long long);
15910 vector signed long long vec_reve (vector signed long long);
15911 vector unsigned long long vec_reve (vector unsigned long long);
15912 vector bool short vec_reve (vector bool short);
15913 vector signed short vec_reve (vector signed short);
15914 vector double vec_reve (vector double);
15915 vector float vec_reve (vector float);
15916 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
15917 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
15918 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
15919 (altivec_vreve): New pattern.
15920 * config/rs6000/altivec.h (vec_reve): New define.
15921 * doc/extend.texi (vec_rev): Update the built-in documentation file
15922 for the new built-in functions.
15924 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15926 PR tree-optimization/71815
15927 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
15929 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
15931 (slsr_process_phi): Likewise.
15932 (replace_uncond_cands_and_profitable_phis): Don't replace a
15933 multiply candidate with a stride of 1 (copy or cast).
15934 (phi_incr_cost): Call uses_consumed_by_stmt rather than
15936 (lowest_cost_path): Likewise.
15937 (total_savings): Likewise.
15939 2017-06-26 Richard Biener <rguenther@suse.de>
15942 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
15943 Use def_builtin_pure for all gather builtins.
15945 2017-06-26 Richard Biener <rguenther@suse.de>
15947 PR tree-optimization/81203
15948 * tree-tailcall.c (find_tail_calls): Do not move stmts into
15949 non-dominating BBs.
15951 2017-06-26 Marek Polacek <polacek@redhat.com>
15954 * doc/invoke.texi: Document -Wmultistatement-macros.
15956 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
15958 * doc/sourcebuild.texi (ARM-specific attributes): Document new
15959 arm_neon_ok_no_float_abi effective target.
15961 2017-06-26 Richard Biener <rguenther@suse.de>
15963 PR tree-optimization/80928
15964 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
15965 (copy_bbs): Set BB_DUPLICATED flag early.
15966 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
15968 (execute_on_shrinking_pred): Likewise.
15969 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
15970 BB_DUPLICATED blocks.
15971 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
15972 iterate over all PHIs considering removal of *gsi.
15974 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
15976 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
15979 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15981 * config/rs6000/rs6000-string.c: (expand_block_clear,
15982 do_load_for_compare, select_block_compare_mode,
15983 compute_current_alignment, expand_block_compare,
15984 expand_strncmp_align_check, expand_strn_compare,
15985 expand_block_move, rs6000_output_load_multiple)
15986 Move functions related to string/block move/compare
15987 to a separate file.
15988 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
15989 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
15990 for this function which is now used in two files.
15991 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
15992 * config.gcc: Add rs6000-string.o to extra_objs for
15993 targets powerpc*-*-* and rs6000*-*-*.
15995 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
15998 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
15999 32-bit, since indexed is not valid for DImode.
16000 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
16001 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
16002 (define_peephole2 for Altivec d-form load): Add 32-bit support.
16003 (define_peephole2 for Altivec d-form store): Likewise.
16006 * multiple_target.c (create_dispatcher_calls): Only create the
16007 dispatcher call if the function is the default clone of a
16008 versioned function.
16010 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
16012 PR middle-end/80902
16013 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
16014 a call, force the call to not be a tail call.
16016 2017-06-23 Jeff Law <law@redhat.com>
16018 * doc/contrib.texi: Add entry for Steven Pemberton's work on
16021 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
16023 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
16024 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
16025 handling for early expansion of vector shifts (sl,sr,sra,rl).
16026 (builtin_function_type): Add vector shift right instructions
16027 to the unsigned argument list.
16029 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
16031 rtl-optimizatoin/79286
16032 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
16033 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
16034 trap. PIC register plus a const unspec without offset can never trap.
16036 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
16038 * tree.h (builtin_structptr_type): New type.
16039 (builtin_structptr_types): Declare new array.
16040 * tree.c (builtin_structptr_types): New array.
16041 (free_lang_data, build_common_tree_nodes): Use it.
16043 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
16046 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
16049 2017-06-22 Matt Turner <mattst88@gmail.com>
16051 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
16052 Lake models to skylake case. Assume skylake for unknown
16053 models with clflushopt.
16055 2017-06-22 Jeff Law <law@redhat.com>
16057 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
16058 frame sizes that do not satisfy aarch64_uimm12_shift.
16060 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
16062 * profile-count.h (apply_probability,
16063 apply_scale, probability_in): Fix checks for zero.
16065 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16067 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
16068 * doc/cppdiropts.texi (-I @var{dir}): Document it.
16070 2016-06-22 Richard Biener <rguenther@suse.de>
16072 * tree-vect-loop.c (vect_model_reduction_cost): Handle
16073 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
16074 REDUC_MAX_EXPR support.
16075 (vectorizable_reduction): Likewise.
16076 (vect_create_epilog_for_reduction): Likewise.
16078 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
16080 * match.pd (A / (1 << B) -> A >> B): New.
16081 * generic-match-head.c: Include optabs-tree.h.
16082 * gimple-match-head.c: Likewise.
16083 * optabs-tree.h (target_supports_op_p): New.
16084 * optabs-tree.c (target_supports_op_p): New.
16086 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16088 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
16089 $gcc_cv_ld --help output.
16090 (gcc_cv_ld_demangle): Likewise.
16091 (gcc_cv_ld_eh_frame_hdr): Likewise.
16092 (gcc_cv_ld_pie): Likewise.
16093 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
16094 (gcc_cv_ld_buildid): Likewise.
16095 (gcc_cv_ld_sysroot): Likewise.
16096 (ld_bndplt_support): Likewise.
16097 (ld_pushpopstate_support): Likewise.
16098 * configure: Regenerate.
16099 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
16101 2017-06-21 Jakub Jelinek <jakub@redhat.com>
16104 * config/i386/sse.md (round<mode>2): Renumber match_dup and
16105 operands indexes to avoid gap between operands and match_dups.
16107 2017-06-21 Andrew Pinski <apinski@cavium.com>
16109 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
16110 Increment Arith_shift and Arith_shift_reg by 1.
16111 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
16113 * config/aarch64/aarch64.c (thunderx_tunings): Enable
16114 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
16115 (aarch64_strip_extend): Add new argument and test for it.
16116 (aarch64_cheap_mult_shift_p): New function.
16117 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
16118 add a cost if it is true.
16119 Update calls to aarch64_strip_extend.
16120 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
16122 2017-06-21 Andrew Pinski <apinski@cavium.com>
16124 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
16126 (thunderxt88): Likewise.
16127 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
16128 (thunderx_prefetch_tune): New variable.
16129 (thunderx2t99_prefetch_tune): Update for the correct values.
16130 (thunderxt88_tunings): New variable.
16131 (thunderx_tunings): Use thunderx_prefetch_tune instead of
16132 generic_prefetch_tune.
16133 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
16135 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16137 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
16138 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
16139 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
16140 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
16141 (aarch64_atomic_cas<mode>, GPI): Likewise.
16143 2017-06-21 Martin Liska <mliska@suse.cz>
16145 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
16146 statements on cold and hot labels.
16147 * predict.c (tree_estimate_probability_bb): Remove the
16148 prediction from this place.
16150 2017-06-21 Martin Liska <mliska@suse.cz>
16152 PR tree-optimization/79489
16153 * gimplify.c (maybe_add_early_return_predict_stmt): New
16155 (gimplify_return_expr): Call the function.
16156 * predict.c (tree_estimate_probability_bb): Remove handling
16158 * predict.def: Update comment about early return predictor.
16159 * gimple-predict.h (is_gimple_predict): New function.
16160 * predict.def: Change default value of early return to 66.
16161 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
16163 * passes.def: Put pass_strip_predict_hints to the beginning of
16166 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
16168 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
16169 FUNCTION_DECL declarations.
16170 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
16172 (dwaf2out_decl): Likewise.
16173 * godump.c (go_early_global_decl): Skip call to the real debug hook
16174 for FUNCTION_DECL declarations.
16175 * passes.c (rest_of_decl_compilation): Skip call to the
16176 early_global_decl debug hook for FUNCTION_DECL declarations, unless
16177 -fdump-go-spec is passed.
16179 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
16181 * config/i386/i386.c (struct builtin_isa): New field pure_p.
16182 Reorder for compactness.
16183 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
16184 (def_builtin_pure, def_builtin_pure2): New functions.
16185 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
16187 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
16189 * match.pd (nop_convert): New predicate.
16190 ((A +- CST1) +- CST2): Allow some NOP conversions.
16192 2017-06-21 Jakub Jelinek <jakub@redhat.com>
16195 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
16196 with ctors/dtors if GOVD_SHARED is set.
16198 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16200 * config/aarch64/aarch64.md (movti_aarch64):
16201 Emit mov rather than orr.
16202 (movtf_aarch64): Likewise.
16203 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
16204 Emit mov rather than orr.
16206 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16208 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
16209 Swap alternatives, make integer dup more expensive.
16211 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16213 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
16214 Return true for non-tls symbols.
16216 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
16218 * config/aarch64/aarch64-cores.def (cortex-a55): New.
16219 (cortex-a75): Likewise.
16220 (cortex-a75.cortex-a55): Likewise.
16221 * config/aarch64/aarch64-tune.md: Regenerate.
16222 * doc/invoke.texi (-mtune): Document new values for -mtune.
16224 2017-06-21 Tom de Vries <tom@codesourcery.com>
16226 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
16227 stack_size feature.
16228 (Effective-Target Keywords, Other attributes): Suggest using
16229 dg-add-options stack_size feature to get stack limit in stack_size
16230 effective target documentation.
16232 2017-06-21 Julian Brown <julian@codesourcery.com>
16233 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16235 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
16236 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
16237 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
16239 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
16240 attribute type list for neon_multiply.
16241 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
16242 attribute type list for neon_multiply.
16243 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
16244 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
16245 attribute type list for neon_multiply.
16246 * config/arm/types.md (crypto_pmull): Add.
16247 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
16248 attribute type list.
16250 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
16252 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
16255 2017-06-20 Jakub Jelinek <jakub@redhat.com>
16257 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
16258 to make sure not to dereference a NULL cost_classes_ptr pointer.
16260 2017-06-20 Carl Love <cel@us.ibm.com>
16262 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16263 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16264 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16265 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16266 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16267 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16268 VMULOSW): New enum "unspec" values.
16269 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16270 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16271 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16272 altivec_vmulosw): New patterns.
16273 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16274 VMULOSW): Add definitions.
16276 2017-06-20 Julia Koval <julia.koval@intel.com>
16278 * config/i386/i386.c: Fix rounding expand for new pattern.
16279 * config/i386/subst.md: Fix pattern (parallel -> unspec).
16281 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16283 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
16284 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
16286 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16288 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
16291 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16293 * config/aarch64/aarch64-cores.def: Rearrange to sort by
16294 architecture, then by implementer ID.
16295 * config/aarch64/aarch64-tune.md: Regenerate.
16297 2017-06-20 Richard Biener <rguenther@suse.de>
16299 PR middle-end/81097
16300 * fold-const.c (split_tree): Fold to type before negating.
16302 2017-06-20 David Malcolm <dmalcolm@redhat.com>
16304 * diagnostic-show-locus.c
16305 (selftest::test_fixit_deletion_affecting_newline): New function.
16306 (selftest::diagnostic_show_locus_c_tests): Call it.
16308 2017-06-20 Andreas Schwab <schwab@suse.de>
16311 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
16314 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16316 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
16317 __ARM_FEATURE_COPROC according to support.
16319 2017-06-20 Jakub Jelinek <jakub@redhat.com>
16321 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
16322 Rewritten to avoid overflow for > 32-bit pointers.
16325 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
16326 by removing enum keyword.
16327 (ubsan_type_descriptor): Likewise. Formatting fix.
16330 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
16331 splitter): Require TARGET_SSE2 in the condition.
16333 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
16336 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
16337 for doing vector set of SFmode on ISA 3.0.
16338 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
16339 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
16341 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
16342 SFmode value into a V4SF variable that was extracted from another
16343 V4SF variable without converting the element to double precision
16344 and back to single precision vector format.
16345 (vsx_insert_extract_v4sf_p9_2): Likewise.
16347 2017-06-19 Jakub Jelinek <jakub@redhat.com>
16349 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
16350 in UWHI to avoid undefined overflow.
16353 * ubsan.h (enum ubsan_encode_value_phase): New.
16354 (ubsan_encode_value): Change second argument to
16355 enum ubsan_encode_value_phase with default value of
16356 UBSAN_ENCODE_VALUE_GENERIC.
16357 * ubsan.c (ubsan_encode_value): Change second argument to
16358 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
16359 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
16360 create_tmp_var_raw instead of create_tmp_var and use a
16362 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
16363 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
16364 ubsan_encode_value callers.
16367 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
16368 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
16369 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
16371 2017-06-19 Richard Biener <rguenther@suse.de>
16373 PR middle-end/81118
16374 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
16375 estimates if we changed anything.
16377 2017-06-19 Richard Biener <rguenther@suse.de>
16379 PR tree-optimization/80887
16380 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
16381 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
16382 simplified lookups, then reset mprts_hook.
16383 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
16385 (try_to_simplify): Likewise.
16387 2017-06-19 Martin Liska <mliska@suse.cz>
16390 * gimplify.c (gimplify_switch_expr):
16391 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
16393 2017-06-19 Martin Liska <mliska@suse.cz>
16395 * doc/install.texi: Document that PGO runs in 4 stages.
16397 2017-06-19 Martin Liska <mliska@suse.cz>
16400 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
16401 to dispatcher function name.
16402 * multiple_target.c (replace_function_decl): New function.
16403 (create_dispatcher_calls): Redirect both edges and references.
16405 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
16407 * profile-count.c (profile_count::dump): Dump quality.
16408 (profile_count::differs_from_p): Update for unsigned val.
16409 * profile-count.h (profile_count_quality): New enum.
16410 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
16412 2017-06-19 Richard Biener <rguenther@suse.de>
16414 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
16415 struct function as arg.
16416 (estimate_numbers_of_iterations): Export overload with loop arg.
16417 (free_numbers_of_iterations_estimates_loop): Use an overload of
16418 free_numbers_of_iterations_estimates instead.
16419 * tree-cfg.c (remove_bb): Adjust.
16420 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
16421 * tree-parloops.c (gen_parallel_loop): Likewise.
16422 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
16424 (tree_unroll_loops_completely): Likewise.
16425 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
16426 Use an overload instead and export.
16427 (estimated_loop_iterations): Adjust.
16428 (max_loop_iterations): Likewise.
16429 (likely_max_loop_iterations): Likewise.
16430 (estimate_numbers_of_iterations): Take struct function as arg
16432 (loop_exits_before_overflow): Adjust.
16433 (free_numbers_of_iterations_estimates_loop): Use an overload.
16434 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
16435 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
16437 2017-06-19 Richard Biener <rguenther@suse.de>
16440 * ipa-prop.c (find_constructor_constant_at_offset): Handle
16441 RANGE_EXPR conservatively.
16443 2017-06-16 Carl Love <cel@us.ibm.com>
16445 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16446 definitions for vec_float, vec_float2, vec_floato,
16447 vec_floate built-ins.
16448 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
16449 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
16451 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
16452 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
16453 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
16454 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
16455 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
16456 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
16457 vec_floato): Add builtin defines.
16458 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
16459 Update the built-in documentation file for the new built-in
16462 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16464 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
16465 (mthumb): Mark as the negative of -marm.
16467 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16469 * doc/invoke.texi (ARM Options, -mcpu): Document supported
16471 (ARM Options, -mtune): Document that this accepts the same
16472 extension options as -mcpu.
16473 (ARM Options, -mfpu): Document addition of -mfpu=auto.
16475 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16477 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
16478 permitted extensions.
16480 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16482 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
16483 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
16484 (armv8-m.main): Add option +nodsp.
16485 * config/arm/arm-cpu-cdata.h: Regenerated.
16487 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16489 * config/arm/t-fuchsia: New file.
16490 * config.gcc (arm*-*-fuchsia*): Use it.
16492 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16494 * config/arm/t-symbian: Rewrite for new option infrastructure.
16496 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16498 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
16499 (MULTILIB_REQUIRED): Likewise.
16501 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16503 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
16504 (MULTILIB_RESUE): Likewise.
16505 (MULTILIB_MATCHES): Likewise.
16506 (MULTLIB_REQUIRED): Likewise.
16508 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16510 * config/arm/t-rtems: Rewrite for new option framework.
16512 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16514 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
16515 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
16516 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
16517 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
16518 * config/arm/t-multilib: ... here.
16519 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
16520 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
16521 armv7-a and armv8*-a when A-profile libraries have not been built.
16522 * config/arm/t-rmprofile: Rewrite.
16524 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16526 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
16527 with a backslash. Remove the backslash after substituting unescaped
16529 * doc/fragments.texi (MULTILIB_REUSE): Document it.
16531 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16533 * config.gcc: (arm*-*-*): When building a-profile libraries, force
16534 the driver to pass through the default setting of -mfloat-abi.
16535 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
16537 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
16538 (all_feat_combs): New rule.
16539 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
16541 * config/arm/t-aprofile: Rewrite.
16543 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16545 * config/arm/arm.h (FPUTYPE_AUTO): Define.
16546 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
16547 fpu is not specified by the user/command-line.
16548 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
16549 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
16550 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
16551 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
16552 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
16553 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
16555 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16557 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
16558 * config/arm/t-arm-elf: Rewritten.
16560 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16562 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
16563 have some floating-point instructions.
16564 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
16565 (TARGET_MAYBE_HARD_FLOAT): New macro.
16566 * config/arm/arm-builtins.c (arm_init_builtins): Use
16567 TARGET_MAYBE_HARD_FLOAT.
16568 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
16570 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16572 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
16573 (configargs.h): Include it.
16574 (arm_print_hint_for_fpu_option): New function.
16575 (arm_parse_fpu_option): New function.
16576 (candidate_extension): New class.
16577 (arm_canon_for_multilib): New function.
16578 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
16579 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
16580 (ARCH_CANONICAL_SPECS): New macro.
16581 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
16583 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16585 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
16586 are set after handling multilib fragments. Set target_cpu_default2
16589 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16591 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
16593 (arm*-*-*): Set target_cpu_default2 to a quoted string.
16594 * config/arm/parsecpu.awk (check_cpu): Validate any extension
16596 (check_arch): Likewise.
16597 * config/arm/arm.c (arm_configure_build_target): Handle
16598 TARGET_CPU_DEFAULT being a string constant. Scan any feature
16599 options in the default.
16601 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16603 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
16604 when an option is an alias of another.
16605 * config/arm/parsecpu.awk (optalias): New parser token.
16606 (gen_comm_data): Mark non-alias options as such. Emit entries
16607 for extension aliases.
16608 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
16609 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
16610 (armv6kz, armv6zk, armv6t2): Likewise.
16611 (armv7): Make vfpv3-d16 an alias.
16612 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
16614 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
16615 Sort in canonical order.
16616 (armv8-a): Sort in canonical order.
16617 (armv8.1-a, armv8.2-a): Likewise.
16618 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
16620 (cortex-a9): Sort in canonical order.
16621 * config/arm/arm.c (selftests.h): Include it.
16622 (arm_test_cpu_arch_data): New function.
16623 (arm_run_self_tests): New function.
16624 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
16625 (targetm): Move declaration to the end of the file.
16626 * arm-cpu-cdata.h: Regenerated.
16628 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16630 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
16631 call to target_mode_check describing the type of option passed.
16632 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
16633 (arm_target_thumb_only): Use arm_parse_arch_option_name or
16634 arm_parse_cpu_option_name to match parameters against list of
16636 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
16637 arm_arch_core_flags data structure.
16638 * config/arm/arm-cpu_cdata.h: Regenerated.
16640 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16642 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
16644 (arm_print_hint_for_cpu_option): Likewise.
16645 (arm_print_hint_for_arch_option): Likewise.
16646 (arm_parse_cpu_option_name): Likewise.
16647 (arm_parse_arch_option_name): Likewise.
16648 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
16649 of entries in the all_fpus list.
16650 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
16651 (arm_parse_cpu_option_name): Declare.
16652 (arm_parse_arch_option_name): Declare.
16653 (arm_parse_option_features): Declare.
16654 (arm_intialize_isa): Declare.
16655 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
16657 (gen_comm_data): ... here. Make definitions non-static.
16658 * config/arm/arm-cpu-data.h: Regenerated.
16659 * config/arm/arm-cpu-cdata.h: Regenerated.
16661 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16663 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
16664 (cpu_arch_extension): New structure.
16665 (cpu_arch_option, arch_option, cpu_option): New structures.
16666 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
16667 architecture types.
16668 (gen_data): Generate new format data tables.
16669 * config/arm/arm.c (cpu_tune): New structure.
16670 (cpu_option, processors): Delete.
16671 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
16672 (arm_print_hint_for_cpu_option): ... this and ...
16673 (arm_print_hint_for_arch_option): ... this.
16674 (arm_parse_arch_cpu_name): Delete. Replace with ...
16675 (arm_parse_cpu_option_name): ... this and ...
16676 (arm_parse_arch_option_name): ... this.
16677 (arm_unrecognized_feature): Change type of target parameter to
16679 (arm_parse_arch_cpu_features): Delete. Replace with ...
16680 (arm_parse_option_features): ... this.
16681 (arm_configure_build_target): Rework to use new configuration data
16683 (arm_print_tune_info): Rework for new configuration data tables.
16684 * config/arm/arm-cpu-data.h: Regenerated.
16685 * config/arm/arm-cpu.h: Regenerated.
16687 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16689 * Makefile.in (OBJS): Move sbitmap.o from here ...
16690 (OBJS-libcommon): ... to here.
16692 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16694 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
16695 (ISA_ALL_CRYPTO): New macro.
16696 (ISA_ALL_SIMD): New macro
16697 (ISA_ALL_FP): New macro.
16698 * config/arm/arm.c (fpu_bitlist): Update initializer.
16699 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
16701 (arm9e): Add fpu. Add option for nofp
16702 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
16703 (arm926ej-s, arm1026ej-s): Likewise.
16704 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
16705 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
16706 neon-fp16, neon-vfpv4, nofp and nosimd.
16707 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
16708 (cortex-a8): Add fpu. Add option for nofp.
16709 (cortex-a9): Add fpu. Add options for nosimd and nofp.
16710 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
16711 (cortex-r4f): Add fpu.
16712 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
16713 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
16715 (cortex-r8): Likewise.
16716 (cortex-m4): Add fpu. Add option for nofp.
16717 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
16718 (cortex-a17.cortex-a7): Likewise.
16719 (cortex-a32): Add fpu. Add options for crypto and nofp.
16720 (cortex-a35, cortex-a53): Likewise.
16721 (cortex-a57): Add fpu. Add option for crypto.
16722 (cortex-a72, cortex-a73): Likewise.
16723 (exynos-m1): Likewise.
16724 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
16725 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
16726 (cortex-m33): Add fpu. Add option for nofp.
16727 * config/arm/arm-cpu-cdata.h: Regenerated
16728 * config/arm/arm-cpu-data.h: Regenerated.
16730 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16732 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
16733 (armv5te, armv5tej): Likewise.
16734 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
16735 (armv7): Add options fp and vfpv3-d16.
16736 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
16737 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
16739 (armv7ve): Likewise.
16740 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
16741 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
16742 (armv8-a): Add nocrypto option.
16743 (armv8.1-a, armv8.2-a): Likewise.
16744 (armv8-m.main): add options fp, fp.dp and nofp.
16746 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16748 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
16750 (armv8-a+crc): Delete.
16751 (armv8.1-a): Add options simd, crypto and nofp.
16752 (armv8.2-a): Add options fp16, simd, crypto and nofp.
16753 (armv8.2-a+fp16): Delete.
16754 (armv8-m.main): Add option dsp.
16755 (armv8-m.main+dsp): Delete.
16756 (cortex-a8): Add fpu. Add nofp option.
16757 (cortex-a9): Add fpu. Add nofp and nosimd options.
16758 * config/arm/parsecpu.awk (gen_data): Generate option tables and
16759 link to main cpu and architecture data structures.
16760 (gen_comm_data): Only put isa attributes from the main architecture
16762 (option): New statement for architecture and CPU entries.
16763 * arm.c (struct cpu_option): New structure.
16764 (struct processors): Add entry for options.
16765 (arm_unrecognized_feature): New function.
16766 (arm_parse_arch_cpu_name): Ignore any characters after the first
16768 (arm_parse_arch_cpu_feature): New function.
16769 (arm_configure_build_target): Separate out any CPU and architecture
16770 features and parse separately. Don't error out if -mfpu=auto is
16771 used with only an architecture string.
16772 (arm_print_asm_arch_directives): New function.
16773 (arm_file_start): Call it.
16774 * config/arm/arm-cpu-cdata.h: Regenerated.
16775 * config/arm/arm-cpu-data.h: Likewise.
16776 * config/arm/arm-tables.opt: Likewise.
16778 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16780 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
16781 assembler when it is not -mfpu=auto.
16783 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16785 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
16786 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
16787 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
16788 (ASM_CPU_SPEC): Rewrite.
16789 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
16790 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
16791 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
16792 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
16793 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
16794 copied string is NUL-terminated. Also strip any characters prefixed
16796 (arm_rewrite_selected_arch): New function.
16797 (arm_rewrite_march): New function.
16799 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16801 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
16802 (x_arm_cpu_string, x_arm_tune_string): Likewise.
16803 (march, mcpu, mtune): Convert to string-based options.
16804 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
16805 (arm_parse_arch_cpu_name): New function.
16806 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
16807 identify selected architecture or CPU.
16808 (arm_option_save): New function.
16809 (TARGET_OPTION_SAVE): Redefine.
16810 (arm_option_restore): Restore string options.
16811 (arm_option_print): Print string options.
16813 2017-06-16 Martin Sebor <msebor@redhat.com>
16815 PR tree-optimization/80933
16816 PR tree-optimization/80934
16817 * builtins.c (fold_builtin_3): Do not handle bcmp here.
16818 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
16819 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
16820 (gimple_fold_builtin): Call them.
16822 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16824 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
16825 as unlikely; update profile.
16827 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16829 * predict.c (force_edge_cold): Handle declaring edges impossible
16832 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16834 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
16836 (try_unroll_loop_completely): Fix reporting.
16838 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16840 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
16842 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
16845 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
16846 if given a non-constant argument for an intrinsic which requires a
16849 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16851 * profile.c (compare_freqs): New function.
16852 (branch_prob): Sort edge list.
16853 (find_spanning_tree): Assume that the list is priority sorted.
16855 2017-06-16 Richard Biener <rguenther@suse.de>
16857 PR tree-optimization/81090
16858 * passes.def (pass_record_bounds): Remove.
16859 * tree-pass.h (make_pass_record_bounds): Likewise.
16860 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
16861 make_pass_record_bounds): Likewise.
16862 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
16863 not free niter estimates at the beginning but at the end.
16864 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
16866 2017-06-16 Richard Biener <rguenther@suse.de>
16868 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
16869 initializer to workaround ICE in host GCC 4.8.
16871 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16873 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
16875 (clone_inlined_nodes): Update.
16877 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16879 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
16880 prefetch settings, and enable prefetching by default at -O3.
16882 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16884 * config/aarch64/aarch64.c (aarch64_override_options_internal):
16885 Set flag_prefetch_loop_arrays according to tuning data.
16887 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16889 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
16890 New tune structure.
16891 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
16892 [Unrelated to main purpose of the patch] Place the pointer field last
16893 to enable type checking errors when tune structure are wrongly merged.
16894 * config/aarch64/aarch64.c (generic_prefetch_tune,)
16895 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
16896 (thunderx2t99_prefetch_tune): New tune constants.
16897 (tune_params *_tunings): Update all tunings (no functional change).
16898 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
16899 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
16900 from tunings structures.
16902 2017-06-16 Jakub Jelinek <jakub@redhat.com>
16905 * ubsan.c (instrument_null): Add T argument, use it instead
16906 of computing it based on IS_LHS.
16907 (instrument_object_size): Likewise.
16908 (pass_ubsan::execute): Adjust instrument_null and
16909 instrument_object_size callers to pass gimple_get_lhs or
16910 gimple_assign_rhs1 result to it. Use instrument_null instead of
16911 calling get_base_address and instrument_mem_ref. Handle
16912 aggregate call arguments for object-size sanitization.
16914 2017-06-16 Yury Gribov <tetra2005@gmail.com>
16916 PR tree-optimization/81089
16917 * tree-vrp.c (is_masked_range_test): Validate operands of
16920 2017-06-15 Martin Sebor <msebor@redhat.com>
16923 * dumpfile.c (dump_register): Avoid calling memset to initialize
16924 a class with a default ctor.
16925 * gcc.c (struct compiler): Remove const qualification.
16926 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
16927 * hash-table.h: Ditto.
16928 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
16930 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
16931 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
16933 * params.h (struct param_info): Make struct members non-const.
16934 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
16935 with default initialization.
16936 * vec.h (vec_copy_construct, vec_default_construct): New helper
16938 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
16939 with vec_copy_construct.
16940 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
16941 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
16942 * doc/invoke.texi (-Wclass-memaccess): Document.
16944 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16946 * emit-rtl.h (is_leaf): Update comment about local
16947 register allocator.
16949 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16952 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
16953 for a variable to have a section before checking if the section has a
16955 Set section to.persistent if persistent attribute is set.
16956 Warn if .persistent attribute is used on an automatic variable.
16958 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
16960 PR rtl-optimization/80474
16961 * reorg.c (update_block): Do not ignore instructions in a delay slot.
16963 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
16965 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
16968 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
16970 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
16971 (casesi): Emit bounds checking as RTL.
16972 (casesi_internal_mips16_<mode>): Remove bounds checking.
16974 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
16976 * config/xtensa/xtensa.c (xtensa_option_override): Append
16977 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
16978 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
16979 xtensa_doloop_hooks): Define unconditionally.
16980 (xtensa_reorg_loops): Only call reorg_loops in the presence of
16982 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
16983 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
16984 for it in xtensa_option_override.
16985 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
16986 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
16988 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
16990 * doc/cppopts.texi: Document '-' special value to -MF.
16992 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
16994 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
16995 (cortex_a53_fconst): Likewise.
16996 (cortex_a53_fpmul): Likewise.
16997 (cortex_a53_f_load_64): Likewise.
16998 (cortex_a53_f_load_many): Likewise.
16999 (cortex_a53_advsimd_alu): Likewise.
17000 (cortex_a53_advsimd_alu_q): Likewise.
17001 (cortex_a53_advsimd_mul): Likewise.
17002 (cortex_a53_advsimd_mul_q): Likewise.
17003 (fpmac bypass): Add new bypass for fpmac-fpmac case.
17004 Add missing fmul, r2f_cvt and fconst cases.
17006 2017-06-14 Richard Biener <rguenther@suse.de>
17008 PR middle-end/81088
17009 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
17011 (fold_binary_loc): When associating do not treat pre-existing
17012 TREE_OVERFLOW on literal constants as a reason to allow
17013 TREE_OVERFLOW on associated literal constants.
17015 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
17017 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
17018 (MASK_FEATURES): New macro.
17019 * config/sparc/sparc.c (sparc_option_override): Remove the special
17020 handling of -mfpu and generalize it to all MASK_FEATURES switches.
17022 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
17024 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
17025 a division of 0 if non-call exceptions are enabled.
17027 2017-06-14 Andrew Pinski <apinski@cavium.com>
17028 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17031 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
17032 Improve vector initialization code gen for only variable case.
17034 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
17036 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
17038 2017-06-14 Richard Biener <rguenther@suse.de>
17040 PR tree-optimization/81083
17041 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
17044 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17046 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
17047 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
17048 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
17049 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
17050 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
17051 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
17053 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17055 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
17056 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
17058 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17060 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
17062 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17064 * config/rs6000/t-rtems: Don't handle SPE.
17066 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17068 * config/rs6000/t-linux: Don't handle SPE.
17070 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17072 * config/rs6000/eabispe.h: Delete file.
17074 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17076 * config/rs6000/t-spe: Delete file.
17078 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17080 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
17081 (rs6000_legitimate_offset_address_p): Return false for anything in
17082 V2SImode or V2SFmode.
17084 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17086 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
17087 except V2SF and V2SI. Rearrange the vector modes, and add comments.
17088 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
17090 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
17091 (rs6000_legitimate_offset_address_p): Ditto.
17092 (rs6000_emit_move): Ditto.
17093 (rs6000_init_builtins): Remove V4HI_type_node.
17095 2017-06-13 Martin Liska <mliska@suse.cz>
17098 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
17099 (gate_asan): Likewise.
17100 * asan.h (asan_no_sanitize_address_p): Remove the function.
17101 (sanitize_flags_p): New function.
17102 * builtins.def: Fix coding style.
17103 * common.opt: Use renamed enum value.
17104 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
17105 * doc/extend.texi: Document no_sanitize attribute.
17106 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
17107 to SANITIZE_UNDEFINED_NONDEFAULT.
17108 * gcc.c (sanitize_spec_function): Use the renamed enum value.
17109 * gimple-fold.c (optimize_atomic_compare_exchange_p):
17110 Use sanitize_flags_p.
17111 * gimplify.c (gimplify_function_tree): Likewise.
17112 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
17113 * opts.c (parse_no_sanitize_attribute): New function.
17114 (common_handle_option): Use renamed enum value.
17115 * opts.h (parse_no_sanitize_attribute): Declare.
17116 * tree.c (sanitize_flags_p): New function.
17117 * tree.h: Declared here.
17118 * tsan.c: Use sanitize_flags_p.
17119 * ubsan.c (ubsan_expand_null_ifn): Likewise.
17120 (instrument_mem_ref): Likewise.
17121 (instrument_bool_enum_load): Likewise.
17122 (do_ubsan_in_current_function): Remove the function.
17123 (pass_ubsan::execute): Use sanitize_flags_p.
17124 * ubsan.h: Remove do_ubsan_in_current_function
17125 * tree-cfg.c (print_no_sanitize_attr_value): New function.
17126 (dump_function_to_file): Use it here.
17128 2017-06-13 Martin Jambor <mjambor@suse.cz>
17130 PR tree-optimization/80803
17131 PR tree-optimization/81063
17132 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
17133 (propagate_subaccesses_across_link): Enqueue subtree whenever
17134 necessary instead of relying on the caller.
17136 2017-06-13 Martin Jambor <mjambor@suse.cz>
17138 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
17139 that have a first_link.
17140 (sort_and_splice_var_accesses): Do not check first_link before
17142 (subtree_mark_written_and_enqueue): Likewise.
17143 (propagate_all_subaccesses): Likewise and do not stop at first
17144 parent with a first_link.
17146 2017-06-13 Martin Jambor <mjambor@suse.cz>
17148 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
17151 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17153 * match.pd: New pattern.
17155 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17157 * tree-vrp.c (is_masked_range_test): New function.
17158 (register_edge_assert_for): Determine ranges for
17161 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17163 PR tree-optimization/67328
17164 * fold-const.c (maskable_range_p): New function.
17165 (build_range_check): Generate bittests if possible.
17167 2017-06-13 Martin Liska <mliska@suse.cz>
17169 * gimple-pretty-print.c (dump_probability): Add new argument.
17170 (dump_edge_probability): Dump both probability and count.
17171 (dump_gimple_label): Likewise.
17172 (dump_gimple_bb_header): Likewise.
17174 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
17177 * config/avr/avr-devices.c: Fix indentation.
17178 * config/avr/gen-avr-mmcu-specs.c: Dito.
17180 2017-06-13 Richard Biener <rguenther@suse.de>
17182 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
17183 instead get vector type from stmt_info.
17184 (vectorizable_reduction): Adjust. Remove dead code.
17186 2017-06-13 Richard Biener <rguenther@suse.de>
17188 PR middle-end/81065
17189 * fold-const.c (extract_muldiv_1): Remove bogus distribution
17190 case of C * (x * C2 + C3).
17191 (fold_addr_of_array_ref_difference): Properly fold index difference.
17193 2017-06-12 David S. Miller <davem@davemloft.net>
17196 * config/sparc/sparc.md (return expander): Emit frame blockage if
17197 function uses alloca.
17199 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
17201 * combine.c (make_field_assignment): Check len rather than the mode
17202 precision when calling force_to_mode.
17204 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
17206 Support multilibs and devices that see flash in RAM address range.
17209 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
17210 (avr_mcu_t) <flash_pm_offset>: New field.
17211 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
17212 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
17213 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
17214 (AVR_TINY_PM_OFFSET): Remove macro.
17215 * config/avr/avr.opt (-mshort-calls): New option.
17216 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17217 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
17218 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
17219 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
17220 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
17221 instead of avr_arch->have_jmp_call.
17222 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
17223 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
17224 avr_arch->flash_pm_offset to define.
17225 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
17226 new field flash_pm_offset. Add entry for avrxmega3.
17227 (avr_texinfo): Add entry for avrxmega3.
17228 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
17229 attiny212, attiny214,
17230 attiny412, attiny414, attiny416, attiny417,
17231 attiny814, attiny816, attiny817,
17232 attiny1614, attiny1616, attiny1617,
17233 attiny3214, attiny3216, attiny3217.
17234 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
17235 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
17236 (avr_print_operand_address) [AVR_TINY]: Same.
17237 (avr_asm_init_sections) <readonly_data_section>: Only patch
17238 callback if avr_arch->flash_pm_offset = 0.
17239 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
17240 for rodata if avr_arch->flash_pm_offset != 0.
17241 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
17242 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
17243 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
17244 (m_options): Append opt_rcall.
17245 (m_dirnames): Append dir_rcall.
17246 * config/avr/t-multilib: Regenerate.
17248 * configure.ac [target=avr]: Check whether avrxmega3 default
17249 linker description file works as needed.
17250 * configure: Regenerate.
17251 * doc/avr-mmcu.texi: Regenerate.
17252 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
17253 <__AVR_ARCH__>: Document avrxmega3 and 103.
17254 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
17255 <__AVR_SHORT_CALLS__>: Document it.
17256 <__AVR_PM_BASE_ADDRESS__>: Document it.
17257 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
17258 (AVR Variable Attributes) <progmem>: Document this is
17259 not needed for avrxmega3.
17260 (AVR Named Address Spaces) <__flash>: Dito.
17262 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
17264 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
17266 2017-06-12 Doug Rupp <rupp@adacore.com>
17268 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
17269 Append vxworks-stdint.h to the tm_file list.
17270 * config/vxworks-stdint.h: New file.
17272 2017-06-12 Martin Liska <mliska@suse.cz>
17274 PR tree-optimization/81041
17275 * tree-profile.c (gimple_gen_ic_func_profiler):
17276 Create an extra BB in profile-generate
17277 (gimple_gen_time_profiler): Likewise.
17279 2017-06-12 Jakub Jelinek <jakub@redhat.com>
17281 PR tree-optimization/81003
17282 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
17283 (update_range_test): Use it instead of force_gimple_operand_gsi.
17285 2017-06-12 Richard Biener <rguenther@suse.de>
17287 PR tree-optimization/81053
17288 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
17289 with backedge value not defined in loop. Simplify def stmt
17292 2017-06-11 Tom de Vries <tom@codesourcery.com>
17295 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
17297 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
17298 nvptx_cannot_force_const_mem.
17300 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17302 * opts.c (finish_options): Move test for flag_split_stack after
17303 it has been initialized.
17305 2017-06-11 Jason Merrill <jason@redhat.com>
17307 * tree.h (id_equal): New.
17308 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
17309 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
17310 instead of strcmp of IDENTIFIER_POINTER.
17312 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17314 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
17315 (mark_all_inlined_calls_cdtor): Fix formating.
17316 (inline_transform): Rescale profile before inlining.
17318 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17320 * cgraph.h (cgraph_edge::clone): Update prototype.
17321 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
17322 (cgraph_node::create_clone): Update.
17323 (cgraph_node::create_version_clone): Update.
17324 * tree-inline.c (copy_bb): Update.
17325 (expand_call_inline): Update.
17327 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
17329 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
17330 factored out from ...
17331 (rs6000_emit_prologue): ... here.
17333 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
17335 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
17336 factored out from ...
17337 (rs6000_emit_prologue): ... here.
17339 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17341 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
17343 (handle_missing_profiles): Fix computation of tp_first_run.
17344 (counts_to_freqs): Do not touch freqs when count is 0.
17346 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17348 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
17351 2017-06-10 Tom de Vries <tom@codesourcery.com>
17353 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
17354 attributes): Document signal effective target.
17356 2017-06-10 Tom de Vries <tom@codesourcery.com>
17358 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
17359 Document effective target stack_size.
17361 2017-06-09 David Malcolm <dmalcolm@redhat.com>
17363 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
17364 to the edit_context if they can be auto-applied.
17366 2017-06-9 Ian Lance Taylor <iant@golang.org>
17368 * opts.c (finish_options): If -fsplit-stack, disable implicit
17369 -forder-blocks-and-partition.
17370 * doc/invoke.texi (Optimize Options): Document that when using
17371 -fsplit-stack -forder-blocks-and-partition is not implicitly
17374 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17376 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
17377 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
17378 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
17379 * builtins.def (abort, trap, unreachable): Declare cold.
17380 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
17381 * tree-core.h (ECF_COLD): New.
17382 * tree.c (set_call_expr_flags): Handle ECF_COLD.
17383 (build_common_builtin_nodes): Mark unreachable and abort as cold.
17385 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17387 * predict.c (unlikely_executed_stmt_p): Cleanup.
17389 2017-06-09 Richard Biener <rguenther@suse.de>
17391 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
17392 model if the ref is always written to.
17394 2017-06-09 Tamar Christina <tamar.christina@arm.com>
17396 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
17398 2017-06-09 Tamar Christina <tamar.christina@arm.com>
17400 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
17403 2017-06-09 Tom de Vries <tom@codesourcery.com>
17406 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
17407 "target cannot support label values" when encountering LABEL_REF.
17409 2017-06-09 Martin Liska <mliska@suse.cz>
17411 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
17412 (gimple_gen_ic_func_profiler): Emit direct comparison
17413 of __gcov_indirect_call_callee with NULL.
17414 (gimple_gen_time_profiler): Change probability from
17415 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
17417 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17419 * profile.c (edge_gcov_counts): Turn to pointer.
17420 (compute_branch_probabilities, compute_branch_probabilities): Update.
17421 (branch_prob): Do not clear edge_gcov_count.
17422 * profile.h (edge_gcov_counts): Turn to pointer.
17423 (edge_gcov_count): Update.
17425 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17427 * gimple.h (gimple_check_failed): Mark cold.
17429 2017-06-09 Richard Biener <rguenther@suse.de>
17431 PR tree-optimization/66623
17432 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
17433 refactor check_reduction into two parts, properly computing
17434 whether we have to check reduction validity for outer loop
17437 2017-06-09 Richard Biener <rguenther@suse.de>
17439 PR tree-optimization/79483
17440 * graphite-scop-detection.c (order): New global.
17441 (get_order): Compute bb to order mapping that satisfies code
17442 generation constraints.
17443 (cmp_pbbs): New helper.
17444 (build_scops): Start domwalk at entry block, sort generated
17447 2017-06-09 Richard Biener <rguenther@suse.de>
17449 PR middle-end/81007
17450 * ipa-polymorphic-call.c
17451 (ipa_polymorphic_call_context::restrict_to_inner_class):
17452 Skip FIELD_DECLs with error_mark_node type.
17453 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
17456 2017-06-09 Martin Liska <mliska@suse.cz>
17458 * predict.c (struct branch_predictor): New struct.
17459 (test_prediction_value_range): New test.
17460 (predict_c_tests): New function.
17461 * selftest-run-tests.c (selftest::run_tests): Run the function.
17462 * selftest.h: Declare new tests.
17464 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
17467 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
17468 gen_add3_insn did not fail.
17469 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
17470 r0, construct that number in a temporary reg and add that reg to r0.
17471 If asked to put the result in r0 as well, fail.
17473 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
17475 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17476 for early expansion of vec_eqv.
17478 2017-06-08 Jakub Jelinek <jakub@redhat.com>
17480 PR middle-end/81005
17481 * ubsan.c (instrument_null): Avoid pointless code temporary.
17482 (pass_ubsan::execute): Instrument aggregate arguments of calls.
17484 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
17488 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
17491 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
17494 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17496 * predict.c (unlikely_executed_edge_p): Move ahead.
17497 (probably_never_executed_edge_p): Use it.
17499 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
17501 PR middle-end/79988
17502 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
17503 gimple_call_builtin_p call.
17505 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17507 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
17508 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
17509 rtl_check_failed_type2, rtl_check_failed_code1,
17510 rtl_check_failed_code2, rtl_check_failed_code_mode,
17511 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
17512 rtvec_check_failed_bounds, rtl_check_failed_flag,
17513 _fatal_insn_not_found, _fatal_insn): Likewise.
17514 * tree.h (tree_contains_struct_check_failed,
17515 tree_check_failed, tree_not_check_failed,
17516 tree_class_check_failed, tree_range_check_failed,
17517 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
17518 tree_vec_elt_check_failed, phi_node_elt_check_failed,
17519 tree_operand_check_failed, omp_clause_check_failed,
17520 omp_clause_operand_check_failed, omp_clause_range_check_failed):
17523 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17525 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
17526 flag_branch_probabilities.
17527 * ipa-inline.c (edge_badness): Likewise.
17528 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
17529 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
17530 * predict.c (maybe_hot_frequency_p): Likewise.
17531 (probably_never_executed): Likewise.
17532 * sched-ebb.c (schedule_ebbs): Likewise.
17533 * sched-rgn.c (find_single_block_region): Likewise.
17534 * tracer.c (tail_duplicate): Likewise.
17536 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17538 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
17539 longer requires x_flag_profile_use.
17541 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17543 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
17544 instead of flag_reorder_blocks_and_partition.
17545 * dbxout.c (dbxout_function_end): Likewise.
17546 * dwarf2out.c (gen_subprogram_die): Likewise.
17547 * haifa-sched.c (sched_create_recovery_edges): Likewise.
17548 * hw-doloop.c (reorg_loops): Likewise.
17549 * varasm.c (assemble_start_function,
17550 assemble_end_function): Likewise.
17551 (decide_function_section): Do not check for
17552 flag_reorder_blocks_and_partition.
17554 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
17556 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
17558 (chkp_get_hard_register_fake_addr_expr): Ditto.
17559 (chkp_build_addr_expr): Add check for hard reg case.
17560 (chkp_parse_array_and_component_ref): Ditto.
17561 (chkp_find_bounds_1): Ditto.
17562 (chkp_process_stmt): Don't generate bounds store for
17565 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17567 * predict.c (maybe_hot_bb_p): Do not check profile status.
17568 (maybe_hot_edge_p): Likewise.
17569 (probably_never_executed): Check for zero counts even if profile
17571 (unlikely_executed_edge_p): New function.
17572 (unlikely_executed_stmt_p): New function.
17573 (unlikely_executed_bb_p): New function.
17574 (set_even_probabilities): Use unlikely predicates.
17575 (combine_predictions_for_bb): Likewise.
17576 (predict_paths_for_bb): Likewise.
17577 (predict_paths_leading_to_edge): Likewise.
17578 (determine_unlikely_bbs): New function.
17579 (estimate_bb_frequencies): Use it.
17580 (compute_function_frequency): Use zero counts even if profile is
17582 * profile-count.h: Fix typo.
17584 2017-08-08 Julia Koval <julia.koval@intel.com>
17586 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
17587 _mm512_mask_cvtsepi16_storeu_epi8,
17588 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
17589 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
17590 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
17591 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
17592 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
17593 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
17594 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
17595 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
17596 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
17597 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
17598 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
17599 __builtin_ia32_pmovuswb256mem_mask,
17600 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
17601 __builtin_ia32_pmovwb512mem_mask): New builtins.
17603 2017-08-08 Julia Koval <julia.koval@intel.com>
17605 PR target/73350,80862
17606 * config/i386/subst.md (round): Fix round pattern.
17607 * config/i386/i386.c (ix86_erase_embedded_rounding):
17608 Fix erasing rounding for the fixed pattern.
17610 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17612 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
17614 2017-06-08 Martin Liska <mliska@suse.cz>
17616 PR gcov-profile/80911
17617 * gcov.c (block_info::block_info): New constructor.
17619 2017-06-07 Carl Love <cel@us.ibm.com>
17621 * config/rs6000/rs6000-c: The return type of the following
17622 built-in functions was implemented as int not long long. Fix sign
17623 of return value for the unsigned version of vec_mulo and vec_mule.
17624 vector unsigned long long vec_bperm (vector unsigned long long,
17625 vector unsigned char)
17626 vector signed long long vec_mule (vector signed int,
17628 vector unsigned long long vec_mule (vector unsigned int,
17629 vector unsigned int)
17630 vector signed long long vec_mulo (vector signed int,
17632 vector unsigned long long vec_mulo (vector unsigned int,
17633 vector unsigned int)
17634 * doc/extend.texi: Fix the documentation for the built-in
17637 2017-06-07 Carl Love <cel@us.ibm.com>
17640 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
17643 2017-06-07 Carl Love <cel@us.ibm.com>
17645 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
17646 support, Generate doublehv for signed int/float for BE case only.
17648 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
17650 * doc/invoke.texi (mcx16): Rewrite.
17652 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17654 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
17655 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
17656 *mov<mode>_softfloat, and an anonymous splitter): Use
17657 nonimmediate_operand instead of rs6000_nonimmediate_operand.
17659 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17661 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
17663 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
17664 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
17665 (rs6000_debug_reg_global): Adjust.
17666 (rs6000_init_hard_regno_mode_ok): Adjust.
17667 (rs6000_dbx_register_number): Adjust.
17668 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
17669 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
17670 Remove SPE_ACC and SPEFSCR.
17671 (REG_ALLOC_ORDER): Ditto.
17672 (FRAME_POINTER_REGNUM): Change to 111.
17673 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
17674 (REG_CLASS_NAMES): Ditto.
17675 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
17676 (REGISTER_NAMES): Ditto.
17677 (ADDITIONAL_REG_NAMES): Ditto.
17678 (rs6000_reg_names): Ditto.
17679 * config/rs6000/rs6000.md: Renumber some register number
17682 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17684 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
17686 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
17687 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
17689 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
17690 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
17691 Delete the SPE high registers.
17692 (REG_ALLOC_ORDER): Ditto.
17693 (enum reg_class): Remove SPE_HIGH_REGS.
17694 (REG_CLASS_NAMES): Ditto.
17695 (REG_CLASS_CONTENTS): Delete the SPE high registers.
17696 (REGISTER_NAMES): Ditto.
17697 (rs6000_reg_names): Ditto.
17698 * doc/tm.texi.in: Remove SPE as example.
17699 * doc/tm.texi: Regenerate.
17701 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17703 * config/rs6000/8540.md (ppc8540_brinc): Delete.
17704 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
17705 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
17706 * config/rs6000/rs6000.md (type): Remove "brinc".
17708 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17710 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
17711 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
17712 * config/rs6000/linuxspe.h: Delete file.
17713 * config/rs6000/rs6000.md: Don't include spe.md.
17714 * config/rs6000/spe.h: Delete file.
17715 * config/rs6000/spe.md: Delete file.
17716 * config/rs6000/t-rs6000: Remove spe.md.
17718 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17720 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
17721 (reg_or_none500mem_operand): Delete.
17722 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
17723 instead of reg_or_none500mem_operand.
17725 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17727 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
17728 handling of SPE flags.
17729 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
17731 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17733 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
17735 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
17736 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
17737 paired_divv2sf3): Similar.
17738 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
17739 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
17740 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
17742 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
17743 Rename the paired_* instruction patterns.
17744 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
17746 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
17747 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
17748 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
17749 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
17750 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
17751 PAIRED_VECTOR_MODE.
17752 (struct machine_function): Delete field spe_insn_chain_scanned_p.
17753 (spe_func_has_64bit_regs_p): Delete.
17754 (spe_expand_predicate_builtin): Delete.
17755 (spe_expand_evsel_builtin): Delete.
17756 (TARGET_DWARF_REGISTER_SPAN): Do not define.
17757 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
17758 (invalid_e500_subreg): Delete.
17759 (rs6000_legitimize_address): Always force_reg op2 as well, for
17760 paired single memory accesses.
17761 (rs6000_member_type_forces_blk): Delete.
17762 (rs6000_spe_function_arg): Delete.
17763 (rs6000_expand_unop_builtin): Delete SPE handling.
17764 (rs6000_expand_binop_builtin): Ditto.
17765 (spe_expand_stv_builtin): Delete.
17766 (bdesc_2arg_spe): Delete.
17767 (spe_expand_builtin): Delete.
17768 (spe_expand_predicate_builtin): Delete.
17769 (spe_expand_evsel_builtin): Delete.
17770 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
17771 (spe_init_builtins): Delete.
17772 (spe_func_has_64bit_regs_p): Delete.
17773 (savres_routine_name): Delete "info" parameter. Adjust callers.
17774 (rs6000_emit_stack_reset): Ditto.
17775 (rs6000_dwarf_register_span): Delete.
17776 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
17777 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
17778 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
17780 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
17782 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
17783 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
17784 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
17785 mulv2sf3, divv2sf3): Delete expanders.
17787 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17789 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
17791 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17793 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
17794 * config/rs6000/rs6000.c: Ditto.
17796 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17798 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
17799 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
17800 comparison_operator.
17802 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17804 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
17805 * config/rs6000/rs6000.opt: Ditto.
17806 * config/rs6000/t-rtems: Ditto.
17808 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17810 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
17811 TARGET_E500_SINGLE by 0, simplify.
17812 * config/rs6000/rs6000.c: Ditto.
17813 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
17814 (spe_build_register_parallel): Delete.
17815 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
17816 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
17817 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
17818 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
17819 (E500_CONVERT): Delete.
17820 * config/rs6000/spe.md: Remove many patterns and all define_constants.
17822 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17824 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
17825 * config/rs6000/dfp.md: Ditto.
17826 (negdd2, *negdd2_fpr): Merge.
17827 (absdd2, *absdd2_fpr): Merge.
17828 (negtd2, *negtd2_fpr): Merge.
17829 (abstd2, *abstd2_fpr): Merge.
17830 * config/rs6000/e500.h: Delete file.
17831 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
17832 TARGET_FPRS by 1 and simplify.
17833 * config/rs6000/rs6000-c.c: Ditto.
17834 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
17835 TARGET_DF_SPE by 0.
17836 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
17838 * config/rs6000/rs6000.md: Ditto.
17839 (floatdidf2, *floatdidf2_fpr): Merge.
17840 (move_from_CR_gt_bit): Delete.
17841 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
17842 (E500_CR_IOR_COMPARE): Delete.
17843 (All patterns that require !TARGET_FPRS): Delete.
17844 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
17846 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17848 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
17850 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17852 * graphds.c (add_edge): Intitialize edge's attached data.
17853 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
17854 pointer parameter. Call pointed function on each edge during
17855 graph traversing. Skip traversing the edge when the function
17857 (graphds_dfs, graphds_scc): Ditto.
17858 (for_each_edge): New parameter. Pass the new parameter to callback
17860 * graphds.h (skip_edge_callback): New function pointer type.
17861 (graphds_dfs, graphds_scc): New function pointer parameter.
17862 (graphds_edge_callback, for_each_edge): New parameter.
17864 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17866 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
17867 out code checking if runtime alias check is possible to below ...
17868 Call the new function.
17869 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
17870 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
17872 2017-06-07 Marek Polacek <polacek@redhat.com>
17875 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
17876 TYPE_OVERFLOW_WRAPS check.
17878 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17880 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
17881 if versioning is required.
17882 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
17883 peeling with the check for versioning.
17885 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17887 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
17888 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
17889 Set true to new parameter if new ssa variable is defined.
17890 (vect_gen_vector_loop_niters): Refactor. Set range information
17891 for the new vector loop bound variable.
17892 (vect_do_peeling): Ditto.
17894 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17896 * tree-affine.c (ssa.h): Include header file.
17897 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
17898 has wrapping overflow behavior.
17900 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17902 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
17904 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17906 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
17907 (tree_to_aff_combination): ... here.
17909 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17911 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
17912 reg_pressure model function.
17913 (ivopts_global_cost_for_size): Delete.
17914 (determine_set_costs, iv_ca_recount_cost): Call new model function
17915 ivopts_estimate_reg_pressure.
17917 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17919 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
17920 expensive than udiv. Remove floating point cases from mod.
17922 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17924 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
17925 Increase idiv cost.
17927 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17929 * config/aarch64/aarch64.md
17930 (copysignsf3): Fix mask generation.
17932 2017-06-07 Jakub Jelinek <jakub@redhat.com>
17934 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
17936 (class dump_manager): Add register_dumps method.
17937 * dumpfile.c: Include langhooks.h.
17938 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
17939 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
17940 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
17941 (dump_manager::dump_register): Start with 512 entries instead of 32.
17942 (dump_manager::register_dumps): New method.
17943 * toplev.c (general_init): Instead of invoking register_dumps
17944 langhook, invoke register_dumps method on the dump manager.
17945 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
17948 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
17950 * doc/md.texi: Clarify the restrictions on a define_insn condition.
17951 Say that # requires an associated define_split to exist, and that
17952 the define_split must be suitable for use after register allocation.
17954 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17956 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
17957 (compute_outgoing_frequencies): Also initialize zero counts.
17958 (find_many_sub_basic_blocks): Do not produce uninitialized profile
17959 around loops; preserve more of profile when nothing changes.
17961 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
17963 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
17965 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
17966 * config/arm/arm-cpu-cdata.h: Regenerate.
17967 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
17968 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
17969 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
17970 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
17971 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
17973 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
17975 * config/arm/t-rmprofile: Likewise.
17976 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
17978 2017-06-06 David S. Miller <davem@davemloft.net>
17981 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
17982 blockage if function uses alloca.
17984 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17986 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
17987 New "uid" fields to hold pretty-print IDs of group and ref.
17988 Memory references are now identified as <group_id>:<ref_id>
17989 instead of using [random] addresses.
17990 (dump_mem_details): Simplify, no functional change.
17991 (dump_mem_ref): Simplify and make output more concise.
17992 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
17993 (find_or_create_group): Initialize group uid.
17994 (record_ref): Initialize ref uid. Improve debug output.
17995 (prune_group_by_reuse, should_issue_prefetch_p,)
17996 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
17997 (mark_nontemporal_store, determine_loop_nest_reuse):
17998 Improve debug output.
18000 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18002 * dbgcnt.def (prefetch): New debug counter.
18003 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
18004 (schedule_prefetches): Stop issueing prefetches if debug counter
18007 2017-06-06 Tom de Vries <tom@codesourcery.com>
18009 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
18010 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
18012 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18014 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
18015 Use aarch64_reg_or_zero predicate for operand 4.
18016 (aarch64_compare_and_swap<mode> define_insn_and_split):
18017 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
18018 (aarch64_store_exclusive<mode>): Likewise for operand 2.
18020 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18022 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
18023 (arm_compute_save_core_reg_mask): This.
18024 (thumb1_compute_save_reg_mask): Rename into ...
18025 (thumb1_compute_save_core_reg_mask): This.
18026 (arm_compute_save_reg0_reg12_mask): Adapt comment.
18027 (arm_compute_frame_layout): Likewise.
18029 2017-06-06 Richard Biener <rguenther@suse.de>
18031 PR tree-optimization/80974
18032 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
18033 keep or clear leaders SSA info.
18035 2017-06-06 Tom de Vries <tom@codesourcery.com>
18037 * config/nvptx/nvptx.c (split_mode_p): New function.
18038 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
18040 2017-06-06 Tom de Vries <tom@codesourcery.com>
18042 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
18044 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18047 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
18050 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18052 * shrink-wrap.c (handle_simple_exit): Update profile.
18053 (try_shrink_wrapping): Upate profile.
18055 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18057 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
18058 (tree_guess_outgoing_edge_probabilities): New.
18059 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
18060 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
18062 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18064 * ipa-split.c (split_function): Initialize return bb profile.
18066 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18068 * profile.c (compute_branch_probabilities): Also initialize
18069 EXIT_BLOCK profile.
18071 2017-06-06 Richard Biener <rguenther@suse.de>
18073 PR tree-optimization/80928
18074 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
18075 (vect_analyze_loop_operations): Properly guard analysis for
18077 (vect_transform_loop): Likewise.
18078 (vect_analyze_loop_2): Also reset SLP type on PHIs.
18079 (vect_model_induction_cost): Do not cost for pure SLP.
18080 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
18081 of induction in inner loop vectorization.
18082 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
18083 (vect_get_and_check_slp_defs): Handle vect_induction_def.
18084 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
18086 (vect_analyze_slp_cost_1): Cost induction.
18087 (vect_detect_hybrid_slp_stmts): Handle PHIs.
18088 (vect_get_slp_vect_defs): Likewise.
18089 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
18090 (vect_transform_stmt): Handle SLP reductions.
18091 * tree-vectorizer.h (vectorizable_induction): Adjust.
18093 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18095 * config/rs6000/rs6000.c (make_resolver_func): Update
18096 init_lowered_empty_function call.
18098 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
18100 * doc/invoke.texi: Document the -fprofile-abs-path option.
18101 * common.opt (fprofile-abs-path): New option.
18102 * gcov-io.h (gcov_write_filename): Declare.
18103 * gcov-io.c (gcov_write_filename): New function.
18104 * coverage.c (coverage_begin_function): Use gcov_write_filename.
18105 * profile.c (output_location): Likewise.
18107 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18109 * shring-wrap.c: Revert accidental commit.
18111 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
18113 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
18115 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18117 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
18119 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
18120 profile in callgraph edge.
18121 * profile-count.h (apply_probability): If THIS is 0, then result is 0
18122 (apply_scale): Likewise.
18123 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
18124 Also scale profile when inlining function with zero profile.
18125 (initialize_cfun): Update exit block profile even when it is zero.
18126 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
18127 when profile is read.
18129 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18131 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
18132 (CLONE_*): New constants to define the processors we can generate
18133 code for with the target_clone attribute.
18134 (rs6000_clone_map): New array to identify which clone processors
18135 the current program is running on.
18136 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
18137 target_clone attribute.
18138 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
18139 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
18140 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
18141 (cpu_expand_builtin): Add support for target_clone attribute.
18142 (rs6000_valid_attribute_p): Allow "default" attribute.
18143 (get_decl_name): New debug function to simplify printing the
18144 current function name in debugging statements.
18145 (rs6000_clone_priority): New functions to support the target_clone
18146 attribute, and be able to generate code to switch between ISA 2.05
18147 through ISA 3.0 (power6 through power9).
18148 (rs6000_compare_version_priority): Likewise.
18149 (rs6000_get_function_versions_dispatcher): Likewise.
18150 (make_resolver_func): Likewise.
18151 (add_condition_to_bb): Likewise.
18152 (dispatch_function_versions): Likewise.
18153 (rs6000_generate_version_dispatcher_body): Likewise.
18154 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
18155 (fusion_gpr_load_p): Fix a spacing issue.
18156 * doc/extend.texi (Common Function Attributes): Document that the
18157 PowerPC supports the target_clone attribute.
18159 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
18161 * config/arm/arm.h: explain F symbol found in description of ARM
18162 register allocation in its legend.
18164 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18166 * config/mips/frame-header-opt.c: Include profile-count.h.
18167 * config/riscv/riscv.c: Include profile-count.h
18169 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18171 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
18173 (sm_set_flag_if_changed): Add bbs field.
18174 (execute_sm_if_changed_flag_set): Pass BBS.
18175 (execute_sm): Update.
18177 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18179 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
18182 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18184 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
18185 (peephole2): New peephole2 to emit the above.
18186 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
18188 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18190 * config/aarch64/aarch64.c (define_peephole2 above
18191 *sub_<shift>_<mode>): New peephole.
18193 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18195 * config/i386/i386.c (make_resolver_func): Update.
18196 * Makefile.in: Add profile-count.h and profile-count.o
18197 * auto-profile.c (afdo_indirect_call): Update to new API.
18198 (afdo_set_bb_count): Update.
18199 (afdo_propagate_edge): Update.
18200 (afdo_propagate_circuit): Update.
18201 (afdo_calculate_branch_prob): Update.
18202 (afdo_annotate_cfg): Update.
18203 * basic-block.h: Include profile-count.h
18204 (struct edge_def): Turn count to profile_count.
18205 (struct basic_block_def): Likewie.
18206 (REG_BR_PROB_BASE): Move to profile-count.h
18207 (RDIV): Move to profile-count.h
18208 * bb-reorder.c (max_entry_count): Turn to profile_count.
18209 (find_traces): Update.
18210 (rotate_loop):Update.
18211 (connect_traces):Update.
18212 (sanitize_hot_paths):Update.
18213 * bt-load.c (migrate_btr_defs): Update.
18214 * cfg.c (RDIV): Remove.
18215 (init_flow): Use alloc_block.
18216 (alloc_block): Uninitialize count.
18217 (unchecked_make_edge): Uninitialize count.
18218 (check_bb_profile): Update.
18219 (dump_edge_info): Update.
18220 (dump_bb_info): Update.
18221 (update_bb_profile_for_threading): Update.
18222 (scale_bbs_frequencies_int): Update.
18223 (scale_bbs_frequencies_gcov_type): Update.
18224 (scale_bbs_frequencies_profile_count): New.
18225 * cfg.h (update_bb_profile_for_threading): Update.
18226 (scale_bbs_frequencies_profile_count): Declare.
18227 * cfgbuild.c (compute_outgoing_frequencies): Update.
18228 (find_many_sub_basic_blocks): Update.
18229 * cfgcleanup.c (try_forward_edges): Update.
18230 (try_crossjump_to_edge): Update.
18231 * cfgexpand.c (expand_gimple_tailcall): Update.
18232 (construct_exit_block): Update.
18233 * cfghooks.c (verify_flow_info): Update.
18234 (dump_bb_for_graph): Update.
18235 (split_edge): Update.
18236 (make_forwarder_block): Update.
18237 (duplicate_block): Update.
18238 (account_profile_record): Update.
18239 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
18240 (get_estimated_loop_iterations): Update.
18241 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
18242 (single_likely_exit): Update.
18243 * cfgloopmanip.c (scale_loop_profile): Update.
18245 (set_zero_probability): Update.
18246 (lv_adjust_loop_entry_edge): Update.
18247 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
18248 (purge_dead_edges): Update.
18249 (rtl_account_profile_record): Update.
18250 * cgraph.c (cgraph_node::create): Uninitialize count.
18251 (symbol_table::create_edge): Uninitialize count.
18252 (cgraph_update_edges_for_call_stmt_node): Update.
18253 (cgraph_edge::dump_edge_flags): Update.
18254 (cgraph_node::dump): Update.
18255 (cgraph_edge::maybe_hot_p): Update.
18256 * cgraph.h: Include profile-count.h
18257 (create_clone), create_edge, create_indirect_edge): Update.
18258 (cgraph_node): Turn count to profile_count.
18259 (cgraph_edge0: Likewise.
18260 (make_speculative, clone): Update.
18261 (create_edge): Update.
18262 (init_lowered_empty_function): Update.
18263 * cgraphclones.c (cgraph_edge::clone): Update.
18264 (duplicate_thunk_for_node): Update.
18265 (cgraph_node::create_clone): Update.
18266 * cgraphunit.c (cgraph_node::analyze): Update.
18267 (cgraph_node::expand_thunk): Update.
18268 * final.c (dump_basic_block_info): Update.
18269 * gimple-streamer-in.c (input_bb): Update.
18270 * gimple-streamer-out.c (output_bb): Update.
18271 * graphite.c (print_global_statistics): Update.
18272 (print_graphite_scop_statistics): Update.
18273 * hsa-brig.c: Include basic-block.h.
18274 * hsa-dump.c: Include basic-block.h.
18275 * hsa-gen.c (T sum_slice): Update.
18276 (convert_switch_statements):Update.
18277 * hsa-regalloc.c: Include basic-block.h.
18278 * ipa-chkp.c (chkp_produce_thunks): Update.
18279 * ipa-cp.c (struct caller_statistics): Update.
18280 (init_caller_stats): Update.
18281 (gather_caller_stats): Update.
18282 (ipcp_cloning_candidate_p): Update.
18283 (good_cloning_opportunity_p): Update.
18284 (get_info_about_necessary_edges): Update.
18285 (dump_profile_updates): Update.
18286 (update_profiling_info): Update.
18287 (update_specialized_profile): Update.
18288 (perhaps_add_new_callers): Update.
18289 (decide_about_value): Update.
18290 (ipa_cp_c_finalize): Update.
18291 * ipa-devirt.c (struct odr_type_warn_count): Update.
18292 (struct decl_warn_count): Update.
18293 (struct final_warning_record): Update.
18294 (possible_polymorphic_call_targets): Update.
18295 (ipa_devirt): Update.
18296 * ipa-fnsummary.c (redirect_to_unreachable): Update.
18297 * ipa-icf.c (sem_function::merge): Update.
18298 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
18299 * ipa-inline.c (compute_uninlined_call_time): Update.
18300 (compute_inlined_call_time): Update.
18301 (want_inline_small_function_p): Update.
18302 (want_inline_self_recursive_call_p): Update.
18303 (edge_badness): Update.
18304 (lookup_recursive_calls): Update.
18305 (recursive_inlining): Update.
18306 (inline_small_functions): Update.
18307 (dump_overall_stats): Update.
18308 (dump_inline_stats): Update.
18309 * ipa-profile.c (ipa_profile_generate_summary): Update.
18310 (ipa_propagate_frequency): Update.
18311 (ipa_profile): Update.
18312 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18313 * ipa-utils.c (ipa_merge_profiles): Update.
18314 * loop-doloop.c (doloop_modify): Update.
18315 * loop-unroll.c (report_unroll): Update.
18316 (unroll_loop_runtime_iterations): Update.
18317 * lto-cgraph.c (lto_output_edge): Update.
18318 (lto_output_node): Update.
18319 (input_node): Update.
18320 (input_edge): Update.
18321 (merge_profile_summaries): Update.
18322 * lto-streamer-in.c (input_cfg): Update.
18323 * lto-streamer-out.c (output_cfg): Update.
18324 * mcf.c (create_fixup_graph): Update.
18325 (adjust_cfg_counts): Update.
18326 (sum_edge_counts): Update.
18327 * modulo-sched.c (sms_schedule): Update.
18328 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
18329 * predict.c (maybe_hot_count_p): Update.
18330 (probably_never_executed): Update.
18331 (dump_prediction): Update.
18332 (combine_predictions_for_bb): Update.
18333 (propagate_freq): Update.
18334 (handle_missing_profiles): Update.
18335 (counts_to_freqs): Update.
18336 (rebuild_frequencies): Update.
18337 (force_edge_cold): Update.
18338 * predict.h: Include profile-count.h
18339 (maybe_hot_count_p, counts_to_freqs): UPdate.
18340 * print-rtl-function.c: Do not include cfg.h
18341 * print-rtl.c: Include basic-block.h
18342 * profile-count.c: New file.
18343 * profile-count.h: New file.
18344 * profile.c (is_edge_inconsistent): Update.
18345 (correct_negative_edge_counts): Update.
18346 (is_inconsistent): Update.
18347 (set_bb_counts): Update.
18348 (read_profile_edge_counts): Update.
18349 (compute_frequency_overlap): Update.
18350 (compute_branch_probabilities): Update; Initialize and deinitialize
18352 (branch_prob): Update.
18353 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
18354 (edge_gcov_count): New.
18355 (bb_gcov_count): New.
18356 * shrink-wrap.c (try_shrink_wrapping): Update.
18357 * tracer.c (better_p): Update.
18358 * trans-mem.c (expand_transaction): Update.
18359 (ipa_tm_insert_irr_call): Update.
18360 (ipa_tm_insert_gettmclone_call): Update.
18361 * tree-call-cdce.c: Update.
18362 * tree-cfg.c (gimple_duplicate_sese_region): Update.
18363 (gimple_duplicate_sese_tail): Update.
18364 (gimple_account_profile_record): Update.
18365 (execute_fixup_cfg): Update.
18366 * tree-inline.c (copy_bb): Update.
18367 (copy_edges_for_bb): Update.
18368 (initialize_cfun): Update.
18369 (freqs_to_counts): Update.
18370 (copy_cfg_body): Update.
18371 (expand_call_inline): Update.
18372 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18373 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
18374 (try_unroll_loop_completely): Update.
18375 (try_peel_loop): Update.
18376 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18377 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
18378 * tree-ssa-loop-split.c (connect_loops): Update.
18379 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
18380 * tree-ssa-reassoc.c (branch_fixup): Update.
18381 * tree-ssa-tail-merge.c (replace_block_by): Update.
18382 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
18383 (compute_path_counts): Update.
18384 (update_profile): Update.
18385 (recompute_probabilities): Update.
18386 (update_joiner_offpath_counts): Update.
18387 (estimated_freqs_path): Update.
18388 (freqs_to_counts_path): Update.
18389 (clear_counts_path): Update.
18390 (ssa_fix_duplicate_block_edges): Update.
18391 (duplicate_thread_path): Update.
18392 * tree-switch-conversion.c (case_bit_test_cmp): Update.
18393 (struct switch_conv_info): Update.
18394 * tree-tailcall.c (decrease_profile): Update.
18395 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18396 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18397 * value-prof.c (check_counter): Update.
18398 (gimple_divmod_fixed_value): Update.
18399 (gimple_mod_pow2): Update.
18400 (gimple_mod_subtract): Update.
18401 (gimple_ic_transform): Update.
18402 (gimple_stringop_fixed_value): Update.
18403 * value-prof.h (gimple_ic): Update.
18405 2017-06-02 Carl Love <cel@us.ibm.com>
18407 * config/rs6000/rs6000-c: Add support for built-in functions
18408 vector double vec_doublee (vector signed int);
18409 vector double vec_doublee (vector unsigned int);
18410 vector double vec_doublee (vector float);
18411 vector double vec_doubleh (vector signed int);
18412 vector double vec_doubleh (vector unsigned int);
18413 vector double vec_doubleh (vector float);
18414 vector double vec_doublel (vector signed int);
18415 vector double vec_doublel (vector unsigned int);
18416 vector double vec_doublel (vector float);
18417 vector double vec_doubleo (vector signed int);
18418 vector double vec_doubleo (vector unsigned int);
18419 vector double vec_doubleo (vector float);.
18420 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
18421 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
18423 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
18424 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
18425 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
18427 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
18428 vec_doublel, vec_doubleh.
18429 * doc/extend.texi: Update the built-in documentation file for the
18430 new built-in functions.
18432 2017-06-02 David Malcolm <dmalcolm@redhat.com>
18435 * ipa-inline-analysis.c (free_growth_caches): Set
18436 edge_removal_hook_holder to NULL after removing it.
18438 2017-06-02 Sudakshina Das <sudi.das@arm.com>
18440 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
18441 comparision with zero.
18443 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
18444 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18445 for early expansion of vec_min and vec_max builtins.
18446 (builtin_function_type): Add min/max unsigned variants to those
18447 identified as having unsigned arguments.
18449 2017-06-02 Olivier Hainque <hainque@adacore.com>
18451 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
18453 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18455 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
18456 Use VALL_F16 iterator rather than VALL.
18458 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18460 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
18461 Emit CBNZ inside loop when doing a strong exchange and comparing
18462 against zero. Generate the CC flags after the loop.
18464 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
18466 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
18467 (dl_section_ref): New.
18468 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
18469 On AIX, append an expression to subtract the size of the
18470 section length to dl_section_ref.
18472 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
18474 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18475 for early expansion of vector absolute builtins.
18477 2017-06-02 Richard Biener <rguenther@suse.de>
18479 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
18480 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
18482 2017-06-02 Richard Biener <rguenther@suse.de>
18484 PR tree-optimization/80948
18485 * tree-tailcall.c (find_tail_calls): Track stmts to move in
18486 stmt order as well.
18488 2017-06-02 Richard Biener <rguenther@suse.de>
18490 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
18492 * tree-vect-stmts.c (process_use): Do not mark backedge defs
18493 for inductions as relevant.
18495 2017-06-02 Richard Biener <rguenther@suse.de>
18497 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
18498 (vectorizable_induction): ... this. Remove dead code.
18500 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
18502 * builtins. (expand_builtin_alloca): Remove second parameter and
18503 infer its value from the first parameter instead.
18504 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
18506 2017-06-02 Jakub Jelinek <jakub@redhat.com>
18508 PR rtl-optimization/80903
18509 * loop-doloop.c (add_test): Unshare sequence.
18511 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
18513 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
18515 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
18517 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
18519 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
18520 xlogue_layout::get_instance, logue_layout::xlogue_layout,
18521 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
18522 (xlogue_layout::get_stub_rtx): Make static.
18523 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
18524 (xlogue_layout::compute_stub_managed_regs): Rename to...
18525 (xlogue_layout::count_stub_managed_regs): ...this.
18526 (xlogue_layout::is_stub_managed_reg): New function.
18527 (xlogue_layout::m_stub_names): Rename to...
18528 (xlogue_layout::s_stub_names): ...this, make static.
18529 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
18530 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
18531 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
18532 xlogue_layout::s_stub_names): Instantiate statics.
18533 (stub_managed_regs): Remove.
18534 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
18535 (disable_call_ms2sysv_xlogues): Rename to...
18536 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
18537 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
18539 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
18540 change after reload_completed.
18541 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
18543 (ix86_expand_prologue): Likewise.
18544 (ix86_expand_epilogue): Likewise.
18545 (ix86_expand_split_stack_prologue): Likewise.
18546 (ix86_compute_frame_layout): Remove frame parameter ...
18547 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
18548 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
18550 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
18551 (ix86_frame): Move from here ...
18552 * config/i386/i386.h (ix86_frame): ... to here.
18553 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
18554 complete ix86_frame data structure instead. Remove some_ld_name.
18556 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
18558 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
18559 symbols that hold a DECL_VALUE_EXPR.
18561 2017-06-01 Martin Jambor <mjambor@suse.cz>
18563 PR tree-optimization/80898
18564 * tree-sra.c (process_subtree_disqualification): Removed.
18565 (disqualify_candidate): Do not acll
18566 process_subtree_disqualification.
18567 (subtree_mark_written_and_enqueue): New function.
18568 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
18569 RHS has been disqualified and re-queue LHS if necessary. Apart
18570 from that, ignore disqualified RHS.
18572 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18574 * config/s390/s390.c (s390_emit_epilogue): Disable early return
18575 address fetch for z10 or later.
18577 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18579 * config/arc/arc.md (tst_movb): Add guard when splitting.
18581 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18583 * config/arc/arc.c (arc_can_eliminate): Test against
18584 arc_frame_pointer_needed.
18586 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18588 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
18589 to prevent store reordering.
18590 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
18591 (type): Add block type.
18592 (stack_tie): Define special instruction to be used in
18595 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18597 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
18598 constraint. It is not valid for the pattern.
18599 (noncommutative_binary_comparison): Likewise.
18601 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18603 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
18606 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18608 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
18609 be used by the reg-alloc.
18611 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18613 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
18614 reg-alloc when having mul64 or mul32x16 instructions.
18615 (mulsidi3): Likewise.
18616 (umulsidi3): Likewise.
18617 (mulsi32x16): New pattern.
18618 (mulsi64): Likewise.
18619 (mulsidi64): Likewise.
18620 (umulsidi64): Likewise.
18621 (MUL32x16_REG): Define.
18622 (mul64_600): Use MUL32x16_REG.
18623 (mac64_600): Likewise.
18624 (umul64_600): Likewise.
18625 (umac64_600): Likewise.
18627 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18629 * config/arc/arc.md (mulsi3_700): Make it commutative.
18631 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
18633 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
18635 (*sign_extendsidi2_insn): Likewise for movstosw.
18637 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
18639 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
18640 the type of the input discriminant value. Convert the
18641 discriminant value of signedness vary.
18643 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
18645 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
18646 Add to -Wall section.
18648 2017-06-01 Richard Biener <rguenther@suse.de>
18650 PR middle-end/66313
18651 * fold-const.c (fold_plusminus_mult_expr): If the factored
18652 factor may be zero use a wrapping type for the inner operation.
18653 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
18654 and handle moved defs.
18655 (process_assignment): Properly guard the unary op case. Return a
18656 tri-state indicating that moving the stmt before the call may allow
18657 to continue. Pass through to_move.
18658 (find_tail_calls): Handle moving unrelated defs before
18661 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
18664 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
18665 splitter result in the canonical way.
18667 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
18669 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
18670 also for 32bit target. Update insn attributes.
18671 (zero-extendsidi2 splitter): Allow all registers for operand 1.
18673 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
18675 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
18676 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
18677 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
18678 (_mm_maskz_min_ss): New intrinsics.
18680 2017-05-31 Martin Liska <mliska@suse.cz>
18682 * tree-vect-loop.c (vect_create_epilog_for_reduction):
18683 Change comment style to one we normally use.
18684 (vectorizable_reduction): Likewise.
18685 (vectorizable_induction): Likewise.
18686 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
18687 (vectorizable_call): Likewise.
18688 (vectorizable_simd_clone_call): Likewise.
18689 (vectorizable_conversion): Likewise.
18690 (vectorizable_assignment): Likewise.
18691 (vectorizable_shift): Likewise.
18692 (vectorizable_operation): Likewise.
18693 (vectorizable_store): Likewise.
18694 (vectorizable_load): Likewise.
18695 * tree-vectorizer.h: Likewise.
18697 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
18699 * passes.c (emergency_dump_function): New.
18700 * tree-pass.h (emergency_dump_function): Declare.
18701 * plugin.c (plugins_internal_error_function): Remove.
18702 * plugin.h (plugins_internal_error_function): Remove declaration.
18703 * toplev.c (internal_error_function): New static function. Use it...
18704 (general_init): ...here.
18706 2017-05-31 Graham Markall <graham.markall@embecosm.com>
18708 * config/arc/arc.c (arc_print_operand): Handle constant operands.
18709 (arc_rtx_costs): Add costs for new patterns.
18710 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
18711 * config/arc/predicates.md: Add _1_2_3_operand predicate.
18713 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
18715 * tree-ssa-strlen.c (get_next_strinfo): New function.
18716 (get_stridx_plus_constant): Use it.
18717 (zero_length_string): Likewise.
18718 (adjust_related_strinfos): Likewise.
18719 (adjust_last_stmt): Likewise.
18721 2017-05-31 Richard Biener <rguenther@suse.de>
18724 * config/i386/i386.c (ix86_expand_builtin): Remove assert
18725 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
18727 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
18729 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
18730 loop_vinfo argument and use of dependence distance vectors.
18731 Check instead whether the two references differ only in their
18732 initial value and assume that they have the same alignment if the
18733 difference is a multiple of the vector alignment.
18734 (vect_analyze_data_refs_alignment): Update call accordingly.
18736 2017-05-31 Martin Liska <mliska@suse.cz>
18739 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
18741 2017-05-31 Bin Cheng <bin.cheng@arm.com>
18743 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
18744 (create_intersect_range_checks): Move from ...
18745 * tree-data-ref.c (create_intersect_range_checks_index)
18746 (create_intersect_range_checks): ... to here.
18747 (create_runtime_alias_checks): New function factored from ...
18748 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
18749 here. Call above function.
18750 * tree-data-ref.h (create_runtime_alias_checks): New function.
18752 2017-05-31 Bin Cheng <bin.cheng@arm.com>
18754 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
18755 segment length for dr_b and compute it in wide_int.
18757 2017-05-31 Richard Biener <rguenther@suse.de>
18759 PR tree-optimization/80906
18760 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
18761 and pass through iv_map.
18762 (copy_bb_and_scalar_dependences): Adjust.
18763 (translate_pending_phi_nodes): Likewise.
18764 (copy_loop_close_phi_args): Handle code-generating IVs instead
18767 2017-05-30 David Malcolm <dmalcolm@redhat.com>
18769 * diagnostic-color.c (color_dict): Add "type-diff".
18770 (parse_gcc_colors): Update comment.
18771 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
18772 -fdiagnostics-show-template-tree and -fno-elide-type.
18773 (GCC_COLORS): Add type-diff to example.
18775 (-fdiagnostics-show-template-tree): New.
18776 (-fno-elide-type): New.
18777 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
18778 the pp_format_decoder callback. Call any m_format_postprocessor's
18780 (pretty_printer::pretty_printer): Initialize
18781 m_format_postprocessor.
18782 (pretty_printer::~pretty_printer): Delete any
18783 m_format_postprocessor.
18784 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
18785 (class format_postprocessor): New class.
18786 (struct pretty_printer::format_decoder): Document the new parameters.
18787 (struct pretty_printer::m_format_postprocessor): New field.
18788 * tree-diagnostic.c (default_tree_printer): Update for new
18789 bool and const char ** params.
18790 * tree-diagnostic.h (default_tree_printer): Likewise.
18792 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
18794 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
18795 (lwa_operand): Delete rs6000_gen_cell_microcode test.
18796 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
18797 rs6000_gen_cell_microcode code.
18798 (rs6000_final_prescan_insn): Delete.
18799 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
18800 "warn-cell-microcode" entries.
18801 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
18802 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
18803 throughout. Change cc_reg_not_micro_cr0_operand to
18804 cc_reg_not_cr0_operand throughout.
18805 (*extendhi<mode>2_noload): Delete.
18806 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
18807 (mwarn-cell-microcode): Delete.
18808 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
18809 -mgen-cell-microcode and -mwarn-cell-microcode.
18811 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
18814 * config/i386/constraints.md (Yd): New constraint.
18816 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
18817 and (?Yd, r) alternatives. Update insn attributes.
18818 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
18819 and (?*Yd, r) alternatives. Update insn attributes.
18820 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
18822 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
18824 * gimplify.c (gimplify_modify_expr): Don't create a
18825 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
18828 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
18830 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
18832 2017-05-30 Richard Biener <rguenther@suse.de>
18834 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
18835 and reduc_def fields.
18836 (STMT_VINFO_REDUC_TYPE): New define.
18837 (STMT_VINFO_REDUC_DEF): Likewise.
18838 (vect_force_simple_reduction): Adjust prototype.
18839 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
18840 (vect_is_simple_reduction): Remove check_reduction argument.
18841 (vect_force_simple_reduction): Adjust and set
18842 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
18843 (vectorizable_reduction): Do not re-do reduction analysis
18844 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
18845 * tree-parloops.c (gather_scalar_reductions): Adjust.
18847 2017-05-30 Richard Biener <rguenther@suse.de>
18849 PR middle-end/80901
18850 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
18853 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
18855 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
18856 Introduce unknown_misalignment parameter and remove vf.
18857 (vect_peeling_hash_get_lowest_cost):
18858 Pass unknown_misalignment parameter.
18859 (vect_enhance_data_refs_alignment):
18860 Fix unsupportable data ref treatment.
18862 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18864 * tree-vect-data-refs.c (vect_get_data_access_cost):
18865 Workaround for SLP handling.
18866 (vect_enhance_data_refs_alignment):
18867 Compute costs for doing no peeling at all, compare to the best
18868 peeling costs so far and avoid peeling if cheaper.
18870 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18872 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
18873 Return peeling info and set costs to zero for unlimited cost
18875 (vect_enhance_data_refs_alignment): Also inspect all datarefs
18876 with unknown misalignment. Compute and costs for unknown
18877 misalignment, compare them to the costs for known misalignment
18878 and choose the cheapest for peeling.
18880 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18882 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
18883 (vect_get_peeling_costs_all_drs): Create function.
18884 (vect_peeling_hash_get_lowest_cost):
18885 Use vect_get_peeling_costs_all_drs.
18886 (vect_peeling_supportable): Create function.
18887 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
18889 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18891 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
18892 DR_HAS_NEGATIVE_STEP.
18893 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
18894 (vect_enhance_data_refs_alignment): Use.
18895 (vect_duplicate_ssa_name_ptr_info): Use.
18896 * tree-vectorizer.h (dr_misalignment): Use.
18897 (known_alignment_for_access_p): Use.
18899 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18902 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
18904 (has_section_name): New function.
18906 2017-05-30 Martin Liska <mliska@suse.cz>
18909 * auto-profile.c (get_function_decl_from_block): Fix
18912 2017-05-30 Richard Biener <rguenther@suse.de>
18914 PR middle-end/80876
18915 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
18917 2017-05-30 Martin Liska <mliska@suse.cz>
18919 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
18920 * dumpfile.h (struct dump_file_info): Remove ctors.
18922 2017-05-30 Martin Liska <mliska@suse.cz>
18924 * predict.def: Fix GNU coding style.
18926 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
18928 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
18929 Mark 'to' argument with ATTRIBUTE_UNUSED.
18931 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
18933 * config/xtensa/xtensa.c (xtensa_emit_call): Use
18934 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
18935 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
18938 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
18940 * doc/install.texi (Options specification): Restore entry of
18941 --enable-sjlj-exceptions.
18943 2017-05-27 Michael Eager <eager@eagercon.com>
18946 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
18948 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
18950 * config/microblaze/microblaze.h
18951 (FIXED_REGISTERS): Update in macro.
18952 (CALL_USED_REGISTERS): Update in macro.
18954 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
18956 * doc/install.texi: Add links to macOS binary distributions.
18958 2017-05-27 Jakub Jelinek <jakub@redhat.com>
18962 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18964 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
18966 2017-05-26 Martin Liska <mliska@suse.cz>
18968 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
18970 2017-05-26 Martin Liska <mliska@suse.cz>
18972 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
18973 always leading ';; '.
18974 (dump_bb_info): Likewise.
18975 (brief_dump_cfg): Likewise.
18976 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
18977 * dumpfile.c: Remove usage of TDF_VERBOSE.
18978 * dumpfile.h (enum dump_kind): Likewise.
18979 (dump_gimple_bb_header): Do not use TDF_COMMENT.
18980 * print-tree.c (debug_verbose): Remove.
18981 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
18982 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
18983 * tree-diagnostic.c (default_tree_printer): Replace
18984 TDF_DIAGNOSTIC with TDF_SLIM.
18986 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18988 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
18989 in parameter loop, rather than loop_vinfo.
18990 (create_intersect_range_checks): Ditto.
18991 (vect_create_cond_for_alias_checks): Update call to above functions.
18993 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18995 PR tree-optimization/80815
18996 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
18997 for merging runtime alias checks. Handle negative DR_STEPs.
18999 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19001 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
19003 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
19004 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
19005 out code pruning runtime alias checks.
19006 * tree-data-ref.c (prune_runtime_alias_test_list): New function
19007 factored out from above.
19008 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
19010 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
19012 (prune_runtime_alias_test_list): New decalaration.
19014 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19016 * tree-vect-data-refs.c (compare_tree): Rename and move ...
19017 * tree-data-ref.c (data_ref_compare_tree): ... to here.
19018 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
19019 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
19020 (operator==, comp_dr_with_seg_len_pair): Ditto.
19021 (vect_prune_runtime_alias_test_list): Ditto.
19023 2017-05-26 Martin Liska <mliska@suse.cz>
19026 * params.def: Bound partial-inlining-entry-probability param.
19028 2017-05-26 Marek Polacek <polacek@redhat.com>
19031 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
19034 2017-05-26 Richard Biener <rguenther@suse.de>
19036 PR tree-optimization/80842
19037 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
19040 2017-05-26 Richard Biener <rguenther@suse.de>
19042 PR tree-optimization/80844
19043 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
19045 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
19047 * doc/md.texi (Machine Constraints): Update x86 family
19048 machine constraints section to match 'config/i386/constraints.md'.
19050 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
19052 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
19054 2017-05-25 Nathan Sidwell <nathan@acm.org>
19056 * doc/invoke.texi (--enable-languages): Update documentation.
19058 2017-05-25 Martin Liska <mliska@suse.cz>
19060 * dumpfile.c: Add TDF_FOLDING.
19061 * dumpfile.h (enum dump_kind): Likewise.
19062 * genmatch.c (dt_simplify::gen_1): Use it.
19064 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19066 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
19068 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19070 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
19071 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
19073 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19075 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
19076 * match.pd (X == C): Rewrite it here.
19077 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
19078 with_certain_nonzero_bits2): New predicates.
19079 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
19081 2017-05-24 Nathan Sidwell <nathan@acm.org>
19083 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
19086 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
19089 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
19091 * config/powerpcspe: New port. Files are copied from the rs6000
19092 port, with "rs6000" in filenames replaced by "powerpcspe".
19094 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
19096 PR rtl-optimization/80754
19097 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
19099 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
19101 * config/sparc/sparc.md (length): Return the correct value for -mflat
19102 sibcalls to match output_sibcall.
19104 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
19108 * config/rs6000/rs6000.c (struct machine_function): Add new field
19110 (rs6000_get_separate_components): Init that field, use it.
19111 (rs6000_components_for_bb): Use the field.
19113 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
19115 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
19117 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
19119 PR middle-end/80823
19120 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
19122 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19125 * config/s390/s390.c (s390_check_qrst_address): Check incoming
19126 address against address_operand predicate.
19127 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
19129 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
19131 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
19132 parameters passed indirectly.
19134 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
19136 * config/i386/i386.md (*movdi_internal): Remove SSE4
19137 alternative 18 (?r, *v). Update insn attributes.
19138 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
19139 Update insn attributes.
19140 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
19141 Update insn attributes.
19142 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
19143 alternative 1 (r, v). Remove isa attribute.
19144 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
19145 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
19146 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
19148 2017-05-23 Tom de Vries <tom@codesourcery.com>
19150 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
19153 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19155 * cgraphunit.c (symbol_table::process_new_functions): Update.
19156 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
19157 (inline_generate_summary): Rename to ...
19158 (ipa_fn_summary_generate): ... this one.
19159 (inline_read_summary): Rename to ...
19160 (ipa_fn_summary_read): ... this one.
19161 (inline_write_summary): Rename to ...
19162 (ipa_fn_summary_write): ... this one.
19163 (inline_free_summary): Rename to ...
19164 (ipa_free_fn_summary): ... this one.
19165 (pass_data_local_fn_summary, pass_local_fn_summary,
19166 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
19167 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
19168 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
19169 make_pass_ipa_fn_summary): New.
19170 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
19171 inline_write_summary, inline_free_summary): Remove.
19172 (ipa_free_fn_summary) : New.
19173 * ipa-inline.c (ipa_inline): Update.
19174 (pass_ipa_inline): Do not generate summaries.
19175 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
19177 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
19178 and add pass_ipa_fn_summary.
19179 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
19181 (make_pass_inline_parameters): Remove.
19183 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
19185 * omp-low.c (struct omp_context): Remove "default_kind" member.
19188 * omp-offload.c (execute_oacc_device_lower): Remove the
19189 parallelism dimensions function attributes for unparallelized
19190 OpenACC kernels constructs.
19192 2017-05-23 Martin Liska <mliska@suse.cz>
19194 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
19196 (cgraph_edge::make_speculative): Likewise.
19197 (cgraph_edge::resolve_speculation): Likewise.
19198 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
19199 (cgraph_node::dump): Likewise.
19200 * cgraph.h: Likewise.
19201 * cgraphunit.c (analyze_functions): Likewise.
19202 (symbol_table::compile): Likewise.
19203 * ipa-cp.c (print_all_lattices): Likewise.
19204 (determine_versionability): Likewise.
19205 (initialize_node_lattices): Likewise.
19206 (ipcp_verify_propagated_values): Likewise.
19207 (estimate_local_effects): Likewise.
19208 (update_profiling_info): Likewise.
19209 (create_specialized_node): Likewise.
19210 (perhaps_add_new_callers): Likewise.
19211 (decide_about_value): Likewise.
19212 (decide_whether_version_node): Likewise.
19213 (identify_dead_nodes): Likewise.
19214 (ipcp_store_bits_results): Likewise.
19215 * ipa-devirt.c (dump_targets): Likewise.
19216 (ipa_devirt): Likewise.
19217 * ipa-icf.c (sem_item::dump): Likewise.
19218 (sem_function::equals): Likewise.
19219 (sem_variable::equals): Likewise.
19220 (sem_item_optimizer::read_section): Likewise.
19221 (sem_item_optimizer::execute): Likewise.
19222 (congruence_class::dump): Likewise.
19223 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
19224 (dump_inline_summary): Likewise.
19225 (estimate_node_size_and_time): Likewise.
19226 (inline_analyze_function): Likewise.
19227 * ipa-inline-transform.c (inline_call): Likewise.
19228 * ipa-inline.c (report_inline_failed_reason): Likewise.
19229 (want_early_inline_function_p): Likewise.
19230 (edge_badness): Likewise.
19231 (update_edge_key): Likewise.
19232 (inline_small_functions): Likewise.
19233 * ipa-profile.c (ipa_profile): Likewise.
19234 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
19235 (ipa_make_edge_direct_to_target): Likewise.
19236 (remove_described_reference): Likewise.
19237 (ipa_impossible_devirt_target): Likewise.
19238 (propagate_controlled_uses): Likewise.
19239 (ipa_print_node_params): Likewise.
19240 (ipcp_transform_function): Likewise.
19241 * ipa-pure-const.c (pure_const_read_summary): Likewise.
19242 (propagate_pure_const): Likewise.
19243 * ipa-reference.c (generate_summary): Likewise.
19244 (read_write_all_from_decl): Likewise.
19245 (propagate): Likewise.
19246 (ipa_reference_read_optimization_summary): Likewise.
19247 * ipa-utils.c (ipa_merge_profiles): Likewise.
19248 * ipa.c (walk_polymorphic_call_targets): Likewise.
19249 (symbol_table::remove_unreachable_nodes): Likewise.
19250 (ipa_single_use): Likewise.
19251 * passes.c (execute_todo): Likewise.
19252 * predict.c (drop_profile): Likewise.
19253 * symtab.c (symtab_node::get_dump_name): New function.
19254 (symtab_node::dump_name): Likewise.
19255 (symtab_node::dump_asm_name): Likewise.
19256 (symtab_node::dump_references): Likewise.
19257 (symtab_node::dump_referring): Likewise.
19258 (symtab_node::dump_base): Likewise.
19259 (symtab_node::debug_symtab): Likewise.
19260 * tree-sra.c (convert_callers_for_node): Likewise.
19261 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
19262 * value-prof.c (init_node_map): Likewise.
19264 2017-05-23 Martin Liska <mliska@suse.cz>
19266 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
19267 and symtab_node::debug_symtab to symbol_table::debug.
19268 * cgraphunit.c (analyze_functions): Use the renamed function.
19269 (symbol_table::compile): Likewise.
19270 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
19271 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
19272 * passes.c (execute_todo): Likewise.
19273 * symtab.c (symbol_table::dump): New function.
19274 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
19276 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19278 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
19279 that nonconst implies exec.
19281 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19283 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
19284 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
19285 (inline_edge_summary_vec): Turn into ...
19286 (ipa_call_summaries): ... this one.
19287 (redirect_to_unreachable, edge_set_predicate,
19288 evaluate_properties_for_edge, inline_summary_alloc,
19289 reset_ipa_call_summary, reset_inline_summary,
19290 inline_summary_t::duplicate): Update.
19291 (inline_edge_duplication_hook): Turn to ...
19292 (ipa_call_summary_t::duplicate): ... this one.
19293 (inline_edge_removal_hook): Turn to ...
19294 (ipa_call_summary_t::remove): ... this one.
19295 (dump_inline_edge_summary): Turn to ...
19296 (dump_ipa_call_summary): ... this one.
19297 (estimate_function_body_sizes): Update.
19298 (inline_update_callee_summaries): Update.
19299 (remap_edge_change_prob): Update.
19300 (remap_edge_summaries): Update.
19301 (inline_merge_summary): Update.
19302 (do_estimate_edge_time): Update.
19303 (inline_generate_summary): Update.
19304 (inline_read_section): Update.
19305 (inline_read_summary): Update.
19306 (inline_free_summary): Update.
19307 * ipa-inline.c (can_inline_edge_p): Update.
19308 (compute_inlined_call_time): Update.
19309 (want_inline_small_function_p): Update.
19310 (edge_badness): Update.
19311 (early_inliner): Update.
19312 * ipa-inline.h (inline_edge_summary): Turn to ...
19313 (ipa_call_summary): ... this one.
19314 (ipa_call_summary_t): New class.
19315 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
19316 (ipa_call_summaries): New.
19317 (inline_edge_summary): Remove.
19318 (estimate_edge_growth): Update.
19319 * ipa-profile.c (ipa_propagate_frequency_1): Update.
19320 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19321 * ipa-split.c (execute_split_functions): Update.
19322 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
19324 2017-05-23 Tom de Vries <tom@codesourcery.com>
19326 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
19327 attributes): Document rdrand effective target.
19329 2017-05-23 Tom de Vries <tom@codesourcery.com>
19331 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
19332 attributes): Sort alphabetically.
19334 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
19336 * config/avr/genmultilib.awk: Use gsub instead of gensub.
19338 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
19341 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
19342 V2DF/V2DI splat into two separate patterns, one that handles
19343 registers, and the other that only handles memory. Drop support
19344 for splatting from a GPR on ISA 2.07 and then splitting the
19345 splat into direct move and splat.
19346 (vsx_splat_<mode>_reg): Likewise.
19347 (vsx_splat_<mode>_mem): Likewise.
19349 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
19351 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
19353 2017-05-22 Jakub Jelinek <jakub@redhat.com>
19355 PR middle-end/80809
19356 * omp-low.c (finish_taskreg_remap): New function.
19357 (finish_taskreg_scan): If unit size of ctx->record_type
19358 is non-constant, unshare the size expression and replace
19359 decls in it with possible outer var refs.
19361 PR middle-end/80809
19362 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
19363 GOVD_SHARED rather than GOVD_PRIVATE with it.
19364 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
19365 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
19367 PR middle-end/80853
19368 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
19369 as last argument to build_outer_var_ref for pointer bases of array
19370 section reductions.
19372 2017-05-19 Martin Sebor <msebor@redhat.com>
19374 * print-tree.c (print_node): Print DECL_READ_P flag.
19376 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19378 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
19379 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
19380 * cgraph.c: Likewise.
19381 * cgraphunit.c: Likewise.
19382 * gengtype.c: Likewise.
19383 * ipa-cp.c: Likewise.
19384 * ipa-devirt.c: Likewise.
19385 * ipa-icf.c: Likewise.
19386 * ipa-predicate.c: Likewise.
19387 * ipa-profile.c: Likewise.
19388 * ipa-prop.c: Likewise.
19389 * ipa-split.c: Likewise.
19391 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
19392 edge_predicate_pool, dump_inline_hints,
19393 inline_summary::account_size_time, redirect_to_unreachable,
19394 edge_set_predicate, set_hint_predicate,
19395 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
19396 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
19397 inline_summary_t::remove, remap_hint_predicate_after_duplication,
19398 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
19399 ipa_call_summary_t::remove, initialize_growth_caches,
19400 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
19401 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
19402 mark_modified, unmodified_parm_1, unmodified_parm,
19403 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
19404 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
19405 compute_bb_predicates, will_be_nonconstant_expr_predicate,
19406 will_be_nonconstant_predicate, record_modified_bb_info,
19407 get_minimal_bb, record_modified, param_change_prob,
19408 phi_result_unknown_predicate, predicate_for_phi_result,
19409 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
19410 estimate_function_body_sizes, compute_inline_parameters,
19411 compute_inline_parameters_for_curren, pass_data_inline_parameters,
19412 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
19413 inline_update_callee_summaries, remap_edge_change_prob,
19414 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
19415 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
19416 inline_analyze_function, inline_summary_t::insert,
19417 inline_generate_summary, read_ipa_call_summary, inline_read_section,
19418 inline_read_summary, write_ipa_call_summary, inline_write_summary,
19419 inline_free_summary): Move to ipa-fnsummary.h
19420 (predicate_t): Remove.
19421 * ipa-fnsummary.c: New file.
19422 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
19423 (enum inline_hints_vals, inline_hints, agg_position_info,
19424 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
19425 inline_summaries, ipa_call_summary, ipa_call_summary_t,
19426 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
19427 dump_inline_summary, dump_inline_hints, inline_generate_summary,
19428 inline_read_summary, inline_write_summary, inline_free_summary,
19429 inline_analyze_function, initialize_inline_failed,
19430 inline_merge_summary, inline_update_overall_summary,
19431 compute_inline_parameters): Move to ipa-fnsummary.h
19432 * ipa-fnsummary.h: New file.
19433 * ipa-inline-transform.h: Include ipa-inline.h.
19434 * ipa-inline.c: LIkewise.
19436 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19438 * ipa-inline.c (edge_badness): Use inlined_time instead of
19439 inline_summaries->get.
19441 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19443 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
19445 2017-05-22 Nathan Sidwell <nathan@acm.org>
19447 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
19448 (fdump-lang): Document 'raw' option.
19449 * dumpfile.h (TDI_tu): Delete.
19450 * dumpfile.c (dump_files): Remove translation-unit.
19451 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
19453 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
19455 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
19456 command option from $(AWK) call.
19457 * config/avr/genmultilib.awk: Simplify and rewrite so that it
19458 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
19459 [FORMAT]: Remove handling of variable.
19460 * config/avr/t-multilib: Regenerate.
19462 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19464 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
19466 (dump_inline_summary): Do not print self_time.
19467 (estimate_function_body_sizes): Do not set self_time.
19468 (compute_inline_parameters): Likewise.
19469 (inline_read_section, inline_write_summary): Do not stream self_time.
19470 * ipa-inline.h (inline_summary): Drop self_time.
19472 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19474 * ipa-inline-analysis.c (account_size_time): Rename to ...
19475 (inline_summary::account_size_time): ... this one.
19476 (reset_ipa_call_summary): Turn to ...
19477 (ipa_call_summary::reset): ... this one.
19478 (reset_inline_summary): Turn to ...
19479 (inline_summary::reset): ... this one.
19480 (inline_summary_t::remove): Update.
19481 (inline_summary_t::duplicate): Update.
19482 (ipa_call_summary_t::remove): Update.
19483 (dump_inline_summary): Update.
19484 (estimate_function_body_sizes): Update.
19485 (compute_inline_parameters): Update.
19486 (estimate_node_size_and_time): Update.
19487 (inline_merge_summary): Update.
19488 (inline_update_overall_summary): Update.
19489 (inline_read_section): Update.
19490 (inline_write_summary): Update.
19491 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
19492 add account_size_time and reset member functions.
19493 (ipa_call_summary): Add reset function.
19494 * ipa-predicate.h (predicate::operator &): Constify.
19496 2017-05-22 Richard Biener <rguenther@suse.de>
19498 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
19500 2017-05-19 Jason Merrill <jason@redhat.com>
19502 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
19504 2017-05-19 Marek Polacek <polacek@redhat.com>
19507 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
19508 TYPE_OVERFLOW_WRAPS checks.
19510 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
19512 * tree-core.h (enum omp_clause_default_kind): Add
19513 "OMP_CLAUSE_DEFAULT_PRESENT".
19514 * tree-pretty-print.c (dump_omp_clause): Handle it.
19515 * gimplify.c (enum gimplify_omp_var_data): Add
19516 "GOVD_MAP_FORCE_PRESENT".
19517 (gimplify_adjust_omp_clauses_1): Map it to
19518 "GOMP_MAP_FORCE_PRESENT".
19519 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
19521 * gimplify.c (oacc_default_clause): Clarify.
19523 2017-05-19 Nathan Sidwell <nathan@acm.org>
19525 LANG_HOOK_REGISTER_DUMPS
19526 * toplev.c (general_init): Call register dump lang hook.
19527 * doc/invoke.texi: Document -fdump-lang option family.
19528 * dumpfile.c (dump_files): Remove class dump here.
19529 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
19530 * dumpfile.h (tree_dump_index): Remove TDI_class.
19531 * langhooks-def.h (lhd_register_dumps): Declare.
19532 (LANG_HOOKS_REGISTER_DUMPS): Define.
19533 (LANG_HOOKS_INITIALIZER): Add it.
19534 * langhooks.c (lhd_register_dumps): Define.
19535 * langhooks.h (struct lang_hooks): Add register_dumps.
19537 2017-05-19 Nathan Sidwell <nathan@acm.org>
19539 * context.h (context::set_passes): New.
19540 * context.c (context::context): Do not create pass manager.
19541 * toplev.c (general_init): Create pass manager here.
19543 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
19545 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
19546 use this splitter if two add or or instructions would also work for
19547 the constant we want to generate.
19549 2017-05-19 Richard Biener <rguenther@suse.de>
19552 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
19553 predicate evaluation.
19555 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
19557 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
19559 * ipa-inline.c (want_inline_small_function_p): Do not cast to
19562 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
19564 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
19565 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
19566 (inline_edge_summary_vec): Turn into ...
19567 (ipa_call_summaries): ... this one.
19568 (redirect_to_unreachable, edge_set_predicate,
19569 evaluate_properties_for_edge, inline_summary_alloc,
19570 reset_ipa_call_summary, reset_inline_summary,
19571 inline_summary_t::duplicate): Update.
19572 (inline_edge_duplication_hook): Turn to ...
19573 (ipa_call_summary_t::duplicate): ... this one.
19574 (inline_edge_removal_hook): Turn to ...
19575 (ipa_call_summary_t::remove): ... this one.
19576 (dump_inline_edge_summary): Turn to ...
19577 (dump_ipa_call_summary): ... this one.
19578 (estimate_function_body_sizes): Update.
19579 (inline_update_callee_summaries): Update.
19580 (remap_edge_change_prob): Update.
19581 (remap_edge_summaries): Update.
19582 (inline_merge_summary): Update.
19583 (do_estimate_edge_time): Update.
19584 (inline_generate_summary): Update.
19585 (inline_read_section): Update.
19586 (inline_read_summary): Update.
19587 (inline_free_summary): Update.
19588 * ipa-inline.c (can_inline_edge_p): Update.
19589 (compute_inlined_call_time): Update.
19590 (want_inline_small_function_p): Update.
19591 (edge_badness): Update.
19592 (early_inliner): Update.
19593 * ipa-inline.h (inline_edge_summary): Turn to ...
19594 (ipa_call_summary): ... this one.
19595 (ipa_call_summary_t): New class.
19596 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
19597 (ipa_call_summaries): New.
19598 (inline_edge_summary): Remove.
19599 (estimate_edge_growth): Update.
19600 * ipa-profile.c (ipa_propagate_frequency_1): Update.
19601 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19602 * ipa-split.c (execute_split_functions): Update.
19603 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
19605 2017-05-19 Richard Biener <rguenther@suse.de>
19607 PR middle-end/80764
19608 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
19610 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
19612 * config/rs6000/rs6000.c (struct machine_function): Add field
19613 fpr_is_wrapped_separately.
19614 (rs6000_get_separate_components): Use 64 components. Handle the
19615 new FPR components.
19616 (rs6000_components_for_bb): Handle the FPR components.
19617 (rs6000_emit_prologue_components): Handle the FPR components.
19618 (rs6000_emit_epilogue_components): Handle the FPR components.
19619 (rs6000_set_handled_components): Handle the FPR components.
19620 (rs6000_emit_prologue): Don't output prologue code for those FPRs
19621 that are already separately shrink-wrapped.
19622 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
19623 that are already separately shrink-wrapped.
19625 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
19628 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
19631 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
19632 (define_peephole2 for Altivec d-form load): Add peepholes to catch
19633 cases where the register allocator uses a move and an offsettable
19634 memory operation to/from a FPR register on ISA 2.06/2.07.
19635 (define_peephole2 for Altivec d-form store): Likewise.
19637 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
19640 * config/i386/mmx.md (*mov<mode>_internal): Enable
19641 alternatives 11, 12, 13 and 14 also for 32bit targets.
19642 Remove alternatives 15, 16, 17 and 18.
19643 * config/i386/sse.md (vec_concatv2di): Change
19644 alternative (!x, *y) to (x, ?!*Yn).
19646 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
19648 * dumpfile.h (enum dump_kind): Remove stray comma.
19650 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
19652 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
19653 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
19654 predicate::num_conditions
19655 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
19656 (CHANGED): turn into predicate::changed.
19657 (agg_position_info): Move to ipa-predicate.h
19658 (add_condition, predicate::add_clause, predicate::operator &=,
19659 predicate::or_with, predicate::evaluate, predicate::probability,
19660 dump_condition, dump_clause, predicate::dump,
19661 predicate::remap_after_duplication, predicate::remap_after_inlining,
19662 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
19663 (evaluate_conditions_for_known_args): Update.
19664 (set_cond_stmt_execution_predicate): Update.
19665 * ipa-inline.h: Include ipa-predicate.h
19666 (condition, inline_param_summary, conditions, agg_position_info,
19667 predicate): Move to ipa-predicate.h
19668 * ipa-predicate.c: New file.
19669 * ipa-predicate.h: New file.
19671 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
19673 * final.c (leaf_function_p): Check we are not in a sequence.
19675 2017-05-18 Martin Liska <mliska@suse.cz>
19677 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
19678 * dumpfile.c (dump_register): Use new enum dump_kind.
19679 (get_dump_file_name): Likewise.
19680 (dump_enable_all): Likewise.
19681 (dump_switch_p_1): Likewise.
19682 (enable_rtl_dump_file): Remove usage of TDF_RTL.
19683 * dumpfile.h (enum dump_kind): New enum type.
19684 (struct dump_file_info): Create constructor and
19685 format fields and comments.
19686 * passes.c (pass_manager::register_one_dump_file):
19688 * statistics.c (statistics_early_init): Likewise.
19689 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
19690 TDF_TREE with TDF_SLIM.
19691 (gather_memory_references_ref): Likewise.
19693 2017-05-18 Martin Liska <mliska@suse.cz>
19695 * vec.h (struct vnull): Use it.
19697 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
19699 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
19700 (true_predicate, false_predicate, true_predicate_p,
19701 false_predicate_p): Remove.
19702 (single_cond_predicate, not_inlined_predicate): Turn to member function
19704 (add_condition): Update.
19705 (add_clause): Turn to...
19706 (predicate::add_clause): ... this one; update; allow passing NULL
19708 (and_predicates): Turn to ...
19709 (predicate::operator &=): ... this one.
19710 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
19711 (or_predicates): Turn to ...
19712 (predicate::or_with): ... this one.
19713 (evaluate_predicate): Turn to ...
19714 (predicate::evaluate): ... this one.
19715 (predicate_probability): Turn to ...
19716 (predicate::probability): ... this one.
19717 (dump_condition): Update.
19718 (dump_predicate): Turn to ...
19719 (predicate::dump): ... this one.
19720 (account_size_time): Update.
19721 (edge_set_predicate): Update.
19722 (set_hint_predicate): UPdate.
19723 (evaluate_conditions_for_known_args): Update.
19724 (evaluate_properties_for_edge): Update.
19725 (remap_predicate_after_duplication): Turn to...
19726 (predicate::remap_after_duplication): ... this one.
19727 (remap_hint_predicate_after_duplication): Update.
19728 (inline_summary_t::duplicate): UPdate.
19729 (dump_inline_edge_summary): Update.
19730 (dump_inline_summary): Update.
19731 (set_cond_stmt_execution_predicate): Update.
19732 (set_switch_stmt_execution_predicate): Update.
19733 (compute_bb_predicates): Update.
19734 (will_be_nonconstant_expr_predicate): Update.
19735 (will_be_nonconstant_predicate): Update.
19736 (phi_result_unknown_predicate): Update.
19737 (predicate_for_phi_result): Update.
19738 (array_index_predicate): Update.
19739 (estimate_function_body_sizes): Update.
19740 (estimate_node_size_and_time): Update.
19741 (estimate_ipcp_clone_size_and_time): Update.
19742 (remap_predicate): Rename to ...
19743 (predicate::remap_after_inlining): ... this one.
19744 (remap_hint_predicate): Update.
19745 (inline_merge_summary): Update.
19746 (inline_update_overall_summary): Update.
19747 (estimate_size_after_inlining): Update.
19748 (read_predicate): Rename to ...
19749 (predicate::stream_in): ... this one.
19750 (read_inline_edge_summary): Update.
19751 (write_predicate): Rename to ...
19752 (predicate::stream_out): ... this one.
19753 (write_inline_edge_summary): Update.
19754 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
19755 (clause_t): Turn to uint32_t
19756 (predicate): Turn to class; implement constructor and operators
19758 (size_time_entry): Update.
19759 (inline_summary): Update.
19760 (inline_edge_summary): Update.
19762 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
19764 * fold-const.c (fold_binary_loc): Move transformation...
19765 * match.pd (C - X CMP X): ... here.
19767 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
19769 * config/sparc/sparc.c (sparc_option_override): Set function
19770 alignment for -mcpu=niagara7 to 64 to match the I$ line.
19771 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
19773 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
19775 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
19777 2017-05-18 Marek Polacek <polacek@redhat.com>
19780 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
19781 (pass_ubsan::execute): Call gimple_assign_single_p instead of
19782 gimple_assign_load_p.
19784 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
19786 PR middle-end/80692
19787 * real.c (do_compare): Give decimal_do_compare preference over
19788 comparing just the signs.
19790 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
19792 * doc/md.texi (Canonicalization of Instructions): Describe the
19793 canonical form of instructions that inherently set a condition
19796 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
19798 PR middle-end/80775
19799 * tree-cfg.c: Move deletion of unreachable case statements to after
19800 the merging of consecutive case labels.
19802 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
19804 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
19805 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
19806 restoring of callee-saved registers.
19808 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
19810 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
19811 * config/visium/visium.c (single_set_and_flags): Likewise.
19812 * config/visium/visium.md (Substitutions): Likewise.
19814 2017-05-17 Martin Liska <mliska@suse.cz>
19816 * cfg.c: Introduce dump_flags_t type and
19817 use it instead of int type.
19819 * cfghooks.c: Likewise.
19820 * cfghooks.h (struct cfg_hooks): Likewise.
19821 * cfgrtl.c: Likewise.
19822 * cfgrtl.h: Likewise.
19823 * cgraph.c (cgraph_node::get_body): Likewise.
19824 * coretypes.h: Likewise.
19825 * domwalk.c: Likewise.
19826 * domwalk.h: Likewise.
19827 * dumpfile.c (struct dump_option_value_info): Likewise.
19828 (dump_enable_all): Likewise.
19829 (dump_switch_p_1): Likewise.
19830 (opt_info_switch_p): Likewise.
19831 * dumpfile.h (enum tree_dump_index): Likewise.
19832 (struct dump_file_info): Likewise.
19833 * genemit.c: Likewise.
19834 * generic-match-head.c: Likewise.
19835 * gengtype.c (open_base_files): Likewise.
19836 * gimple-pretty-print.c: Likewise.
19837 * gimple-pretty-print.h: Likewise.
19838 * graph.c (print_graph_cfg): Likewise.
19839 * graphite-scop-detection.c (dot_all_sese): Likewise.
19840 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
19841 * loop-unroll.c (report_unroll): Likewise.
19842 * passes.c (pass_manager::register_one_dump_file): Likewise.
19843 * print-tree.c: Likewise.
19844 * statistics.c: Likewise.
19845 * tree-cfg.c: Likewise.
19846 * tree-cfg.h: Likewise.
19847 * tree-dfa.c: Likewise.
19848 * tree-dfa.h: Likewise.
19849 * tree-dump.c (dump_function): Likewise.
19850 * tree-dump.h (struct dump_info): Likewise.
19851 * tree-pretty-print.c: Likewise.
19852 * tree-pretty-print.h: Likewise.
19853 * tree-ssa-live.c: Likewise.
19854 * tree-ssa-live.h: Likewise.
19855 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
19856 * tree-vect-loop.c: Likewise.
19857 * tree-vect-slp.c: Likewise.
19859 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
19860 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19862 PR tree-optimization/80457
19863 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
19864 of all arguments to a statement as scalar_to_vec operations.
19865 (vectorizable_call): Adjust call to vect_model_simple_cost for
19867 (vectorizable_conversion): Likewise.
19868 (vectorizable_assignment): Likewise.
19869 (vectorizable_shift): Likewise.
19870 (vectorizable_operation): Likewise.
19871 (vectorizable_comparison): Likewise.
19872 (vect_is_simple_cond): Record the def types for operands.
19873 (vectorizable_condition): Likewise, call vect_model_simple_cost.
19874 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
19875 for statement argument count.
19877 2017-05-16 Carl Love <cel@us.ibm.com>
19879 * config/rs6000/rs6000-c: Add support for built-in functions
19880 vector unsigned long long vec_bperm (vector unsigned long long,
19881 vector unsigned char)
19882 vector signed long long vec_mule (vector signed int,
19884 vector unsigned long long vec_mule (vector unsigned int,
19885 vector unsigned int)
19886 vector signed long long vec_mulo (vector signed int,
19888 vector unsigned long long vec_mulo (vector unsigned int,
19889 vector unsigned int)
19890 vector signed char vec_sldw (vector signed char,
19891 vector signed char,
19893 vector unsigned char vec_sldw (vector unsigned char,
19894 vector unsigned char,
19896 vector signed short vec_sldw (vector signed short,
19897 vector signed short,
19899 vector unsigned short vec_sldw (vector unsigned short,
19900 vector unsigned short,
19902 vector signed int vec_sldw (vector signed int,
19905 vector unsigned int vec_sldw (vector unsigned int,
19906 vector unsigned int,
19908 vector signed long long vec_sldw (vector signed long long,
19909 vector signed long long,
19911 vector unsigned long long vec_sldw (vector unsigned long long,
19912 vector unsigned long long,
19914 * config/rs6000/rs6000-c: Add support for built-in functions
19915 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
19916 * config/rs6000/altivec.h: Add defintion for vec_sldw.
19917 * doc/extend.texi: Update the built-in documentation for the
19918 new built-in functions.
19920 2017-05-16 Marek Polacek <polacek@redhat.com>
19924 * tree.c (save_expr): Don't fold the expression.
19926 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
19928 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
19929 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
19930 and (?*y,m). Update insn attributes.
19932 2017-05-16 Martin Liska <mliska@suse.cz>
19934 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
19935 flags argument of print_gimple_stmt, print_gimple_expr,
19936 print_generic_stmt and print_generic_expr.
19937 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
19938 * coretypes.h: Likewise.
19939 * except.c (dump_eh_tree): Likewise.
19940 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
19941 * gimple-pretty-print.h: Likewise.
19942 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
19943 (backprop::push_to_worklist): Likewise.
19944 (backprop::pop_from_worklist): Likewise.
19945 (backprop::process_use): Likewise.
19946 (backprop::intersect_uses): Likewise.
19947 (note_replacement): Likewise.
19948 * gimple-ssa-store-merging.c
19949 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
19950 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
19951 (pass_store_merging::execute): Likewise.
19952 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
19953 (ssa_base_cand_dump_callback): Likewise.
19954 (dump_incr_vec): Likewise.
19955 (replace_refs): Likewise.
19956 (replace_mult_candidate): Likewise.
19957 (create_add_on_incoming_edge): Likewise.
19958 (create_phi_basis): Likewise.
19959 (insert_initializers): Likewise.
19960 (all_phi_incrs_profitable): Likewise.
19961 (introduce_cast_before_cand): Likewise.
19962 (replace_one_candidate): Likewise.
19963 * gimplify.c (gimplify_expr): Likewise.
19964 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
19965 (set_rename): Likewise.
19966 (rename_uses): Likewise.
19967 (copy_loop_phi_nodes): Likewise.
19968 (add_close_phis_to_merge_points): Likewise.
19969 (copy_loop_close_phi_args): Likewise.
19970 (copy_cond_phi_args): Likewise.
19971 (graphite_copy_stmts_from_block): Likewise.
19972 (translate_pending_phi_nodes): Likewise.
19973 * graphite-poly.c (print_pdr): Likewise.
19974 (dump_gbb_cases): Likewise.
19975 (dump_gbb_conditions): Likewise.
19976 (print_scop_params): Likewise.
19977 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
19978 (build_cross_bb_scalars_use): Likewise.
19979 (gather_bbs::before_dom_children): Likewise.
19980 * hsa-dump.c (dump_hsa_immed): Likewise.
19981 * ipa-cp.c (print_ipcp_constant_value): Likewise.
19982 (get_replacement_map): Likewise.
19983 * ipa-inline-analysis.c (dump_condition): Likewise.
19984 (estimate_function_body_sizes): Likewise.
19985 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
19986 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
19987 * ipa-prop.c (ipa_dump_param): Likewise.
19988 (ipa_print_node_jump_functions_for_edge): Likewise.
19989 (ipa_modify_call_arguments): Likewise.
19990 (ipa_modify_expr): Likewise.
19991 (ipa_dump_param_adjustments): Likewise.
19992 (ipa_dump_agg_replacement_values): Likewise.
19993 (ipcp_modif_dom_walker::before_dom_children): Likewise.
19994 * ipa-pure-const.c (check_stmt): Likewise.
19995 (pass_nothrow::execute): Likewise.
19996 * ipa-split.c (execute_split_functions): Likewise.
19997 * omp-offload.c (dump_oacc_loop_part): Likewise.
19998 (dump_oacc_loop): Likewise.
19999 * trans-mem.c (tm_log_emit): Likewise.
20000 (tm_memopt_accumulate_memops): Likewise.
20001 (dump_tm_memopt_set): Likewise.
20002 (dump_tm_memopt_transform): Likewise.
20003 * tree-cfg.c (gimple_verify_flow_info): Likewise.
20004 (print_loop): Likewise.
20005 * tree-chkp-opt.c (chkp_print_addr): Likewise.
20006 (chkp_gather_checks_info): Likewise.
20007 (chkp_get_check_result): Likewise.
20008 (chkp_remove_check_if_pass): Likewise.
20009 (chkp_use_outer_bounds_if_possible): Likewise.
20010 (chkp_reduce_bounds_lifetime): Likewise.
20011 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
20012 (chkp_mark_completed_bounds): Likewise.
20013 (chkp_register_incomplete_bounds): Likewise.
20014 (chkp_mark_invalid_bounds): Likewise.
20015 (chkp_maybe_copy_and_register_bounds): Likewise.
20016 (chkp_build_returned_bound): Likewise.
20017 (chkp_get_bound_for_parm): Likewise.
20018 (chkp_build_bndldx): Likewise.
20019 (chkp_get_bounds_by_definition): Likewise.
20020 (chkp_generate_extern_var_bounds): Likewise.
20021 (chkp_get_bounds_for_decl_addr): Likewise.
20022 * tree-chrec.c (chrec_apply): Likewise.
20023 * tree-data-ref.c (dump_data_reference): Likewise.
20024 (dump_subscript): Likewise.
20025 (dump_data_dependence_relation): Likewise.
20026 (analyze_overlapping_iterations): Likewise.
20027 * tree-inline.c (expand_call_inline): Likewise.
20028 (tree_function_versioning): Likewise.
20029 * tree-into-ssa.c (dump_defs_stack): Likewise.
20030 (dump_currdefs): Likewise.
20031 (dump_names_replaced_by): Likewise.
20032 (dump_update_ssa): Likewise.
20033 (update_ssa): Likewise.
20034 * tree-object-size.c (pass_object_sizes::execute): Likewise.
20035 * tree-parloops.c (build_new_reduction): Likewise.
20036 (try_create_reduction_list): Likewise.
20037 (ref_conflicts_with_region): Likewise.
20038 (oacc_entry_exit_ok_1): Likewise.
20039 (oacc_entry_exit_single_gang): Likewise.
20040 * tree-pretty-print.h: Likewise.
20041 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
20042 (get_scalar_evolution): Likewise.
20043 (add_to_evolution): Likewise.
20044 (get_loop_exit_condition): Likewise.
20045 (analyze_evolution_in_loop): Likewise.
20046 (analyze_initial_condition): Likewise.
20047 (analyze_scalar_evolution): Likewise.
20048 (instantiate_scev): Likewise.
20049 (number_of_latch_executions): Likewise.
20050 (gather_chrec_stats): Likewise.
20051 (final_value_replacement_loop): Likewise.
20052 (scev_const_prop): Likewise.
20053 * tree-sra.c (dump_access): Likewise.
20054 (disqualify_candidate): Likewise.
20055 (create_access): Likewise.
20056 (reject): Likewise.
20057 (maybe_add_sra_candidate): Likewise.
20058 (create_access_replacement): Likewise.
20059 (analyze_access_subtree): Likewise.
20060 (analyze_all_variable_accesses): Likewise.
20061 (sra_modify_assign): Likewise.
20062 (initialize_constant_pool_replacements): Likewise.
20063 (find_param_candidates): Likewise.
20064 (decide_one_param_reduction): Likewise.
20065 (replace_removed_params_ssa_names): Likewise.
20066 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
20067 * tree-ssa-copy.c (dump_copy_of): Likewise.
20068 (copy_prop_visit_cond_stmt): Likewise.
20069 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
20070 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
20071 (record_equivalences_from_stmt): Likewise.
20072 * tree-ssa-dse.c (compute_trims): Likewise.
20073 (delete_dead_call): Likewise.
20074 (delete_dead_assignment): Likewise.
20075 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
20076 (forward_propagate_into_cond): Likewise.
20077 (pass_forwprop::execute): Likewise.
20078 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
20079 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
20081 (move_computations_worker): Likewise.
20082 (execute_sm): Likewise.
20083 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
20084 (remove_exits_and_undefined_stmts): Likewise.
20085 (remove_redundant_iv_tests): Likewise.
20086 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
20087 (adjust_iv_update_pos): Likewise.
20088 * tree-ssa-math-opts.c (bswap_replace): Likewise.
20089 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
20090 (value_replacement): Likewise.
20091 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
20092 * tree-ssa-pre.c (print_pre_expr): Likewise.
20093 (get_representative_for): Likewise.
20094 (create_expression_by_pieces): Likewise.
20095 (insert_into_preds_of_block): Likewise.
20096 (eliminate_insert): Likewise.
20097 (eliminate_dom_walker::before_dom_children): Likewise.
20098 (eliminate): Likewise.
20099 (remove_dead_inserted_code): Likewise.
20100 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
20101 * tree-ssa-reassoc.c (get_rank): Likewise.
20102 (eliminate_duplicate_pair): Likewise.
20103 (eliminate_plus_minus_pair): Likewise.
20104 (eliminate_not_pairs): Likewise.
20105 (undistribute_ops_list): Likewise.
20106 (eliminate_redundant_comparison): Likewise.
20107 (update_range_test): Likewise.
20108 (optimize_range_tests_var_bound): Likewise.
20109 (optimize_vec_cond_expr): Likewise.
20110 (rewrite_expr_tree): Likewise.
20111 (rewrite_expr_tree_parallel): Likewise.
20112 (linearize_expr): Likewise.
20113 (break_up_subtract): Likewise.
20114 (linearize_expr_tree): Likewise.
20115 (attempt_builtin_powi): Likewise.
20116 (attempt_builtin_copysign): Likewise.
20117 (transform_stmt_to_copy): Likewise.
20118 (transform_stmt_to_multiply): Likewise.
20119 (dump_ops_vector): Likewise.
20120 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
20121 (print_scc): Likewise.
20122 (set_ssa_val_to): Likewise.
20123 (visit_reference_op_store): Likewise.
20124 (visit_use): Likewise.
20125 (sccvn_dom_walker::before_dom_children): Likewise.
20126 (run_scc_vn): Likewise.
20127 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
20129 (expr_hash_elt::print): Likewise.
20130 (const_and_copies::pop_to_marker): Likewise.
20131 (const_and_copies::record_const_or_copy_raw): Likewise.
20132 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
20133 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
20134 (dump_predicates): Likewise.
20135 (find_uninit_use): Likewise.
20136 (warn_uninitialized_phi): Likewise.
20137 (pass_late_warn_uninitialized::execute): Likewise.
20138 * tree-ssa.c (verify_vssa): Likewise.
20139 (verify_ssa): Likewise.
20140 (maybe_optimize_var): Likewise.
20141 * tree-vrp.c (dump_value_range): Likewise.
20142 (dump_all_value_ranges): Likewise.
20143 (dump_asserts_for): Likewise.
20144 (register_edge_assert_for_2): Likewise.
20145 (vrp_visit_cond_stmt): Likewise.
20146 (vrp_visit_switch_stmt): Likewise.
20147 (vrp_visit_stmt): Likewise.
20148 (vrp_visit_phi_node): Likewise.
20149 (simplify_cond_using_ranges_1): Likewise.
20150 (fold_predicate_in): Likewise.
20151 (evrp_dom_walker::before_dom_children): Likewise.
20152 (evrp_dom_walker::push_value_range): Likewise.
20153 (evrp_dom_walker::pop_value_range): Likewise.
20154 (execute_early_vrp): Likewise.
20156 2017-05-16 Richard Biener <rguenther@suse.de>
20158 * dwarf2out.c (loc_list_from_tree_1): Do not create
20159 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
20161 2017-05-16 Richard Biener <rguenther@suse.de>
20163 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
20165 (note_variable_value_in_expr): If we resolved the decl ref
20166 do not push to the stack.
20168 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
20170 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
20171 operations in fast-math mode.
20172 (vaddq_f16): Likewise.
20173 (vmul_f16): Likewise.
20174 (vmulq_f16): Likewise.
20175 (vsub_f16): Likewise.
20176 (vsubq_f16): Likewise.
20177 * config/arm/neon.md (add<mode>3): New.
20179 (fma:<VH:mode>3): New. Also remove outdated comment.
20182 2017-05-16 Martin Liska <mliska@suse.cz>
20186 * ipa-devirt.c (warn_types_mismatch): Fix typo.
20187 (odr_types_equivalent_p): Likewise.
20189 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
20191 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
20193 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
20196 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
20197 non-interunit SSE move alternatives with '?'.
20198 (zero-extendsidi peephole2): New peephole to skip intermediate
20199 general register in SSE zero-extend sequence.
20201 2017-05-15 Jeff Law <law@redhat.com>
20203 * reorg.c (relax_delay_slots): Create a new variable to hold
20204 the temporary target rather than clobbering TARGET_LABEL.
20206 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
20207 missing argument to extract_bit_field call.
20208 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
20210 2017-05-15 Martin Liska <mliska@suse.cz>
20213 * gcc.c (process_command): Do not allow empty argument of -o option.
20215 2017-05-15 Renlin Li <renlin.li@arm.com>
20217 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
20218 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
20219 * config/aarch64/constraints.md (Usf): Add long call check.
20220 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
20221 (call_value): Likewise.
20222 (sibcall): Likewise.
20223 (sibcall_value): Likewise.
20225 (call_value_insn): New.
20226 (sibcall_insn): Update rtx pattern.
20227 (sibcall_value_insn): Likewise.
20228 (call_internal): Remove.
20229 (call_value_internal): Likewise.
20230 (sibcall_internal): Likewise.
20231 (sibcall_value_internal): Likewise.
20232 (call_reg): Likewise.
20233 (call_symbol): Likewise.
20234 (call_value_reg): Likewise.
20235 (call_value_symbol): Likewise.
20237 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
20240 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
20242 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
20244 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
20245 compatible with CCGOCmode and with CCZmode.
20247 2017-05-14 Martin Sebor <msebor@redhat.com>
20249 PR middle-end/77671
20250 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
20251 (gimple_fold_builtin_snprintf): Same.
20252 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
20253 (gimple_fold_builtin_snprintf): Same.
20254 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
20255 of character types.
20256 (is_call_safe): New function.
20257 (try_substitute_return_value): Call it.
20258 (try_simplify_call): New function.
20259 (pass_sprintf_length::handle_gimple_call): Call it.
20261 2017-05-14 Martin Sebor <msebor@redhat.com>
20263 PR middle-end/80669
20264 * builtins.c (expand_builtin_stpncpy): Simplify.
20266 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
20268 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
20269 * config/i386/i386.h
20270 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
20271 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
20272 (struct machine_function): Add new members call_ms2sysv,
20273 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
20274 (struct machine_frame_state): New fields sp_realigned and
20275 sp_realigned_offset.
20276 * config/i386/i386.c
20277 (enum xlogue_stub): New enum.
20278 (enum xlogue_stub_sets): New enum.
20279 (class xlogue_layout): New class.
20280 (struct ix86_frame): New fields stack_realign_allocate_offset,
20281 stack_realign_offset and outlined_save_offset. Modify comments to
20282 detail stack layout when using out-of-line stubs.
20283 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
20284 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
20285 -mcall-ms2sysv-xlogues.
20286 (stub_managed_regs): New static variable.
20287 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
20288 registers managed by out-of-line stub.
20289 (disable_call_ms2sysv_xlogues): New function.
20290 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
20291 m->call_ms2sysv when appropriate and compute frame layout for
20293 (sp_valid_at, fp_valid_at): New inline functions.
20294 (choose_basereg): New function.
20295 (choose_baseaddr): Add align parameter, use choose_basereg and modify
20297 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
20298 Use align parameter of choose_baseaddr to generated aligned SSE movs
20300 (pro_epilogue_adjust_stack): Modify to track
20301 machine_frame_state::sp_realigned.
20302 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
20303 (ix86_nsaved_sseregs): Likewise.
20304 (ix86_emit_save_regs): Likewise.
20305 (ix86_emit_save_regs_using_mov): Likewise.
20306 (ix86_emit_save_sse_regs_using_mov): Likewise.
20307 (get_scratch_register_on_entry): Likewise.
20308 (gen_frame_set): New function.
20309 (gen_frame_load): Likewise.
20310 (gen_frame_store): Likewise.
20311 (emit_outlined_ms2sysv_save): Likewise.
20312 (emit_outlined_ms2sysv_restore): Likewise.
20313 (ix86_expand_prologue): Modify stack re-alignment code and call
20314 emit_outlined_ms2sysv_save when appropriate.
20315 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
20316 parameter rtx_insn *insn, which allows the function to be used to only
20317 generate the notes.
20318 (ix86_expand_epilogue): Modify validity checks of frame and stack
20319 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
20320 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
20321 * config/i386/predicates.md
20322 (save_multiple): New predicate.
20323 (restore_multiple): Likewise.
20324 * config/i386/sse.md
20325 (save_multiple<mode>): New pattern.
20326 (save_multiple_realign<mode>): Likewise.
20327 (restore_multiple<mode>): Likewise.
20328 (restore_multiple_and_return<mode>): Likewise.
20329 (restore_multiple_leave_return<mode>): Likewise.
20330 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
20332 2017-05-14 Julia Koval <julia.koval@intel.com>
20334 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
20335 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
20336 (__builtin_ia32_xsetbv): New builtins.
20337 * config/i386/i386.c (ix86_expand_special_args_builtin):
20339 (ix86_expand_builtin): Special expand for new intrinsics.
20340 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
20341 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
20342 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
20344 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20346 * cfganal.c (inverted_post_order_compute): Change argument type
20348 * cfganal.h (inverted_post_order_compute): Adjust prototype.
20349 * df-core.c (rest_of_handle_df_initialize): Adjust.
20350 (rest_of_handle_df_finish): Likewise.
20351 (df_analyze_1): Likewise.
20352 (df_analyze): Likewise.
20353 (loop_inverted_post_order_compute): Change argument to be a vec *.
20354 (df_analyze_loop): Adjust.
20355 (df_get_n_blocks): Likewise.
20356 (df_get_postorder): Likewise.
20357 * df.h (struct df_d): Change field to be a vec.
20358 * lcm.c (compute_laterin): Adjust.
20359 (compute_available): Likewise.
20360 * lra-lives.c (lra_create_live_ranges_1): Likewise.
20361 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
20362 * tree-ssa-pre.c (compute_antic): Likewise.
20364 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20366 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
20367 (depth_first_search::depth_first_search): Change structure init
20368 function to this constructor.
20369 (depth_first_search::add_bb): Rename function to this member.
20370 (depth_first_search::execute): Likewise.
20371 (flow_dfs_compute_reverse_finish): Adjust.
20373 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20375 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
20376 (longest_simple_path): Likewise.
20377 * shrink-wrap.c (spread_components): Likewise.
20378 (disqualify_problematic_components): Likewise.
20379 (emit_common_heads_for_components): Likewise.
20380 (emit_common_tails_for_components): Likewise.
20381 (insert_prologue_epilogue_for_components): Likewise.
20383 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20385 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
20388 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20390 * df-core.c (df_set_blocks): Start using auto_bitmap.
20391 (df_compact_blocks): Likewise.
20392 * df-problems.c (df_rd_confluence_n): Likewise.
20393 * df-scan.c (df_insn_rescan_all): Likewise.
20394 (df_process_deferred_rescans): Likewise.
20395 (df_update_entry_block_defs): Likewise.
20396 (df_update_exit_block_uses): Likewise.
20397 (df_entry_block_bitmap_verify): Likewise.
20398 (df_exit_block_bitmap_verify): Likewise.
20399 (df_scan_verify): Likewise.
20400 * lra-constraints.c (lra_constraints): Likewise.
20401 (undo_optional_reloads): Likewise.
20402 (lra_undo_inheritance): Likewise.
20403 * lra-remat.c (calculate_gen_cands): Likewise.
20404 (do_remat): Likewise.
20405 * lra-spills.c (assign_spill_hard_regs): Likewise.
20406 (spill_pseudos): Likewise.
20407 * tree-ssa-pre.c (bitmap_set_and): Likewise.
20408 (bitmap_set_subtract_values): Likewise.
20410 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20412 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
20413 management with auto_bitmap.
20414 (fix_inter_tick): Likewise.
20415 (fix_recovery_deps): Likewise.
20416 * ira.c (add_store_equivs): Likewise.
20417 (find_moveable_pseudos): Likewise.
20418 (split_live_ranges_for_shrink_wrap): Likewise.
20419 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
20420 (rtx_reuse_manager::seen_def_p): Likewise.
20421 (rtx_reuse_manager::set_seen_def): Likewise.
20422 * print-rtl.h (class rtx_reuse_manager): Likewise.
20424 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20426 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
20428 (migrate_btr_def): Likewise.
20429 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
20430 * df-core.c (loop_post_order_compute): Likewise.
20431 (loop_inverted_post_order_compute): Likewise.
20432 * hsa-common.h: Likewise.
20433 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
20434 * init-regs.c (initialize_uninitialized_regs): Likewise.
20435 * ipa-inline.c (resolve_noninline_speculation): Likewise.
20436 (inline_small_functions): Likewise.
20437 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
20438 * ira.c (combine_and_move_insns): Likewise.
20439 (build_insn_chain): Likewise.
20440 * loop-invariant.c (find_invariants): Likewise.
20441 * lower-subreg.c (propagate_pseudo_copies): Likewise.
20442 * predict.c (tree_predict_by_opcode): Likewise.
20443 (predict_paths_leading_to): Likewise.
20444 (predict_paths_leading_to_edge): Likewise.
20445 (estimate_loops_at_level): Likewise.
20446 (estimate_loops): Likewise.
20447 * shrink-wrap.c (try_shrink_wrapping): Likewise.
20448 (spread_components): Likewise.
20449 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
20450 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
20451 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
20452 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
20453 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
20454 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
20455 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
20456 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
20457 (mark_threaded_blocks): Likewise.
20458 (thread_through_all_blocks): Likewise.
20459 * tree-ssa.c (verify_ssa): Likewise.
20460 (execute_update_addresses_taken): Likewise.
20461 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
20463 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20465 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
20467 (post_order_compute): Likewise.
20468 (inverted_post_order_compute): Likewise.
20469 (pre_and_rev_post_order_compute_fn): Likewise.
20471 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20473 * genrecog.c (int_set::int_set): Explicitly construct our
20474 auto_vec base class.
20475 * vec.h (auto_vec::auto_vec): New constructor.
20477 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20479 * bitmap.h (class auto_bitmap): New constructor taking
20480 bitmap_obstack * argument.
20482 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20484 * bitmap.h (class auto_bitmap): Change type of m_bits to
20485 bitmap_head, and adjust ctor / dtor and member operators.
20487 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
20489 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
20490 when returned register mode doesn't match original mode.
20492 2017-05-12 Jeff Law <law@redhat.com>
20493 Jakub Jelinek <jakub@redhat.com>
20495 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
20496 we look for cc setter after the compare-elim changes.
20497 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
20498 within the vector to match what compare-elim now expects.
20499 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
20500 (xorsi3_flags, one_cmplsi2_flags): Likewise.
20502 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
20503 after the compare-elim changes.
20504 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
20505 the vector to match what compare-elim now expects.
20506 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
20507 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
20508 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
20509 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
20510 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
20512 * config/visium/visium.c (single_set_and_flags): Fix where
20513 we look for cc setter after the compare-elim changes.
20514 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
20515 with the vector to match what compare-elim now expects.
20516 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
20517 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
20518 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
20519 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
20520 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
20521 (neg<mode>2_insn_set_overflow): Likewise.
20523 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
20525 PR middle-end/79794
20526 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
20527 maybe_expand_insn call, set ops[0].target. If still set after call,
20528 set alt_rtl. Add extra arg to recursive calls.
20529 (extract_bit_field): Add alt_rtl argument. Pass to
20531 * expmed.h (extract_bit_field): Fix prototype.
20532 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
20533 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
20534 to extract_bit_field_calls.
20535 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
20536 Pass alt_rtl to extract_bit_field calls.
20537 * calls.c (store_unaligned_arguments_into_psuedos)
20538 load_register_parameters): Pass extra NULL to extract_bit_field calls.
20539 * optabs.c (maybe_legitimize_operand): Clear op->target when call
20541 * optabs.h (struct expand_operand): Add target bitfield.
20543 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
20545 * compare-elim.c (try_eliminate_compare): Canonicalize
20546 operation with embedded compare to
20547 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
20548 (set (reg) (operation)].
20550 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
20552 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
20555 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
20556 cost of adding a carry flag for ADC instruction.
20557 [case MINUS]: Ignore the cost of subtracting a carry flag
20558 for SBB instruction.
20560 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
20562 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
20564 * config/rs6000/bmiintrin.h: New file.
20565 * config/rs6000/bmi2intrin.h: New file.
20566 * config/rs6000/x86intrin.h: New file.
20568 2017-05-12 Jeff Law <law@redhat.com>
20570 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
20573 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
20575 PR middle-end/80707
20576 * tree-cfg.c: Remove cfg edges of unreachable case statements.
20578 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20580 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
20581 early expansion of vector divide builtins.
20582 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
20583 builtins identified as having unsigned arguments.
20585 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20587 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
20588 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
20589 expansion of vector logical operations (and, andc, or, xor,
20592 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20594 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
20595 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
20597 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20599 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
20600 early GIMPLE expansion of vector multiplies.
20602 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20604 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
20605 TARGET_HAVE_MOVT conditional.
20606 (movt splitter): Likewise.
20608 2017-05-12 Richard Biener <rguenther@suse.de>
20610 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
20611 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
20612 Fold all stmts not inplace.
20614 2017-05-12 Richard Biener <rguenther@suse.de>
20616 PR tree-optimization/80713
20617 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
20618 inserted_exprs bit for not removed stmts.
20620 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
20622 PR middle-end/69921
20623 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
20624 parallelized" attribute for parallelized OpenACC kernels.
20625 * omp-offload.c (execute_oacc_device_lower): Use it.
20627 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
20628 Set "oacc kernels" attribute.
20629 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
20630 parameter. Adjust all users.
20631 (oacc_fn_attrib_kernels_p): Remove function.
20632 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
20633 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
20634 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
20635 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
20636 assert "oacc kernels" attribute is set.
20638 2017-05-11 Carl Love <cel@us.ibm.com>
20640 * config/rs6000/rs6000-c: Add support for built-in functions
20641 vector unsigned char vec_popcnt (vector signed char)
20642 vector unsigned char vec_popcnt (vector unsigned char)
20643 vector unsigned short vec_popcnt (vector signed short)
20644 vector unsigned short vec_popcnt (vector unsigned short)
20645 vector unsigned int vec_popcnt (vector signed int)
20646 vector unsigned int vec_popcnt (vector unsigned int)
20647 vector unsigned long long vec_popcnt (vector signed long long)
20648 vector unsigned long long vec_popcnt (vector unsigned long long)
20649 vector signed long long vec_slo (vector signed long long,
20650 vector signed char)
20651 vector signed long long vec_slo (vector signed long long,
20652 vector unsigned char)
20653 vector unsigned long long vec_slo (vector unsigned long long,
20654 vector signed char)
20655 vector unsigned long long vec_slo (vector unsigned long long,
20656 vector unsigned char)
20657 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
20658 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
20659 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
20660 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
20661 * doc/extend.texi: Update the built-in documentation file for the
20662 new built-in functions.
20664 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
20666 * attribs.h (sorted_attr_string): Move machine independent
20667 functions for target clone support from the i386 port to common
20668 code. Rename ix86_function_versions to common_function_versions.
20669 Rename make_name to make_unique_name.
20670 (common_function_versions): Likewise.
20671 (make_unique_name): Likewise.
20672 (make_dispatcher_decl): Likewise.
20673 (is_function_default_version): Likewise.
20674 * attribs.c (attr_strcmp): Likewise.
20675 (sorted_attr_string): Likewise.
20676 (common_function_versions): Likewise.
20677 (make_unique_name): Likewise.
20678 (make_dispatcher_decl): Likewise.
20679 (is_function_default_version): Likewise.
20680 * config/i386/i386.c (attr_strcmp): Likewise.
20681 (sorted_attr_string): Likewise.
20682 (ix86_function_versions): Likewise.
20683 (make_name): Likewise.
20684 (make_dispatcher_decl): Likewise.
20685 (is_function_default_version): Likewise.
20686 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
20688 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20691 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20692 Account for direct move costs for vec_construct of integer
20695 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
20698 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
20699 (UNSPEC_STX_ATOMIC): Ditto.
20700 (loaddi_via_sse): New insn.
20701 (storedi_via_sse): Ditto.
20702 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
20703 Update corresponding peephole2 patterns.
20704 (atomic_storedi_fpu): Ditto.
20706 2017-05-11 Julia Koval <julia.koval@intel.com>
20708 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
20709 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
20711 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
20712 (__builtin_ia32_rsqrt14ss_mask): New builtins.
20713 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
20715 2017-05-11 Nathan Sidwell <nathan@acm.org>
20717 * graphite-poly.c: Include dumpfile.h.
20719 * dumpfle.h (dump_function): Declare here ...
20720 * tree-dump.h (dump_function): ... not here.
20721 * dumpfile.c: #include tree-cfg.h.
20722 (dump_function): Move here from ...
20723 * tree-dump.c (dump_function): ... here.
20724 * gimplify.c: #include splay-tree.h, not tree-dump.h.
20725 * graphite-poly.c: Don't include tree-dump.h.
20726 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
20727 * print-tree.c: Likewise.
20728 * stor-layout.c: Likewise.
20729 * tree-nested.c: Likewise.
20731 * dumpfile.c (dump_start): Use TDF_FLAGS.
20732 (dump_enable_all): Fix TDF_KIND check thinko.
20734 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
20736 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20737 array entries to represent two legal parameterizations of the
20738 overloaded __builtin_cmpb function, as represented by the
20739 P6_OV_BUILTIN_CMPB constant.
20740 (altivec_resolve_overloaded_builtin): Add special case handling
20741 for the __builtin_cmpb function, as represented by the
20742 P6_OV_BUILTIN_CMPB constant.
20743 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
20744 (BU_P6_64BIT_2): New macro.
20745 (BU_P6_OVERLOAD_2): New macro
20746 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
20747 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
20748 (CMPB): Add overload support to represent both 32-bit and 64-bit
20749 compare-bytes function.
20750 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
20751 support for TARGET_CMPB.
20752 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
20753 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20754 documentation of the __builtin_cmpb overloaded built-in function.
20756 2017-05-11 Richard Biener <rguenther@suse.de>
20758 PR tree-optimization/80705
20759 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
20760 bases are not vectorizable.
20762 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20764 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
20765 when counting register pressure.
20767 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20769 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
20770 (iv_ca_more_deps): Renamed to ...
20771 (iv_ca_compare_deps): ... this.
20772 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
20774 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20776 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
20778 (determine_group_iv_costs): ... here.
20779 (find_inv_vars_cb): Record inv var if it's not recorded before.
20781 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20783 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
20784 (get_shiftadd_cost): Ditto.
20786 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20788 * tree-ssa-address.c: Include header file.
20789 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
20791 (add_to_parts): Refactor.
20792 (addr_to_parts): New parameter. Update use of move_hint_to_base.
20793 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
20796 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20798 PR tree-optimization/53090
20799 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
20801 (extract_cond_operands): Detect condition with IV on both sides
20802 and return COMP_IV_EXPR_2.
20803 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
20804 (rewrite_use_compare): Simplify by removing call to function
20805 extract_cond_operands.
20807 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20809 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
20810 (extract_cond_operands): Detect condition comparing against non-
20811 invariant bound and return appropriate enum value.
20812 (find_interesting_uses_cond): Update use of extract_cond_operands.
20813 Handle its return value accordingly.
20814 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
20816 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20818 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
20819 nonlinear iv_use computation in loop invariant sensitive way.
20821 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20823 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
20824 (find_iv_candidates): Call relate_compare_use_with_all_cands.
20826 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20828 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
20829 (dump_cand): Support iv_cand.inv_exprs.
20830 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
20832 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
20835 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20837 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
20839 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
20840 as local function. Include necessary header files.
20841 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
20843 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20845 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
20847 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20849 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
20850 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
20851 RSHIFT_EXPR and BIT_NOT_EXPR.
20853 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20855 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
20856 (adjust_setup_cost): New parameter supporting round up adjustment.
20857 (struct address_cost_data): Delete.
20858 (force_expr_to_var_cost): Don't bound cost with spill_cost.
20859 (split_address_cost, ptr_difference_cost): Delete.
20860 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
20861 (struct ainc_cost_data): New struct.
20862 (get_address_cost_ainc): New function.
20863 (get_address_cost, get_computation_cost): Reimplement.
20864 (determine_group_iv_cost_address): Record inv_expr for all uses of
20866 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
20867 (iv_ca_has_deps): Reimplemented to ...
20868 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
20870 (iv_ca_extend): Call iv_ca_more_deps.
20872 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20874 * tree-ssa-address.c (struct mem_address): Move to header file.
20875 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
20876 * tree-ssa-address.h (struct mem_address): Move from C file.
20877 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
20879 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20881 * tree-affine.h (aff_combination_type): New interface.
20882 (aff_combination_zero_p): Remove static.
20883 (aff_combination_const_p): New interface.
20884 (aff_combination_singleton_var_p): New interfaces.
20886 2017-05-11 Richard Biener <rguenther@suse.de>
20888 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
20889 Skip unreachable blocks and destinations.
20890 (eliminate): Move stmt removal and fixup ...
20891 (fini_eliminate): ... here. Skip inserted exprs.
20892 (pass_pre::execute): Move fini_pre after fini_eliminate.
20893 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
20894 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
20895 PRE to get rid of dead code that has invalid SSA form and
20896 split critical edges again.
20898 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20900 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20902 2017-05-11 Richard Biener <rguenther@suse.de>
20904 * passes.c (execute_function_todo): Verify loops if they are
20905 said to be up-to-date.
20906 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
20907 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
20909 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
20912 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
20913 handle calling assemble_external ourself.
20916 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
20917 modes with zero size. Enhance comment.
20919 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20921 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
20922 built-ins for vec_xl and vec_xst with short and char pointer
20925 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
20927 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
20928 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
20929 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
20930 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
20931 (_mm_maskz_min_round_ss): New intrinsics.
20932 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
20933 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
20934 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
20935 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
20936 (__builtin_ia32_minss_mask_round): New builtins.
20937 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20938 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20939 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
20941 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
20942 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
20944 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
20947 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
20949 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
20950 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
20951 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
20952 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
20953 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
20954 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
20955 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
20956 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20957 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
20958 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
20959 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
20960 (__builtin_ia32_mulss_mask_round): New builtins.
20961 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20962 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20963 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
20965 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
20966 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
20968 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
20971 2017-05-10 Julia Koval <julia.koval@intel.com>
20973 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
20974 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
20975 (_mm256_setr_m128i): New intrinsics.
20977 2017-05-10 Julia Koval <julia.koval@intel.com>
20979 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
20980 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
20981 (_mm_maskz_rcp14_ss): New intrinsics.
20982 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
20983 (__builtin_ia32_rcp14ss_mask): New builtins.
20984 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
20986 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
20988 PR tree-optimization/51513
20989 * tree-cfg.c (gimple_seq_unreachable_p): New function.
20990 (assert_unreachable_fallthru_edge_p): Use it.
20991 (group_case_labels_stmt): Likewise.
20992 * tree-cfg.h: Prototype it.
20993 * stmt.c: Include cfghooks.h and tree-cfg.h.
20994 (emit_case_dispatch_table) <gap_label>: New local variable.
20995 Use it to fill dispatch table gaps.
20996 Test for default_label before updating probabilities.
20997 (expand_case) <default_label>: Remove unneeded initialization.
20998 Test for unreachable default case statement and remove its edge.
20999 Set default_label accordingly.
21000 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
21002 2017-05-10 Carl Love <cel@us.ibm.com>
21004 * config/rs6000/rs6000-c: Add support for built-in functions
21005 vector signed char vec_neg (vector signed char)
21006 vector signed short int vec_neg (vector short int)
21007 vector signed int vec_neg (vector signed int)
21008 vector signed long long vec_neg (vector signed long long)
21009 vector float vec_neg (vector float)
21010 vector double vec_neg (vector double)
21011 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
21013 * config/rs6000/altivec.h: Add define for vec_neg
21014 * doc/extend.texi: Update the built-in documentation for the
21015 new built-in functions.
21017 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21019 PR tree-optimization/77644
21020 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
21022 2017-05-10 Nathan Sidwell <nathan@acm.org>
21024 * dumpfile.h (TDI_lang_all): New.
21025 (TDF_KIND): New. Renumber others
21026 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
21028 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
21030 (get_dump_file_name): Adjust suffix generation.
21031 (dump_enable_all): Use TDF_KIND.
21032 * doc/invoke.texi (-fdump-lang-all): Document.
21034 * dumpfile.h: Tabify.
21036 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
21039 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
21040 Move member access before delete.
21042 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
21044 * tree-inline.c (expand_call_inline): Split block at stmt
21047 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
21050 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
21051 are now unused after splitting mov{sf,sd}_hardfloat.
21052 (f32_lr2): Likewise.
21053 (f32_lm): Likewise.
21054 (f32_lm2): Likewise.
21055 (f32_li): Likewise.
21056 (f32_li2): Likewise.
21057 (f32_lv): Likewise.
21058 (f32_sr): Likewise.
21059 (f32_sr2): Likewise.
21060 (f32_sm): Likewise.
21061 (f32_sm2): Likewise.
21062 (f32_si): Likewise.
21063 (f32_si2): Likewise.
21064 (f32_sv): Likewise.
21065 (f32_dm): Likewise.
21066 (f32_vsx): Likewise.
21067 (f32_av): Likewise.
21068 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
21069 For movsf, order stores so the VSX stores occur before the GPR
21070 store which encourages the register allocator to use a traditional
21071 FPR instead of a GPR. For movsd, order the stores so that the GPR
21072 store comes before the VSX stores to allow the power6 to work.
21073 This is due to the power6 not having a 32-bit integer store
21074 instruction from a FPR.
21075 (movsf_hardfloat): Likewise.
21076 (movsd_hardfloat): Likewise.
21078 2017-05-09 Martin Sebor <msebor@redhat.com>
21080 PR translation/80280
21081 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
21084 PR translation/80280
21085 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
21086 data member added in r247778.
21087 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
21089 2017-05-09 Nathan Sidwell <nathan@acm.org>
21091 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
21093 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
21096 2017-05-09 Marek Polacek <polacek@redhat.com>
21098 * doc/invoke.texi: Fix typo.
21100 2017-05-09 Richard Biener <rguenther@suse.de>
21102 * tree-vrp.c (vrp_val_is_max): Adjust comment.
21103 (vrp_val_is_min): Likewise.
21104 (set_value_range_to_value): Likewise.
21105 (set_value_range_to_nonnegative): Likewise.
21106 (gimple_assign_nonzero_p): Likewise.
21107 (gimple_stmt_nonzero_p): Likewise.
21108 (vrp_int_const_binop): Likewise. Remove unreachable case.
21109 (adjust_range_with_scev): Adjust comments.
21110 (compare_range_with_value): Likewise.
21111 (extract_range_from_phi_node): Likewise.
21112 (test_for_singularity): Likewise.
21114 2017-05-09 Richard Biener <rguenther@suse.de>
21116 * tree-vrp.c (get_single_symbol): Add assert that we don't
21117 get overflowed constants as invariant part.
21118 (compare_values_warnv): Add comment before the TREE_NO_WARNING
21119 checks. Use wi::cmp instead of recursing for integer constants.
21120 (compare_values): Just ignore whether we assumed undefined
21121 overflow instead of failing the compare.
21122 (extract_range_for_var_from_comparison_expr): Add comment before the
21123 TREE_NO_WARNING sets.
21124 (test_for_singularity): Likewise.
21125 (extract_range_from_comparison): Do not disable optimization
21126 when we assumed undefined overflow.
21127 (extract_range_basic): Remove init of unused var.
21129 2017-05-09 Richard Biener <rguenther@suse.de>
21131 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
21132 (extract_range_from_multiplicative_op_1): Adjust.
21133 (extract_range_from_binary_expr_1): Use int_const_binop.
21135 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
21138 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
21139 rs6000_store_data_bypass_p in seven define_bypass directives and
21140 in several comments.
21141 * config/rs6000/rs6000-protos.h: Add prototype for
21142 rs6000_store_data_bypass_p function.
21143 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
21144 function implements slightly different (rs6000-specific) semantics
21145 than store_data_bypass_p, returning false rather than aborting
21146 with assertion error when arguments do not satisfy the
21147 requirements of store data bypass.
21148 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
21149 rs6000_store_data_bypass_p.
21151 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
21153 * config/xtensa/xtensa-protos.h
21154 (xtensa_initial_elimination_offset): New declaration.
21155 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21156 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
21157 macro definition, add case for FRAME_POINTER_REGNUM when
21158 FRAME_GROWS_DOWNWARD.
21159 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
21160 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
21161 xtensa_initial_elimination_offset.
21163 2017-05-08 Nathan Sidwell <nathan@acm.org>
21165 * doc/invoke.texi: Alphabetize -fdump options.
21167 2017-05-08 Martin Sebor <msebor@redhat.com>
21169 PR translation/80280
21170 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
21172 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
21174 * target.def (compute_frame_layout): New optional target hook.
21175 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
21176 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
21177 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
21179 * reload1.c (verify_initial_elim_offsets): Likewise.
21180 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
21181 (use_simple_return_p): Call arm_compute_frame_layout if needed.
21182 (arm_get_frame_offsets): Split up into this ...
21183 (arm_compute_frame_layout): ... and this function.
21185 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
21187 * config/aarch64/constraints.md (Usa): New constraint.
21188 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
21190 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
21192 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
21193 with_multilib_list after it has been checked.
21195 2017-05-08 Richard Biener <rguenther@suse.de>
21197 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
21198 (bitmap_set_subtract_values): Likewise.
21200 2017-05-08 Richard Biener <rguenther@suse.de>
21202 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
21203 (gimple_assign_nonzero): ... this and remove strict_overflow_p
21205 (gimple_stmt_nonzero_warnv_p): Rename to ...
21206 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
21208 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
21209 (extract_range_basic): Adjust, do not disable propagation on
21210 strict overflow sensitive simplification.
21211 (vrp_visit_cond_stmt): Likewise.
21213 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
21215 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
21216 body size unconditionally.
21218 2017-05-07 Jeff Law <law@redhat.com>
21221 2017-05-06 Jeff Law <law@redhat.com>
21222 PR tree-optimization/78496
21223 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
21226 PR tree-optimization/78496
21227 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
21228 (simplify_stmt_using_ranges): Call it.
21229 (vrp_dom_walker::before_dom_children): Extract equivalences
21230 from an ASSERT_EXPR with an equality comparison against a
21233 2017-05-06 Jeff Law <law@redhat.com>
21235 PR tree-optimization/78496
21236 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
21239 PR tree-optimization/78496
21240 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
21241 (simplify_stmt_using_ranges): Call it.
21242 (vrp_dom_walker::before_dom_children): Extract equivalences
21243 from an ASSERT_EXPR with an equality comparison against a
21246 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
21248 * lra-constraints.c (lra_copy_reg_equiv): New function.
21249 (split_reg): Use it to copy equivalence information from the
21250 original register to the spill register.
21252 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
21254 PR rtl-optimization/75964
21255 * simplify-rtx.c (simplify_const_relational_operation): Remove
21256 invalid handling of comparisons of integer ABS.
21258 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
21260 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
21261 initialize to zero.
21262 (init_regs): Remove declaration.
21263 (function_arg_advance_32): Initialize error_p as boolean variable.
21265 2017-05-05 Nathan Sidwell <nathan@acm.org>
21267 * store-motion.c (remove_reachable_equiv_notes): Reformat long
21268 lines. Use for (;;).
21270 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21272 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
21273 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
21274 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
21275 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
21276 VF=2 that require versioning.
21278 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21280 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
21283 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21285 * diagnostic.h (diagnostic_override_option_index): Convert from
21286 macro to inline function.
21288 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21290 * diagnostic.c (last_module_changed_p): New function.
21291 (set_last_module): New function.
21292 (diagnostic_report_current_module): Convert macro usage to
21293 the above functions.
21294 * diagnostic.h (diagnostic_context::last_module): Strengthen
21295 from const line_map * to const line_map_ordinary *.
21296 (diagnostic_last_module_changed): Delete macro.
21297 (diagnostic_set_last_module): Delete macro.
21299 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21301 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
21302 with diagnostic_report_diagnostic.
21303 (diagnostic_n_impl_richloc): Likewise.
21304 * diagnostic.h (report_diagnostic): Delete macro.
21305 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
21306 with diagnostic_report_diagnostic.
21307 * substring-locations.c (format_warning_va): Likewise.
21309 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21311 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
21312 save/restor of format_spec. Move option-printing code to...
21313 (print_option_information): ...this new function, and
21314 reimplement by simply printing to the pretty_printer,
21315 rather than appending to the format string.
21317 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21319 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
21320 handling logic into...
21321 (update_effective_level_from_pragmas): ...this new function.
21323 2017-05-04 Andrew Waterman <andrew@sifive.com>
21325 * config/riscv/riscv.opt (mstrict-align): New option.
21326 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
21327 (SLOW_UNALIGNED_ACCESS): Define.
21328 (riscv_slow_unaligned_access): Declare.
21329 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
21331 (riscv_slow_unaligned_access): New variable.
21332 (rocket_tune_info): Set slow_unaligned_access to true.
21333 (optimize_size_tune_info): Set slow_unaligned_access to false.
21334 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
21335 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
21336 (riscv_option_override): Set riscv_slow_unaligned_access.
21337 * doc/invoke.texi: Add -mstrict-align to RISC-V.
21339 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
21341 * config/riscv/riscv.md: Unify indentation.
21343 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21348 * config/rs6000/rs6000.md (u code attribute): Add FIX and
21350 (extendsi<mode>2): Add support for doing sign extension via
21351 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
21352 don't have ISA 3.0 instructions.
21353 (extendsi<mode>2 splitter): Likewise.
21354 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
21355 generate the normal insns since SImode can now go in vector
21356 registers. Disallow the special UNSPECs needed for previous
21357 machines to hide SImode being used. Add new insns
21358 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
21359 (fix_trunc<mode>si2_stfiwx): Likewise.
21360 (fix_trunc<mode>si2_internal): Likewise.
21361 (fixuns_trunc<mode>si2): Likewise.
21362 (fixuns_trunc<mode>si2_stfiwx): Likewise.
21363 (fctiw<u>z_<mode>_smallint): Likewise.
21364 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
21365 of floating point to 32-bit integer from doing a direct move to
21366 the GPR registers to do a store.
21367 (fctiwz_<mode>): Break long line.
21369 2017-05-05 Bin Cheng <bin.cheng@arm.com>
21371 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
21372 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
21373 (addr_list, addr_offset_valid_p): New.
21374 (split_address_groups): Check offset validity with above function.
21375 (gt-tree-ssa-loop-ivopts.h): Include header file.
21377 2017-05-05 Nathan Sidwell <nathan@acm.org>
21379 * config.gcc (arm*-*-*): Add missing 'fi'.
21381 2017-05-05 Steve Ellcey <sellcey@cavium.com>
21383 * doc/invoke.texi (-fopt-info): Explicitly say order of options
21384 included in -fopt-info does not matter.
21385 * doc/optinfo.texi (-fopt-info): Fix description of default
21386 behavour. Explicitly say order of options included in -fopt-info
21389 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21391 * config.gcc: Allow combinations of aprofile and rmprofile values for
21392 --with-multilib-list.
21393 * config/arm/t-multilib: New file.
21394 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
21395 variables. Remove setting of ISA and floating-point ABI in
21396 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
21397 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
21398 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
21399 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
21401 * config/arm/t-rmprofile: Likewise except for the matches changes.
21402 * doc/install.texi (--with-multilib-list): Document the combination of
21403 aprofile and rmprofile values and warn about pitfalls in doing that.
21405 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
21407 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
21408 (movdi_aarch64): Likewise.
21410 2017-05-05 Jakub Jelinek <jakub@redhat.com>
21412 PR tree-optimization/80632
21413 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
21415 (build_arrays): Initialize it for virtual phis.
21416 (fix_phi_nodes): Use it for virtual phis.
21418 PR tree-optimization/80558
21419 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
21420 [x, y] op z into [x op, y op z] for op & or | if conditions
21423 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
21424 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21427 * config/arm/arm.md (use_literal_pool): Remove.
21428 (64-bit immediate split): No longer takes cost into consideration
21429 if arm_disable_literal_pool is enabled.
21430 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
21431 used when arm_disable_literal_pool is enabled.
21432 (arm_max_const_double_inline_cost): Remove use of
21433 arm_disable_literal_pool.
21434 (push_minipool_fix): Add assert.
21435 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
21436 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
21437 (no_literal_pool_sf_immediate): New.
21439 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21441 PR tree-optimization/80613
21442 * tree-ssa-dce.c (propagate_necessity): Remove cases for
21443 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
21445 2017-05-05 Richard Biener <rguenther@suse.de>
21447 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
21449 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
21451 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
21452 of this flag from insn conditions due to removal from r247495.
21454 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
21456 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
21458 (arm_early_store_addr_dep_ptr): Likewise.
21459 * config/arm/aarch-common-protos.h
21460 (arm_early_load_addr_dep_ptr): Add prototype.
21461 (arm_early_store_addr_dep_ptr): Likewise.
21462 * config/arm/cortex-a53.md: Add new bypasses.
21464 2017-05-05 Jakub Jelinek <jakub@redhat.com>
21466 * tree.c (next_type_uid): Change type to unsigned.
21467 (type_hash_canon): Decrement back next_type_uid if
21468 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
21469 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
21472 2017-05-04 Martin Sebor <msebor@redhat.com>
21474 * builtins.c: Fix a trivial typo in a comment.
21476 PR middle-end/79234
21477 * builtins.c (check_sizes): Adjust to handle reading past the end.
21478 Avoid printing excessive upper bound of ranges. Use %E to print
21479 tree nodes instead of converting them to %wu.
21480 (expand_builtin_memchr): New function.
21481 (compute_dest_size): Rename...
21482 (compute_objsize): ...to this.
21483 (expand_builtin_memcpy): Adjust.
21484 (expand_builtin_mempcpy): Adjust.
21485 (expand_builtin_strcat): Adjust.
21486 (expand_builtin_strcpy): Adjust.
21487 (check_strncat_sizes): Adjust.
21488 (expand_builtin_strncat): Adjust.
21489 (expand_builtin_strncpy): Adjust and simplify.
21490 (expand_builtin_memset): Adjust.
21491 (expand_builtin_bzero): Adjust.
21492 (expand_builtin_memcmp): Adjust.
21493 (expand_builtin): Handle memcmp.
21494 (maybe_emit_chk_warning): Check strncat just once.
21496 2017-05-04 Martin Sebor <msebor@redhat.com>
21498 PR preprocessor/79214
21499 PR middle-end/79222
21500 PR middle-end/79223
21501 * builtins.c (check_sizes): Add inlining context and issue
21502 warnings even when -Wno-system-headers is set.
21503 (check_strncat_sizes): Same.
21504 (expand_builtin_strncat): Same.
21505 (expand_builtin_memmove): New function.
21506 (expand_builtin_stpncpy): Same.
21507 (expand_builtin): Handle memmove and stpncpy.
21509 2017-05-04 Bin Cheng <bin.cheng@arm.com>
21511 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
21512 which is not used any more.
21514 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21516 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
21518 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21520 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
21521 (cortexa53_tunings): Likewise.
21522 (cortexa57_tunings): Likewise.
21523 (cortexa72_tunings): Likewise.
21524 (cortexa73_tunings): Likewise.
21526 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21528 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
21529 Set loop alignment to 8.
21531 2017-05-04 Martin Sebor <msebor@redhat.com>
21533 PR translation/80280
21534 * builtins.c (expand_builtin_object_size): Add missing quoting to
21535 %D and like directives.
21536 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
21537 (hsa_type_for_tree_type): Same.
21538 (verify_function_arguments): Same.
21539 * symtab.c (symbol_table::change_decl_assembler_name): Same.
21540 * varasm.c (get_section): Same.
21543 2017-05-04 Martin Sebor <msebor@redhat.com>
21545 PR translation/80280
21546 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
21548 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21550 * config/aarch64/aarch64.c (generic_addrcost_table):
21551 Change HI/TI mode setting.
21553 2017-05-04 Martin Jambor <mjambor@suse.cz>
21555 PR tree-optimization/80622
21556 * tree-sra.c (comes_initialized_p): New function.
21557 (build_accesses_from_assign): Only set write lazily when
21558 comes_initialized_p is false.
21559 (analyze_access_subtree): Use comes_initialized_p.
21560 (propagate_subaccesses_across_link): Assert !comes_initialized_p
21561 instead of testing for PARM_DECL.
21563 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21565 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
21566 constraint on operand 0 to allow more general addressing modes.
21567 Adjust output template.
21568 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
21570 * config/aarch64/aarch64-protos.h
21571 (aarch64_address_valid_for_prefetch_p): Declare prototype.
21572 * config/aarch64/constraints.md (Dp): New address constraint.
21573 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
21576 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
21578 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
21579 update use of estimate_ipcp_clone_size_and_time.
21580 (estimate_local_effects): Update use of
21581 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
21582 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
21583 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
21584 Return nonspecialized time.
21586 2017-05-04 Richard Biener <rguenther@suse.de>
21588 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
21589 for the last VUSE which def dominates the PHI. Directly call
21591 (get_continuation_for_phi_1): Remove.
21593 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
21595 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
21596 to explain the use of truncating division. Cap the number of
21597 iterations to the maximum given by nb_iterations_upper_bound,
21600 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21602 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
21603 * configure: Regenerate.
21604 * config.in: Regenerate.
21605 * config/i386/driver-mingw32.c: new file.
21606 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
21607 * config.host: Link driver-mingw32.o on MinGW host.
21608 * doc/install.texi: Document new --enable-mingw-wildcard configure
21611 2017-05-04 Marek Polacek <polacek@redhat.com>
21613 PR tree-optimization/80612
21614 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
21616 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21617 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
21619 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
21620 (movt splitter): Likewise.
21621 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
21622 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
21623 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
21624 block for Thumb-1 with MOVT.
21625 (thumb2_legitimate_address_p): Move code block ...
21626 (can_avoid_literal_pool_for_label_p): ... into this new function.
21627 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
21629 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
21630 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
21631 "M-profile targets with the MOVT instruction".
21633 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21635 * config/arm/arm-builtins.c (arm_init_builtins): Rename
21636 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
21637 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
21639 2017-05-04 Martin Liska <mliska@suse.cz>
21641 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
21642 variable cond_code.
21644 2017-05-04 Richard Biener <rguenther@suse.de>
21646 * tree.c (array_at_struct_end_p): Handle arrays at struct
21647 end with flexarrays more conservatively. Refactor and treat
21648 arrays of arrays or aggregates more strict. Fix
21649 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
21650 * tree.c (array_at_struct_end_p): Adjust prototype.
21651 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
21652 * gimple-fold.c (get_range_strlen): Likewise.
21653 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
21655 2017-05-04 Richard Biener <rguenther@suse.de>
21657 PR tree-optimization/31130
21658 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
21660 (supports_overflow_infinity): Likewise.
21661 (is_negative_overflow_infinity): Likewise.
21662 (is_positive_overflow_infinity): Likewise.
21663 (is_overflow_infinity): Likewise.
21664 (stmt_overflow_infinity): Likewise.
21665 (overflow_infinity_range_p): Likewise.
21666 (usable_range_p): Remove as always returning true.
21667 (make_overflow_infinity): Remove.
21668 (negative_overflow_infinity): Likewise.
21669 (positive_overflow_infinity): Likewise.
21670 (avoid_overflow_infinity): Likewise.
21671 (set_value_range): Adjust accordingly.
21672 (set_value_range_to_nonnegative): Likewise, remove now unused
21673 overflow_infinity arg.
21674 (vrp_operand_equal_p): Adjust.
21675 (update_value_range): Likewise.
21676 (range_int_cst_singleton_p): Likewise.
21677 (operand_less_p): Likewise.
21678 (compare_values_warnv): Likewise.
21679 (extract_range_for_var_from_comparison_expr): Likewise.
21680 (vrp_int_const_binop): Likewise.
21681 (zero_nonzero_bits_from_vr): Likewise.
21682 (extract_range_from_multiplicative_op_1): Likewise.
21683 (extract_range_from_binary_expr_1): Likewise.
21684 (extract_range_from_unary_expr): Likewise.
21685 (extract_range_from_comparison): Likewise.
21686 (extract_range_basic): Likewise.
21687 (adjust_range_with_scev): Likewise.
21688 (compare_ranges): Likewise.
21689 (compare_range_with_value): Likewise.
21690 (dump_value_range): Likewise.
21691 (test_for_singularity): Likewise, remove strict_overflow_p parameter
21693 (simplify_cond_using_ranges): Adjust.
21695 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
21697 * brig-builtins.def: Added a builtin for class_f64.
21698 * builtin-types.def: Added a builtin type needed by class_f64.
21700 2017-05-03 Jason Merrill <jason@redhat.com>
21702 * timevar.def: Add TV_CONSTEXPR.
21704 2017-05-03 David Malcolm <dmalcolm@redhat.com>
21706 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
21708 2017-05-03 Martin Jambor <mjambor@suse.cz>
21710 * ipa-prop.c (ipa_update_after_lto_read): Removed.
21711 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
21712 * ipa-cp.c (ipcp_propagate_stage): Do not call
21713 ipa_update_after_lto_read.
21714 * ipa-inline.c (ipa_inline): Likewise.
21716 2017-05-03 Martin Jambor <mjambor@suse.cz>
21718 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
21719 tag. Added a default constructor and a destructor.
21720 (ipa_edge_args_sum_t): New class;
21721 (ipa_edge_args_sum): Declare.
21722 (ipa_edge_args_vector): Remove declaration.
21723 (IPA_EDGE_REF): Use ipa_edge_args_sum.
21724 (ipa_free_edge_args_substructures): Remove declaration.
21725 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
21726 (ipa_edge_args_info_available_for_edge_p): Likewise.
21727 * ipa-prop.c (ipa_edge_args_vector): Removed.
21728 (edge_removal_hook_holder): Likewise.
21729 (edge_duplication_hook_holder): Likewise.
21730 (ipa_edge_args_sum): New variable.
21731 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
21732 ipa_edge_args_vector.
21733 (ipa_free_edge_args_substructures): Likewise.
21734 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
21735 ipa_edge_args_vector.
21736 (ipa_edge_removal_hook): Turned into method
21737 ipa_edge_args_sum_t::remove.
21738 (ipa_edge_duplication_hook): Turned into method
21739 ipa_edge_args_sum_t::duplicate.
21740 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
21741 registering edge hooks.
21742 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
21743 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
21744 ipa_edge_args_sum instead of ipa_edge_args_vector.
21745 * ipa-profile.c (ipa_profile): Likewise.
21747 2017-05-03 Martin Jambor <mjambor@suse.cz>
21749 * symbol-summary.h (function_summary): New method exists.
21750 (function_summary::symtab_removal): Deallocate through release.
21751 (call_summary): New class.
21752 (gt_ggc_mx): New overload.
21753 (gt_pch_nx): Likewise.
21754 (gt_pch_nx): Likewise.
21756 2017-05-03 Jeff Law <law@redhat.com>
21758 PR tree-optimization/78496
21759 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
21760 from simplify_cond_using_ranges. Split off code to walk
21761 backwards through casts into ...
21762 (simplify_cond_using_ranges_2): New function.
21763 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
21764 (execute_vrp): After identifying jump threads, call
21765 simplify_cond_using_ranges_2.
21767 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
21770 * ipa-inline.h (inline_summary): Add ctor.
21771 (create_ggc): Do not use ggc_cleared_alloc.
21773 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
21774 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21776 * gcc.c (handle_braces): Support escaping in switch matching
21778 * doc/invoke.texi (Spec Files): Document it.
21779 Remove superfluous @code markup in items.
21781 2017-05-03 David Malcolm <dmalcolm@redhat.com>
21783 * diagnostic-show-locus.c (struct column_range): New struct.
21784 (get_affected_columns): New function.
21785 (get_printed_columns): New function.
21786 (struct correction): New struct.
21787 (correction::ensure_capacity): New function.
21788 (correction::ensure_terminated): New function.
21789 (struct line_corrections): New struct.
21790 (line_corrections::~line_corrections): New dtor.
21791 (line_corrections::add_hint): New function.
21792 (layout::print_trailing_fixits): Reimplement in terms of the new
21794 (selftest::test_overlapped_fixit_printing): New function.
21795 (selftest::diagnostic_show_locus_c_tests): Call it.
21797 2017-05-03 Nathan Sidwell <nathan@acm.org>
21799 Canonicalize canonical type hashing
21800 * tree.h (type_hash_canon_hash): Declare.
21801 * tree.c (type_hash_list, attribute_hash_list): Move into
21802 type_hash_canon_hash.
21803 (build_type_attribute_qual_variant): Break out hash code calc into
21804 type_hash_canon_hash.
21805 (type_hash_canon_hash): New. Generic type hash computation.
21806 (build_range_type_1, build_array_type_1, build_function_type,
21807 build_method_type_directly, build_offset_type, build_complex_type,
21808 make_vector_type): Call it.
21810 2017-05-03 Richard Biener <rguenther@suse.de>
21812 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
21813 When all DRs have unknown misaligned do not always peel
21814 when there is a store but apply the same costing model as if
21815 there were only loads.
21817 2017-05-03 Richard Biener <rguenther@suse.de>
21820 PR tree-optimization/80492
21821 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
21822 compare_base_decls returning dont-know properly.
21824 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
21826 * config/arm/iterators.md (CCSI): New mode iterator.
21827 (arch): New mode attribute.
21828 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
21829 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
21830 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
21831 code iterator for success result mode.
21832 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
21833 the corresponding new insn generators.
21835 2017-05-03 Bin Cheng <bin.cheng@arm.com>
21838 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21839 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21841 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
21843 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
21844 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
21845 (DDR_A): Wrap DDR argument in brackets.
21846 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
21847 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
21848 (DDR_REVERSED_P): Likewise.
21850 2017-05-03 Jakub Jelinek <jakub@redhat.com>
21852 PR tree-optimization/79472
21853 * tree-switch-conversion.c (struct switch_conv_info): Add
21854 contiguous_range and default_case_nonstandard fields.
21855 (collect_switch_conv_info): Compute contiguous_range and
21856 default_case_nonstandard fields, don't clear final_bb if
21857 contiguous_range and only the default case doesn't have the required
21859 (check_all_empty_except_final): Set default_case_nonstandard instead
21860 of failing if contiguous_range and the default case doesn't have empty
21862 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
21863 and only the default case doesn't have the required constants. Skip
21865 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
21866 if default_case_nonstandard.
21867 (build_constructors): Build constant 1 just once. Assert that default
21868 values aren't inserted in between cases if contiguous_range. Skip
21870 (build_arrays): Skip virtual phis.
21871 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
21872 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
21873 Handle virtual phis.
21874 (gen_inbound_check): Handle default_case_nonstandard case.
21875 (process_switch): Adjust check_final_bb caller. Call
21876 gather_default_values with the first non-default case instead of
21877 default case if default_case_nonstandard.
21879 2017-05-02 Nathan Sidwell <nathan@acm.org>
21881 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
21882 check. Fix formatting.
21884 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
21886 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
21887 errors when comparing specialized and unspecialized times.
21889 2017-05-02 David Malcolm <dmalcolm@redhat.com>
21891 * diagnostic-show-locus.c
21892 (layout::should_print_annotation_line_p): Make private.
21893 (layout::print_annotation_line): Make private.
21894 (layout::annotation_line_showed_range_p): Make private.
21895 (layout::show_ruler): Make private.
21896 (layout::print_source_line): Make private. Pass in line and
21897 line_width, rather than calling location_get_source_line. Drop
21899 (layout::print_leading_fixits): New method.
21900 (layout::print_any_fixits): Rename to...
21901 (layout::print_trailing_fixits): ...this, and make private.
21902 Don't print newline fixits.
21903 (diagnostic_show_locus): Move logic for printing one row into...
21904 (layout::print_line): ...this new function. Move the
21905 location_get_source_line call and error-handling from
21906 print_source_line to here. Call print_leading_fixits, and rename
21907 print_any_fixits to print_trailing_fixits.
21908 (selftest::test_fixit_insert_containing_newline): Update now that
21909 newlines are partially supported.
21910 (selftest::test_fixit_insert_containing_newline_2): New test.
21911 (selftest::test_fixit_replace_containing_newline): Update comments.
21912 (selftest::diagnostic_show_locus_c_tests): Call the new test.
21913 * edit-context.c (class added_line): New class.
21914 (class edited_line): Describe newline handling in comment.
21915 (edited_line::actually_edited_p): New method.
21916 (edited_line::print_content): Delete redundant decl.
21917 (edited_line::m_predecessors): New field.
21918 (edited_file::print_content): Call edited_line::print_content.
21919 (edited_file::print_diff): Update to support newlines.
21920 (edited_file::print_diff_hunk): Likewise.
21921 (edited_file::print_run_of_changed_lines): New function.
21922 (edited_file::print_diff_line): Convert to...
21923 (print_diff_line): ...this.
21924 (edited_file::get_effective_line_count): New function.
21925 (edited_line::edited_line): Initialize new field m_predecessors.
21926 (edited_line::~edited_line): Clean up m_predecessors.
21927 (edited_line::apply_fixit): Handle newlines.
21928 (edited_line::get_effective_line_count): New function.
21929 (edited_line::print_content): New function.
21930 (edited_line::print_diff_lines): New function.
21931 (selftest::test_applying_fixits_insert_containing_newline): New
21933 (selftest::test_applying_fixits_replace_containing_newline): New
21935 (selftest::insert_line): New function.
21936 (selftest::test_applying_fixits_multiple_lines): Add example of
21938 (selftest::edit_context_c_tests): Call the new tests.
21940 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21942 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
21943 parameter cand. Update dump information.
21944 (get_computation_cost): Update uses.
21946 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21948 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
21949 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
21950 (get_computation_at, rewrite_use_address): Update use of
21951 get_computation_aff.
21953 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21955 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
21956 (get_computation): Delete.
21957 (get_computation_cost): Implement like get_computation_cost_at.
21958 Use get_computation_at.
21959 (get_computation_cost_at): Delete.
21960 (rewrite_use_nonlinear_expr): Use get_computation_at.
21961 (rewrite_use_compare, remove_unused_ivs): Ditto.
21963 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21965 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
21967 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21969 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
21970 (ivopts_global_cost_for_size): Rename parameter and update uses.
21971 (iv_ca_recount_cost): Update uses.
21972 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
21973 candidates seperately in n_invs and n_cands.
21974 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
21976 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21978 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
21979 (find_inv_vars_cb): New.
21980 (find_depends): Renamed to ...
21981 (find_inv_vars): ... this.
21982 (add_candidate_1, force_var_cost): Call find_inv_vars.
21983 (split_address_cost, determine_group_iv_cost_cond): Ditto.
21985 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21987 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
21988 inv_vars. Add inv_exprs.
21989 (struct iv_cand): Rename depends_on to inv_vars.
21990 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
21991 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
21992 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
21993 (dump_cand): Dump inv_vars.
21994 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
21995 (record_invariant, find_depends, add_candidate_1): Ditto.
21996 (set_group_iv_cost, force_var_cost): Ditto.
21997 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
21998 (get_computation_cost_at, get_computation_cost): Ditto.
21999 (determine_group_iv_cost_generic): Ditto.
22000 (determine_group_iv_cost_address): Ditto.
22001 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
22002 (determine_group_iv_costs): Ditto.
22003 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
22004 (iv_ca_set_remove_invariants): Renamed to ...
22005 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
22006 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
22007 (iv_ca_set_add_invariants): Renamed to ...
22008 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
22009 (iv_ca_set_cp): Use iv_ca_set_add_invs.
22010 (iv_ca_has_deps): Support inv_vars and inv_exprs.
22011 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
22012 (create_new_ivs): Remove useless dump.
22014 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22016 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
22018 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
22019 (iv_ca_set_no_cp, create_new_iv): Ditto.
22021 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22023 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22025 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
22027 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
22028 function tree_check2.
22030 2017-05-02 Martin Liska <mliska@suse.cz>
22032 * doc/gcov.texi: Add missing preposition.
22033 * gcov.c (function_info::function_info): Properly fill up
22034 all member variables.
22036 2017-05-02 Tamar Christina <tamar.christina@arm.com>
22038 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
22040 2017-05-02 Tamar Christina <tamar.christina@arm.com>
22042 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
22044 2017-05-02 Martin Liska <mliska@suse.cz>
22047 * lto-streamer-in.c (lto_read_tree_1): Remove
22048 LTO_STREAMER_DEBUG.
22049 * lto-streamer.c (struct tree_hash_entry): Likewise.
22050 (struct tree_entry_hasher): Likewise.
22051 (tree_entry_hasher::hash): Likewise.
22052 (tree_entry_hasher::equal): Likewise.
22053 (lto_streamer_init): Likewise.
22054 (lto_orig_address_map): Likewise.
22055 (lto_orig_address_get): Likewise.
22056 (lto_orig_address_remove): Likewise.
22057 * lto-streamer.h: Likewise.
22058 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
22059 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
22061 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
22063 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
22064 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
22065 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
22066 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
22067 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
22068 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
22069 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
22070 (mm_maskz_sub_ss): New intrinsics.
22071 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22072 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
22073 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
22074 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
22075 (__builtin_ia32_subss_mask_round): New builtins.
22076 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22077 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
22078 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
22080 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
22081 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
22083 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
22086 2017-05-02 Martin Jambor <mjambor@suse.cz>
22088 PR tree-optimization/78687
22089 * tree-sra.c (access): New field parent.
22090 (process_subtree_disqualification): New function.
22091 (disqualify_candidate): Call it.
22092 (build_accesses_from_assign): Reset write flag if creating an
22094 (build_access_subtree): Fill in parent field and also prpagate
22095 down grp_write flag.
22096 (create_artificial_child_access): New parameter set_grp_write, set
22097 grp_write to its value.
22098 (propagate_subaccesses_across_link): Also propagate grp_write flag
22100 (propagate_all_subaccesses): Push the closest parent back to work
22101 queue if add_access_to_work_queue returned true.
22103 2017-05-02 Richard Biener <rguenther@suse.de>
22105 * common.opt (fstrict-overflow): Alias negative to fwrapv.
22106 * doc/invoke.texi (fstrict-overflow): Remove all traces of
22107 -fstrict-overflow documentation.
22108 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
22109 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
22110 flag_strict_overflow.
22111 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
22112 * lto-opts.c (lto_write_options): Do not stream it.
22113 * lto-wrapper.c (merge_and_complain): Do not handle it.
22114 * opts.c (default_options_table): Do not set -fstrict-overflow.
22115 (finish_options): Likewise do not clear it when sanitizing.
22116 * simplify-rtx.c (simplify_const_relational_operation): Do not
22117 test flag_strict_overflow.
22119 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
22121 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
22122 using enabled attribute.
22123 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
22124 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
22125 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
22126 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
22127 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
22128 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
22129 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
22130 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
22131 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
22132 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
22134 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
22136 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
22138 2017-05-02 Richard Biener <rguenther@suse.de>
22140 PR tree-optimization/80591
22142 2017-04-10 Richard Biener <rguenther@suse.de>
22144 * tree-ssa-structalias.c (find_func_aliases): Properly handle
22147 2017-05-02 Richard Biener <rguenther@suse.de>
22149 PR tree-optimization/80549
22150 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
22151 (cleanup_tree_cfg_noloop): Create forwarders to known loop
22152 headers if they do not have a preheader.
22154 2017-05-02 Martin Liska <mliska@suse.cz>
22157 * common.opt: Fix typo.
22158 * doc/invoke.texi: Likewise.
22160 2017-05-01 Jan Beulich <jbeulich@suse.com>
22162 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
22163 swapping, add (x,x,m,x,n) alternative.
22165 2017-05-01 Nathan Sidwell <nathan@acm.org>
22167 * calls.c (combine_pending_stack_adjustment_and_call): Remove
22168 unnecessary unadjusted_alignment check.
22170 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
22173 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
22175 * gimplify.c (gimplify_cilk_detach): New function.
22176 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
22177 * tree-core.h: Document EXPR_CILK_SPAWN.
22178 * tree.h (EXPR_CILK_SPAWN): Define.
22180 2017-05-01 David Malcolm <dmalcolm@redhat.com>
22182 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
22183 to use new fixit_hint representation, using the "replace" logic.
22184 (get_line_span_for_fixit_hint): Likewise.
22185 (layout::print_any_fixits): Likewise.
22186 (selftest::test_one_liner_many_fixits): Rename to...
22187 (selftest::test_one_liner_many_fixits_1): ...this, and update
22188 comment and expected output to reflect that the multiple fix-it
22189 hints are now consolidated into one insertion.
22190 (selftest::test_one_liner_many_fixits_2): New test.
22191 (selftest::test_diagnostic_show_locus_one_liner): Update for
22193 (selftest::test_fixit_consolidation): Update for fix-it API
22195 * diagnostic.c (print_parseable_fixits): Likewise.
22196 * edit-context.c (edited_line::m_line_events): Convert from
22197 auto_vec <line_event *> to auto_vec <line_event>.
22198 (class line_event): Convert from abstract base class to a concrete
22199 class, taking over the role of replace_event.
22200 (class insert_event): Delete.
22201 (class replace_event): Rename to class line_event. Convert to
22203 (edit_context::add_fixits): Reimplement.
22204 (edit_context::apply_insert): Delete.
22205 (edit_context::apply_replace): Rename to...
22206 (edit_context::apply_fixit): ...this. Convert to half-open range.
22207 (edited_file::apply_insert): Delete.
22208 (edited_file::apply_replace): Rename to...
22209 (edited_file::apply_fixit): ...this.
22210 (edited_line::~edited_line): Drop deletion of events.
22211 (edited_line::apply_insert): Delete.
22212 (edited_line::apply_replace): Rename to...
22213 (edited_line::apply_fixit): ...this. Convert to half-open range.
22214 Update for change to type of m_line_events.
22215 * edit-context.h (edit_context::apply_insert): Delete.
22216 (edit_context::apply_replace): Rename to...
22217 (edit_context::apply_fixit): ...this.
22219 2017-05-01 Martin Sebor <msebor@redhat.com>
22221 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
22224 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
22227 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
22228 __get_cpuid_max returns 0.
22229 (__get_cpuid_count): Ditto.
22231 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
22233 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
22234 replacement expression is another instance of one of its arguments.
22236 2017-05-01 Jakub Jelinek <jakub@redhat.com>
22239 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
22240 check for stack push/pop autoinc.
22241 * config/i386/i386.c (ix86_agi_dependent): Return false
22242 if the only reason why modified_in_p returned true is that
22243 addr is SP based and set_insn is a push or pop.
22245 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
22247 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
22250 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
22253 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
22254 (account_size_time): Use two predicates - exec_pred and
22256 (evaluate_conditions_for_known_args): Compute both clause and
22258 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
22259 (inline_summary_t::duplicate): Update.
22260 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
22262 (compute_inline_parameters): Likewise.
22263 (estimate_edge_size_and_time): Update caluclation of time.
22264 (estimate_node_size_and_time): Compute both time and nonspecialized
22266 (estimate_ipcp_clone_size_and_time): Update.
22267 (inline_merge_summary): Update.
22268 (do_estimate_edge_time): Update.
22269 (do_estimate_edge_size): Update.
22270 (do_estimate_edge_hints): Update.
22271 (inline_read_section, inline_write_summary): Stream both new predicates.
22272 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
22274 (compute_inlined_call_time): Cleanup.
22275 (big_speedup_p): Update.
22276 (edge_badness): Update.
22277 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
22278 (size_time_entry): Replace predicate by exec_predicate and
22279 nonconst_predicate.
22280 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
22281 (estimate_edge_time): Return also nonspec_time.
22282 (reset_edge_growth_cache): Update.
22284 2017-04-29 Jakub Jelinek <jakub@redhat.com>
22286 PR rtl-optimization/80491
22287 * ifcvt.c (noce_process_if_block): When looking for x setter
22288 with missing else_bb, don't check only the insn right before
22289 cond_earliest, but look for the last insn that x is modified in
22290 within the same bb.
22292 PR rtl-optimization/80491
22293 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
22295 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
22297 PR tree-optimization/80487
22298 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
22300 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22302 PR tree-optimization/79697
22303 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
22304 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
22305 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
22307 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
22308 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
22310 2017-04-28 Martin Sebor <msebor@redhat.com>
22312 PR tree-optimization/80523
22313 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
22314 (init_target_to_host_charmap, target_to_host, target_strtol10): New
22316 (maybe_warn, format_directive, parse_directive): Use new functions.
22317 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
22319 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
22321 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
22323 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
22325 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
22326 target_header_dir): Set correctly.
22327 * configure: Regenerated.
22328 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22329 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22330 instead of SYSTEM_HEADER_DIR.
22332 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
22334 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
22335 (estimate_local_effects): Likewise.
22336 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
22337 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
22338 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
22339 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
22340 do_estimate_edge_time, estimate_edge_time): Likewise.
22341 * ipa-inline-analysis.c (estimate_node_size_and_time,
22342 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
22343 (estimate_time_after_inlining): Remove.
22345 2017-04-28 Martin Liska <mliska@suse.cz>
22347 * doc/gcov.texi: Enhance documentation of gcov.
22349 2017-04-28 Martin Liska <mliska@suse.cz>
22351 * doc/gcov.texi: Sort options in alphabetic order.
22352 * doc/gcov-dump.texi: Likewise.
22353 * doc/gcov-tool.texi: Likewise.
22354 * gcov.c (print_usage): Likewise.
22355 * gcov-dump.c (print_usage): Likewise.
22356 * gcov-tool.c (print_merge_usage_message): Likewise.
22357 (print_rewrite_usage_message): Likewise.
22358 (print_overlap_usage_message): Likewise.
22360 2017-04-28 Martin Liska <mliska@suse.cz>
22362 PR gcov-profile/53915
22363 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
22365 2017-04-28 Martin Liska <mliska@suse.cz>
22367 PR gcov-profile/79891
22368 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
22369 is marked by compiler as living on a line.
22370 (get_cycles_count): Remove usage of the union.
22371 (output_intermediate_file): Likewise.
22372 (find_source): Fix GNU coding style.
22373 (accumulate_line_counts): Remove old non-all block mode.
22374 (output_lines): Remove usage of the union.
22375 * profile.c (output_location): Include all BBs, even if
22376 belonging to a same line (and file) as a previous BB.
22378 2017-04-28 Martin Liska <mliska@suse.cz>
22380 * gcov.c (process_args): Handle new argument 'w'.
22381 (read_graph_file): Assign ID to BBs.
22382 (output_branch_count): Display BB # if verbose flag is set.
22383 (output_lines): Likewise for arcs.
22384 (print_usage): Add '--verbose' option help.
22385 * doc/gcov.texi: Document --verbose (-w) option.
22387 2017-04-28 Martin Liska <mliska@suse.cz>
22389 * gcov.c (struct block_location_info): New struct.
22390 (process_file): Fill up the new structure.
22391 (read_graph_file): Replace usage of encoding by the newly added
22393 (add_line_counts): Likewise.
22394 (accumulate_line_counts): Remove usage of the union.
22395 (function_info::function_info): New function.
22396 (function_info::~function_info): Likewise.
22397 (process_file): Call delete instead of release_function.
22398 (release_function): Release the function.
22399 (release_structures): Call delete instead of release_function.
22400 (solve_flow_graph): Replace usage of num_blocks.
22401 (find_exception_blocks): Likewise.
22402 (output_lines): Fix GNU coding style.
22404 2017-04-28 Martin Liska <mliska@suse.cz>
22407 * coverage.c (coverage_remove_note_file): New function.
22408 * coverage.h: Declare the function.
22409 * toplev.c (finalize): Clean if an error has been seen.
22411 2017-04-28 Martin Liska <mliska@suse.cz>
22413 PR gcov-profile/80031
22414 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
22415 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
22416 * gcov.c (read_graph_file): Read just number of blocks.
22417 * profile.c (branch_prob): Do not stream 0 flags per a basic
22420 2017-04-28 Martin Liska <mliska@suse.cz>
22422 * gcov-dump.c (tag_*): Add new argument to declarations.
22423 (dump_gcov_file): Likewise.
22424 (tag_blocks): Add and use new argument depth.
22425 (tag_arcs): Likewise.
22426 (tag_lines): Likewise.
22427 (tag_counters): Likewise.
22428 (tag_summary): Likewise.
22429 (dump_working_sets): Use depth to do a proper indentation.
22431 2017-04-28 Jakub Jelinek <jakub@redhat.com>
22434 * cgraph.h (symtab_node::debug_symtab): No longer inline.
22435 * symtab.c (symtab_node::debug_symtab): Move definition here.
22437 2017-04-28 Richard Biener <rguenther@suse.de>
22439 * lto-streamer.h (LTO_major_version): Bump to 7.
22441 2017-04-28 Richard Biener <rguenther@suse.de>
22443 * tree-vrp.c (assert_info): New struct.
22444 (add_assert_info): New helper.
22445 (register_edge_assert_for_2): Refactor to add asserts to a vector
22447 (register_edge_assert_for_1): Likewise.
22448 (register_edge_assert_for): Likewise.
22449 (finish_register_edge_assert_for): New helper actually registering
22450 asserts where live on edge.
22451 (find_conditional_asserts): Adjust.
22452 (find_switch_asserts): Likewise.
22453 (evrp_dom_walker::try_find_new_range): Generalize.
22454 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
22456 2017-04-27 Marek Polacek <polacek@redhat.com>
22459 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
22460 arg10 and arg11 to itype.
22462 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
22464 * doc/extend.texi (Object Size Checking): Improve grammar.
22466 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
22469 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
22470 that the logic for permitting reciprocal estimates matches that
22473 2017-04-27 Jakub Jelinek <jakub@redhat.com>
22476 * tree.c (type_cache_hasher::equal): Only compare
22477 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
22478 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
22479 non-aggregate element types.
22480 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
22481 about the flag on ARRAY_TYPEs in the comment, formatting fix.
22483 2017-04-27 Richard Biener <rguenther@suse.de>
22485 PR middle-end/80533
22486 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
22487 stripping ARRAY_REFs from MEM_EXPR make sure we're not
22488 keeping a reference to a trailing array.
22490 2017-04-27 Richard Biener <rguenther@suse.de>
22492 PR middle-end/80539
22493 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
22494 being in loop-closed SSA form conservatively.
22495 (chrec_fold_multiply_poly_poly): Likewise.
22497 2017-04-27 Tamar Christina <tamar.christina@arm.com>
22499 PR middle-end/79665
22500 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
22501 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
22503 2017-04-27 Jakub Jelinek <jakub@redhat.com>
22506 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
22507 (aarch64_function_arg_alignment): Return unsigned int again, but still
22508 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
22509 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
22510 Don't emit -Wpsabi note.
22511 (aarch64_function_arg_boundary): Likewise.
22512 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
22515 2017-04-26 Nathan Sidwell <nathan@acm.org>
22517 * tree.h (crc32_unsigned_n): Declare.
22518 (crc32_unsigned, crc32_unsigned): Make inline.
22519 * tree.c (crc32_unsigned_bits): Replace with ...
22520 (crc32_unsigned_n): ... this.
22521 (crc32_unsigned, crc32_byte): Remove.
22522 (crc32_string): Remove unnecessary braces.
22524 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
22526 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
22527 * ipa-inline-analysis.c (MAX_TIME): Remove.
22528 (account_size_time): Use sreal for time.
22529 (dump_inline_summary): Update.
22530 (estimate_function_body_sizes): Update.
22531 (estimate_edge_size_and_time): Update.
22532 (estimate_calls_size_and_time): Update.
22533 (estimate_node_size_and_time): Update.
22534 (inline_merge_summary): Update.
22535 (inline_update_overall_summary): Update.
22536 (estimate_time_after_inlining): Update.
22537 (inline_read_section): Update.
22538 (inline_write_summary): Update.
22539 * ipa-inline.c (compute_uninlined_call_time): Update.
22540 (compute_inlined_call_time): Update.
22541 (recursive_inlining): Update.
22542 (inline_small_functions): Update.
22543 (dump_overall_stats): Update.
22544 * ipa-inline.h: Include sreal.h.
22545 (size_time_entry): Turn time to sreal.
22546 (inline_summary): Turn self_time nad time to sreal.
22548 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
22550 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
22552 (sreal::stream_out, sreal::stream_in): New.
22553 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
22555 2017-04-25 Jakub Jelinek <jakub@redhat.com>
22557 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
22560 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
22563 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
22565 (dimode_scalar_chain::compute_convert_gain): Ditto.
22566 (dimode_scalar_chain::make_vector_copies): Ditto.
22567 (dimode_scalar_chain::convert_reg): Ditto.
22568 (dimode_scalar_chain::convert_insn): Ditto.
22569 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
22570 (VI248_AVX512BW_1): New mode iterator.
22571 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
22572 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
22575 2017-04-25 Martin Sebor <msebor@redhat.com>
22577 PR tree-optimization/80497
22578 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
22579 constants are representable in HOST_WIDE_INT.
22580 (parse_directive): Ditto.
22582 2017-04-25 Martin Sebor <msebor@redhat.com>
22585 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
22586 (new_zero_array): Adjust signature.
22587 (dom_info::dom_init): Used unsigned rather that size_t.
22588 (dom_info::dom_info): Same.
22590 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22591 Jakub Jelinek <jakub@redhat.com>
22594 * config/arm/arm.c: Include gimple.h.
22595 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
22596 returns negative, increment ncrn only if it returned positive.
22597 (arm_needs_doubleword_align): Return int instead of bool,
22598 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
22599 members, but if there is any such non-FIELD_DECL
22600 > PARM_BOUNDARY aligned decl, return -1 instead of false.
22601 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
22602 returns negative, increment nregs only if it returned positive.
22603 (arm_setup_incoming_varargs): Likewise.
22604 (arm_function_arg_boundary): Emit -Wpsabi note if
22605 arm_needs_doubleword_align returns negative, return
22606 DOUBLEWORD_ALIGNMENT only if it returned positive.
22608 2017-04-25 Marek Polacek <polacek@redhat.com>
22611 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
22612 first argument to type.
22614 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
22617 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
22618 type checks to test for compatibility instead of equality.
22620 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22621 Jakub Jelinek <jakub@redhat.com>
22624 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
22626 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
22627 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
22628 the alignment computation, but return their maximum in warn_alignment.
22629 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
22630 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
22632 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
22633 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
22636 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22638 * config/arc/simdext.md (dmpyh): Fix typo.
22640 2017-04-25 Richard Biener <rguenther@suse.de>
22642 PR tree-optimization/80492
22643 * alias.c (compare_base_decls): Handle registers with asm
22644 specification conservatively.
22645 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
22646 compare_base_decls returning dont-know properly.
22648 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22650 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
22651 (legitimate_offset_address_p): New function.
22652 (arc_legitimate_address_p): Use above function.
22654 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22656 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
22658 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22660 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
22661 ACCH registers whenever they are available.
22663 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22665 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
22666 double regs fix when not used.
22668 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22670 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
22672 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
22673 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
22675 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22677 * config/arc/arc.c (arc_output_addsi): Check for h-register class
22678 when emitting short ADD instructions.
22680 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22682 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
22684 (cmpsi_cc_c_insn): Likewise.
22685 (cbranchsi4_scratch): Compute proper instruction length using
22686 compact_hreg_operand.
22687 * config/arc/predicates.md (compact_hreg_operand): New predicate.
22689 2017-04-25 Richard Biener <rguenther@suse.de>
22691 PR middle-end/80509
22692 * passes.c (pass_manager::pass_manager): Initialize
22693 m_name_to_pass_map.
22695 2017-04-25 Richard Biener <rguenther@suse.de>
22697 PR tree-optimization/79201
22698 * tree-ssa-sink.c (statement_sink_location): Handle calls.
22700 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22703 * config/s390/vector.md: Split MEM->GPR vector moves for
22704 non-s_operand addresses.
22706 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22709 * config/s390/predicates.md (reload_const_wide_int_operand): New
22711 * config/s390/s390.md ("movti"): Remove d/P alternative.
22712 ("movti_bigconst"): New pattern definition.
22714 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22717 * s390-protos.h (s390_expand_cs_hqi): Removed.
22718 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
22719 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
22720 modes as well as CCZ1mode and CCZmode.
22721 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
22722 signature of s390_emit_compare_and_swap.
22723 (s390_expand_cs_hqi): Likewise, make static.
22724 (s390_expand_cs_tdsi): Generate an explicit compare before trying
22725 compare-and-swap, in some cases.
22726 (s390_expand_cs): Wrapper function.
22727 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
22729 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
22730 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
22731 patterns for small and large integers. Forbid symref memory operands.
22732 Move expander to s390.c. Require cc register.
22733 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
22734 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
22735 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
22736 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
22737 symref memory operands. Remove CC mode and call s390_match_ccmode
22739 ("atomic_exchange<mode>"): Allow and implement all integer modes.
22741 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22743 * config/s390/s390.md (define_peephole2): New peephole to help
22744 combining the load-and-test pattern with volatile memory.
22746 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22748 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
22749 with CCZmode for TARGET_Z196.
22751 2017-04-25 Jakub Jelinek <jakub@redhat.com>
22753 PR rtl-optimization/80501
22754 * combine.c (make_compound_operation_int): Set subreg_code to SET
22755 even for AND with mask of the sign bit of mode.
22757 PR rtl-optimization/80500
22758 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
22759 sum's initial value.
22761 2017-04-25 Julian Brown <julian@codesourcery.com>
22762 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22764 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
22766 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
22768 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
22770 2017-04-25 Julian Brown <julian@codesourcery.com>
22771 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22773 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
22774 (thunderx2t99_sha): New Reservation.
22776 2017-04-25 Julian Brown <julian@codesourcery.com>
22777 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22779 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
22780 type for 1-element load.
22782 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
22784 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
22786 2017-04-24 Martin Jambor <mjambor@suse.cz>
22788 PR tree-optimization/80293
22789 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
22790 char arrays not totally scalarizable if it is false.
22791 (analyze_all_variable_accesses): Pass correct value in the new
22792 parameter. Add a statistics counter.
22794 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
22796 PR middle-end/79931
22797 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
22799 2017-04-24 Richard Biener <rguenther@suse.de>
22801 PR tree-optimization/80494
22802 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
22803 out for complex types.
22805 2017-04-24 Richard Biener <rguenther@suse.de>
22807 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
22808 * tree-ssa-sccvn.c (print_scc): Print SCC size.
22809 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
22810 (DFS): Adjust and never fail.
22811 (sccvn_dom_walker::fail): Remove.
22812 (sccvn_dom_walker::before_dom_children): Adjust.
22813 (run_scc_vn): Likewise and never fail.
22814 * tree-ssa-pre.c (pass_pre::execute): Adjust.
22815 (pass_fre::execute): Likewise.
22817 2017-04-24 Richard Biener <rguenther@suse.de>
22819 PR tree-optimization/79725
22820 * tree-ssa-sink.c (statement_sink_location): Return whether
22821 failure reason was zero uses. Move that check later.
22822 (sink_code_in_bb): Deal with zero uses by removing the stmt
22825 2017-04-24 Richard Biener <rguenther@suse.de>
22828 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
22829 pointer-based references.
22831 2017-04-24 Richard Biener <rguenther@suse.de>
22834 * pass_manager.h (pass_manager::operator new): Remove.
22835 (pass_manager::operator delete): Likewise.
22836 * passes.c (pass_manager::operator new): Remove.
22837 (pass_manager::operator delete): Likewise.
22838 (pass_manager::pass_manager): Zero individual pass members.
22840 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
22843 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
22844 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
22845 Check "XEXP (src, 1)" operand here.
22846 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
22847 Check "XEXP (src, 1)" operand here.
22848 (dimode_scalar_chain::make_vector_copies): Detect count register
22849 of a shift instruction. Zero extend count register from QImode
22850 to DImode to satisfy vector shift pattern count operand predicate.
22851 Substitute vector shift count operand with a DImode copy.
22852 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
22855 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
22857 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
22858 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
22859 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
22860 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
22861 (UNSPEC_NOREX_MEM): Remove definition.
22863 2017-04-21 Richard Biener <rguenther@suse.de>
22865 PR tree-optimization/79547
22866 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
22867 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
22868 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
22869 without any constraints.
22871 2017-04-21 Richard Biener <rguenther@suse.de>
22873 PR tree-optimization/78847
22874 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
22876 2017-04-21 Richard Biener <rguenther@suse.de>
22878 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
22879 (build_distinct_type_copy): Likewise.
22880 (build_variant_type_copy): Likewise.
22881 * tree.c (build_qualified_type): Pass down mem-stat info.
22882 (build_distinct_type_copy): Likewise.
22883 (build_variant_type_copy): Likewise.
22885 2017-04-21 Richard Biener <rguenther@suse.de>
22887 PR tree-optimization/80237
22888 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
22890 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
22891 for a simplified result.
22893 2016-04-21 Richard Biener <rguenther@suse.de>
22895 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
22896 sth as strict as a simple_iv but a chrec without symbols and an
22897 operand defined in the loop we are peeling (and not some subloop).
22898 (propagate_constants_for_unrolling): Propagate all constants.
22900 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
22903 * config/i386/i386.c (print_reg): Remove assert for disalowed
22904 regno values, call output_operand_lossage instead.
22906 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
22909 * config/i386/constraints.md (Yc): New register constraint.
22910 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
22911 Use Yc constraint for alternative 2 of operand 0. Remove
22912 preferred_for_speed attribute.
22914 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
22916 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
22917 lastprivate clauses in SIMT case.
22919 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
22921 * doc/invoke.texi (-Wextra-semi): Document new warning option.
22923 2017-04-20 Richard Biener <rguenther@suse.de>
22925 PR tree-optimization/57796
22926 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
22927 as N scalar stores.
22928 (vect_model_load_cost): Cost gathers as N scalar loads.
22930 2017-04-20 Richard Biener <rguenther@suse.de>
22932 * ggc-page.c (ggc_allocated_p): Rename to ...
22933 (safe_lookup_page_table_entry): ... this and return the lookup
22935 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
22937 2017-04-20 Richard Biener <rguenther@suse.de>
22939 PR tree-optimization/80453
22940 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
22941 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
22942 from the conditions.
22943 (vn_phi_eq): Pass them down.
22944 (vn_phi_lookup): Record them.
22945 (vn_phi_insert): Likewise.
22947 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
22949 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
22950 uninitialized variable warning to avoid buffer overrun.
22952 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
22955 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
22956 is suppressed for '{ 0 }' in C.
22958 2017-04-20 Jakub Jelinek <jakub@redhat.com>
22960 * BASE-VER: Set to 8.0.0.
22962 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
22964 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
22965 priority .init_array and .fini_array section with SECTION_NOTYPE
22968 2017-04-20 Jakub Jelinek <jakub@redhat.com>
22970 PR middle-end/80423
22971 * tree.h (build_array_type): Add typeless_storage default argument.
22972 * tree.c (type_cache_hasher::equal): Also compare
22973 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
22974 (build_array_type): Add typeless_storage argument, set
22975 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
22977 (build_nonshared_array_type): Adjust build_array_type_1 caller.
22978 (build_array_type): Likewise. Add typeless_storage argument.
22980 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
22981 Jakub Jelinek <jakub@redhat.com>
22983 PR tree-optimization/80426
22984 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
22985 operation on symbolic operands, also compute the overflow for the
22986 invariant part when the operation degenerates into a negation.
22988 2017-04-19 Jakub Jelinek <jakub@redhat.com>
22991 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
22992 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
22995 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
22997 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
23000 * config/avr/avr.c (tree.h): Include it.
23001 (cgraph.h): Include it.
23002 (avr_encode_section_info): Don't warn for uninitialized progmem
23003 variable if it's just an alias.
23005 2017-04-19 Richard Biener <rguenther@suse.de>
23008 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
23009 when needed by AutoPGO.
23011 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
23014 * doc/lto.texi: Remove an extra 'that'.
23016 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
23018 PR rtl-optimization/80429
23019 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
23020 are only used in debug insns.
23022 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
23023 Vladimir Makarov <vmakarov@redhat.com>
23025 * config/sparc/predicates.md (input_operand): Add comment. Return
23026 true for any memory operand when LRA is in progress.
23027 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
23029 2017-04-18 Jeff Law <law@redhat.com>
23032 * mips.md ({return,simple_return}_internal): Do not overwrite
23035 2017-04-18 Jakub Jelinek <jakub@redhat.com>
23037 PR tree-optimization/80443
23038 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
23039 instead of adding 1, subtract -1 and similarly instead of subtracting
23042 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
23044 PR rtl-optimization/80357
23045 * haifa-sched.c (tmp_bitmap): New variable.
23046 (model_recompute): Handle duplicate use records.
23047 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
23048 (free_global_sched_pressure_data): Free it.
23050 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
23053 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
23054 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23055 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23056 instead of SYSTEM_HEADER_DIR.
23058 2017-04-18 Jeff Law <law@redhat.com>
23060 PR middle-end/80422
23061 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
23062 predecessors after walking up the insn chain.
23064 2017-04-18 Jakub Jelinek <jakub@redhat.com>
23067 * dwarf2out.c (modified_type_die): Try harder not to emit internal
23068 sizetype type into debug info.
23070 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23073 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
23074 unneeded test for TARGET_UPPER_REGS_SF.
23075 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
23077 2017-04-18 Jakub Jelinek <jakub@redhat.com>
23080 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
23081 instead of gsi_after_labels.
23083 2017-04-18 Jeff Law <law@redhat.com>
23085 * regcprop.c (maybe_mode_change): Avoid creating copies of the
23089 2017-04-13 Jeff Law <law@redhat.com>
23090 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
23091 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
23093 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
23096 * config/avr/avr.c (intl.h): Include it.
23097 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
23099 2017-04-18 Martin Liska <mliska@suse.cz>
23101 PR gcov-profile/78783
23102 * gcov-tool.c (gcov_output_files): Validate that destination
23103 file is either removed by the tool or by a user.
23105 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
23106 Guy Benyei <guybe@mellanox.com>
23108 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
23109 block, and do not negate it, the stored id is already negative.
23111 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
23113 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
23115 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
23118 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
23119 masks of options that should be turned off if the VSX vector
23120 options are turned off.
23121 (OTHER_P8_VECTOR_MASKS): Likewise.
23122 (OTHER_VSX_VECTOR_MASKS): Likewise.
23123 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
23124 rs6000_disable_incompatible_switches to validate no type switches
23126 (rs6000_incompatible_switch): New function to disallow turning on
23127 other vector options if -mno-vsx, -mno-power8-vector, or
23128 -mno-power9-vector are specified.
23130 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23132 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
23134 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23136 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
23137 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
23138 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
23139 (ARG_POINTER_CFA_OFFSET): Likewise.
23141 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23143 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
23144 conditions to take advantage of various optimizations.
23146 2017-04-13 Jeff Law <law@redhat.com>
23148 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
23149 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
23150 (zero_extendsidi2_dext): Likewise.
23152 2017-04-13 Jakub Jelinek <jakub@redhat.com>
23155 * fold-const.c (fold_ternary_loc): Revert
23156 use op0 instead of fold_convert_loc (loc, type, arg0) part of
23159 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
23161 PR rtl-optimization/80343
23162 * lra-remat.c (update_scratch_ops): Assign original hard reg to
23163 new scratch pseudo.
23165 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
23168 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
23169 to ubsan_encode_value.
23171 2017-04-13 Jeff Law <law@redhat.com>
23173 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
23174 appearing in DEBUG_INSNs.
23176 2017-04-13 Martin Liska <mliska@suse.cz>
23178 PR gcov-profile/80413
23179 * gcov-io.c (gcov_write_string): Copy to buffer just when
23180 allocated size is greater than zero.
23182 2017-04-13 Jakub Jelinek <jakub@redhat.com>
23185 * dwarf2out.c (decls_for_scope): Ignore declarations of
23186 current_function_decl in BLOCK_NONLOCALIZED_VARS.
23188 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
23191 * ipa-visibility.c (non_local_p): Fix typos.
23192 (localize_node): When localizing symbol in same comdat group,
23193 dissolve the group only when we know external symbols are going
23195 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
23197 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23199 PR tree-optimization/79390
23200 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
23201 order does not result in usable sequence, retry with reversed operand
23207 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
23208 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
23209 op0 instead of fold_convert_loc (loc, type, arg0).
23211 2017-04-12 Jeff Law <law@redhat.com>
23213 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
23214 has a delay slot in the generated code.
23216 * config/cris/cris.md (cris_preferred_reload_class): Return
23217 GENNONACR_REGS rather than GENERAL_REGS.
23219 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23222 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
23223 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
23224 signedness of the result type.
23226 2017-04-12 Richard Biener <rguenther@suse.de>
23227 Jeff Law <law@redhat.com>
23229 PR tree-optimization/80359
23230 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
23231 trim stores to TARGET_MEM_REFs.
23233 2017-04-12 Richard Biener <rguenther@suse.de>
23235 PR tree-optimization/79390
23236 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
23237 threading case even more.
23239 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
23242 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
23243 for quad_address_p for TImode, instead of just not indexed_address.
23245 2017-04-12 Richard Biener <rguenther@suse.de>
23246 Bernd Edlinger <bernd.edlinger@hotmail.de>
23248 PR middle-end/79671
23249 * alias.c (component_uses_parent_alias_set_from): Handle
23250 TYPE_TYPELESS_STORAGE.
23251 (get_alias_set): Likewise.
23252 * tree-core.h (tree_type_common): Add typeless_storage flag.
23253 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
23254 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
23255 for types containing members with TYPE_TYPELESS_STORAGE.
23256 (place_field): Likewise.
23257 (layout_type): Likewise for ARRAY_TYPE.
23258 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
23259 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
23260 TYPE_TYPELESS_STORAGE.
23261 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
23263 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23266 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
23267 first argument to type.
23269 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23273 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
23274 CONST0_RTX (mode) rather than const0_rtx where appropriate.
23275 (rs6000_expand_binop_builtin): Likewise.
23276 (rs6000_expand_ternop_builtin): Likewise; also add missing
23277 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
23278 vshasigma built-ins.
23279 * doc/extend.texi: Document that vec_xxpermdi's third argument
23280 must be a constant.
23282 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
23284 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
23285 Use shift_const cost parameter when calculating gain of STV shifts.
23287 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
23289 PR rtl-optimization/70478
23290 * lra-constraints.c (process_alt_operands): Check memory for
23291 disfavoring memory insn operand.
23293 2017-04-11 Jakub Jelinek <jakub@redhat.com>
23295 PR middle-end/80100
23296 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
23297 left shift in unsigned HOST_WIDE_INT type.
23299 PR rtl-optimization/80385
23300 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
23301 (not (neg X)) into (plus X -1) for complex or non-integral modes.
23304 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
23305 if they have any depend clauses.
23307 2017-04-11 Martin Liska <mliska@suse.cz>
23310 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
23311 * ipa-split.c (split_function): Create a local comdat symbol
23312 if caller is in a comdat group.
23314 2017-04-11 Martin Liska <mliska@suse.cz>
23317 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
23320 2017-04-11 Martin Sebor <msebor@redhat.com>
23322 PR middle-end/80364
23323 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
23324 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
23326 (directive::set_width, directive::set_precision, format_character):
23328 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
23331 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
23334 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
23335 conflict, set target->arch_name instead of target->cpu_name.
23337 2017-04-11 Richard Biener <rguenther@suse.de>
23339 PR tree-optimization/80374
23340 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
23341 build_zero_cst, remove fold_convertible_p check again.
23343 2017-04-11 Martin Liska <mliska@suse.cz>
23346 * ubsan.c (instrument_object_size): Do not instrument register
23349 2017-04-11 Jakub Jelinek <jakub@redhat.com>
23352 * config/i386/i386-builtin-types.def
23353 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
23354 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
23355 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
23356 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
23357 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
23358 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
23359 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
23360 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
23361 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
23362 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
23363 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
23364 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
23365 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
23366 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
23367 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
23368 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
23369 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
23370 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
23371 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
23372 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
23373 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
23374 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
23375 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
23376 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
23377 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
23378 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
23379 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
23380 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
23381 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
23382 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
23383 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
23384 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
23385 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
23386 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
23387 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
23388 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
23389 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
23390 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
23391 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
23392 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
23393 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
23394 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
23395 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
23396 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
23397 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
23398 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
23400 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
23401 flag to second_arg_count, handle 4 argument function type _COUNT
23402 aliases, handle second_arg_count on second argument rather than last.
23404 2017-04-10 Jeff Law <law@redhat.com>
23406 PR tree-optimization/80374
23407 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
23408 record anything if we can not convert integer_zero_node to the
23411 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
23414 * config/rs6000/rs6000.c (rs6000_option_override_internal):
23415 Enhance special handling given to the TARGET_P9_MINMAX option in
23416 relation to certain other options.
23418 2017-04-10 Bin Cheng <bin.cheng@arm.com>
23420 PR tree-optimization/80153
23421 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
23422 remove POINTER_PLUS_EXPR's base part directly, rather than through
23425 2017-04-10 Richard Biener <rguenther@suse.de>
23426 Bin Cheng <bin.cheng@arm.com>
23428 PR tree-optimization/80153
23429 * tree-affine.c (aff_combination_to_tree): Get base pointer from
23430 the first element of pointer type aff_tree. Build result expr in
23432 (add_elt_to_tree): Convert to type unconditionally. Remove other
23433 fold_convert calls.
23434 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
23435 (rewrite_use_nonlinear_expr): Check invariant using iv information.
23437 2017-04-10 Richard Biener <rguenther@suse.de>
23439 * tree-ssa-structalias.c (find_func_aliases): Properly handle
23442 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
23444 PR rtl-optimization/70478
23445 * lra-constraints.c (curr_small_class_check): New.
23446 (update_and_check_small_class_inputs): New.
23447 (process_alt_operands): Update curr_small_class_check. Disfavor
23448 alternative insn memory operands. Check available regs for small
23451 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
23454 * config/mips/mips.opt (-mvirt): Update description.
23455 * doc/invoke.texi (-mvirt): Likewise.
23457 2017-04-10 Richard Biener <rguenther@suse.de>
23459 PR middle-end/80362
23460 * fold-const.c (fold_binary_loc): Look at unstripped ops when
23461 looking for NEGATE_EXPR in -A / -B to A / B folding.
23463 2017-04-10 Martin Liska <mliska@suse.cz>
23465 PR gcov-profile/80224
23466 * gcov.c (print_usage): Fix usage string.
23467 (get_gcov_intermediate_filename): Remove.
23468 (output_gcov_file): Use both for normal and intermediate format.
23469 (generate_results): Do not initialize special file for
23470 intermediate format.
23472 2017-04-10 Richard Biener <rguenther@suse.de>
23474 PR tree-optimization/80304
23475 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
23478 2017-04-10 Nathan Sidwell <nathan@acm.org>
23481 * config/rs6000/rs6000.c (rs6000_vector_type): New.
23482 (rs6000_init_builtins): Use it.
23484 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23486 * config/arm/arm.md (<mrc>): Add mode to SET source.
23487 (<mrrc>): Likewise.
23489 2017-04-10 Richard Biener <rguenther@suse.de>
23491 PR middle-end/80344
23492 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
23494 2017-04-10 Jakub Jelinek <jakub@redhat.com>
23497 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
23498 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
23499 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
23500 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
23501 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
23502 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
23503 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
23504 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
23505 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
23506 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
23507 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
23508 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
23509 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
23510 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
23511 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
23512 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
23513 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
23514 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
23515 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
23516 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
23517 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
23518 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
23519 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
23521 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
23523 PR rtl-optimization/70478
23524 * lra-constraints.c: Reverse the last patch.
23526 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
23528 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
23529 Add comment for WCHAR_T.
23531 2017-04-08 Martin Liska <mliska@suse.cz>
23534 2017-04-07 Martin Liska <mliska@suse.cz>
23537 * ipa-split.c (split_function): Add function part to a same comdat
23540 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23543 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
23545 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
23547 * rs6000/rs6000.c (vec_load_pendulum): Rename...
23548 (vec_pairing): ...to this.
23549 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
23550 (rs6000_sched_init): Adjust for name change.
23551 (struct rs6000_sched_context): Likewise.
23552 (rs6000_init_sched_context): Likewise.
23553 (rs6000_set_sched_context): Likewise.
23555 2017-04-07 Jakub Jelinek <jakub@redhat.com>
23561 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
23563 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
23564 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
23565 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
23567 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
23569 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
23571 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
23573 PR rtl-optimization/70703
23574 * ira-color.c (update_conflict_hard_regno_costs): Use
23575 int64_t instead of HOST_WIDE_INT.
23577 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
23579 PR rtl-optimization/70478
23580 * lra-constraints.c (process_alt_operands): Disfavor alternative
23581 insn memory operands.
23583 2017-04-07 Jeff Law <law@redhat.com>
23585 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
23586 CALL and NOTE_INSN_CALL_ARG_LOCATION.
23588 2017-04-07 Martin Liska <mliska@suse.cz>
23591 * config/aarch64/aarch64.c (aarch64_process_target_attr):
23592 Show error message instead of an ICE.
23594 2017-04-07 Martin Liska <mliska@suse.cz>
23597 * ipa-split.c (split_function): Add function part to a same comdat
23600 2017-04-07 Richard Biener <rguenther@suse.de>
23602 PR middle-end/80341
23603 * tree.c (get_unwidened): Also handle ! for_type case for
23605 * convert.c (do_narrow): Split out from ...
23606 (convert_to_integer_1): ... here. Do not pass final truncation
23607 type to get_unwidened for TRUNC_DIV_EXPR.
23609 2017-04-07 Richard Biener <rguenther@suse.de>
23611 * tree-affine.c (wide_int_ext_for_comb): Take type rather
23613 (aff_combination_const): Adjust.
23614 (aff_combination_scale): Likewise.
23615 (aff_combination_add_elt): Likewise.
23616 (aff_combination_add_cst): Likewise.
23617 (aff_combination_convert): Likewise.
23618 (add_elt_to_tree): Likewise. Remove unused argument.
23619 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
23621 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
23623 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
23625 * config/arm/arm.c (arm_default_short_enums): Use
23626 ARM_DEFAULT_SHORT_ENUMS.
23627 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
23629 2017-04-06 Jakub Jelinek <jakub@redhat.com>
23632 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
23633 members with redundant out-of-class redeclaration.
23635 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
23638 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
23639 * config/i386/i386.md (*zero_extendsidi2):
23640 Add (?*x,*x) and (?*v,*v) alternatives.
23642 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
23645 * config/i386/i386.c (ix86_expand_builtin)
23646 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
23647 mode from insn data. Convert operands to insn operand mode.
23648 Copy operands that don't satisfy insn predicate to a register.
23650 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
23652 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
23655 2017-04-06 Richard Biener <rguenther@suse.de>
23657 PR tree-optimization/80334
23658 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
23659 preserve alignment of accesses.
23661 2017-04-06 Richard Biener <rguenther@suse.de>
23663 PR tree-optimization/80262
23664 * tree-sra.c (build_ref_for_offset): Preserve address-space
23666 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
23667 Drop useless address-space information on MEM_REF offsets.
23669 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
23671 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
23673 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
23675 PR rtl-optimization/70703
23676 * ira-color.c (update_conflict_hard_regno_costs): Use
23677 HOST_WIDE_INT instead of long.
23679 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
23682 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
23683 not defined for x86_64 target. Add -mmmx target option when __SSE2__
23685 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
23686 for x86_64 target. Handle -m3dnowa option.
23688 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
23690 PR rtl-optimization/70703
23691 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
23692 (update_conflict_hard_regno_costs): Use long instead of unsigned
23693 arithmetic for cost calculation.
23695 2017-04-05 Jakub Jelinek <jakub@redhat.com>
23696 Bernd Edlinger <bernd.edlinger@hotmail.de>
23699 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
23702 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
23705 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
23706 ptr_mode with Pmode throughout.
23707 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
23708 into probe_stack_range and use DImode.
23710 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
23713 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
23714 call_eh_return is true.
23716 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23718 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
23719 Initialize last_match_fntype_index.
23721 2017-04-05 Jakub Jelinek <jakub@redhat.com>
23724 * tree-nvr.c: Include internal-fn.h.
23725 (pass_return_slot::execute): Ignore internal calls without
23728 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23729 Richard Biener <rguenther@suse.de>
23732 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
23733 captures used multiple times, except for the last use.
23734 * generic-match-head.c: Include gimplify.h.
23736 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23738 PR tree-optimization/79390
23739 * target.h (struct noce_if_info): Declare.
23740 * targhooks.h (default_noce_conversion_profitable_p): Declare.
23741 * target.def (noce_conversion_profitable_p): New target hook.
23742 * ifcvt.h (struct noce_if_info): New type, moved from ...
23743 * ifcvt.c (struct noce_if_info): ... here.
23744 (noce_conversion_profitable_p): Renamed to ...
23745 (default_noce_conversion_profitable_p): ... this. No longer
23747 (noce_try_store_flag_constants, noce_try_addcc,
23748 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
23749 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
23750 instead of noce_conversion_profitable_p.
23751 * config/i386/i386.c: Include ifcvt.h.
23752 (ix86_option_override_internal): Don't override
23753 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
23754 (ix86_noce_conversion_profitable_p): New function.
23755 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
23756 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
23757 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
23758 * doc/tm.texi: Regenerated.
23760 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23762 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
23765 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23768 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
23769 instructions for small multiply cores.
23771 2017-04-04 Jeff Law <law@redhat.com>
23773 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
23775 (mips_expand_vec_perm_const): Initialize elements in orig_perm
23776 that are not set by the loop over the elements.
23778 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23781 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
23782 int mode, convert_modes it to mode as unsigned, otherwise use
23783 lowpart_subreg to mode rather than SImode.
23784 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
23785 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
23786 Use DImode instead of SImode for the shift count operand.
23787 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
23790 2017-04-04 Richard Biener <rguenther@suse.de>
23792 PR middle-end/80281
23793 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
23794 arithmetic done for the negate or the plus. Simplify.
23795 (A - (-B) -> A + B): Likewise.
23796 * fold-const.c (split_tree): Make sure to not negate pointers.
23798 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
23800 PR rtl-optimization/60818
23801 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
23802 a compare of comparisons with the thing compared if this results
23803 in a different machine mode.
23805 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
23807 * alias.c (base_alias_check): Fix typo in comment.
23808 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
23809 * cgraphunit.c (symbol_table::compile): Likewise.
23810 * collect2.c (maybe_run_lto_and_relink): Likewise.
23811 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
23812 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
23813 * config/avr/avr.c (avr_map_op_t): Likewise.
23814 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
23815 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
23816 * config/epiphany/epiphany.md (movcc): Likewise.
23817 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
23818 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
23820 * config/mips/mips.c (mips_save_restore_reg): Likewise.
23821 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
23822 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
23823 * config/sh/sh.c (sh_rtx_costs): Likewise.
23824 * fold-const.c (fold_truth_andor): Likewise.
23825 * genautomata.c (collapse_flag): Likewise.
23826 * gengtype.h (struct type::u::s): Likewise.
23827 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
23828 * input.c (FORMAT_AMOUNT): Likewise.
23829 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
23830 (known_aggs_to_agg_replacement_list): Likewise.
23831 * ipa-inline-analysis.c: Likewise.
23832 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
23833 * ipa-polymorphic-call.c
23834 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
23835 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
23836 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
23838 * modulo-sched.c (apply_reg_moves): Likewise.
23839 * omp-expand.c (build_omp_regions_1): Likewise.
23840 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
23841 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
23842 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
23843 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
23844 * value-prof.c: Likewise.
23845 * var-tracking.c (val_reset): Likewise.
23847 2017-04-03 Richard Biener <rguenther@suse.de>
23849 PR tree-optimization/80275
23850 * fold-const.c (split_address_to_core_and_offset): Handle
23853 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
23855 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
23856 descriptors is at least equal to that of functions.
23858 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
23860 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
23862 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
23865 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
23866 (mov<IMOD4:mode>): New expander.
23867 (*mov<IMOD4:mode>_internal): New insn and split pattern.
23869 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
23871 PR rtl-optimization/79405
23872 * fwprop.c (propagations_left): New variable.
23873 (forward_propagate_into): Decrement it.
23874 (fwprop_init): Initialize it.
23875 (fw_prop): If the variable has reached zero, stop propagating.
23876 (fwprop_addr): Ditto.
23878 2017-03-31 Jakub Jelinek <jakub@redhat.com>
23881 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
23882 a FUNCTION_DECL, pass it as decl instead of origin to
23885 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
23887 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
23890 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
23893 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
23894 TARGET_VSX_SMALL_INTEGER.
23896 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23898 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23899 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
23901 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
23903 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
23904 extraction from odd-numbered MSA register.
23906 2017-03-31 Jakub Jelinek <jakub@redhat.com>
23908 PR middle-end/80173
23909 * expmed.c (store_bit_field_1): Don't attempt to create
23910 a word subreg out of hard registers wider than word if they
23911 have HARD_REGNO_NREGS of 1 for their mode.
23913 PR middle-end/80163
23914 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
23915 conversions to integer types wider than word and pointer.
23918 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
23919 (rtx_equal_for_cselib_p): Pass 0 to it.
23920 * cselib.c (cselib_hasher::equal): Likewise.
23921 (rtx_equal_for_cselib_1): Add depth argument. If depth
23922 is 128, don't look up VALUE locs and punt. Increment
23923 depth in recursive calls when walking VALUE locs.
23925 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
23927 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
23928 (make_gcov_file_name): Use the canonical path name for generating
23930 (read_line): Fix handling of files with ascii null bytes.
23932 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
23934 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
23935 to initialise a vector register instead
23936 of using a const_int.
23938 2017-03-30 Jakub Jelinek <jakub@redhat.com>
23940 PR translation/80189
23941 * gimplify.c (omp_default_clause): Use %qs instead of %s in
23942 diagnostic messages.
23944 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
23947 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
23948 (dfp_diex_<mode>): Update mode of operand 1.
23949 * doc/extend.texi (dxex, dxexq): Document change to return type.
23950 (diex, diexq): Document change to argument type.
23952 2017-03-30 Martin Jambor <mjambor@suse.cz>
23955 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
23956 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
23957 it reflects the signature changes performed at the callee side.
23958 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
23959 to cgraph_build_function_type_skip_args.
23960 (build_function_decl_skip_args): Adjust call to the above function.
23962 2017-03-30 Jakub Jelinek <jakub@redhat.com>
23965 * config/i386/sse.md
23966 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
23967 register as dest whenever it is a MEM not rtx_equal_p to the
23968 corresponding dup operand, and when forcing into reg move the
23969 reg into the memory afterwards.
23970 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
23971 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
23972 for the force_reg mode.
23973 (avx512vl_vextractf128<mode>): Use register as dest either
23974 always when a MEM, or when it is a MEM not rtx_equal_p to the
23975 corresponding dup operand, or even not when it is a CONST_VECTOR
23976 depending on the mode and lo vs. hi.
23977 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
23979 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
23980 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
23981 Likewise. Require that operands[2] is even.
23982 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
23983 Remove extraneous parens. Require that operands[2] is a multiple
23985 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
23986 operands[0] is a MEM if <mask_applied>, the predicates/constraints
23987 disallow memory then.
23989 2017-03-30 Richard Biener <rguenther@suse.de>
23991 PR tree-optimization/77498
23992 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
23993 to non-constants over backedges.
23995 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
23997 PR rtl-optimization/80233
23998 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
23999 as last_combined_insn. Do not test for BARRIER_P separately.
24001 2017-03-29 Andreas Schwab <schwab@suse.de>
24004 * calls.c (prepare_call_address): Convert funexp to Pmode before
24005 copying to temp reg.
24007 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24009 PR tree-optimization/80158
24010 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
24011 Handle possible future case of more than one alternate
24013 (replace_rhs_if_not_dup): Likewise.
24014 (replace_one_candidate): Likewise.
24016 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
24018 PR rtl-optimization/80193
24019 * ira.c (ira): Do not check allocation for LRA.
24021 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
24023 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
24024 (nvptx_output_simt_exit): Declare.
24025 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
24026 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
24027 (init_softstack_frame): Move initialization of crtl->is_leaf to...
24028 (nvptx_declare_function_name): ...here. Emit declaration of local
24029 memory space buffer for omp_simt_enter insn.
24030 (nvptx_output_unisimt_switch): New.
24031 (nvptx_output_softstack_switch): New.
24032 (nvptx_output_simt_enter): New.
24033 (nvptx_output_simt_exit): New.
24034 * config/nvptx/nvptx.h (struct machine_function): New fields
24035 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
24036 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
24037 (UNSPECV_SIMT_EXIT): Ditto.
24038 (omp_simt_enter_insn): New insn.
24039 (omp_simt_enter): New expansion.
24040 (omp_simt_exit): New insn.
24041 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
24043 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
24044 (expand_GOMP_SIMT_ENTER_ALLOC): New.
24045 (expand_GOMP_SIMT_EXIT): New.
24046 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
24047 (GOMP_SIMT_ENTER_ALLOC): Ditto.
24048 (GOMP_SIMT_EXIT): Ditto.
24049 * target-insns.def (omp_simt_enter): New insn.
24050 (omp_simt_exit): Ditto.
24051 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
24053 (lower_rec_simd_input_clauses): Implement SIMT privatization.
24054 (lower_rec_input_clauses): Likewise.
24055 (lower_lastprivate_clauses): Handle SIMT privatization.
24057 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
24058 (ompdevlow_adjust_simt_enter): New.
24059 (find_simtpriv_var_op): New.
24060 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
24061 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
24063 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
24064 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
24065 (copy_decl_for_dup_finish): Ditto.
24067 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
24069 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
24072 * config/i386/i386.c (ix86_option_override_internal): Always
24073 allow -mpreferred-stack-boundary=3 for 64-bit targets.
24075 2017-03-28 Bin Cheng <bin.cheng@arm.com>
24077 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
24079 2017-03-28 Bin Cheng <bin.cheng@arm.com>
24081 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
24082 mark new edge's irreducible flag accordign to it.
24083 (vect_do_peeling): Check loop preheader edge's irreducible flag
24084 and pass it to function slpeel_add_loop_guard.
24086 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
24088 PR tree-optimization/80218
24089 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
24090 Update block frequencies and counts.
24092 2017-03-28 Richard Biener <rguenther@suse.de>
24094 PR tree-optimization/78644
24095 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
24096 of a simplification result we may not use it at all.
24098 2017-03-28 Richard Biener <rguenther@suse.de>
24101 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
24102 without arguments, generate default definition of a SSA name.
24104 2017-03-28 Richard Biener <rguenther@suse.de>
24106 PR middle-end/80222
24107 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
24108 TYPE_REF_CAN_ALIAS_ALL references.
24109 * fold-const.c (fold_indirect_ref_1): Likewise.
24111 2017-03-28 Martin Liska <mliska@suse.cz>
24114 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
24115 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
24117 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
24118 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
24120 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
24121 (EXTRA_SPECS): Define.
24122 (SUBTARGET_EXTRA_SPECS): Likewise.
24123 (SUBTARGET_CPP_SPEC): Likewise.
24124 * config/arc/elf.h (EXTRA_SPECS): Renamed to
24125 SUBTARGET_EXTRA_SPECS.
24126 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
24128 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
24130 * config/arc/simdext.md (vst64_insn): Update pattern.
24131 (vld32wh_insn): Likewise.
24132 (vld32wl_insn): Likewise.
24133 (vld64_insn): Likewise.
24134 (vld32_insn): Likewise.
24136 2017-03-28 Marek Polacek <polacek@redhat.com>
24139 * fold-const.c (fold_comparison): Use protected_set_expr_location
24140 instead of SET_EXPR_LOCATION.
24142 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
24144 * tree.c (add_expr): Avoid name lookup warning.
24146 2017-03-27 Jeff Law <law@redhat.com>
24148 PR tree-optimization/80216
24149 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
24150 function name. Limit recursion depth.
24151 (record_temporary_equivalences): Corresponding changes.
24153 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
24155 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
24158 2017-03-27 Jakub Jelinek <jakub@redhat.com>
24161 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
24163 * cfgcleanup.c (reg_note_cfa_p): New array.
24164 (insns_have_identical_cfa_notes): New function.
24165 (old_insns_match_p): Don't cross-jump in between /f
24166 and non-/f instructions. If both i1 and i2 are frame related,
24167 verify all CFA notes, their order and content.
24169 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
24172 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
24173 HImode and SImode with zero extend to DImode to one insn.
24174 (bswap<mode>2_extenddi): Likewise.
24175 (bswapsi2_extenddi): Likewise.
24176 (bswaphi2_extendsi): Likewise.
24177 (bswaphi2): Combine bswap HImode and SImode into one insn.
24178 Separate memory insns from swapping register.
24179 (bswapsi2): Likewise.
24180 (bswap<mode>2): Likewise.
24181 (bswaphi2_internal): Delete, no longer used.
24182 (bswapsi2_internal): Likewise.
24183 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
24184 store, and gpr<-gpr swap insns.
24185 (bswap<mode>2_store): Likewise.
24186 (bswaphi2_reg): Register only splitter, combine with the splitter.
24187 (bswaphi2 splitter): Likewise.
24188 (bswapsi2_reg): Likewise.
24189 (bswapsi2 splitter): Likewise.
24190 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
24191 the insns into load, store, and register/register insns.
24192 (bswapdi2_ldbrx): Likewise.
24193 (bswapdi2_load): Likewise.
24194 (bswapdi2_store): Likewise.
24195 (bswapdi2_reg): Likewise.
24197 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
24199 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
24200 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
24202 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
24205 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
24207 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24208 special handling for target option conflicts between dform
24209 options (-mpower9-dform, -mpower9-dform-vector,
24210 -mpower9-dform-scalar) and -mno-direct-move.
24212 2017-03-27 Richard Biener <rguenther@suse.de>
24214 PR tree-optimization/80181
24215 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
24217 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24219 * config/arc/predicates.md (move_double_src_operand): Replace the
24220 call to move_double_src_operand with a call to address_operand.
24222 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24224 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
24225 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
24226 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
24228 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24230 * config/arc/predicates.md (long_immediate_loadstore_operand):
24231 Consider scaled addresses cases.
24233 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24235 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
24236 restored when in interrupt.
24237 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
24238 doesn't have delay slot.
24240 2017-03-27 Richard Biener <rguenther@suse.de>
24243 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
24244 inlined thunk clones.
24246 2017-03-27 Jakub Jelinek <jakub@redhat.com>
24249 * asan.c (instrument_derefs): Copy over last operand from
24250 original COMPONENT_REF to the new COMPONENT_REF with
24251 DECL_BIT_FIELD_REPRESENTATIVE.
24252 * ubsan.c (instrument_object_size): Likewise.
24254 2017-03-27 Richard Biener <rguenther@suse.de>
24256 PR tree-optimization/80170
24257 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
24258 sure DR/SCEV didnt fold in constants we do not see when looking
24259 at the reference base alignment.
24261 2017-03-27 Richard Biener <rguenther@suse.de>
24263 PR middle-end/80171
24264 * gimple-fold.c (fold_ctor_reference): Properly guard against
24265 NULL return value from canonicalize_constructor_val.
24267 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
24270 * config/i386/i386.c (ix86_expand_builtin)
24271 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
24272 flags reg setting and flags reg using instructions.
24273 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
24274 clobbering instructions to zero extend op2.
24276 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
24278 * doc/install.texi (Configuration) <--with-aix-soname>:
24279 Update link to AIX ld.
24281 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
24283 PR rtl-optimization/80160
24284 PR rtl-optimization/80159
24285 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
24286 reg_alternate_class into account.
24288 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
24291 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
24292 to consider in curr_insn_transform.
24294 2017-03-24 Jakub Jelinek <jakub@redhat.com>
24296 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
24297 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
24298 and emit_mode_inner.
24300 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24302 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
24303 argument to the overloaded builtin variants. Use the new flag to
24304 deprecate certain builtin variants.
24305 * config/s390/s390-builtin-types.def: Add new builtin types.
24306 * config/s390/s390-builtins.h: Support new flags field for
24307 overloaded builtins.
24308 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
24309 (s390_macro_to_expand): Enable vector float data type.
24310 (s390_cpu_cpp_builtins_internal): Indicate support of the new
24311 builtins by incrementing the __VEC__ version number.
24312 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
24314 (s390_resolve_overloaded_builtin): Emit error messages depending
24315 on the builtin flags.
24316 * config/s390/s390.c (s390_expand_builtin): Support additional
24317 flags argument. Change error message to match the messages
24318 emitted in s390-c.c.
24319 * config/s390/s390.md: New UNSPEC_* constants.
24320 (op_type): Add new instruction types.
24321 * config/s390/vecintrin.h: Add new builtins and test data class
24323 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
24324 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
24325 (VEC_INEXACT, VEC_NOINEXACT): New constants.
24326 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
24327 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
24328 ("vec_mergel<mode>"): V_HW -> VEC_HW.
24330 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
24331 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
24332 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
24333 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
24335 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
24336 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
24337 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
24338 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
24340 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
24341 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
24342 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
24343 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
24344 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
24345 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
24346 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
24348 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
24349 ("vec_scatter_element<V_HW_4:mode>_DI")
24350 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
24351 ("vec_fpint<mode>", "vflls")
24352 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
24353 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
24354 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
24355 ("*vec_cmphe<mode>_cc"): ... these.
24357 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
24358 mode constant instead of magic value.
24360 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24362 * config/s390/s390.c (s390_expand_vec_compare): Support other
24363 vector floating point modes than just V2DF.
24364 (s390_expand_vcond): Likewise.
24365 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
24366 (s390_cannot_change_mode_class): Prevent mode changes between TF
24367 and V1TF in vector registers.
24368 * config/s390/s390.md (DF, SF): New mode attributes.
24369 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
24370 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
24371 SFmode support for VRs.
24372 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
24374 (VFT, VF_HW): New mode iterators.
24375 (vw, sdx): New mode attributes.
24376 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
24377 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
24378 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
24379 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
24380 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
24381 also the new vector floating point modes. Renaming to ...
24383 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
24384 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
24385 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
24386 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
24387 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
24388 ("vec_unordered<mode>"): ... these.
24390 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
24391 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
24392 ("*vec_extendv2df"): New insn definitions.
24394 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24396 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
24397 ("mulditi3_2", "*muldi3_sign"): New patterns.
24398 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
24399 rename the pattern definition.
24401 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24403 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
24405 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
24407 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24409 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
24410 instruction if possible.
24411 * config/s390/vector.md (vec_halfnumelts): New mode
24413 ("*vec_vllezlf<mode>"): New pattern.
24415 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24417 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
24418 ("popcountv4si2", "popcountv2di2"): Rename to ...
24419 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
24420 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
24422 ("popcount<mode>2_vxe"): New pattern.
24424 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24426 * common/config/s390/s390-common.c (processor_flags_table): Add
24428 * config.gcc: Add arch12.
24429 * config/s390/driver-native.c (s390_host_detect_local_cpu):
24430 Default to arch12 for unknown CPU model numbers.
24431 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
24432 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
24433 PROCESSOR_max sanity check.
24434 * config/s390/s390-opts.h (enum processor_type): Add
24436 * config/s390/s390.c (processor_table): Add arch12.
24437 (s390_expand_builtin): Add check for B_VXE flag.
24438 (s390_issue_rate): Add PROCESSOR_ARCH12.
24439 (s390_get_sched_attrmask): Likewise.
24440 (s390_get_unit_mask): Likewise.
24441 (s390_sched_score): Enable z13 scheduling for arch12.
24442 (s390_sched_reorder): Likewise.
24443 (s390_sched_variable_issue): Likewise.
24444 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
24446 (s390_tune_attr): Use z13 scheduling also for arch12.
24447 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
24448 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
24449 (TARGET_VXE_P): New macros.
24450 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
24451 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
24452 * config/s390/s390.opt: Add arch12 as processor_type.
24454 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24456 * config/s390/s390.md
24457 ("fixuns_truncdddi2", "fixuns_trunctddi2")
24458 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
24459 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
24461 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
24462 Rename expanders to ...
24464 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
24465 ("fixuns_truncdddi2_emu"): ... these.
24467 ("fixuns_trunc<mode>si2_emu"): New expander.
24469 ("*fixuns_truncdfdi2_z13"): Rename to ...
24470 ("*fixuns_truncdfdi2_vx"): ... this.
24472 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24474 * config/s390/2964.md: Remove the single element vector compare
24475 instructions which are no longer used.
24476 * config/s390/s390.c (s390_select_ccmode): Remove handling of
24478 (s390_canonicalize_comparison): Remove handling of DFmode
24480 (s390_expand_vec_compare_scalar): Remove function.
24481 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
24482 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
24484 ("*cmp<mode>_ccs"): Add wfcdb instruction.
24486 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24488 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
24490 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
24491 will anyway by matched by mov<mode>_64dfp.
24493 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24495 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
24496 vlef/vstef. Add missing operand to vleif.
24498 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24500 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
24501 pair for all vector types with 64 bit elements.
24502 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
24503 * config/s390/vector.md (V_HW_64): ... here.
24504 (V_128_NOSINGLE): New mode iterator.
24505 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
24506 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
24507 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
24508 ("*vec_load_pairv2di"): Change to ...
24509 ("*vec_load_pair<mode>"): ... this one.
24511 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24513 * config/s390/constraints.md: Add comments.
24514 (jKK): Reject element sizes > 8 bytes.
24515 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
24517 * config/s390/s390.md: Add the s_operand checks formerly in
24518 s390_split_ok_p to various splitters where they are still
24520 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
24521 for 128 bit vectors. Plus two splitters.
24523 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24525 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
24528 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24531 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
24532 error if the boundary argument is not constant.
24534 2017-03-24 Jakub Jelinek <jakub@redhat.com>
24536 PR rtl-optimization/80112
24537 * loop-doloop.c (doloop_condition_get): Don't check condition
24538 if cmp isn't SET with IF_THEN_ELSE src.
24540 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24542 PR tree-optimization/80158
24543 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
24544 replacing a candidate statement, also replace it for the
24545 candidate's alternate interpretation.
24546 (replace_rhs_if_not_dup): Likewise.
24547 (replace_one_candidate): Likewise.
24549 2017-03-24 Richard Biener <rguenther@suse.de>
24551 PR tree-optimization/80167
24552 * graphite-isl-ast-to-gimple.c
24553 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
24555 (translate_isl_ast_to_gimple::get_rename): Likewise.
24557 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
24559 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
24560 handling of certain combinations of target options, including the
24561 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
24562 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
24564 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24567 * config/arm/arm.md (*load_multiple): Add reload_completed to
24568 matching condition.
24570 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24571 Richard Biener <rguenth@suse.de>
24573 PR tree-optimization/79908
24574 PR tree-optimization/80136
24575 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
24576 been cast away, gimplify_and_add suffices.
24578 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
24580 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
24582 2017-03-23 Richard Biener <rguenther@suse.de>
24584 PR tree-optimization/80032
24585 * gimplify.c (gimple_push_cleanup): Forced unconditional
24586 cleanups still have to go to the conditional_cleanups
24589 2017-03-22 Jakub Jelinek <jakub@redhat.com>
24591 PR tree-optimization/80072
24592 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
24594 (next_operand_entry_id): Change type to unsigned int.
24595 (sort_by_operand_rank): Make sure to return the right return value
24596 even if unsigned fields are bigger than INT_MAX.
24597 (struct oecount): Change cnt and id type to unsigned int.
24598 (oecount_hasher::equal): Formatting fix.
24599 (oecount_cmp): Make sure to return the right return value
24600 even if unsigned fields are bigger than INT_MAX.
24601 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
24604 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
24605 TREE_READONLY on result if writing it more than once.
24608 * doc/invoke.texi (-fsanitize=thread): Document that with
24609 -fnon-call-exceptions atomics are not able to throw
24613 * tsan.c: Include tree-eh.h.
24614 (instrument_builtin_call): Call maybe_clean_eh_stmt or
24615 maybe_clean_or_replace_eh_stmt where needed.
24616 (instrument_memory_accesses): Add cfg_changed argument.
24617 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
24618 if it returned true.
24619 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
24621 PR rtl-optimization/63191
24622 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
24623 wrapper function, moved the whole old content into ...
24624 (ix86_delegitimize_address_1): ... this. New inline function.
24625 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
24626 true as last argument instead of ix86_delegitimize_address.
24628 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
24630 * config/aarch64/aarch64.c (generic_branch_cost): Copy
24631 cortexa57_branch_cost.
24633 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
24635 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
24637 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24640 * doc/md.texi (Constraints): Document wA constraint.
24641 * config/rs6000/constraints.md (wA): New.
24642 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
24643 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
24644 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
24645 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
24647 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
24650 * gimplify.c (is_oacc_declared): New function.
24651 (oacc_default_clause): Use it to set default flags for acc declared
24652 variables inside parallel regions.
24653 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
24654 declared variables.
24655 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
24656 declare attribute to any decl as necessary.
24658 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
24661 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
24662 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
24663 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
24664 (arm_arch_lpae): This.
24665 * config/arm/arm.c (arm_arch7ve): Rename into ...
24666 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
24667 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
24670 2017-03-22 Martin Liska <mliska@suse.cz>
24673 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
24674 error message instead of an ICE.
24676 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24678 * doc/extend.texi (6.11 Additional Floating Types): Revise.
24680 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
24682 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
24684 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24687 2017-03-21 Martin Sebor <msebor@redhat.com>
24689 * doc/extend.texi: Use "cannot" instead of "can't."
24690 * doc/hostconfig.texi: Same.
24691 * doc/install.texi: Same.
24692 * doc/invoke.texi: Same.
24693 * doc/loop.texi: Same.
24694 * doc/md.texi: Same.
24695 * doc/objc.texi: Same.
24696 * doc/rtl.texi: Same.
24697 * doc/tm.texi: Same.
24698 * doc/tm.texi.in: Same.
24699 * doc/trouble.texi: Same.
24701 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
24704 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
24705 (collect_checksum_attributes): Set it.
24706 (die_checksum_ordered): Use it.
24708 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24710 PR tree-optimization/79908
24711 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
24712 change: For a VA_ARG whose LHS has been cast away, use
24713 force_gimple_operand to construct the side effects.
24715 2017-03-21 David Malcolm <dmalcolm@redhat.com>
24717 PR translation/80001
24718 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
24719 more amenable to translation.
24720 (oacc_loop_auto_partitions): Likewise.
24722 2017-03-21 Marek Polacek <polacek@redhat.com>
24723 Martin Sebor <msebor@redhat.com>
24725 PR tree-optimization/80109
24726 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
24727 on INTEGRAL_TYPE_P.
24729 2017-03-21 Jakub Jelinek <jakub@redhat.com>
24730 Segher Boessenkool <segher@kernel.crashing.org>
24733 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
24734 check reg_used_between_p between insn and one of succ or succ2
24735 depending on if succ is artificial insn not inserted into insn
24738 2017-03-21 Martin Liska <mliska@suse.cz>
24740 PR gcov-profile/80081
24741 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
24742 * doc/gcc.texi: Include gcov-dump stuff.
24743 * doc/gcov-dump.texi: New file.
24745 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
24747 PR rtl-optimization/79150
24748 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
24749 conditional jump, if the jump is the last insn of the loop.
24751 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24752 Richard Biener <rguenth@suse.de>
24754 PR tree-optimization/79908
24755 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
24756 been cast away, use force_gimple_operand to construct the side
24759 2017-03-21 Martin Liska <mliska@suse.cz>
24761 PR libfortran/79956
24762 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
24765 2017-03-21 Brad Spengler <spender@grsecurity.net>
24768 * plugin.c (htab_hash_plugin): New function.
24769 (add_new_plugin): Use it and adjust.
24770 (parse_plugin_arg_opt): Adjust.
24771 (init_one_plugin): Likewise.
24773 2017-03-21 Richard Biener <rguenther@suse.de>
24775 PR tree-optimization/80032
24776 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
24777 if set force the cleanup to happen unconditionally.
24778 (gimplify_target_expr): Push inserted clobbers with force_uncond
24779 to avoid them being removed by control-dependent DCE.
24781 2017-03-21 Richard Biener <rguenther@suse.de>
24783 PR tree-optimization/80122
24784 * tree-inline.c (copy_bb): Do not expans va-arg packs or
24785 va_arg_pack_len when the inlined call stmt requires pack
24787 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
24789 2017-03-21 Jakub Jelinek <jakub@redhat.com>
24792 * tsan.c (instrument_builtin_call): If the memory model argument
24793 is not a constant, assume it is valid.
24796 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
24799 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
24801 PR rtl-optimization/79910
24802 * combine.c (can_combine_p): Do not allow combining an I0 or I1
24803 if its dest is used by an insn before I2 (other than the combined
24804 insns themselves, which are properly handled already).
24806 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
24809 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
24811 * combine.c (record_used_regs): New static function.
24812 (try_combine): Handle situations where there is an additional
24813 instruction between I2 and I3 which needs to have a LOG_LINK
24817 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
24819 * combine.c (try_combine): Delete redundant i1 test. Call
24820 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
24822 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24825 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
24826 alternatives 13/14.
24828 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24830 PR tree-optimization/80054
24831 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
24832 the optimization if a PHI or any of its arguments is not dominated
24833 by the candidate's basis. Use gphi* rather than gimple* as
24835 (replace_profitable_candidates): Clean up a gimple* variable that
24836 should be a gphi* variable.
24838 2017-03-20 Martin Sebor <msebor@redhat.com>
24841 * doc/extend.texi (attribute constructor): Document present limitation.
24843 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
24846 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
24847 __POWER9_VECTOR__ #ifdef control, change template definition to
24848 use Power9-specific built-in function.
24849 (vec_any_eq): Likewise.
24850 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
24851 to control outcomes from this test.
24852 (vector_ae_<mode>p): For VEC_F modes, likewise.
24854 2017-03-20 Ian Lance Taylor <iant@google.com>
24856 * config/i386/i386.c (ix86_function_regparm): Save an extra
24857 register for -fsplit-stack with DECL_STATIC_CHAIN.
24859 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
24862 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
24863 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
24865 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
24867 * config/riscv/riscv.c (riscv_print_operand): Use "fence
24869 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
24872 2017-03-20 Marek Polacek <polacek@redhat.com>
24875 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
24877 2017-03-20 Richard Biener <rguenther@suse.de>
24879 PR tree-optimization/80113
24880 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
24881 allocate extra SSA name for PHI def.
24882 (add_close_phis_to_outer_loops): Likewise.
24883 (add_close_phis_to_merge_points): Likewise.
24884 (copy_loop_close_phi_args): Likewise.
24885 (copy_cond_phi_nodes): Likewise.
24887 2017-03-20 Martin Liska <mliska@suse.cz>
24889 PR middle-end/79753
24890 * tree-chkp.c (chkp_build_returned_bound): Do not build
24891 returned bounds for a LHS that's not a BOUNDED_P type.
24893 2017-03-20 Martin Liska <mliska@suse.cz>
24897 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
24898 COMPLEX_CST and VECTOR_CST.
24900 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24903 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
24904 target operand. A new splitter adds the clobber statement in case
24905 the target operand is dead anyway.
24907 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
24909 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
24910 to age-old versions of binutils and glibc.
24912 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
24914 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
24916 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
24918 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
24920 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
24922 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
24923 requirement for binutils 2.13.
24925 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
24927 * combine.c (try_combine): Delete redundant i1 test. Call
24928 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
24930 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
24932 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
24933 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
24935 <riscv64-*-elf>: Re-arrange section
24936 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
24937 <riscv32-*-linux>: Likewise.
24938 <riscv64-*-elf>: Likewise
24939 <riscv64-*-linux>: Likewise.
24941 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
24944 * aarch64.opt(verbose-cost-dump): Fix typo.
24946 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
24949 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
24950 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
24952 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
24954 * reload.c (find_reloads): When reloading a nonoffsettable address,
24955 use RELOAD_OTHER for it and its address reloads.
24957 PR rtl-optimization/79910
24958 * combine.c (record_used_regs): New static function.
24959 (try_combine): Handle situations where there is an additional
24960 instruction between I2 and I3 which needs to have a LOG_LINK
24963 2017-03-17 Jeff Law <law@redhat.com>
24965 PR tree-optimization/71437
24966 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
24967 conditional in the hash table first.
24968 (vrp_dom_walker::before_dom_children): Extract condition from
24969 ASSERT_EXPR. Record condition, its inverion and any implied
24970 conditions as well.
24972 2017-03-17 Marek Polacek <polacek@redhat.com>
24973 Markus Trippelsdorf <markus@trippelsdorf.de>
24975 PR tree-optimization/80079
24976 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
24979 2017-03-17 Richard Biener <rguenther@suse.de>
24981 PR middle-end/80075
24982 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
24983 Properly verify the LHS before the RHS possibly claims to be
24985 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
24988 2017-03-17 Martin Jambor <mjambor@suse.cz>
24990 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
24991 (List of -O2 options): Likewise.
24992 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
24995 2017-03-17 Tom de Vries <tom@codesourcery.com>
24997 * gcov-dump.c (print_usage): Print bug_report_url.
24999 2017-03-17 Richard Biener <rguenther@suse.de>
25001 PR middle-end/80050
25002 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
25003 (parser::peek): Likewise.
25005 2017-03-17 Richard Biener <rguenther@suse.de>
25007 PR tree-optimization/80048
25008 * sese.c (free_sese_info): Properly release rename_map and
25009 copied_bb_map elements.
25011 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
25013 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
25014 Add linked-list forward and backlinks. Insert on
25015 construction, remove on destruction.
25016 (class pass_store_merging): Add m_stores_head field.
25017 (pass_store_merging::terminate_and_process_all_chains):
25018 Iterate over m_stores_head list.
25019 (pass_store_merging::terminate_all_aliasing_chains):
25021 (pass_store_merging::execute): Check for debug stmts first.
25022 Push new chains onto the m_stores_head stack.
25024 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
25027 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
25028 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
25029 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
25031 2017-03-16 Jeff Law <law@redhat.com>
25033 PR tree-optimization/71437
25034 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
25035 member function. Implementation moved into after_dom_children
25036 member function and into the threader's thread_outgoing_edges
25038 (dom_opt_dom_walker::after_dom_children): Simplify by moving
25039 some code into new thread_outgoing_edges.
25040 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
25041 definition. Simplify marker handling (do it here). Assume we always
25042 have the available expression and the const/copies tables.
25043 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
25045 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
25046 * tree-vrp.c (equiv_stack): No longer file scoped.
25047 (vrp_dom_walker): New class.
25048 (vrp_dom_walker::before_dom_children): New member function.
25049 (vrp_dom_walker::after_dom_children): Likewise.
25050 (identify_jump_threads): Setup domwalker. Use it rather than
25051 walking edges in a random order by hand. Simplify setup/finalization.
25052 (finalize_jump_threads): Remove.
25053 (vrp_finalize): Do not call identify_jump_threads here.
25054 (execute_vrp): Do it here instead and call thread_through_all_blocks
25057 PR tree-optimization/71437
25058 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
25060 (simplify_stmt_for_jump_threading): Add basic_block argument. All
25062 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
25063 (dom_opt_dom_walker::thread_across_edge): Remove
25064 handle_dominating_asserts argument. All callers changed.
25065 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
25066 changes. Remove calls to lhs_of_dominating_assert. Other
25067 uses of handle_dominating_asserts turn into unconditional code
25068 (simplify_control_stmt_condition_1): Likewise.
25069 (simplify_control_stmt_condition): Likewise.
25070 (thread_through_normal_block, thread_across_edge): Likewise.
25071 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
25072 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
25073 object if it is not an SSA_NAME.
25074 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
25075 before calling into the VRP specific simplifiers.
25076 (identify_jump_threads): Remove handle_dominating_asserts
25079 2017-03-16 Jakub Jelinek <jakub@redhat.com>
25082 * tree-diagnostic.c (default_tree_printer): No longer static.
25083 * tree-diagnostic.h (default_tree_printer): New prototype.
25085 2017-03-16 Tamar Christina <tamar.christina@arm.com>
25087 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
25088 Change ins into fmov.
25090 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25092 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
25093 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
25094 Use h_con constraint for operand 1.
25095 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
25096 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
25098 2017-03-15 Jeff Law <law@redhat.com>
25100 PR tree-optimization/71437
25101 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
25102 (record_temporary_equivalences): Use it.
25104 PR tree-optimization/71437
25105 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
25106 tree-ssa-scopedtables.
25107 (lookup_avail_expr, build_and_record_new_cond): Likewise.
25108 (record_conditions, record_cond, vuse_eq): Likewise.
25109 (record_edge_info): Adjust to API tweak of record_conditions.
25110 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
25111 (record_temporary_equivalences, optimize_stmt): Likewise.
25112 (eliminate_redundant_computations): Likewise.
25113 (record_equivalences_from_stmt): Likewise.
25114 * tree-ssa-scopedtables.c: Include options.h and params.h.
25115 (vuse_eq): New function, moved from tree-ssa-dom.c
25116 (build_and_record_new_cond): Likewise.
25117 (record_conditions): Likewise. Accept vector of conditions rather
25118 than edge_equivalence structure for first argument.
25119 for the first argument.
25120 (avail_exprs_stack::lookup_avail_expr): New member function, moved
25121 from tree-ssa-dom.c.
25122 (avail_exprs_stack::record_cond): Likewise.
25123 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
25124 from tree-ssa-dom.c.
25125 (avail_exprs_stack): Add new member functions lookup_avail_expr
25127 (record_conditions): Declare.
25129 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
25132 * lra-constraints.c (process_alt_operands): Increase reject for
25133 reloading an input/output operand.
25135 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
25138 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
25139 insns to convert from signed/unsigned char/short to IEEE 128-bit
25141 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
25143 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
25146 * config/i386/i386.c (ix86_vector_duplicate_value): Create
25147 subreg of inner mode for values already in registers.
25149 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
25151 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
25152 iteration reg is used after the loop.
25154 2017-03-14 Martin Sebor <msebor@redhat.com>
25156 PR tree-optimization/79800
25157 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
25158 precision in negative-positive range.
25159 (format_floating): Call non-const overload with adjusted precision.
25161 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25164 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
25167 2017-03-14 Martin Sebor <msebor@redhat.com>
25169 PR middle-end/80020
25170 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
25171 * builtins.def (aligned_alloc): Use it.
25174 * Makefile.in (GTFILES): Add calls.c.
25175 * calls.c: Include "gt-calls.h".
25177 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
25179 PR rtl-optimization/79728
25180 * regs.h (struct target_regs): New field
25181 x_contains_allocatable_regs_of_mode.
25182 (contains_allocatable_regs_of_mode): New macro.
25183 * reginfo.c (init_reg_sets_1): Initialize it, and change
25184 contains_reg_of_mode so it includes global regs as well.
25185 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
25186 rather than contains_regs_of_mode.
25188 2017-03-14 Martin Liska <mliska@suse.cz>
25190 * doc/invoke.texi: Document options that can't be combined with
25191 -fcheck-pointer-bounds.
25193 2017-03-14 Martin Liska <mliska@suse.cz>
25195 PR middle-end/79831
25196 * doc/invoke.texi (-Wchkp): Document the option.
25198 2017-03-14 Martin Liska <mliska@suse.cz>
25200 * Makefile.in: Install gcov-dump.
25202 2017-03-14 Martin Liska <mliska@suse.cz>
25204 * multiple_target.c (expand_target_clones): Bail out for
25205 an invalid attribute.
25207 2017-03-14 Richard Biener <rguenther@suse.de>
25209 * alias.c (struct alias_set_entry): Pack properly.
25210 * cfgloop.h (struct loop): Likewise.
25211 * cse.c (struct set): Likewise.
25212 * ipa-utils.c (struct searchc_env): Likewise.
25213 * loop-invariant.c (struct invariant): Likewise.
25214 * lra-remat.c (struct cand): Likewise.
25215 * recog.c (struct change_t): Likewise.
25216 * rtl.h (struct address_info): Likewise.
25217 * symbol-summary.h (function_summary): Likewise.
25218 * tree-loop-distribution.c (struct partition): Likewise.
25219 * tree-object-size.c (struct object_size_info): Likewise.
25220 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
25221 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
25222 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
25223 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
25224 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
25225 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
25226 (struct _stmt_vec_info): Likewise.
25228 2017-03-14 Martin Liska <mliska@suse.cz>
25231 * multiple_target.c (create_dispatcher_calls): Check that
25232 a target can create a function dispatcher.
25234 2017-03-14 Martin Liska <mliska@suse.cz>
25237 * multiple_target.c (expand_target_clones): Drop local.local
25238 flag for default implementation.
25240 2017-03-14 Richard Biener <rguenther@suse.de>
25242 PR tree-optimization/80030
25243 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
25245 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
25247 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
25248 gcc_fallthrough() instead of __attribute__((fallthrough));
25250 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
25252 * doc/gcc.texi: Remove "up" link to (DIR).
25253 * doc/gccint.texi: Ditto.
25255 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
25257 * doc/install.texi (Specific) <avr>: Remove reference to
25260 2017-03-13 Jeff Law <law@redhat.com>
25262 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
25263 attribute rather than comments.
25265 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
25266 match_scratch operand is highest.
25268 2017-03-13 Martin Liska <mliska@suse.cz>
25270 PR middle-end/78339
25271 * ipa-pure-const.c (warn_function_noreturn): If the declarations
25272 is a CHKP clone, use original declaration.
25274 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25276 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
25277 (arc_conditional_register_usage): Use a different allocation order
25278 when optimizing for size.
25279 * common/config/arc/arc-common.c (arc_option_optimization_table):
25280 Section anchors default on when optimizing for size.
25282 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25284 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
25286 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25288 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
25289 * config/arc/arc.md (cpu_facility): Add cd variant.
25290 (*movqi_insn): Add code density variant.
25291 (*movhi_insn): Likewise.
25292 (*movqi_insn): Likewise.
25293 (*addsi3_mixed): Likewise.
25294 (subsi3_insn): Likewise.
25296 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25298 * config/arc/arc.md (movsi_cond_exec): Update constraint.
25300 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25302 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
25303 expressions with MINUS and UNARY ops.
25305 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25308 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
25310 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
25311 between vec_select and vector argument.
25312 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
25313 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
25314 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
25315 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
25316 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
25317 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
25319 2017-03-13 Richard Biener <rguenther@suse.de>
25322 * params.def (vect-max-peeling-for-alignment): Fix typo.
25324 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25326 * doc/install.texi (Specific) <mips-*-*>: Remove description of
25327 issue that only occurred with binutils below 2.18.
25329 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25331 * doc/install.texi (Specific) <cris-axis-elf>: No longer
25332 refer to binutils 2.11/2.12 minimum.
25334 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25336 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
25337 ftp.kernel.org and simplify binutils requirement.
25339 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
25341 * doc/invoke.texi (Warning Options): Fix spelling of link-time
25343 (Optimize Options): Ditto. Also remove redundancy.
25345 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25347 PR translation/79848
25348 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
25350 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
25351 to G_ to avoid double translation.
25353 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25355 PR translation/79923
25356 * auto-profile.c (get_combined_location): Convert leading
25357 character of diagnostics to lower case and remove trailing period.
25358 (read_profile): Likewise for various diagnostics.
25359 * config/arm/arm.c (arm_option_override): Remove trailing period
25360 from various diagnostics.
25361 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
25362 (msp430_expand_delay_cycles): Likewise.
25364 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25367 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
25368 full command-line argument, rather than just "str".
25369 (aarch64_validate_march): Likewise.
25370 (aarch64_validate_mtune): Likewise.
25372 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
25374 PR rtl-optimization/78911
25375 * lra-assigns.c (must_not_spill_p): New function.
25376 (spill_for): Use it.
25378 2017-03-10 Jakub Jelinek <jakub@redhat.com>
25380 PR tree-optimization/79981
25381 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
25382 ATOMIC_COMPARE_EXCHANGE ifn result.
25383 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
25384 IFN_ATOMIC_COMPARE_EXCHANGE.
25386 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25389 * opts.c (parse_sanitizer_options): Add missing question mark to
25390 "did you mean" message.
25392 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25394 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
25396 (VMULEUH_UNS): Likewise.
25397 (VMULOUB_UNS): Likewise.
25398 (VMULOUH_UNS): Likewise.
25399 * config/rs6000/rs6000.c (builtin_function_type): Remove
25400 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
25402 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25405 * read-rtl-function.c (function_reader::read_rtx_operand): Update
25406 x with result of extra_parsing_for_operand_code_0.
25407 (function_reader::extra_parsing_for_operand_code_0): Convert
25408 return type from void to rtx, returning x. When reading
25409 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
25410 larger size containing struct block_symbol.
25412 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
25414 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
25415 -mfloat128-hardware without -m64.
25417 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
25420 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
25421 entries to the case statement that marks unsigned arguments to
25422 overloaded functions.
25424 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25426 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
25427 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
25429 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
25432 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
25433 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
25435 2017-03-10 Martin Liska <mliska@suse.cz>
25439 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
25440 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
25443 2017-03-10 Olivier Hainque <hainque@adacore.com>
25445 * tree-switch-conversion (array_value_type): Start by resetting
25446 candidate type to it's main variant.
25448 2017-03-10 Jakub Jelinek <jakub@redhat.com>
25450 PR rtl-optimization/79909
25451 * combine.c (try_combine): Use simplify_replace_rtx on individual
25452 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
25453 of the whole CALL_INSN_FUNCTION_USAGE.
25455 PR tree-optimization/79972
25456 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
25457 get_range_info on SSA_NAMEs. Formatting fixes.
25459 2017-03-10 Richard Biener <rguenther@suse.de>
25460 Jakub Jelinek <jakub@redhat.com>
25462 PR tree-optimization/77975
25463 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
25464 edge to be constant.
25465 (get_val_for): For constant x return it. Formatting fix.
25466 (loop_niter_by_eval): Avoid pointless looping if the next iteration
25467 would use the same bases as the current one.
25469 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25471 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
25472 instead of vec_select for V1TImode.
25473 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
25475 (VSX_LE_128): Add V1TI to this mode iterator.
25476 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
25477 (*vsx_le_perm_store_<mode>): Likewise.
25478 (pre-reload splitter for VSX stores): Likewise.
25479 (post-reload splitter for VSX stores): Likewise.
25480 (*vsx_xxpermdi2_le_<mode>): Likewise.
25481 (*vsx_lxvd2x2_le_<mode>): Likewise.
25482 (*vsx_stxvd2x2_le_<mode>): Likewise.
25484 2017-03-09 Michael Eager <eager@eagercon.com>
25486 Correct failures with --enable-checking=yes,rtl.
25488 * config/microblaze/microblaze.c (microblaze_expand_shift):
25489 Replace GET_CODE test with CONST_INT_P and INTVAL test with
25490 test for const0_rtx.
25491 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
25492 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
25494 2017-03-09 Richard Biener <rguenther@suse.de>
25496 PR tree-optimization/79977
25497 * graphite-scop-detection.c (scop_detection::merge_sese):
25498 Handle the case of extra exits to blocks dominating the entry.
25500 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
25502 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25505 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
25507 PR rtl-optimization/79949
25508 * lra-constraints.c (process_alt_operands): Check memory when
25509 trying to predict a cycle. Print about the overall increase.
25511 2017-03-09 Richard Biener <rguenther@suse.de>
25513 PR middle-end/79971
25514 * gimple-expr.c (useless_type_conversion_p): Preserve
25515 TYPE_SATURATING for fixed-point types.
25517 2017-03-09 Richard Biener <rguenther@suse.de>
25520 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
25521 alignment of BLKmode params.
25523 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25526 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
25527 (VALL_NO_V2Q): Likewise.
25529 * config/aarch64/aarch64-simd.md
25530 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
25532 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
25533 VALL_NO_V2Q mode iterator.
25534 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
25536 2017-03-09 Martin Liska <mliska@suse.cz>
25538 PR tree-optimization/79631
25539 * tree-chkp-opt.c (chkp_is_constant_addr): Call
25540 tree_int_cst_sign_bit just for INTEGER constants.
25542 2017-03-09 Martin Liska <mliska@suse.cz>
25546 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
25549 2017-03-09 Marek Polacek <polacek@redhat.com>
25552 * tree.c (inchash::add_expr): Handle TREE_VEC.
25554 2017-03-09 Martin Liska <mliska@suse.cz>
25557 (chkp_narrow_size_and_offset): New function.
25558 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
25559 (void chkp_parse_bit_field_ref): New function.
25560 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
25561 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
25563 2017-03-09 Martin Liska <mliska@suse.cz>
25566 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
25567 (chkp_find_bounds_1): Remove gcc_unreachable.
25569 2017-03-09 Jakub Jelinek <jakub@redhat.com>
25572 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
25573 BUILT_IN_SYNC*, determine the access type from the size suffix and
25574 always build a MEM_REF with that type. Handle forgotten
25575 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
25578 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
25579 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
25580 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
25581 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
25582 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
25583 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
25584 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
25585 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
25586 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
25587 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
25588 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
25589 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
25590 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
25591 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
25592 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
25593 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
25594 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
25595 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
25596 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
25597 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
25598 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
25599 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
25600 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
25601 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
25602 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
25603 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
25604 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
25605 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
25606 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
25607 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
25608 definitions outside of __OPTIMIZE__ guarded section.
25611 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
25612 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
25613 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
25614 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
25617 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25619 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
25620 ("vfenez<mode>"): Add missing constraints.
25622 2017-03-08 Martin Sebor <msebor@redhat.com>
25625 * config/nds32/nds32.c (nds32_option_override):
25626 Fix misspelled diagnostic.
25628 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25631 * gimplify.c (gimplify_omp_for): Replace index var in outer
25632 taskloop statement with an artificial variable and add
25633 OMP_CLAUSE_PRIVATE clause for it.
25635 2017-03-08 Richard Biener <rguenther@suse.de>
25637 PR tree-optimization/79955
25638 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
25639 for accesses that are completely outside of the variable.
25641 2017-03-08 Andrew Haley <aph@redhat.com>
25643 PR tree-optimization/79943
25644 * tree-ssa-loop-split.c (compute_new_first_bound): When
25645 calculating the new upper bound, (END-BEG) should be added, not
25648 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25650 * config/avr/avr.md (setmemhi): Make sure match_dup
25651 operand number comes before match_scratch.
25653 2017-03-08 Richard Biener <rguenther@suse.de>
25655 PR tree-optimization/79920
25656 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
25657 with ncopies == 1 to ...
25658 (vect_transform_slp_perm_load): ... here. Properly compute
25659 all element loads by iterating VF times over the group. Do
25660 not handle ncopies (computed in a broken way) in
25661 vect_create_mask_and_perm.
25663 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25666 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
25667 is a uniform vector, use uniform_vector_p return value instead of
25668 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
25670 2017-03-07 Marek Polacek <polacek@redhat.com>
25672 PR middle-end/79809
25673 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
25674 (alloca_call_type): Likewise.
25676 2017-03-07 Martin Liska <mliska@suse.cz>
25678 * gcov.c (process_args): Put comment to correct location.
25680 2017-03-07 Martin Liska <mliska@suse.cz>
25682 PR middle-end/68270
25683 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
25684 Use array_at_struct_end_p instead of DECL_CHAIN (field).
25685 (chkp_narrow_bounds_for_field): Likewise.
25686 (chkp_parse_array_and_component_ref): Pass one more argument to
25689 2017-03-07 Richard Biener <rguenther@suse.de>
25691 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
25694 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
25696 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
25697 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
25699 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25702 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
25703 to end of description.
25704 (PARAM_MAX_STORES_TO_MERGE): Likewise.
25706 2017-03-07 Jakub Jelinek <jakub@redhat.com>
25708 PR rtl-optimization/79901
25709 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
25711 (*avx512f_<code><mode>3<mask_name>): ... this.
25712 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
25713 iterator instead of VI8_AVX2_AVX512BW.
25715 PR rtl-optimization/79901
25716 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
25717 min/max expander, expand it using expand_vec_cond_expr.
25720 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
25723 2017-03-06 Jakub Jelinek <jakub@redhat.com>
25726 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
25727 to void * for PCH reasons.
25728 * dwarf2out.c (output_loc_operands, output_die): Cast
25729 v.val_vec.array to unsigned char *.
25731 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
25734 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
25737 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
25739 PR rtl-optimization/79571
25740 * lra-constraints.c (process_alt_operands): Calculate static
25741 reject and subtract it from overall when only addresses will be
25744 2017-03-06 Julia Koval <julia.koval@intel.com>
25747 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
25748 incoming stack boundary to 128 for 64-bit targets.
25750 2017-03-06 Richard Biener <rguenther@suse.de>
25752 PR tree-optimization/79894
25753 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
25754 to NULL after folding it.
25756 2017-03-06 Richard Biener <rguenther@suse.de>
25758 PR tree-optimization/79824
25759 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
25760 check disabling peeling for gaps.
25762 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
25764 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
25765 attributes): Document gettimeofday.
25767 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
25769 * config/s390/s390.c (s390_option_override_internal): Set
25770 PARAM_MIN_VECT_LOOP_BOUND
25772 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
25774 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
25775 * config/s390/s390.md: Likewise.
25777 2017-03-06 Jakub Jelinek <jakub@redhat.com>
25780 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
25781 (<avx2_avx512>_perm<mode>): Rename to ...
25782 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
25784 (<avx512>_perm<mode>_mask): Rename to ...
25785 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
25787 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
25788 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
25789 instead of VI8F_256_512.
25790 (avx512f_perm<mode>): New define_expand.
25791 (avx512f_perm<mode>_mask): Likewise.
25792 (avx512f_perm<mode>_1<mask_name>): New define_insn.
25793 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
25795 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25797 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
25798 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
25800 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
25802 2017-03-06 Martin Liska <mliska@suse.cz>
25805 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
25806 when having a SSA NAME w/o VAR_DECL assigned to it.
25808 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25810 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
25811 msa_dpsub_<su>_d): Fix MODE for vec_select.
25813 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25815 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
25817 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
25819 2017-03-06 Richard Biener <rguenther@suse.de>
25821 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
25822 * plugin.c (register_plugin_info): Likewise.
25823 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
25825 2017-03-05 Jakub Jelinek <jakub@redhat.com>
25827 * config/i386/sse.md (sse_storehps, sse_storelps,
25828 avx_<castmode><avxsizesuffix>_<castmode>,
25829 avx512f_<castmode><avxsizesuffix>_<castmode>,
25830 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
25831 in condition that at least one operand is not a MEM.
25833 2017-03-03 Jakub Jelinek <jakub@redhat.com>
25835 PR middle-end/79805
25836 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
25837 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
25839 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
25840 gimple_call_nothrow_p flag based on whether original builtin can throw.
25841 If it can, emit following stmts on the fallthrough edge.
25842 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
25843 don't create new bb if inserting just debug stmts on the edge, try to
25844 insert them on the fallthru bb or just reset debug stmts.
25846 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
25849 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
25850 restore recog_data (including the operand rtxes inside it) around
25851 the call to get_insn_template.
25853 2017-03-03 Martin Sebor <msebor@redhat.com>
25855 PR tree-optimization/79699
25856 * context.c (context::~context): Free MPFR caches to avoid
25857 a memory leak on program exit.
25859 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25861 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
25862 Use wide_int::ulow () instead of .elt (0).
25864 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
25866 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
25867 (*pushxf): Limit oF constraint to 32bit targets and add oC
25868 constraint for 64bit targets.
25869 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
25870 (*pushdf): Change rmF constraint to rmC.
25872 2017-03-03 Martin Liska <mliska@suse.cz>
25874 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
25875 Remove unused variable.
25877 2017-03-03 Jakub Jelinek <jakub@redhat.com>
25880 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
25881 is a memory operand, increase num_memory.
25882 (ix86_expand_args_builtin): Likewise.
25884 2017-03-03 Jan Hubicka <jh@suse.cz>
25887 * ipa-devirt.c (maybe_record_node): Properly handle
25888 __cxa_pure_virtual visibility.
25890 2017-03-03 Martin Liska <mliska@suse.cz>
25892 PR tree-optimization/79803
25893 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
25895 (pass_loop_prefetch::execute): Disabled optimization if an
25896 assumption about L1 cache size is not met.
25898 2017-03-03 Martin Liska <mliska@suse.cz>
25900 PR rtl-optimization/79574
25901 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
25902 (hash_scan_set): Likewise.
25903 (dump_hash_table): Likewise.
25904 (hoist_code): Likewise.
25906 2017-03-03 Richard Biener <rguenther@suse.de>
25908 * fixed-value.c (fixed_from_string): Restore use of elt (1)
25909 in place of uhigh ().
25910 (fixed_convert_from_real): Likewise.
25912 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
25915 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
25917 2017-03-03 Richard Biener <rguenther@suse.de>
25919 PR middle-end/79818
25920 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
25921 TYPE_OVERFLOW_UNDEFINED check.
25923 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25925 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
25927 (vector_ae_<mode>_p): Likewise.
25928 (vector_nez_<mode>_p): Likewise.
25929 (vector_ne_v2di_p): Likewise.
25930 (vector_ae_v2di_p): Likewise.
25931 (vector_ne_<mode>_p): Likewise.
25932 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
25934 (vsx_tsqrt<mode>2_fe): Likewise.
25936 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
25939 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
25941 2017-03-02 Jakub Jelinek <jakub@redhat.com>
25943 PR rtl-optimization/79780
25944 * cprop.c (one_cprop_pass): When second and further conditional trap
25945 in a single basic block is turned into an unconditional trap, turn it
25946 into a deleted note to avoid RTL verification failures.
25948 2017-03-02 Richard Biener <rguenther@suse.de>
25950 * fold-const.c (const_binop): Use ulow () instead of elt (0).
25952 2017-03-02 Richard Biener <rguenther@suse.de>
25954 PR tree-optimization/79345
25956 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
25957 param and abort the walk, returning -1 if it is hit.
25958 (walk_aliased_vdefs): Take a limit param and pass it on.
25959 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
25960 defaulting to 0 and return a signed int.
25961 * tree-ssa-uninit.c (struct check_defs_data): New struct.
25962 (check_defs): New helper.
25963 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
25964 about uninitialized memory.
25965 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
25966 bogus uninitialized warning.
25967 (fixed_convert_from_real): Likewise.
25969 2017-03-02 Bin Cheng <bin.cheng@arm.com>
25971 PR tree-optimization/66768
25972 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
25973 iv_use if base object can't be determined.
25975 2017-03-02 Jakub Jelinek <jakub@redhat.com>
25977 PR tree-optimization/79345
25978 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
25979 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
25980 (get_pattern_stats): Initialize it.
25981 * genemit.c (gen_expand): Verify match_scratch numbers come after
25982 match_operand/match_dup numbers.
25983 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
25984 match_scratch numbers.
25985 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
25987 * config/s390/s390.md (trunctdsd2): Likewise.
25989 2017-03-02 Richard Biener <rguenther@suse.de>
25991 * wide-int.h (wide_int_storage::operator=): Implement in terms
25994 2017-03-02 Richard Biener <rguenther@suse.de>
25996 PR tree-optimization/79777
25997 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
25998 the to insert expression to sth existing.
26000 2017-03-01 Martin Sebor <msebor@redhat.com>
26002 PR middle-end/79692
26003 * gimple-ssa-sprintf.c
26004 (directive::known_width_and_precision): New function.
26005 (format_integer): Use it.
26006 (get_mpfr_format_length): Consider the full range of precision
26007 when computing %g output with the # flag. Set the likely byte
26008 count to 3 rather than 1 when precision is indeterminate.
26009 (format_floating): Correct the lower bound of precision.
26011 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26013 * doc/invoke.texi: Document default code model for 64-bit Linux.
26015 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26018 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
26019 udiv rather than div since input pattern is unsigned.
26021 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
26023 * config/i386/i386.c (print_reg): Warn for values of
26024 unsupported size in integer register.
26026 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
26029 * config/rs6000/predicates.md (current_file_function_operand): Do
26030 not allow self calls to be local if the function is replaceable.
26032 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
26035 * config/rs6000/altivec.h (vec_ctz and others): Change the
26036 preprocessor macro that controls conditional compilation from
26037 _ARCH_PWR9 to __POWER9_VECTOR__.
26038 (vec_all_ne): Change parameterization of __altivec_scalar_pred
26039 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
26040 control (instead of _ARCH_PWR9 control) so that template
26041 definition uses power9-specific function.
26042 (vec_any_eq): Likewise.
26043 (vec_all_ne): Change macro definition to use a power9-specific
26044 expansion under #ifdef __POWER9_VECTOR__ control (instead of
26045 _ARCH_PWR9 control).
26046 (vec_any_eq) Likewise.
26047 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
26048 expansion for CMPNEF to remove support for xvcmpnesp instruction.
26049 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
26050 support for xvcmpnedp instruction.
26051 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
26052 macro expansion so that Power9 implementation of vec_all_ne does
26053 not use the AltiVec predicate framework.
26054 (VCMPNEH_P): Likewise.
26055 (VCMPNEW_P): Likewise.
26056 (VCMPNED_P): Likewise.
26057 (VCMPNEFP_P): Likewise.
26058 (VCMPNEDP_P): Likewise.
26059 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
26060 implementation of vec_any_eq to not use AltiVec predicate
26062 (VCMPAEH_P): Likewise.
26063 (VCMPAEW_P): Likewise.
26064 (VCMPAED_P): Likewise.
26065 (VCMPAEFP_P): Likewise.
26066 (VCMPAEDP_P): Likewise.
26067 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
26068 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
26069 not use the AltiVec predicate framework.
26070 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
26071 of vec_any_eq to not use AltiVec predicate framework.
26072 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
26073 support for predefined __POWER9_VECTOR__ macro to indicate that
26074 Power9 instruction selection is enabled.
26075 (altivec_overloaded_builtins): Remove extraneous
26076 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
26077 function argument types RS6000_BTI_bool_V16QI and
26078 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
26079 entry for overloaded function argument types RS6000_BTI_bool_V4SI
26080 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
26081 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
26082 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
26083 Power9 for implementations of vec_cmpne. Change the signature for
26084 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
26085 (representing vec_all_ne) to remove the previously described first
26086 argument of type RS6000_BTI_INTSI, as this was an artifact of
26087 reliance on the AltiVec predicate framework, which is no longer
26088 used in the implementation of these functions. Add
26089 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
26090 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
26091 since, unlike the AltiVec predicate framework implementation, we
26092 do not share function descriptors between vec_alle and vec_anyeq.
26093 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
26094 set of modes that receive special treatment even when
26095 TARGET_P9_VECTOR is true. The special treatment emits code that
26096 does not depend on Power9 instructions.
26097 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
26098 define_expand to not rely on AltiVec predicate framework.
26099 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
26101 (vector_ne_v2di_p): Change this define_expand to not rely on
26102 AltiVec predicate framework.
26103 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
26105 (vector_ne_<mode>_p): Change this define_expand to not rely on
26106 AltiVec predicate framework.
26107 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
26109 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
26110 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
26111 define_insn pattern.
26112 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
26113 define_insn pattern because the xvcmpne<VSs>. instruction is not
26115 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
26116 instruction is not supported.
26118 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26120 * config/nvptx/nvptx.c: Include intl.h.
26122 2017-03-01 Martin Jambor <mjambor@suse.cz>
26125 * ipa-prop.h (ipa_bits): Removed field known.
26126 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
26127 to pointers. Adjusted their comments to warn about their sharing.
26128 (ipcp_transformation_summary): Change bits to a vector of pointers.
26129 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
26130 (ipa_get_ipa_bits_for_value): Declare.
26131 * tree-vrp.h (value_range): Mark as GTY((for_user)).
26132 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
26133 (ipa_bits_hash_table): Likewise.
26134 (ipa_vr_ggc_hash_traits): Likewise.
26135 (ipa_vr_hash_table): Likewise.
26136 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
26137 being pointers and vr_known being removed.
26138 (ipa_set_jf_unknown): Likewise.
26139 (ipa_get_ipa_bits_for_value): New function.
26140 (ipa_set_jfunc_bits): Likewise.
26141 (ipa_get_value_range): New overloaded functions.
26142 (ipa_set_jfunc_vr): Likewise.
26143 (ipa_compute_jump_functions_for_edge): Use the above functions to
26144 construct bits and vr parts of jump functions.
26145 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
26146 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
26148 (ipcp_grow_transformations_if_necessary): Also allocate
26149 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
26151 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
26152 them. Fix too long lines.
26153 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
26154 vr_known being removed.
26155 (ipa_read_jump_function): Use new setter functions to construct bits
26156 and vr parts of jump functions or set them to NULL.
26157 (write_ipcp_transformation_info): Adjust for bits being pointers.
26158 (read_ipcp_transformation_info): Likewise.
26159 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
26161 Include gt-ipa-prop.h.
26162 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
26164 (ipcp_store_bits_results): Likewise.
26165 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
26166 Do not write to existing jump functions but use a temporary instead.
26168 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26171 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
26172 attempt to use its first operand as BIT_FIELD_REF base.
26174 2017-03-01 Richard Biener <rguenther@suse.de>
26176 PR middle-end/79721
26177 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
26178 interpolating formula in wrapping arithmetic.
26179 (chrec_apply): Convert chrec_evaluate return value to wanted type.
26181 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26183 PR tree-optimization/79734
26184 * tree-vect-generic.c (expand_vector_condition): Optimize
26185 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
26186 Handle VEC_COND_EXPR where comparison has different inner width from
26187 type's inner width.
26189 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
26191 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
26192 markup, and similar issues. Remove @opindex entries for things
26193 that aren't options. Add missing -mmpy-option entries.
26195 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26197 PR tree-optimization/79737
26198 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
26199 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
26200 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
26201 instead of byte_size. Formatting fix.
26202 (shift_bytes_in_array_right): Formatting fix.
26204 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
26207 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
26208 condition on optimize for the leaf function test.
26210 2017-02-28 Martin Liska <mliska@suse.cz>
26213 * read-rtl-function.c (function_reader::handle_unknown_directive):
26214 Bail out when one uses -flto.
26216 2017-02-28 Martin Liska <mliska@suse.cz>
26218 * common.opt: Replace space with tabular for options of <number>
26220 * config/i386/i386.opt: Show <number> value for
26221 -mlarge-data-threshold.
26222 * opts.c (print_filtered_help): Do not display number in hexadecimal
26225 2017-02-28 Martin Liska <mliska@suse.cz>
26227 * common.opt: Fix --help=option -Q for options which are of
26230 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
26232 * config/i386/i386.c (print_reg): Error out for values
26233 of 8-bit size in invalid integer register.
26235 2017-02-28 Martin Sebor <msebor@redhat.com>
26237 PR tree-optimization/79691
26238 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
26240 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26243 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
26244 gcc_unreachable with output_operand_lossage.
26246 2017-02-28 Richard Biener <rguenther@suse.de>
26248 PR tree-optimization/79740
26249 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
26251 (visit_nary_op): Insert the nary into the hashtable if we
26252 pattern-matched sth.
26253 * tree-ssa-pre.c (eliminate_insert): Robustify.
26255 2017-02-28 Richard Biener <rguenther@suse.de>
26257 PR middle-end/79731
26258 * fold-const.c (decode_field_reference): Reject out-of-bound
26261 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26263 * config/i386/i386.c: Include intl.h.
26264 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
26265 instead of just cond ? "..." : "...".
26266 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
26267 * coverage.c (read_counts_file): Likewise.
26268 * omp-offload.c: Include intl.h.
26269 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
26270 of just cond ? "..." : "...".
26271 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
26272 of just cond ? "..." : "...".
26274 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
26277 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
26279 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
26280 'tune for' CPU name.
26281 * config/arm/arm-cpu-data.h: Regenerated.
26283 2017-02-28 Richard Biener <rguenther@suse.de>
26285 PR tree-optimization/79732
26286 * tree-inline.c (expand_call_inline): Do not shadow var.
26288 2017-02-28 Richard Biener <rguenther@suse.de>
26290 PR tree-optimization/79723
26291 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
26292 address-space properly.
26294 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
26296 * doc/optinfo.texi (Optimization groups): Fix option used for
26298 * doc/invoke.texi (-fopt-info): Document "omp".
26299 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
26300 (OPTGROUP_ALL): Add OPTGROUP_OMP.
26301 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
26302 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
26303 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
26305 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
26307 * dumpfile.c (optgroup_options): Instead of "openmp", associate
26308 OPTGROUP_OMP with "omp".
26310 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
26313 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
26314 for arithmetic shift of unsigned V2DI.
26316 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
26318 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
26319 arc/linux.h headers.
26320 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
26321 (LINK_SPEC): Likewise.
26322 (ARC_TLS_EXTRA_START_SPEC): Likewise.
26323 (EXTRA_SPECS): Likewise.
26324 (STARTFILE_SPEC): Likewise.
26325 (ENDFILE_SPEC): Likewise.
26326 (LIB_SPEC): Likewise.
26327 (TARGET_SDATA_DEFAULT): Likewise.
26328 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
26329 (MULTILIB_DEFAULTS): Likewise.
26330 (DWARF2_UNWIND_INFO): Likewise.
26331 * config/arc/big.h: New file.
26332 * config/arc/elf.h: Likewise.
26333 * config/arc/linux.h: Likewise.
26334 * config/arc/t-uClibc: Remove.
26336 2017-02-27 Bin Cheng <bin.cheng@arm.com>
26338 PR tree-optimization/77536
26339 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
26340 (tree_transform_and_unroll_loop): Use above function to compute the
26341 estimated niter of unrolled loop and use it when scaling profile.
26342 Also use count info rather than frequency if it's non-zero.
26343 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
26344 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
26345 (vect_transform_loop): Call above function.
26347 2017-02-27 Richard Biener <rguenther@suse.de>
26349 PR tree-optimization/45397
26350 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
26351 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
26352 (visit_nary_op): Add pattern matching for CSEing sign-changed
26353 or truncated operations with wider ones.
26355 2017-02-27 Richard Biener <rguenther@suse.de>
26357 PR tree-optimization/79690
26358 * tree-vect-stmts.c (vectorizable_store): Use vector type
26359 built from the DR with address-space.
26361 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
26363 * doc/invoke.texi (Optimize Options): Refine the description
26364 of asan-use-after-return.
26366 2017-02-25 Alan Modra <amodra@gmail.com>
26368 PR rtl-optimization/79584
26369 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
26370 base, not ad->base_term, the reg within base. Remove assertion
26371 that ad->base == ad->base_term. Replace gen_int_mode using
26372 bogus mode with const0_rtx.
26374 2017-02-25 Jakub Jelinek <jakub@redhat.com>
26376 PR middle-end/79396
26377 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
26378 FMA_EXPR like tcc_binary or tcc_unary.
26380 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
26383 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
26385 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
26386 (output_loc_operands): Handle DW_OP_call_ref and
26387 DW_OP_GNU_variable_value.
26388 (struct variable_value_struct): New type.
26389 (struct variable_value_hasher): Likewise.
26390 (variable_value_hash): New variable.
26391 (string_types): Remove.
26392 (copy_loc_descr): New function.
26393 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
26394 (prepend_loc_descr_to_each): New function.
26395 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
26396 instead of add_loc_descr_to_each if the first argument is single
26397 location list and the second has multiple.
26398 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
26399 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
26400 when looking for variable value which doesn't have other location info.
26401 (loc_list_from_tree): Formatting fix.
26402 (gen_array_type_die): Simplify DW_AT_string_length handling.
26403 (adjust_string_types): Remove.
26404 (gen_subprogram_die): Don't call adjust_string_types nor test/set
26405 string_types. Call resolve_variable_values.
26406 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
26407 (resolve_addr_in_expr): Likewise. Add A argument.
26408 (copy_deref_exprloc): Remove deref argument. Adjust for the
26409 original expression being DW_OP_GNU_variable_value with optionally
26410 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
26411 optionally after it.
26412 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
26413 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
26414 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
26415 (variable_value_hasher::hash, variable_value_hasher::equal): New
26417 (resolve_variable_value_in_expr, resolve_variable_value,
26418 resolve_variable_values, note_variable_value_in_expr,
26419 note_variable_value): New functions.
26420 (dwarf2out_early_finish): Call note_variable_value on all toplevel
26423 2017-02-24 Jakub Jelinek <jakub@redhat.com>
26426 * opts.h (handle_generated_option): Add GENERATED_P argument.
26427 * opts-common.c (handle_option): Adjust function comment.
26428 (handle_generated_option): Add GENERATED_P argument, pass it to
26430 (control_warning_option): Pass false to handle_generated_option
26432 * opts.c (maybe_default_option): Pass true to handle_generated_option
26434 * optc-gen.awk: Likewise.
26436 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26438 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
26439 a REG, look at the REG it is a SUBREG of.
26440 (splitter for cmpeqsi_t): Ditto.
26442 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26444 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
26445 the special USEs with the pattern of the insn, not the insn itself.
26447 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
26450 * doc/invoke.texi: Document -mload-store-pairs.
26452 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26453 Sandra Loosemore <sandra@codesourcery.com>
26455 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
26456 argument isn't a CONST_INT.
26457 (nios2_alternate_compare_const): Assert op is a CONST_INT.
26458 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
26459 (nios2_validate_compare): Bypass alternate compare logic if *op2
26460 is not a CONST_INT.
26461 (ldstwm_operation_p): Return false if first_base is not a REG or
26462 if first_offset is not a CONST_INT.
26464 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26466 * config/cris/cris.md: Use correct operand in a define_peephole2.
26468 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26470 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
26472 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26474 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
26475 this_insn if it is an INSN or JUMP_INSN.
26476 (force_offsettable): Look at base, not at addr.
26477 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
26478 on things that aren't necessarily CONST_INTs.
26480 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
26482 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
26483 -mfpmath=sse is the default also for x86-32 targets with SSE2
26484 instruction set when @option{-ffast-math} is enabled
26486 2017-02-24 Jeff Law <law@redhat.com>
26488 PR rtl-optimizatoin/79286
26489 * ira.c (update_equiv_regs): Drop may_trap_p exception to
26492 2017-02-24 Richard Biener <rguenther@suse.de>
26494 PR tree-optimization/79389
26495 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
26498 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
26500 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
26501 function comment to reflect reality.
26502 (loop_exits_before_overflow): Fix typo in function description.
26504 2017-02-24 Richard Biener <rguenther@suse.de>
26506 PR tree-optimization/79389
26507 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
26508 properly that a threading opportunity exists. Detect conditional
26509 copy/constant propagation opportunities.
26511 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
26513 * config/visium/visium.md (type): Add trap.
26514 (b): New mode attribute.
26515 (*btst): Rename into...
26516 (*btst<mode>): ...this and adjust.
26517 (*cbranchsi4_btst_insn): Rename into...
26518 (*cbranch<mode>4_btst_insn): ...this and adjust.
26519 (trap): New define_insn.
26521 2017-02-23 Jakub Jelinek <jakub@redhat.com>
26523 PR tree-optimization/79389
26524 * ifcvt.c (struct noce_if_info): Add rev_cond field.
26525 (noce_reversed_cond_code): New function.
26526 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
26527 reversed_comparison_code. Formatting fix.
26528 (noce_try_store_flag): Test rev_cond != NULL in addition to
26529 reversed_comparison_code.
26530 (noce_try_store_flag_constants): Likewise.
26531 (noce_try_store_flag_mask): Likewise.
26532 (noce_try_addcc): Use rev_cond if non-NULL instead of
26533 reversed_comparison_code.
26534 (noce_try_cmove_arith): Likewise. Formatting fixes.
26535 (noce_try_minmax, noce_try_abs): Clear rev_cond.
26536 (noce_find_if_block): Initialize rev_cond.
26537 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
26538 instead of false as last argument never attempt to reverse it
26541 2017-02-23 Bin Cheng <bin.cheng@arm.com>
26543 PR tree-optimization/79663
26544 * tree-predcom.c (combine_chains): Process refs in reverse order
26545 only for ZERO length chains, and add explaining comment.
26547 2017-02-23 Jeff Law <law@redhat.com>
26549 PR tree-optimization/79578
26550 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
26551 in call to operand_equal_p.
26553 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
26556 * config/i386/cpuid.h: Fix another undefined behavior.
26558 2017-02-23 Richard Biener <rguenther@suse.de>
26560 PR tree-optimization/79683
26561 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
26562 vector types for data-refs.
26564 2017-02-23 Martin Liska <mliska@suse.cz>
26566 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
26568 2017-02-23 Jakub Jelinek <jakub@redhat.com>
26570 PR middle-end/79665
26571 * internal-fn.c (get_range_pos_neg): Moved to ...
26572 * tree.c (get_range_pos_neg): ... here. No longer static.
26573 * tree.h (get_range_pos_neg): New prototype.
26574 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
26575 are known to be in between 0 and signed maximum inclusive, try to
26576 expand both unsigned and signed divmod and use the cheaper one from
26579 2017-02-22 Jeff Law <law@redhat.com>
26581 PR tree-optimization/79578
26582 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
26583 to compare base operands.
26585 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
26588 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
26589 gpc_reg_operand instead of fpr_reg_operand.
26591 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
26593 * config/mips/mips.c (mips_return_in_memory): Force FP
26594 vector types to be returned in memory for o32 ABI.
26596 2017-02-22 Jakub Jelinek <jakub@redhat.com>
26598 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
26599 instead of DW_TAG_member for static data member declarations and don't
26600 set no_linkage_name for static inline data members.
26601 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
26604 2017-02-22 Martin Liska <mliska@suse.cz>
26606 * doc/invoke.texi: Replace inequality signs with square brackets
26609 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26611 PR tree-optimization/68644
26612 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
26614 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
26617 * lra-constraints.c (simplify_operand_subreg): Handle
26618 WORD_REGISTER_OPERATIONS targets.
26620 2017-02-22 Jakub Jelinek <jakub@redhat.com>
26623 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
26624 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
26625 elimination by swapping fld*.
26627 2017-02-22 Richard Biener <rguenther@suse.de>
26629 PR tree-optimization/79673
26630 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
26631 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
26632 irrelevant address-space qualifiers and avoiding a
26633 ADDR_SPACE_CONVERT_EXPR from fold_convert.
26635 2017-02-22 Richard Biener <rguenther@suse.de>
26637 PR tree-optimization/79666
26638 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
26639 to not symbolically negate if that may introduce undefined
26642 2017-02-22 Martin Liska <mliska@suse.cz>
26645 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
26646 * data-streamer-out.c (streamer_write_gcov_count_stream):
26648 * value-prof.c (stream_out_histogram_value): Make assert more
26649 precise based on type of counter.
26651 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
26654 * config/i386/i386.md (standard_x87sse_constant_load splitter):
26655 Use nonimmediate_operand instead of memory_operand for operand 1.
26656 (float-extend standard_x87sse_constant_load splitter): Ditto.
26658 2017-02-21 Jeff Law <law@redhat.com>
26660 PR tree-optimization/79621
26661 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
26662 blocks with edges to themselves.
26664 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26667 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
26668 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
26669 Use gimple_call_builtin_p.
26672 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
26673 on temporarily removed DEBUG_INSNs.
26675 PR tree-optimization/79649
26676 * tree-loop-distribution.c (classify_partition): Give up on
26677 non-generic address space loads/stores.
26679 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
26681 * doc/loop.texi (Loop manipulation): Remove nonexistent
26682 tree_ssa_loop_version from the documentation.
26683 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
26685 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26688 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
26689 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
26690 * config/rs6000/rs6000.c: Include except.h.
26691 (rs6000_expand_split_stack_prologue): Call
26692 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
26694 2017-02-21 Martin Jambor <mjambor@suse.cz>
26697 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
26698 have been analyzed.
26700 2017-02-21 Martin Jambor <mjambor@suse.cz>
26702 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
26703 for backward compatibility only.
26704 * doc/invoke.texi (Option Summary): Remove all references to
26705 -fipa-cp-alignment.
26707 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
26711 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26713 * lra-constraints.c (curr_insn_transform): Handle
26714 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
26716 2017-02-21 Martin Liska <mliska@suse.cz>
26718 * config/i386/i386.opt: Replace -masm-dialect with -masm.
26720 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
26722 PR translation/79638
26723 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
26725 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
26728 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
26729 (arm_function_ok_for_sibcall): Return false for an indirect call by
26730 descriptor if all the argument registers are used.
26731 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
26732 alignment of the function.
26734 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26736 PR tree-optimization/61441
26737 * simplify-rtx.c (simplify_const_unary_operation): For
26738 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
26739 the sNaN unmodified.
26741 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26743 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26744 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26745 instead of SYSTEM_HEADER_DIR.
26747 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
26748 Martin Liška <mliska@suse.cz>
26750 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
26751 Fix typos and grammar, use active voice, and clarify.
26753 2017-02-20 Marek Polacek <polacek@redhat.com>
26755 PR middle-end/79537
26756 * gimplify.c (gimplify_expr): Handle unused *&&L;.
26759 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
26761 2017-02-20 Jakub Jelinek <jakub@redhat.com>
26764 * config/i386/i386.c (ix86_expand_builtin): Handle
26765 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
26766 ix86_builtins_isa[fcode].isa as a requirement of those
26767 flags and any other flag in the bitmask.
26768 (ix86_init_mmx_sse_builtins): Use 0 instead of
26769 ~OPTION_MASK_ISA_64BIT as mask.
26770 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
26771 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
26772 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
26773 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
26775 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26778 * lra-constraints.c (split_reg): Check requested split mode
26779 is supported by the register.
26781 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26783 * lra-constraints.c (simplify_operand_subreg): Remove early
26786 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26789 * lra-constraints.c (curr_insn_transform): Tighten condition
26790 for converting SUBREG reloads from OP_OUT to OP_INOUT.
26792 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26795 * lra-constraints.c (curr_insn_transform): Handle
26796 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
26798 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
26801 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
26803 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
26805 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
26808 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
26811 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26813 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
26816 2017-02-18 Jakub Jelinek <jakub@redhat.com>
26818 * final.c (last_columnnum, override_columnnum): New variables.
26819 (final_start_function): Set last_columnnum, pass it to begin_prologue
26820 hook and pass 0 to dwarf2out_begin_prologue.
26821 (final_scan_insn): Update override_columnnum. Pass last_columnnum
26822 to source_line debug hook.
26823 (notice_source_line): Compute last_columnnum and for debug_column_info
26824 return true on column changes.
26825 * debug.h (struct gcc_debug_hooks): Add column argument to
26826 source_line and begin_prologue hooks.
26827 (debug_nothing_int_charstar_int_bool): Remove prototype.
26828 (debug_nothing_int_int_charstar,
26829 debug_nothing_int_int_charstar_int_bool): New prototypes.
26830 (dwarf2out_begin_prologue): Add column argument.
26831 * debug.c (do_nothing_debug_hooks): Adjust source_line and
26832 begin_prologue hooks.
26833 (debug_nothing_int_charstar_int_bool): Remove.
26834 (debug_nothing_int_int_charstar,
26835 debug_nothing_int_int_charstar_int_bool): New functions.
26836 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
26837 through to dwarf2out_source_line.
26838 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
26839 (dwarf2out_source_line): Add column argument, emit it if requested.
26840 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
26842 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
26843 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
26844 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
26845 through to dwarf2out_begin_prologue.
26846 (vmsdbgout_source_line): Add column argument, pass it through to
26847 dwarf2out_source_line.
26848 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
26849 dbxout_source_line caller.
26850 (dbxout_source_line): Add column argument.
26852 * common.opt (gno-column-info, gcolumn-info): New options.
26853 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
26854 (check_die): Also test for multiple DW_AT_decl_column attributes.
26855 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
26856 DW_AT_decl_column if requested.
26857 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
26859 (gen_variable_die): Likewise.
26860 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
26861 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
26864 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
26865 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
26866 (ix86_handle_option): Handle OPT_m3dnowa.
26867 * doc/invoke.texi (-m3dnowa): Document.
26868 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
26869 -m3dnowa instead of -m3dnow -march=athlon.
26872 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
26873 instead of gcc_assert for K, r and R code checks. Formatting fixes.
26875 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26878 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
26879 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
26880 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
26881 generator for vsx_xxpermdi_<mode>_be.
26882 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
26883 force big-endian semantics.
26884 (vsx_xxpermdi_<mode>_be): New define_expand with same
26885 implementation as previous version of vsx_xxpermdi_<mode>.
26887 2017-02-17 Jakub Jelinek <jakub@redhat.com>
26889 PR tree-optimization/79327
26890 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
26891 variable, its initialization and use.
26893 2017-02-17 Julia Koval <julia.koval@intel.com>
26895 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
26896 (OPTION_MASK_ISA_PKU_UNSET): New.
26897 (ix86_handle_option): Handle -mrdpid.
26898 * config/i386/cpuid.h (bit_RDPID): New.
26899 * config/i386/driver-i386.c (host_detect_local_cpu):
26900 Detect RDPID feature.
26901 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
26902 * config/i386/i386-c.c (ix86_target_macros_internal):
26904 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
26905 (ix86_valid_target_attribute_inner_p): Add "rdpid".
26906 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
26907 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
26908 * config/i386/i386.md (define_insn "rdpid"): New.
26909 * config/i386/i386.opt Add -mrdpid.
26910 * config/i386/immintrin.h (_rdpid_u32): New.
26912 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
26914 PR rtl-optimization/79541
26915 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
26916 instead of transforming it into USE.
26918 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
26920 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
26921 If HONOR_SNANS (SFmode) force the input to a register.
26922 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
26923 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
26924 an frsp or similar insn.
26926 2017-02-17 Martin Liska <mliska@suse.cz>
26928 PR rtl-optimization/79577
26929 * params.def (selsched-max-sched-times): Increase minimum to 1.
26931 2017-02-17 Martin Liska <mliska@suse.cz>
26933 PR rtl-optimization/79574
26934 * gcse.c (want_to_gcse_p): Prevent integer overflow.
26936 2017-02-17 Martin Liska <mliska@suse.cz>
26938 PR tree-optimization/79529
26939 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
26940 ssa_defined_default_def_p to handle cases which are implicitly
26942 * tree-ssa.c (ssa_defined_default_def_p): New function.
26943 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
26944 which are implicitly defined.
26945 * tree-ssa.h (ssa_defined_default_def_p): Declare.
26947 2017-02-17 Richard Biener <rguenther@suse.de>
26949 PR middle-end/79576
26950 * params.def (max-ssa-name-query-depth): Limit to 10.
26952 2017-02-17 Richard Biener <rguenther@suse.de>
26954 PR tree-optimization/79552
26955 * tree-ssa-structalias.c (visit_loadstore): Properly verify
26958 2017-02-17 Richard Biener <rguenther@suse.de>
26961 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
26963 2017-02-17 Marek Polacek <polacek@redhat.com>
26965 PR middle-end/79536
26966 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
26967 (fold_negate_expr): New wrapper.
26969 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
26971 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
26972 Correct terminology and de-emphasize pre-standard behavior.
26974 2017-02-16 Alan Modra <amodra@gmail.com>
26976 PR rtl-optimization/79286
26977 * ira.c (def_dominates_uses): New function.
26978 (update_equiv_regs): Don't create an equivalence for insns that
26979 may trap where the register def does not dominate the use.
26981 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
26983 PR rtl-optimization/78127
26984 * lra.c (lra): Call lra_eliminate before finish the loop after
26987 2017-02-16 Richard Biener <rguenther@suse.de>
26989 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
26991 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
26992 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
26993 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
26994 (isl_val_int_from_wi): New function.
26995 (extract_affine_gmp): Rename to ...
26996 (extract_affine_wi): ... this, take a widest_int.
26997 (extract_affine_int): Just wrap extract_affine_wi.
26998 (add_param_constraints): Use isl_val_int_from_wi.
26999 (add_loop_constraints): Likewise, and extract_affine_wi.
27001 2017-02-15 Jeff Law <law@redhat.com>
27003 PR middle-end/79521
27004 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
27005 ira_init_register_move_cost_if_necessary.
27007 2017-02-15 Martin Sebor <msebor@redhat.com>
27009 PR middle-end/32003
27010 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
27011 removed in a prior commit.
27013 2017-02-15 Bin Cheng <bin.cheng@arm.com>
27015 PR tree-optimization/79347
27016 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
27017 counters during peeling.
27019 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
27021 * Makefile.in (site.exp): Remove "set ISLVER".
27023 2017-02-15 Jakub Jelinek <jakub@redhat.com>
27026 * real.c (real_from_integer): Call real_convert even for decimal.
27028 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
27031 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
27033 2017-02-14 Andrew Pinski <apinski@cavium.com>
27035 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
27036 cores and change the partno/implementer to be correct.
27037 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
27038 the 'B" as the implementer.
27039 * config/aarch64/aarch64-tune.md: Regenerate.
27041 2017-02-14 Carl Love <cel@us.ibm.com>
27043 * config/rs6000/rs6000.c: Add case statement entry to make the
27044 xvcvuxdsp built-in argument unsigned.
27045 * config/rs6000/vsx.md: Fix the source and return operand types so they
27046 match the instruction definitions from the ISA document. Fix typo
27047 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
27050 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
27053 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
27054 member early_clobber_alts.
27055 * lra-lives.c (reg_early_clobber_p): New.
27056 (process_bb_lives): Use it.
27057 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
27058 (debug_operand_data): Initialize early_clobber_alts.
27059 (setup_operand_alternative): Set up early_clobber_alts.
27060 (collect_non_operand_hard_regs): Ditto. Pass early clobber
27061 alternatives to new_insn_reg.
27062 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
27064 (lra_update_insn_regno_info): Pass the new arg.
27066 2017-02-14 Jakub Jelinek <jakub@redhat.com>
27068 PR middle-end/79505
27069 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
27070 (new_oacc_loop_raw): Don't clear already cleared fields.
27073 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
27074 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
27075 _mm512_prefetch_i64gather_ps): New inline functions and macros.
27077 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
27080 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
27082 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
27085 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
27086 the extra instruction to the right place to store 128-bit constant
27089 2017-02-14 Martin Sebor <msebor@redhat.com>
27091 PR middle-end/79448
27092 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
27093 warning for strings of unknown length.
27095 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
27097 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
27099 2017-02-14 Jeff Law <law@redhat.com>
27102 * ira-costs.c (scan_one_insn): Initialize register move costs
27103 for pseudos seen in USE/CLOBBER insns.
27105 PR tree-optimization/79095
27106 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
27107 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
27108 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
27109 if the operands are known to be not equal, then the resulting range
27111 (intersect_ranges): If the new range is ~[0,0] and the old range is
27112 wide, then prefer ~[0,0].
27113 * tree-vrp.c (overflow_comparison_p_1): New function.
27114 (overflow_comparison_p): New function.
27115 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
27116 if NAME is used in an overflow test.
27117 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
27118 overflow check that can be expressed as an equality test, then adjust
27119 ops to be that equality test.
27121 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27123 * config/s390/s390-builtin-types.def: Remove flags argument.
27124 * config/s390/s390.c (s390_init_builtins): Likewise.
27126 2017-02-14 Martin Liska <mliska@suse.cz>
27128 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
27129 vector. Fix trailing white spaces.
27131 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
27133 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
27136 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27138 PR rtl-optimization/68664
27139 * config/arm/arm.c (arm_sched_can_speculate_insn):
27140 New function. Declare prototype.
27141 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
27143 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27145 PR rtl-optimization/68664
27146 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
27148 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
27150 2017-02-14 Amit Pawar <amit.pawar@amd.com>
27152 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
27153 max skip bytes for function, loop and jump.
27155 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27157 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
27158 ABS_EXPR for gimple dump.
27160 2017-02-14 Jakub Jelinek <jakub@redhat.com>
27163 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
27165 PR tree-optimization/79408
27166 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
27167 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
27168 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
27169 also if rhs1 is INTEGER_CST.
27171 2017-02-14 Richard Biener <rguenther@suse.de>
27173 PR middle-end/79432
27174 * tree-into-ssa.c (insert_phi_nodes): When the function can
27175 have abnormal edges rewrite SSA names with broken use-def
27176 dominance out of SSA and register them for PHI insertion.
27178 2017-02-13 Martin Sebor <msebor@redhat.com>
27180 PR middle-end/79496
27181 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
27182 clearing info.nowrite flag when snprintf size argument is a range.
27184 2017-02-13 Jakub Jelinek <jakub@redhat.com>
27186 * cprop.c (cprop_jump): Add missing space in string literal.
27187 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
27188 (get_constraint_for_component_ref): Likewise.
27189 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
27190 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
27191 * lra-constraints.c (process_alt_operands): Likewise.
27192 * ipa-inline.c (inline_small_functions): Likewise.
27193 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
27194 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
27195 * trans-mem.c (diagnose_tm_1_op): Likewise.
27196 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
27197 (grid_parallel_clauses_gridifiable): Likewise.
27199 * config/nvptx/mkoffload.c (process): Add space in between
27202 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
27203 "MOD4_SSE_REGS" and "ALL_REGS".
27205 * spellcheck.c (test_data): Add , in between "foo" and "food".
27207 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27210 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
27211 boundary crossing check and subsequent code generation agree.
27213 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27215 * config/aarch64/aarch64.c (has_memory_op): Delete.
27216 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
27219 2017-02-13 Jakub Jelinek <jakub@redhat.com>
27221 PR rtl-optimization/79388
27222 PR rtl-optimization/79450
27223 * combine.c (distribute_notes): When removing TEM_INSN for which
27224 corresponding dest has last value recorded, invalidate that last
27227 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27229 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
27230 of explicit '@'. Add missing assembly comment marker on branch costs
27233 2017-02-13 Nathan Sidwell <nathan@acm.org>
27235 * gengtype-lex.l (<in_struct>): Add '/'.
27237 2017-02-13 Martin Liska <mliska@suse.cz>
27240 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
27242 2017-02-13 Richard Biener <rguenther@suse.de>
27244 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
27246 * configure: Re-generate.
27247 * config.in: Likewise.
27248 * graphite-dependences.c: Simplify as if
27249 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
27250 * graphite-isl-ast-to-gimple.c: Likewise.
27251 * graphite-optimize-isl.c: Likewise.
27252 * graphite-poly.c: Likewise.
27253 * graphite-sese-to-poly.c: Likewise.
27254 * graphite.h: Likewise.
27255 * toplev.c: Include isl/version.h and use isl_version () for
27256 printing the ISL version.
27257 * doc/install.texi: Update ISL requirement.
27259 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
27261 * doc/standards.texi (Standards): Update reference to
27264 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
27266 * doc/extend.texi (Named Address Spaces): sourceware.org now
27268 * doc/install.texi (Binaries): Ditto.
27271 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
27273 * doc/cpp.texi: Replace "stringify"/"stringification" with C
27274 standard terminology "stringize"/"stringizing" throughout.
27275 * doc/cppinternals.texi: Likewise.
27277 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
27279 * doc/extend.texi: Fix some spelling mistakes and typos.
27280 * doc/invoke.texi: Likewise.
27282 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
27285 * params.def (inline-min-speedup) Change from 10 to 8.
27287 2017-02-11 Jakub Jelinek <jakub@redhat.com>
27289 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
27292 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
27295 * ipa-inline-analysis.c (get_minimal_bb): New function.
27296 (record_modified): Use it.
27297 (remap_edge_change_prob): Handle also ancestor functions.
27299 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
27301 * doc/contrib.texi (Contributors): Remove broken link into
27302 the Mauve CVS repository.
27304 2017-02-11 Jakub Jelinek <jakub@redhat.com>
27306 PR middle-end/79454
27307 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
27308 result computation whenever lhs doesn't have vector mode, not
27309 just when it has BLKmode.
27311 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
27313 * doc/makefile.texi (profiledbootstrap): Refer to the
27314 installation instructions only in textual form.
27316 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27319 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
27321 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
27323 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
27324 (Specific): Update mingw-w64 reference.
27326 (Specific): Remove broken link to Renesas RX processor.
27328 2017-02-10 Richard Biener <rguenther@suse.de>
27330 * toplev.c (process_options): Do not mention obsolete graphite
27331 options when printing sorry message about missing graphite support.
27332 Mention -floop-nest-optimize.
27334 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
27336 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
27337 (vtst_p16): Likewise.
27338 (vtstq_p8): Likewise.
27339 (vtstq_p16): Likewise.
27341 (vtstq_p64): Likewise.
27342 * config/arm/arm_neon.h (vgetq_lane_p64): New.
27343 (vset_lane_p64): New.
27344 (vsetq_lane_p64): New.
27346 2017-02-10 Jakub Jelinek <jakub@redhat.com>
27348 PR tree-optimization/79411
27349 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
27350 stmt operands are SSA_NAMEs used in abnormal phis.
27351 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
27354 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
27357 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
27360 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
27362 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
27364 2017-02-09 Jakub Jelinek <jakub@redhat.com>
27366 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
27370 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
27371 not arbitrary TREE_CONSTANT.
27374 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
27375 "omp declare target link" attribute unless is_global_var.
27376 * omp-offload.c (find_link_var_op): Likewise.
27378 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
27379 Chung-Lin Tang <cltang@codesourcery.com>
27381 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
27383 (gimplify_adjust_omp_clauses): Don't delete TILE.
27384 (gimplify_omp_for): Deal with TILE.
27385 * internal-fn.c (expand_GOACC_TILE): New function.
27386 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
27388 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
27389 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
27391 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
27392 avoid DIV for outermost collapse var.
27393 (expand_oacc_for): Insert tile element loop as needed. Adjust.
27394 Remove out of date comments, fix whitespace.
27395 * omp-general.c (omp_extract_for_data): Deal with tiling.
27396 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
27397 adjust OLF_DIM_BASE value.
27398 (struct omp_for_data): Add tiling field.
27399 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
27400 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
27401 for auto loops. Remove default auto determining, moved to
27402 oacc_loop_fixed_partitions.
27403 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
27404 stmts, add e_mask field.
27405 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
27406 (oacc_thread_numbers): Use oacc_dim_call.
27407 (oacc_xform_tile): New.
27408 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
27409 (finish_oacc_loop): Adjust for ifns vector.
27410 (oacc_loop_discover_walk): Append loop abstraction sites to list,
27411 add case for GOACC_TILE fns.
27412 (oacc_loop_xform_loop): Delete.
27413 (oacc_loop_process): Iterate over call list directly, and add
27414 handling for GOACC_TILE fns.
27415 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
27417 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
27418 vector partitioning to outer loops. Assign 2 partitions to loops
27419 when available. Add TILE handling.
27420 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
27421 (execite_oacc_device_lower): Process GOACC_TILE fns,
27422 ignore unknown specs.
27423 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
27424 * tree.c (omp_clause_num_ops): Adjust TILE ops.
27425 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
27427 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
27429 * configure.ac (ACX_BUGURL): Update.
27430 * configure: Regenerate.
27432 2017-02-09 Richard Biener <rguenther@suse.de>
27434 PR tree-optimization/69823
27435 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
27436 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
27438 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
27440 * config/arc/arc-c.def: Add __NPS400__ definition.
27441 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
27442 (TARGET_NPS400): Define.
27444 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
27446 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
27448 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
27449 pointer, arch_info.
27450 (arc_cpu_types): Fill the arch_info field with a pointer into the
27451 arc_arch_types table.
27452 (arc_selected_cpu): Declare.
27453 * config/arc/arc.c (arc_selected_cpu): Make global.
27454 (arc_selected_arch): Delete.
27455 (arc_base_cpu): Delete.
27456 (arc_override_options): Remove references to deleted variables,
27457 update access to arch information.
27458 (ARC_OPT): Update access to arch information.
27459 (ARC_OPTX): Likewise.
27460 * config/arc/arc.h (arc_base_cpu): Remove declaration.
27461 (TARGET_ARC600): Update access to arch information.
27462 (TARGET_ARC601): Likewise.
27463 (TARGET_ARC700): Likewise.
27464 (TARGET_EM): Likewise.
27465 (TARGET_HS): Likewise.
27466 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
27469 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
27472 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
27473 condition/operands for integer GE/LE/GEU/LEU operations.
27475 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
27477 PR translation/79397
27478 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
27481 2017-02-08 Martin Jambor <mjambor@suse.cz>
27484 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
27485 whether allocation happened.
27486 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
27487 nothing was allocated.
27489 2017-02-08 Jakub Jelinek <jakub@redhat.com>
27491 PR tree-optimization/79408
27492 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
27493 constant, but SSA_NAME with a known integer range, use the minimum
27494 of that range instead of op1 to determine if modulo can be replaced
27495 with its first operand.
27497 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27499 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
27501 2017-02-08 Richard Biener <rguenther@suse.de>
27503 PR tree-optimization/71824
27504 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
27505 Check all loops contained in the merged region.
27507 2017-02-07 Andrew Pinski <apinski@cavium.com>
27509 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
27511 2017-02-07 Andrew Pinski <apinski@cavium.com>
27513 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
27514 (thunderxt88): Likewise.
27515 (thunderxt81): Disable LSE and change v8.1 to v8.
27516 (thunderxt83): Likewise.
27518 2017-02-07 Jakub Jelinek <jakub@redhat.com>
27519 Richard Biener <rguenther@suse.de>
27521 PR middle-end/79399
27522 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
27523 type from int to size_t.
27524 * ira-costs.c (struct_costs_size): Change type from int to size_t.
27526 2017-02-07 Jakub Jelinek <jakub@redhat.com>
27528 PR rtl-optimization/79386
27529 * cprop.c (bypass_conditional_jumps): Initialize
27530 bypass_last_basic_block already before splitting bbs after
27531 unconditional traps...
27532 (bypass_conditional_jumps): ... rather than here.
27535 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
27536 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
27537 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
27538 fix -masm=intel patterns.
27540 2017-02-07 Richard Biener <rguenther@suse.de>
27542 PR tree-optimization/79256
27543 PR middle-end/79278
27544 * builtins.c (get_object_alignment_2): Use min_align_of_type
27545 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
27546 and ADJUST_FIELD_ALIGN.
27548 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
27550 * doc/tm.texi: Regenerate.
27551 * stor-layout.c (layout_decl): Adjust.
27552 (update_alignment_for_field): Likewise.
27553 (place_field): Likewise.
27554 (min_align_of_type): Likewise.
27555 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
27556 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
27557 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
27558 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
27559 * config/frv/frv.c (frv_adjust_field_align): Likewise.
27560 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
27561 * config/i386/i386.c (x86_field_alignment): Likewise.
27562 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
27563 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
27564 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
27565 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
27566 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
27567 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
27571 2017-01-30 Richard Biener <rguenther@suse.de>
27573 PR tree-optimization/79256
27574 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
27575 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
27578 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
27580 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
27581 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
27582 builtins to SImode and emit a zero-extend, if necessary.
27584 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27586 * docs/invoke.texi (RISC-V Options): Alphabetize.
27588 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27590 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
27593 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27595 * config/riscv/riscv.c: New file.
27596 * common/config/riscv/riscv-common.c: Likewise.
27597 * config.gcc: Likewise.
27598 * config/riscv/constraints.md: Likewise.
27599 * config/riscv/elf.h: Likewise.
27600 * config/riscv/generic.md: Likewise.
27601 * config/riscv/linux.h: Likewise.
27602 * config/riscv/multilib-generator: Likewise.
27603 * config/riscv/peephole.md: Likewise.
27604 * config/riscv/pic.md: Likewise.
27605 * config/riscv/predicates.md: Likewise.
27606 * config/riscv/riscv-builtins.c: Likewise.
27607 * config/riscv/riscv-c.c: Likewise.
27608 * config/riscv/riscv-ftypes.def: Likewise.
27609 * config/riscv/riscv-modes.def: Likewise.
27610 * config/riscv/riscv-opts.h: Likewise.
27611 * config/riscv/riscv-protos.h: Likewise.
27612 * config/riscv/riscv.h: Likewise.
27613 * config/riscv/riscv.md: Likewise.
27614 * config/riscv/riscv.opt: Likewise.
27615 * config/riscv/sync.md: Likewise.
27616 * config/riscv/t-elf-multilib: Likewise.
27617 * config/riscv/t-linux: Likewise.
27618 * config/riscv/t-linux-multilib: Likewise.
27619 * config/riscv/t-riscv: Likewise.
27620 * configure.ac: Likewise.
27621 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
27622 Waterman as RISC-V maintainers.
27623 * doc/install.texi: Add RISC-V entries.
27624 * doc/invoke.texi: Add RISC-V options section.
27625 * doc/md.texi: Add RISC-V constraints section.
27626 * configure: Regenerated.
27628 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
27631 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
27632 false values to be constant vectors with all 0 or all 1 bits set.
27633 (vcondu<mode><mode>): Likewise.
27634 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
27636 (fpmask_comparison_operator): Update comment.
27637 (vecint_comparison_operator): New predicate.
27638 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
27639 vector conditionals when the true and false values are constant
27640 vectors with all 0 bits or all 1 bits set.
27642 2017-02-06 Martin Sebor <msebor@redhat.com>
27644 PR tree-optimization/79376
27645 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
27647 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
27649 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
27650 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
27651 to simplify split condition.
27653 2017-02-06 Jakub Jelinek <jakub@redhat.com>
27655 * omp-expand.c (oxpand_omp_atomic_fetch_op,
27656 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
27659 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
27661 PR rtl-optimization/68664
27662 * target.def (can_speculate_insn): New hook.
27663 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
27664 * doc/tm.texi: Regenerate.
27665 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
27666 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
27667 (rs6000_sched_can_speculate_insn): New function.
27669 2017-02-06 Jakub Jelinek <jakub@redhat.com>
27671 PR tree-optimization/79284
27672 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
27673 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
27674 vectorizable_mask_load_store, vectorizable_operation,
27675 vect_is_simple_cond, get_same_sized_vectype): Use it instead
27676 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
27677 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
27678 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
27679 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
27680 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
27681 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
27682 is_gimple_assign (stmt). Replace another such test with
27683 is_gimple_assign (stmt).
27685 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
27688 * config/avr/avr.c (rtl-iter.h): Include it.
27689 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
27690 (avr_legitimate_combined_insn): ...and implementation.
27692 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
27694 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
27695 * config/s390/s390.c (s390_const_operand_ok)
27696 (s390_canonicalize_comparison, s390_extract_part)
27697 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
27698 (s390_contiguous_bitmask_p, s390_rtx_costs)
27699 (legitimize_pic_address): Likewise.
27700 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
27701 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
27702 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
27703 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
27704 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
27706 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
27708 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
27709 REGNO($0) == REGNO($1).
27711 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27713 * config/s390/linux.h(SIZE_TYPE): Add comment.
27715 2017-02-06 Julian Brown <julian@codesourcery.com>
27716 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27717 Virendra Pathak <virendra.pathak@broadcom.com>
27719 * config/aarch64/aarch64-cores.def: Change the scheduler
27721 * config/aarch64/aarch64.md: Include thunderx2t99.md.
27722 * config/aarch64/thunderx2t99.md: New file.
27724 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27726 * doc/standards.texi (Go Language): Update link to language
27729 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
27731 * tree-eh.c (lower_resx): Sanitize profile.
27732 (cleanup_empty_eh_move_lp): Likewise.
27734 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
27737 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
27739 * cfgloopmanip.h (loop_version): Update prototype.
27740 * modulo-sched.c (sms_schedule): Update call of loop_version.
27741 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
27742 * tree-parloops.c (gen_parallel_loop): Likewise.
27743 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
27744 * tree-ssa-loop-split.c (split_loop): Likewise.
27745 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
27746 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
27748 2017-02-05 Martin Liska <mliska@suse.cz>
27751 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
27753 (print_operand_address): Initialize a struct to zero.
27755 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27757 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
27758 garbage collector only in textual form.
27760 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27762 * doc/extend.texi (x86 specific memory model extensions for
27763 transactional memory): Simplify a phrase.
27765 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
27768 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
27769 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
27770 (atomic_storedi_1): Likewise.
27772 2017-02-04 Jakub Jelinek <jakub@redhat.com>
27774 PR tree-optimization/79338
27775 * tree-parloops.c (gather_scalar_reductions): Don't call
27776 vect_analyze_loop_form for loop->inner before destroying loop's
27779 2017-02-03 Martin Sebor <msebor@redhat.com>
27781 PR tree-optimization/79327
27782 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
27783 when precision has resulted in leading zeros.
27784 (format_integer): Adjust the likely counter to assume an unknown
27785 argument that may be zero is non-zero.
27787 2017-02-03 Jason Merrill <jason@redhat.com>
27790 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
27791 avoid copying non-taken branch.
27793 2017-02-03 Jakub Jelinek <jakub@redhat.com>
27795 PR tree-optimization/79340
27796 * tree-vect-loop.c (vectorizable_reduction): Release
27797 vec_defs elements after safe_splicing them into other vectors.
27800 PR tree-optimization/79327
27801 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
27802 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
27804 (format_integer): Use wide_int_to_tree instead of build_int_cst
27805 + to_?hwi. If argmin is NULL, just set argmin and argmax to
27806 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
27807 of shortest and longest sequence.
27809 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
27811 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
27812 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
27814 2017-02-03 Walter Lee <walt@tilera.com>
27817 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
27818 after initial stackframe link reg save.
27819 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
27821 2017-02-03 Jakub Jelinek <jakub@redhat.com>
27824 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
27825 wu for stxssp alternative.
27827 2017-02-03 Martin Sebor <msebor@redhat.com>
27829 PR tree-optimization/79352
27830 * gimple-fold.c (get_range_strlen): Add argument.
27831 (get_range_strlen): Change return type to bool.
27832 (get_maxval_strlen): Pass in a dummy argument.
27833 * gimple-fold.h (get_range_strlen): Change return type to bool.
27834 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
27835 * tree.h (array_at_struct_end_p): Add argument.
27836 * tree.c (array_at_struct_end_p): Handle it.
27838 2017-02-03 Martin Liska <mliska@suse.cz>
27841 * multiple_target.c (create_dispatcher_calls): Redirect edge
27842 from a caller of a dispatcher.
27843 (expand_target_clones): Make the clones local.
27844 (ipa_target_clone): Do both target clones and resolvers.
27845 (ipa_dispatcher_calls): Remove the pass.
27846 (pass_dispatcher_calls::gate): Likewise.
27847 (make_pass_dispatcher_calls): Likewise.
27848 * passes.def (pass_target_clone): Put as very first IPA early
27851 2017-02-03 Martin Liska <mliska@suse.cz>
27853 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
27854 in case of a function with ifunc attribute.
27856 2017-02-03 Martin Liska <mliska@suse.cz>
27858 * cgraph.c (cgraph_node::dump): Dump function version info.
27859 * symtab.c (symtab_node::dump_base): Add missing new line.
27861 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
27863 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
27864 (ifcombine_ifandif): Use it.
27866 2017-02-03 Martin Liska <mliska@suse.cz>
27868 * doc/invoke.texi: Document default value for
27869 use-after-scope-direct-emission-threshold.
27871 2017-02-03 Martin Liska <mliska@suse.cz>
27873 PR tree-optimization/79339
27874 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
27875 (format_floating): Likewise.
27877 2017-02-03 Martin Liska <mliska@suse.cz>
27880 * ipa-prop.c (ipa_node_params_t::insert): Remove current
27882 (ipa_node_params_t::remove): Likewise.
27883 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
27884 initialization from removed ipa_node_params_t::insert.
27885 (ipa_node_params::~ipa_node_params): Move from removed
27886 ipa_node_params_t::release.
27887 * symbol-summary.h (symbol_summary::m_released): New member.
27888 Do not release a summary twice. Do not allow to call finalizer
27889 for types of a summary that live in GGC memory.
27891 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27893 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
27896 2017-02-02 Martin Sebor <msebor@redhat.com>
27898 PR middle-end/79275
27899 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
27900 (format_string): Tighten up the range of output for non-constant
27901 strings and correct the expected range for wide non-constant strings.
27903 2017-02-02 Martin Sebor <msebor@redhat.com>
27905 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
27907 PR middle-end/32003
27908 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
27910 (-fdump-tree-@var): Add to index and document how to come up
27911 with pass-specific option and dump file names.
27912 (-fdump-passes): Clarify where to look for output.
27914 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
27916 PR middle-end/77445
27917 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
27918 statistics of the analyzed path; allow threading for speed when
27919 any of BBs along the path are optimized for speed.
27921 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
27923 PR middle-end/78468
27924 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
27925 settings of the virtual registers.
27928 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
27930 * explow.c (get_dynamic_stack_size): Take known alignment of stack
27931 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
27934 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27936 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
27937 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
27939 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27941 * config/s390/s390.md: Add missing comments with the expanded
27943 * config/s390/vector.md: Likewise.
27944 * config/s390/vx-builtins.md: Likewise.
27946 2017-02-02 Jakub Jelinek <jakub@redhat.com>
27949 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
27950 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
27951 conditions on a single line.
27953 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27955 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
27956 __S390_VX__ to __VX__.
27958 2017-02-01 Andrew Pinski <apinski@cavium.com>
27960 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
27961 stmt_info to record_stmt_cost.
27962 (vect_get_known_peeling_cost): Pass stmt_info if known to
27964 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
27965 cpu_vector_cost field into
27966 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
27967 field into vec_int_stmt_cost and vec_fp_stmt_cost.
27968 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
27969 splitting of scalar_stmt_cost and vec_stmt_cost.
27970 (thunderx_vector_cost): Likewise.
27971 (cortexa57_vector_cost): LIkewise.
27972 (exynosm1_vector_cost): Likewise.
27973 (xgene1_vector_cost): Likewise.
27974 (thunderx2t99_vector_cost): Improve after the splitting of the two
27976 (aarch64_builtin_vectorization_cost): Update for the splitting of
27977 scalar_stmt_cost and vec_stmt_cost.
27979 2017-02-01 Torvald Riegel <triegel@redhat.com>
27980 Richard Henderson <rth@redhat.com>
27982 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
27983 conditional on existance of a fast atomic load.
27984 * optabs-query.c (can_atomic_load_p): New function.
27985 * optabs-query.h (can_atomic_load_p): Declare it.
27986 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
27987 no fast atomic load is available for the particular size of access.
27988 (expand_atomic_compare_and_swap): Likewise.
27989 (expand_atomic_load): Likewise.
27990 (expand_atomic_store): Likewise.
27991 (expand_atomic_fetch_op): Likewise.
27992 * testsuite/lib/target-supports.exp
27993 (check_effective_target_sync_int_128): Remove x86 because it provides
27994 no fast atomic load.
27995 (check_effective_target_sync_int_128_runtime): Likewise.
27997 2017-02-01 Richard Biener <rguenther@suse.de>
27999 * graphite.c: Include tree-vectorizer.h for find_loop_location.
28000 (graphite_transform_loops): Provide opt-info for optimized nests.
28001 * tree-parloop.c (parallelize_loops): Provide opt-info for
28002 parallelized loops.
28004 2017-02-01 Richard Biener <rguenther@suse.de>
28006 PR middle-end/79315
28007 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
28008 was not set before.
28010 2017-02-01 Richard Biener <rguenther@suse.de>
28012 PR tree-optimization/71824
28013 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
28014 Verify the loops are valid in the merged SESE region.
28015 (scop_detection::can_represent_loop_1): Check analyzing the
28016 evolution of the number of iterations in the region succeeds.
28018 2017-01-31 Ian Lance Taylor <iant@golang.org>
28020 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
28021 REG_ARGS_SIZE note to 32-bit push insns and call insn.
28023 2017-01-31 David Malcolm <dmalcolm@redhat.com>
28025 PR preprocessor/79210
28026 * input.c (get_substring_ranges_for_loc): Replace line_width
28027 assertion with error-handling.
28029 2017-01-31 Richard Biener <rguenther@suse.de>
28031 PR tree-optimization/77318
28032 * graphite-sese-to-poly.c (extract_affine): Fix assert.
28033 (create_pw_aff_from_tree): Take loop parameter.
28034 (add_condition_to_pbb): Pass loop of the condition to
28035 create_pw_aff_from_tree.
28037 2017-01-31 Jakub Jelinek <jakub@redhat.com>
28039 * config/s390/s390.c (s390_asan_shadow_offset): New function.
28040 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
28042 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
28046 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
28048 (convert_int_to_float128): Likewise.
28049 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
28050 (convert_int_to_float128): Likewise.
28051 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
28052 (UNSPEC_IEEE128_CONVERT): Likewise.
28053 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
28054 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
28055 Use local variables for IBM extended format.
28056 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
28057 (fix_trunc<mode>si2_fprs): Likewise.
28058 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
28059 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
28060 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
28061 to know that we can now have integers of all sizes in vector
28063 (fix<uns>_<mode>di2_hw): Likewise.
28064 (float<uns>_<mode>si2_hw): Likewise.
28065 (fix_<mode>si2_hw): Likewise.
28066 (fixuns_<mode>si2_hw): Likewise.
28067 (float<uns>_<mode>di2_hw): Likewise.
28068 (float_<mode>di2_hw): Likewise.
28069 (float_<mode>si2_hw): Likewise.
28070 (floatuns_<mode>di2_hw): Likewise.
28071 (floatuns_<mode>si2_hw): Likewise.
28072 (xscvqp<su>wz_<mode>): Delete, no longer used.
28073 (xscvqp<su>dz_<mode>): Likewise.
28074 (xscv<su>dqp_<mode>): Likewise.
28075 (ieee128_mfvsrd_64bit): Likewise.
28076 (ieee128_mfvsrd_32bit): Likewise.
28077 (ieee128_mfvsrwz): Likewise.
28078 (ieee128_mtvsrw): Likewise.
28079 (ieee128_mtvsrd_64bit): Likewise.
28080 (ieee128_mtvsrd_32bit): Likewise.
28082 2017-01-31 Martin Liska <mliska@suse.cz>
28085 * ipa-prop.c (ipa_free_all_node_params): Call release method
28086 instead of ~sumbol_summary to not to trigger double times
28089 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
28091 PR tree-optimization/71691
28092 * bitmap.h (class auto_bitmap): New.
28093 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
28094 is_maybe_undefined instead of ssa_undefined_value_p.
28096 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28098 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
28099 __S390_ARCH_LEVEL__ to __ARCH__.
28101 2017-01-31 Jakub Jelinek <jakub@redhat.com>
28103 PR tree-optimization/79267
28104 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
28105 if should_remove_lhs_p is true.
28107 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
28110 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
28111 (add_alignment_attribute): New.
28112 (base_type_die): Add alignment attribute.
28113 (subrange_type_die): Likewise.
28114 (modified_type_die): Likewise.
28115 (gen_array_type_die): Likewise.
28116 (gen_descr_array_type_die: Likewise.
28117 (gen_enumeration_type_die): Likewise.
28118 (gen_subprogram_die): Likewise.
28119 (gen_variable_die): Likewise.
28120 (gen_field_die): Likewise.
28121 (gen_ptr_to_mbr_type_die): Likewise.
28122 (gen_struct_or_union_type_die): Likewise.
28123 (gen_subroutine_type_die): Likewise.
28124 (gen_typedef_die): Likewise.
28125 (base_type_cmp): Compare alignment attribute.
28127 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28130 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
28131 (setb_unsigned) New pattern for setb with CCUNS.
28132 * config/rs6000/rs6000.c (expand_block_compare): Use a different
28133 subfc./subfe sequence to avoid overflow problems. Generate a
28134 shorter sequence with cmpld/setb for power9.
28135 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
28136 for generating subfc. instruction.
28137 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
28138 now uses this instruction.
28140 2017-01-30 Ian Lance Taylor <iant@google.com>
28143 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
28144 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
28146 2017-01-30 Martin Sebor <msebor@redhat.com>
28148 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
28149 Move constant to the right of a relational operator.
28150 (get_mpfr_format_length, format_character, format_string): Ditto.
28151 (should_warn_p, maybe_warn): Same.
28153 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
28155 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
28158 * asan.c (get_translation_unit_decl): Remove function.
28159 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
28161 2017-01-30 Martin Liska <mliska@suse.cz>
28163 PR gcov-profile/79259
28164 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
28165 -fprofile-generate.
28167 2017-01-30 Martin Liska <mliska@suse.cz>
28170 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
28171 Initialize variables with NULL value.
28173 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
28176 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
28178 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
28180 2017-01-30 Richard Biener <rguenther@suse.de>
28182 PR tree-optimization/79276
28183 * tree-vrp.c (process_assert_insertions): Properly adjust common
28184 when removing a duplicate.
28186 * gcc.dg/torture/pr79276.c: New testcase.
28188 2017-01-30 Richard Biener <rguenther@suse.de>
28190 PR tree-optimization/79256
28191 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
28192 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
28194 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
28196 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
28199 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
28200 ("*r<noxa>sbg_<mode>_sll_bitmask")
28201 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
28202 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
28203 Use contiguous_bitmask_nowrap_operand.
28205 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28208 * config/rs6000/altivec.h (vec_xl): Revise #define.
28209 (vec_xst): Likewise.
28211 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
28213 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
28215 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
28217 PR rtl-optimization/79194
28218 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
28219 traps before call to bypass_conditional_jumps.
28221 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
28223 PR tree-optimization/71374
28224 * lra-constraints.c (check_conflict_input_operands): New.
28225 (match_reload): Use it.
28227 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
28230 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
28231 account to calculate conflict_set.
28233 2017-01-27 Bin Cheng <bin.cheng@arm.com>
28235 PR rtl-optimization/78559
28236 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
28237 other_insn in combine.
28239 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
28241 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
28242 uint16_type_node for BT_UINT16.
28244 2017-01-27 David Malcolm <dmalcolm@redhat.com>
28246 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
28247 "RTL Tests" to menu.
28248 (GIMPLE Tests): New node.
28249 (RTL Tests): New node.
28251 2017-01-27 Richard Biener <rguenther@suse.de>
28253 PR tree-optimization/79245
28254 * tree-loop-distribution.c (distribute_loop): Apply cost
28255 modeling also to detected patterns.
28257 2017-01-27 Richard Biener <rguenther@suse.de>
28259 PR tree-optimization/71433
28260 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
28261 (compare_assert_loc): New function.
28262 (process_assert_insertions): Sort and optimize assert locations
28263 to remove duplicates and push down identical assertions on
28264 edges to their destination block.
28266 2017-01-27 Richard Biener <rguenther@suse.de>
28268 PR tree-optimization/79244
28269 * tree-vrp.c (remove_range_assertions): Forcefully propagate
28270 out SSA names even if abnormal.
28272 2017-01-27 Jakub Jelinek <jakub@redhat.com>
28274 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
28275 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
28276 instead of MPFR_RNDN.
28278 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
28281 * arm.c (arm_option_override): Don't call build_target_option_node
28282 until after doing all option overrides.
28283 (arm_valid_target_attribute_tree): Likewise.
28285 2017-01-27 Martin Liska <mliska@suse.cz>
28287 * doc/invoke.texi (-fprofile-arcs): Document profiling support
28288 for {cd}tors and C++ {cd}tors.
28290 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
28292 * config/s390/s390.md ("*setmem_long_and")
28293 ("*setmem_long_and_31z"): Use zero_extend instead of and.
28295 2017-01-26 Martin Sebor <msebor@redhat.com>
28297 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
28300 2017-01-26 Martin Sebor <msebor@redhat.com>
28302 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
28303 HAVE_DFmode before using XFmode or DFmode.
28304 (parse_directive): Avoid using the z length modifier to avoid
28305 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
28307 PR middle-end/78703
28308 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
28309 to accept adjustment as an array.
28310 (get_int_range): New function.
28311 (struct directive): Make width and prec arrays.
28312 (directive::set_width, directive::set_precision): Call get_int_range.
28313 (format_integer, format_floating): Handle width and precision ranges.
28314 (format_string, parse_directive): Same.
28316 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28319 * dwarf2out.c (generate_skeleton_bottom_up): For children with
28320 comdat_type_p set, just clone them, but keep the children in the
28324 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
28325 which have direct callers with -fvar-tracking-assignments enabled
28327 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
28328 inside of type units.
28330 2017-01-26 Martin Sebor <msebor@redhat.com>
28332 PR middle-end/78703
28333 * gimple-ssa-sprintf.c (struct result_range): Add likely and
28335 (struct format_result): Replace number_chars, number_chars_min,
28336 and number_chars_max with a single member of struct result_range.
28338 (format_result::operator+=): Adjust.
28339 (struct fmtresult): Remove bounded. Handle likely and unlikely
28341 (fmtresult::adjust_for_width_or_precision): New function.
28342 (fmtresult:type_max_digits): New function.
28343 (bytes_remaining): Handle likely and unlikely counters.
28344 (min_bytes_remaining): Remove.
28345 (format_percent): Simplify.
28346 (format_integer, format_floating): Set likely and unlikely counters.
28347 (get_string_length, format_character, format_string): Same.
28348 (format_plain, should_warn_p): New function.
28349 (maybe_warn): Call should_warn_p. Update diagnostic messages
28350 and handle those for all directives, including plain strings.
28351 (format_directive): Handle likely and unlikely counters.
28352 Remove unnecessary quoting from diagnostics. Add an informational
28354 (add_bytes): Remove.
28355 (pass_sprintf_length::compute_format_length): Simplify.
28356 (try_substitute_return_value): Handle likely and unlikely counters.
28358 2017-01-26 Carl Love <cel@us.ibm.com>
28360 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
28361 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
28363 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
28366 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
28367 endianess for subregs into account.
28368 * lra-constraints.c (lra_constraints): Do risky transformations
28369 always on the first iteration.
28370 * lra-lives.c (check_pseudos_live_through_calls): Add arg
28371 last_call_used_reg_set.
28372 (process_bb_lives): Define and use last_call_used_reg_set.
28373 * lra.c (lra): Always continue after lra_constraints on the first
28376 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
28378 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
28380 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
28382 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28384 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
28385 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
28386 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
28387 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
28388 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
28389 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
28390 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
28391 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
28392 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
28394 2017-01-26 Marek Polacek <polacek@redhat.com>
28397 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
28398 for the third operand.
28400 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28402 PR middle-end/79236
28403 * omp-low.c (struct omp_context): Add simt_stmt field.
28404 (scan_omp_for): Return omp_context *.
28405 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
28406 context to the _simt_ SIMD stmt.
28407 (lower_omp_for): For combined SIMD with sibling _simt_
28408 SIMD, make sure to use the same decls in _looptemp_
28409 clauses as in the sibling.
28411 2017-01-26 David Sherwood <david.sherwood@arm.com>
28413 PR middle-end/79212
28414 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
28417 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28420 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
28421 emit fld b; fld a; if possible.
28423 * brig-builtins.def: Update copyright years.
28424 * config/arm/arm_acle_builtins.def: Update copyright years.
28426 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
28429 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
28430 constraint instead of o for the stxsd instruction.
28432 2017-01-25 Carl Love <cel@us.ibm.com>
28434 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
28435 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
28437 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
28439 * doc/invoke.texi (C++ Dialect Options): Fix typo.
28441 2017-01-25 Richard Biener <rguenther@suse.de>
28443 PR tree-optimization/69264
28444 * target.def (vector_alignment_reachable): Improve documentation.
28445 * doc/tm.texi: Regenerate.
28446 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
28448 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
28449 earlier changes with respect to TYPE_USER_ALIGN.
28450 (vector_alignment_reachable_p): Likewise. Improve dumping.
28452 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28455 * config/arm/arm.md (xordi3): Force constant operand into a register
28458 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28460 * doc/invoke.texi (-fstore-merging): Correct default optimization
28461 levels at which it is enabled.
28462 (-O): Move -fstore-merging from list to...
28465 2017-01-25 Richard Biener <rguenther@suse.de>
28468 * omp-expand.c: Include debug.h.
28469 (expand_omp_taskreg): Make sure to generate early debug before
28470 outlining anything from a function.
28471 (expand_omp_target): Likewise.
28472 (grid_expand_target_grid_body): Likewise.
28474 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
28477 * asan.c (get_translation_unit_decl): New function.
28478 (asan_add_global): Extract modules file name from globals
28479 TRANSLATION_UNIT_DECL name.
28481 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
28484 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
28485 for long calls with APCS frame and VFP.
28487 2017-01-24 David Malcolm <dmalcolm@redhat.com>
28489 * cfg.c (original_copy_tables_initialized_p): New function.
28490 * cfg.h (original_copy_tables_initialized_p): New decl.
28491 * cfgrtl.c (relink_block_chain): Guard the call to
28492 free_original_copy_tables with a call to
28493 original_copy_tables_initialized_p.
28494 * cgraph.h (symtab_node::native_rtl_p): New decl.
28495 * cgraphunit.c (symtab_node::native_rtl_p): New function.
28496 (symtab_node::needed_p): Don't assert for early assembly output
28497 for __RTL functions.
28498 (cgraph_node::finalize_function): Set "force_output" for __RTL
28500 (cgraph_node::analyze): Bail out early for __RTL functions.
28501 (analyze_functions): Update assertion to support __RTL functions.
28502 (cgraph_node::expand): Bail out early for __RTL functions.
28503 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
28505 * function.h (struct function): Update comment for field
28507 * gimple-expr.c: Include "tree-pass.h".
28508 (gimple_has_body_p): Return false for __RTL functions.
28509 * Makefile.in (OBJS): Add run-rtl-passes.o.
28510 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
28512 (gcc::pass_manager::get_clean_slate): New accessor.
28513 * passes.c: Include "insn-addr.h".
28514 (should_skip_pass_p): Add logging. Update logic for running
28515 "expand" to be compatible with both __GIMPLE and __RTL. Guard
28516 property-provider override so it is only done for gimple passes.
28518 (skip_pass): New function.
28519 (execute_one_pass): Call skip_pass when skipping passes.
28520 * read-md.c (md_reader::read_char): Support filtering
28521 the input to a subset of line numbers.
28522 (md_reader::md_reader): Initialize fields
28523 m_first_line and m_last_line.
28524 (md_reader::read_file_fragment): New function.
28525 * read-md.h (md_reader::read_file_fragment): New decl.
28526 (md_reader::m_first_line): New field.
28527 (md_reader::m_last_line): New field.
28528 * read-rtl-function.c (function_reader::create_function): Only
28529 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
28530 curr_properties. Set DECL_INITIAL to a dummy block.
28531 (read_rtl_function_body_from_file_range): New function.
28532 * read-rtl-function.h (read_rtl_function_body_from_file_range):
28534 * run-rtl-passes.c: New file.
28535 * run-rtl-passes.h: New file.
28537 2017-01-24 Jeff Law <law@redhat.com>
28539 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
28542 2017-01-24 Bin Cheng <bin.cheng@arm.com>
28544 PR tree-optimization/79159
28545 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
28546 (record_nonwrapping_iv): Improve boundary using above function if no
28547 value range information.
28549 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
28550 Martin Jambor <mjambor@suse.cz>
28552 * brig-builtins.def: New file.
28553 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
28554 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
28555 (DEF_HSAIL_SAT_BUILTIN): Likewise.
28556 (DEF_HSAIL_INTR_BUILTIN): Likewise.
28557 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
28558 * builtin-types.def (BT_INT8): New.
28559 (BT_INT16): Likewise.
28560 (BT_UINT8): Likewise.
28561 (BT_UINT16): Likewise.
28562 (BT_FN_ULONG): Likewise.
28563 (BT_FN_UINT_INT): Likewise.
28564 (BT_FN_UINT_ULONG): Likewise.
28565 (BT_FN_UINT_LONG): Likewise.
28566 (BT_FN_UINT_PTR): Likewise.
28567 (BT_FN_ULONG_PTR): Likewise.
28568 (BT_FN_INT8_FLOAT): Likewise.
28569 (BT_FN_INT16_FLOAT): Likewise.
28570 (BT_FN_UINT32_FLOAT): Likewise.
28571 (BT_FN_UINT16_FLOAT): Likewise.
28572 (BT_FN_UINT8_FLOAT): Likewise.
28573 (BT_FN_UINT64_FLOAT): Likewise.
28574 (BT_FN_UINT16_UINT32): Likewise.
28575 (BT_FN_UINT32_UINT16): Likewise.
28576 (BT_FN_UINT16_UINT16_UINT16): Likewise.
28577 (BT_FN_INT_PTR_INT): Likewise.
28578 (BT_FN_UINT_PTR_UINT): Likewise.
28579 (BT_FN_LONG_PTR_LONG): Likewise.
28580 (BT_FN_ULONG_PTR_ULONG): Likewise.
28581 (BT_FN_VOID_UINT64_UINT64): Likewise.
28582 (BT_FN_UINT8_UINT8_UINT8): Likewise.
28583 (BT_FN_INT8_INT8_INT8): Likewise.
28584 (BT_FN_INT16_INT16_INT16): Likewise.
28585 (BT_FN_INT_INT_INT): Likewise.
28586 (BT_FN_UINT_FLOAT_UINT): Likewise.
28587 (BT_FN_FLOAT_UINT_UINT): Likewise.
28588 (BT_FN_ULONG_UINT_UINT): Likewise.
28589 (BT_FN_ULONG_UINT_PTR): Likewise.
28590 (BT_FN_ULONG_ULONG_ULONG): Likewise.
28591 (BT_FN_UINT_UINT_UINT): Likewise.
28592 (BT_FN_VOID_UINT_PTR): Likewise.
28593 (BT_FN_UINT_UINT_PTR: Likewise.
28594 (BT_FN_UINT32_UINT64_PTR): Likewise.
28595 (BT_FN_INT_INT_UINT_UINT): Likewise.
28596 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
28597 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
28598 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
28599 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
28600 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
28601 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
28602 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
28603 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
28604 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
28605 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
28606 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
28607 * doc/frontends.texi: List BRIG FE.
28608 * doc/install.texi (Testing): Add BRIG tesring requirements.
28609 * doc/invoke.texi (Overall Options): Mention BRIG.
28610 * doc/standards.texi (Standards): Doucment BRIG HSA version.
28612 2017-01-24 Richard Biener <rguenther@suse.de>
28614 PR translation/79208
28615 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
28617 2017-01-24 Martin Jambor <mjambor@suse.cz>
28620 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
28621 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
28622 and known_contexts.
28624 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
28626 PR middle-end/79123
28627 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
28628 casts from signed to unsigned really don't have a range.
28630 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
28632 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
28633 GMP_RNDx for compatiblity.
28635 2017-01-24 Martin Liska <mliska@suse.cz>
28638 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
28639 that would prevent us to call alloca with -1 as argument.
28641 2017-01-24 Jakub Jelinek <jakub@redhat.com>
28643 * dwarf2out.c (output_compilation_unit_header, output_file_names):
28644 Avoid -Wformat-security warning.
28646 2017-01-23 Andrew Pinski <apinski@cavium.com>
28648 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
28651 2017-01-23 Martin Sebor <msebor@redhat.com>
28653 PR middle-end/78703
28654 * gimple-ssa-sprintf.c (warn_level): New global.
28655 (format_integer): Use it here and throughout the rest of the file.
28656 Use the same switch to compute sign as base.
28657 (maybe_warn): New function.
28658 (format_directive): Factor out warnings into maybe_warn.
28659 Add debugging output. Use warn_level.
28660 (add_bytes): Use warn_level.
28661 (pass_sprintf_length::compute_format_length): Add debugging output.
28662 (try_substitute_return_value): Same.
28663 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
28665 PR middle-end/78703
28666 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
28667 (struct fmtresult, format_integer, format_floating): Adjust.
28668 (fmtresult::fmtresult): Set max correctly in two argument ctor.
28669 (get_string_length, format_string,format_directive): Same.
28670 (pass_sprintf_length::compute_format_length): Same.
28671 (try_substitute_return_value): Simplify slightly.
28673 PR middle-end/78703
28674 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
28675 (fmtresult::operator+=): Outlined.
28676 (struct fmtresult): Add ctors.
28677 (struct conversion_spec): Rename...
28678 (struct directive): ...to this. Add and remove data members.
28679 (directive::set_width, directive::set_precision): New functions.
28680 (format_percent): Use fmtresult ctor.
28681 (get_width_and_precision): Remove.
28682 (format_integer): Make naming changes. Avoid computing width and
28684 (format_floating): Same. Adjust indentation.
28685 (format_character, format_none): New functions.
28686 (format_string): Moved character handling to format_character.
28687 (format_directive): Remove arguments, change return type.
28688 (parse_directive): New function.
28689 (pass_sprintf_length::compute_format_length): Move directive
28690 parsing to parse_directive.
28692 2017-01-23 Jakub Jelinek <jakub@redhat.com>
28694 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
28695 (assign_assembler_name_if_needed): ... this.
28696 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
28697 (assign_assembler_name_if_needed): ... this.
28698 (free_lang_data_in_cgraph): Adjust callers.
28699 * cgraphunit.c (cgraph_node::analyze): Likewise.
28700 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
28703 2017-01-23 Richard Biener <rguenther@suse.de>
28705 PR tree-optimization/79088
28706 PR tree-optimization/79188
28707 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
28708 resetting loop bounds after last path deletion. Reset loop
28709 bounds of the target loop, make code match the comments.
28710 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
28711 Make sure loops need no fixups.
28713 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28715 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
28716 exponent support with double type for first argument.
28717 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
28718 type returned by __builtin_vec_extract_sig,
28719 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
28720 functions from "vector int" to "vector unsigned int" or from
28721 "vector long long int" to "vector unsigned long long int".
28722 Changed type returned by __builtin_vec_extract_exp,
28723 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
28724 functions from "vector int" to "vector unsigned int" or from
28725 "vector long long int" to "vector unsigned long long int".
28726 Changed return type of __builtin_vec_test_data_class,
28727 __builtin_vec_test_data_class_sp, and
28728 __builtin_vec_test_data_class_dp from "vector int" to
28729 "vector bool int" or from "vector long long int" to "vector bool
28730 long long int" and changed second argument type from "unsigned
28731 int" to "int". Added new overloaded function forms "vector float
28732 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
28733 "vector float __builtin_vec_insert_exp_sp (vector float, vector
28734 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
28735 double, vector unsigned long long int)" and "vector double
28736 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
28737 long int)". Changed return type of
28738 __builtin_scalar_test_data_class and
28739 __builtin_scalar_test_data_class_sp and
28740 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
28741 int" and changed second argument from "unsigned int" to "int".
28742 Changed type returned by __builtin_scalar_test_neg,
28743 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
28744 from "int" to "bool int". Added new overloaded function form
28745 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
28746 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
28747 exponent double-precision with floating point first argument.
28748 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
28749 documentation of scalar_test_data_class, scalar_test_neg,
28750 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
28751 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
28752 vec_test_data_class built-in functions to reflect refinements in
28753 their type signatures.
28755 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
28757 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
28759 (aarch64_elf_asm_destructor): Likewise.
28761 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
28763 PR rtl-optimization/78634
28764 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
28765 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
28766 * ifcvt.c (noce_try_cmove): Add missing cost check.
28768 PR rtl-optimization/71724
28769 * combine.c (if_then_else_cond): Look for situations where it is
28770 beneficial to undo the work of one of the recursive calls.
28772 2017-01-23 Bin Cheng <bin.cheng@arm.com>
28774 PR tree-optimization/70754
28775 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
28776 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
28777 combined stmt before it if not NULL.
28778 (combine_chains): Process refs reversely and compute dominance point
28781 2017-01-23 Martin Liska <mliska@suse.cz>
28783 PR tree-optimization/79196
28784 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
28785 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
28787 (strlen_optimize_stmt): Call the renamed function.
28789 2017-01-23 Michael Matz <matz@suse.de>
28791 PR tree-optimization/78384
28792 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
28794 2017-01-23 Richard Biener <rguenther@suse.de>
28796 PR tree-optimization/79186
28797 * tree-vrp.c (register_new_assert_for): Make sure we've seen
28798 both incoming edges before moving an assert.
28800 2017-01-23 Martin Jambor <mjambor@suse.cz>
28802 * ipa-prop.c (load_from_param_1): Removed.
28803 (load_from_unmodified_param): Bits from load_from_param_1 put back
28805 (load_from_param): Removed.
28806 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
28807 with stmt. Reverted back to use of load_from_unmodified_param.
28809 2017-01-23 Martin Jambor <mjambor@suse.cz>
28812 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
28813 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
28814 field a pointer to garbage collected vector, mark lattices and
28815 ipcp_orig_node with GTY((skip)).
28816 (ipa_get_param_count): Adjust to descriptors being a pointer.
28817 (ipa_get_param): Likewise.
28818 (ipa_get_type): Likewise.
28819 (ipa_get_param_move_cost): Likewise.
28820 (ipa_set_param_used): Likewise.
28821 (ipa_get_controlled_uses): Likewise.
28822 (ipa_set_controlled_uses): Likewise.
28823 (ipa_is_param_used): Likewise.
28824 (ipa_node_params_t): Move into garbage collector. New methods insert
28826 (ipa_node_params_sum): Annotate wth GTY(()).
28827 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
28829 (ipa_load_from_parm_agg): Adjust declaration.
28830 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
28831 * ipa-profile.c (ipa_profile): Likewise.
28832 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
28833 (ipa_populate_param_decls): Make descriptors parameter garbage
28835 (ipa_dump_param): Adjust to descriptors being a pointer.
28836 (ipa_alloc_node_params): Likewise.
28837 (ipa_initialize_node_params): Likewise.
28838 (load_from_param_1): Make descriptors parameter garbage collected.
28839 (load_from_unmodified_param): Likewise.
28840 (load_from_param): Likewise.
28841 (ipa_load_from_parm_agg): Likewise.
28842 (ipa_node_params::~ipa_node_params): Removed.
28843 (ipa_free_all_node_params): Remove call to delete operator.
28844 (ipa_node_params_t::insert): New.
28845 (ipa_node_params_t::remove): Likewise.
28846 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
28847 copy known_csts and known_contexts vectors.
28848 (ipa_read_node_info): Adjust to descriptors being a pointer.
28849 (ipcp_modif_dom_walker): Make m_descriptors field garbage
28851 (ipcp_transform_function): Make descriptors variable garbage
28854 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
28856 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
28857 * config/i386/avx512dqintrin.h: Ditto.
28858 * config/i386/avx512fintrin.h: Ditto.
28859 * config/i386/i386.c: Handle new builtins.
28860 * config/i386/i386-builtin.def: Add new builtins.
28861 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
28862 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
28864 2017-01-23 Jakub Jelinek <jakub@redhat.com>
28865 Martin Liska <mliska@suse.cz>
28867 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
28868 * asan.c (asan_expand_poison_ifn): Support stores and use
28869 appropriate ASAN report function.
28870 * internal-fn.c (expand_ASAN_POISON_USE): New function.
28871 * internal-fn.def (ASAN_POISON_USE): Declare.
28872 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
28873 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
28874 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
28875 ASAN_POISON calls w/o LHS.
28876 * tree-ssa.c (execute_update_addresses_taken): Create clobber
28877 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
28878 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
28879 * gimplify.c (asan_poison_variables): Add attribute
28880 use_after_scope_memory to variables that really needs to live
28882 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
28883 having the attribute.
28885 2017-01-23 Martin Liska <mliska@suse.cz>
28887 * asan.c (create_asan_shadow_var): New function.
28888 (asan_expand_poison_ifn): Likewise.
28889 * asan.h (asan_expand_poison_ifn): New declaration.
28890 * internal-fn.c (expand_ASAN_POISON): Likewise.
28891 * internal-fn.def (ASAN_POISON): New builtin.
28892 * sanopt.c (pass_sanopt::execute): Expand
28893 asan_expand_poison_ifn.
28894 * tree-inline.c (copy_decl_for_dup_finish): Make function
28896 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
28897 * tree-ssa.c (is_asan_mark_p): New function.
28898 (execute_update_addresses_taken): Rewrite local variables
28899 (identified just by use-after-scope as addressable) into SSA.
28901 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
28903 * doc/install.texi (Specific): opensource.apple.com uses https
28904 now. Remove trailing slash.
28906 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
28908 * README.Portability: Remove note on an Irix compatibility issue.
28910 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
28912 * gcov.c (INCLUDE_ALGORITHM): Define.
28913 (INCLUDE_VECTOR): Define.
28914 No longer include <vector> and <algorithm> directly.
28916 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
28918 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
28920 * doc/invoke.texi (Code Gen Options): Ditto.
28922 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
28925 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
28927 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
28929 rtl-optimization/79125
28930 * cprop.c (local_cprop_pass): Handle cases where we make an
28931 unconditional trap.
28933 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
28937 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
28938 read from, for big endian.
28940 2017-01-20 Jiong Wang <jiong.wang@arm.com>
28942 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
28943 register pauth builtins for LP64 only.
28945 2017-01-20 Marek Polacek <polacek@redhat.com>
28948 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
28951 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
28953 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
28955 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
28956 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
28957 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
28959 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28962 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
28963 in big-endian mode when they are not a single duplicated value.
28965 2017-01-20 Richard Biener <rguenther@suse.de>
28967 * BASE-VER: Bump to 7.0.1.
28969 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
28971 * omp-low.c (omplow_simd_context): New struct. Use it...
28972 (lower_rec_simd_input_clauses): ...here and...
28973 (lower_rec_input_clauses): ...here to hold common data. Adjust all
28974 references to idx, lane, max_vf, is_simt.
28976 2017-01-20 Graham Markall <graham.markall@embecosm.com>
28978 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
28981 2017-01-20 Martin Jambor <mjambor@suse.cz>
28983 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
28984 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
28986 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
28987 (GTFILES): Rename hsa.c to hsa-common.c.
28988 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
28989 * hsa-dump.c: Likewise.
28990 * hsa-gen.c: Likewise.
28991 * hsa-regalloc.c: Likewise.
28992 * ipa-hsa.c: Likewise.
28993 * omp-expand.c: Likewise.
28994 * omp-low.c: Likewise.
28995 * toplev.c: Likewise.
28997 2017-01-20 Marek Polacek <polacek@redhat.com>
29000 * doc/invoke.texi: Document -Wduplicated-branches.
29001 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
29002 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
29003 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
29004 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
29005 return 0 only when not OEP_LEXICOGRAPHIC.
29006 (fold_build_cleanup_point_expr): Use the expression
29007 location when building CLEANUP_POINT_EXPR.
29008 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
29009 * tree.c (add_expr): Handle error_mark_node.
29011 2017-01-20 Martin Liska <mliska@suse.cz>
29014 * tree-profile.c (init_ic_make_global_vars): Do not call
29016 (gimple_init_gcov_profiler): Likewise.
29018 2017-01-20 Martin Liska <mliska@suse.cz>
29021 * cgraph.h (maybe_create_reference): Remove argument and
29023 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
29025 * ipa-cp.c (create_specialized_node): Likewise.
29026 * symtab.c (symtab_node::maybe_create_reference): Handle
29027 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
29029 2017-01-20 Martin Liska <mliska@suse.cz>
29031 * read-rtl-function.c (function_reader::create_function): Use
29032 build_decl instread of build_decl_stat.
29034 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
29036 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
29037 * config/i386/avx512dqintrin.h: Ditto.
29038 * config/i386/avx512fintrin.h: Ditto.
29039 * config/i386/i386-builtin-types.def: Add new types.
29040 * config/i386/i386.c: Handle new types.
29041 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
29042 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
29043 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
29044 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
29045 (__builtin_ia32_kshiftridi): New.
29046 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
29048 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
29052 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
29053 define to rs6000_init_stack_protect_guard.
29054 (rs6000_init_stack_protect_guard): New function.
29056 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
29057 Yunqiang Su <yunqiang.su@imgtec.com>
29059 * config.gcc (supported_defaults): Add madd4.
29060 (with_madd4): Add validation.
29061 (all_defaults): Add madd4.
29062 * config/mips/mips.opt (mmadd4): New option.
29063 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
29065 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
29067 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
29068 (ISA_HAS_FUSED_MADD4): Likewise.
29069 * doc/invoke.texi (-mmadd4): Document the new option.
29070 * doc/install.texi (--with-madd4): Document the new option.
29072 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29074 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
29075 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
29076 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
29077 (aarch64_init_pauth_hint_builtins): New.
29078 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
29079 (aarch64_expand_builtin): Expand new builtins.
29081 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29083 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
29084 * combine-stack-adj.c (no_unhandled_cfa): Handle
29085 REG_CFA_TOGGLE_RA_MANGLE.
29086 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
29087 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
29088 info for return address signing.
29089 (aarch64_expand_epilogue): Likewise.
29091 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29093 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
29094 * config/aarch64/aarch64-protos.h
29095 (aarch64_return_address_signing_enabled): New declaration.
29096 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
29098 (aarch64_expand_prologue): Sign return address before it's pushed onto
29100 (aarch64_expand_epilogue): Authenticate return address fetched from
29102 (aarch64_override_options): Sanity check for ILP32 and ISA level.
29103 (aarch64_attributes): New function attributes for "sign-return-address".
29104 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
29105 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
29106 ("*do_return"): Generate combined instructions according to key index.
29107 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
29108 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
29110 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
29111 * config/aarch64/aarch64.opt (msign-return-address=): New.
29112 * doc/extend.texi (AArch64 Function Attributes): Documents
29113 "sign-return-address=".
29114 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
29116 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
29118 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
29119 overall option summary.
29121 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29123 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
29124 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
29125 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
29126 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
29128 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
29130 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
29131 -mpower9-minmax by default for -mcpu=power9.
29132 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
29133 128-bit floating point.
29135 2017-01-20 Alan Modra <amodra@gmail.com>
29137 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
29138 optimizing for size.
29140 2017-01-20 Alan Modra <amodra@gmail.com>
29143 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
29144 for strcmp and strncmp from corresponding builtin decl.
29146 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
29148 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
29149 instead of i386/rtems-64.h.
29150 * config/i386/rtems-64.h: Remove.
29152 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
29156 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
29158 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
29160 2017-01-19 Tamar Christina <tamar.christina@arm.com>
29162 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
29163 Change int to HOST_WIDE_INT.
29164 * config/aarch64/aarch64-protos.h
29165 (aarch64_simd_gen_const_vector_dup): Likewise.
29166 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
29168 2017-01-19 David Malcolm <dmalcolm@redhat.com>
29170 * langhooks-def.h (lhd_type_for_size): New decl.
29171 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
29172 * langhooks.c (lhd_type_for_size): New function, taken from
29175 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
29177 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
29178 define_bypass for CR latency.
29179 (power9-cracked-alu): Update bypass latency and remove power9-branch.
29180 (power9-alu2): Add define_bypass for CR latency.
29182 (power9-mul): Update insn latency.
29183 (power9-mul-compare): Update insn latency, bypass latency and remove
29186 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29188 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
29190 * config/aarch64/aarch64.md
29191 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
29192 aarch64_nopcrelative_literal_loads.
29193 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
29195 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
29197 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
29198 TARGET_LOONGSON_3A.
29199 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
29201 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
29204 * config.gcc (supported_defaults): Add lxc1-sxc1.
29205 (with_lxc1_sxc1): Add validation.
29206 (all_defaults): Add lxc1-sxc1.
29207 * config/mips/mips.opt (mlxc1-sxc1): New option.
29208 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
29210 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
29211 __mips_no_lxc1_sxc1.
29212 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
29213 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
29214 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
29216 2017-01-19 Richard Biener <rguenther@suse.de>
29218 PR tree-optimization/72488
29219 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
29220 sure to restore SSA info.
29221 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
29223 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
29225 PR rtl-optimization/79121
29226 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
29227 of the inner type when shifting an extended value.
29229 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
29232 * symtab.c (symtab_node::equal_address_to): Fix comparing of
29233 interposable aliases.
29235 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
29238 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
29239 Use the evmergelohi instruction.
29240 (mov_si<mode>_e500_subreg4_2_le): Likewise.
29241 (mov_sitf_e500_subreg8_2_be): Likewise.
29242 (mov_sitf_e500_subreg12_2_le): Likewise.
29243 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
29244 (mov_si<mode>_e500_subreg4_2_be): Likewise.
29245 (mov_sitf_e500_subreg8_2_le): Likewise.
29246 (mov_sitf_e500_subreg12_2_be): Likewise.
29248 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29250 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
29251 attribute from vecsimple to vecperm.
29252 (altivec_vbpermq2): Likewise.
29254 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29257 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
29259 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29260 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
29261 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
29262 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
29263 case where N arg is SIZE_MAX.
29264 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
29265 (cmpstrsi): Add pattern.
29267 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29269 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
29270 __builtin_vec_revb builtins.
29271 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
29272 built-in functions to support generation of the ISA 3.0 XXBR<x>
29273 vector byte reverse instructions.
29274 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
29275 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
29276 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
29277 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
29278 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
29279 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
29280 (P9V_BUILTIN_VEC_REVB): Likewise.
29281 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
29282 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
29283 (p9_xxbrq_v16qi): Likewise.
29284 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
29285 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
29286 (p9_xxbrh_v8hi): Likewise.
29287 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
29288 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
29289 vec_revb built-in functions.
29291 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
29293 PR rtl-optimization/78952
29294 * config/i386/i386.md (any_extract): New code iterator.
29295 (*insvqi_2): Use any_extract for source operand.
29296 (*insvqi_3): Use any_shiftrt for source operand.
29298 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
29300 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
29302 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
29304 2017-01-18 Matthias Klose <doko@ubuntu.com>
29306 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
29308 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29310 * config/rs6000/altivec.h (vec_bperm): Change #define.
29311 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
29312 (altivec_vbpermq2): New define_insn.
29313 (altivec_vbpermd): Likewise.
29314 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
29315 function interface.
29316 (VBPERMD): Likewise.
29317 (VBPERM): New polymorphic function interface.
29318 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
29319 Add entries for P9V_BUILTIN_VEC_VBPERM.
29320 * doc/extend.texi: Add interfaces for vec_bperm.
29322 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29324 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
29325 first letter of error messages.
29326 (s390_resolve_overloaded_builtin): Likewise.
29327 * config/s390/s390.c (s390_expand_builtin): Likewise.
29328 (s390_invalid_arg_for_unprototyped_fn): Likewise.
29329 (s390_valid_target_attribute_inner_p): Likewise.
29330 * config/s390/s390.md ("tabort"): Likewise.
29332 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
29334 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
29335 (ISA_AVOID_DIV_HILO): New macro.
29336 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
29337 (ISA_HAS_DDIV): Likewise.
29339 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29341 * doc/invoke.texi (fabi-version): Correct number of occurrences.
29343 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29345 * doc/invoke.texi (fabi-version): Spelling fix.
29347 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29350 * doc/invoke.texi (fabi-version): Mention mangling fix for
29353 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29356 * doc/invoke.texi (fabi-version): Document discriminator mangling.
29358 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
29361 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
29362 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
29364 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
29366 (stack_protect_test): Ditto.
29367 * config/rs6000/rs6000.opt (mstack-protector-guard=,
29368 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
29370 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
29371 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
29372 -mstack-protector-guard-offset=.
29373 (RS/6000 and PowerPC Options): Ditto.
29375 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
29377 * config/i386/i386.h (MASK_CLASS_P): New define.
29378 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
29379 there are no registers from different register sets also when
29380 mask registers are used. Update function comment.
29381 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
29382 to (*k/*r) and (*k/*km) alternatives.
29384 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
29386 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
29387 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
29388 (EH_RETURN_HANDLER_RTX): New define.
29389 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
29390 Force frame pointer in EH return functions.
29391 (aarch64_expand_epilogue): Add barrier for eh_return.
29392 (aarch64_final_eh_return_addr): Remove.
29393 (aarch64_eh_return_handler_rtx): New function.
29394 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
29396 (aarch64_eh_return_handler_rtx): New prototype.
29398 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29400 * config/rs6000/altivec.h (vec_rlmi): New #define.
29401 (vec_vrlnm): Likewise.
29402 (vec_rlnm): Likewise.
29403 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
29404 (UNSPEC_VRLNM): Likewise.
29405 (VIlong): New mode iterator.
29406 (altivec_vrl<VI_char>mi): New define_insn.
29407 (altivec_vrl<VI_char>nm): Likewise.
29408 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
29410 (VRLDNM): Likewise.
29411 (RLNM): New polymorphic function entry.
29412 (VRLWMI): New monomorphic function entry.
29413 (VRLDMI): Likewise.
29414 (RLMI): New polymorphic function entry.
29415 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
29416 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
29417 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
29420 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29423 * dwarf2out.c (field_byte_offset): Restore the
29424 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
29425 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
29426 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
29429 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
29432 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
29434 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29437 * dwarf2out.c (add_data_member_location_attribute): For constant
29438 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
29439 instead of DW_AT_data_member_location, DW_AT_bit_offset and
29440 DW_AT_byte_size attributes.
29442 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
29444 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
29445 after forcing to constant memory when the code model is medium.
29447 2017-01-17 Julia Koval <julia.koval@intel.com>
29450 * config/i386/avx512fintrin.h
29451 (_mm512_i32gather_ps): Change __addr type to void const*.
29452 (_mm512_mask_i32gather_ps): Ditto.
29453 (_mm512_i32gather_pd): Ditto.
29454 (_mm512_mask_i32gather_pd): Ditto.
29455 (_mm512_i64gather_ps): Ditto.
29456 (_mm512_mask_i64gather_ps): Ditto.
29457 (_mm512_i64gather_pd): Ditto.
29458 (_mm512_mask_i64gather_pd): Ditto.
29459 (_mm512_i32gather_epi32): Ditto.
29460 (_mm512_mask_i32gather_epi32): Ditto.
29461 (_mm512_i32gather_epi64): Ditto.
29462 (_mm512_mask_i32gather_epi64): Ditto.
29463 (_mm512_i64gather_epi32): Ditto.
29464 (_mm512_mask_i64gather_epi32): Ditto.
29465 (_mm512_i64gather_epi64): Ditto.
29466 (_mm512_mask_i64gather_epi64): Ditto.
29467 (_mm512_i32scatter_ps): Change __addr type to void*.
29468 (_mm512_mask_i32scatter_ps): Ditto.
29469 (_mm512_i32scatter_pd): Ditto.
29470 (_mm512_mask_i32scatter_pd): Ditto.
29471 (_mm512_i64scatter_ps): Ditto.
29472 (_mm512_mask_i64scatter_ps): Ditto.
29473 (_mm512_i64scatter_pd): Ditto.
29474 (_mm512_mask_i64scatter_pd): Ditto.
29475 (_mm512_i32scatter_epi32): Ditto.
29476 (_mm512_mask_i32scatter_epi32): Ditto.
29477 (_mm512_i32scatter_epi64): Ditto.
29478 (_mm512_mask_i32scatter_epi64): Ditto.
29479 (_mm512_i64scatter_epi32): Ditto.
29480 (_mm512_mask_i64scatter_epi32): Ditto.
29481 (_mm512_i64scatter_epi64): Ditto.
29482 (_mm512_mask_i64scatter_epi64): Ditto.
29483 * config/i386/avx512pfintrin.h
29484 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
29485 (_mm512_mask_prefetch_i32gather_ps): Ditto.
29486 (_mm512_mask_prefetch_i64gather_pd): Ditto.
29487 (_mm512_mask_prefetch_i64gather_ps): Ditto.
29488 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
29489 (_mm512_prefetch_i32scatter_ps): Ditto.
29490 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
29491 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
29492 (_mm512_prefetch_i64scatter_pd): Ditto.
29493 (_mm512_prefetch_i64scatter_ps): Ditto.
29494 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
29495 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
29496 * config/i386/avx512vlintrin.h
29497 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
29498 (_mm_mmask_i32gather_ps): Ditto.
29499 (_mm256_mmask_i32gather_pd): Ditto.
29500 (_mm_mmask_i32gather_pd): Ditto.
29501 (_mm256_mmask_i64gather_ps): Ditto.
29502 (_mm_mmask_i64gather_ps): Ditto.
29503 (_mm256_mmask_i64gather_pd): Ditto.
29504 (_mm_mmask_i64gather_pd): Ditto.
29505 (_mm256_mmask_i32gather_epi32): Ditto.
29506 (_mm_mmask_i32gather_epi32): Ditto.
29507 (_mm256_mmask_i32gather_epi64): Ditto.
29508 (_mm_mmask_i32gather_epi64): Ditto.
29509 (_mm256_mmask_i64gather_epi32): Ditto.
29510 (_mm_mmask_i64gather_epi32): Ditto.
29511 (_mm256_mmask_i64gather_epi64): Ditto.
29512 (_mm_mmask_i64gather_epi64): Ditto.
29513 (_mm256_i32scatter_ps): Change __addr type to void*.
29514 (_mm256_mask_i32scatter_ps): Ditto.
29515 (_mm_i32scatter_ps): Ditto.
29516 (_mm_mask_i32scatter_ps): Ditto.
29517 (_mm256_i32scatter_pd): Ditto.
29518 (_mm256_mask_i32scatter_pd): Ditto.
29519 (_mm_i32scatter_pd): Ditto.
29520 (_mm_mask_i32scatter_pd): Ditto.
29521 (_mm256_i64scatter_ps): Ditto.
29522 (_mm256_mask_i64scatter_ps): Ditto.
29523 (_mm_i64scatter_ps): Ditto.
29524 (_mm_mask_i64scatter_ps): Ditto.
29525 (_mm256_i64scatter_pd): Ditto.
29526 (_mm256_mask_i64scatter_pd): Ditto.
29527 (_mm_i64scatter_pd): Ditto.
29528 (_mm_mask_i64scatter_pd): Ditto.
29529 (_mm256_i32scatter_epi32): Ditto.
29530 (_mm256_mask_i32scatter_epi32): Ditto.
29531 (_mm_i32scatter_epi32): Ditto.
29532 (_mm_mask_i32scatter_epi32): Ditto.
29533 (_mm256_i32scatter_epi64): Ditto.
29534 (_mm256_mask_i32scatter_epi64): Ditto.
29535 (_mm_i32scatter_epi64): Ditto.
29536 (_mm_mask_i32scatter_epi64): Ditto.
29537 (_mm256_i64scatter_epi32): Ditto.
29538 (_mm256_mask_i64scatter_epi32): Ditto.
29539 (_mm_i64scatter_epi32): Ditto.
29540 (_mm_mask_i64scatter_epi32): Ditto.
29541 (_mm256_i64scatter_epi64): Ditto.
29542 (_mm256_mask_i64scatter_epi64): Ditto.
29543 (_mm_i64scatter_epi64): Ditto.
29544 (_mm_mask_i64scatter_epi64): Ditto.
29545 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
29546 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
29547 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
29548 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
29549 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
29550 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
29551 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
29552 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
29553 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
29554 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
29555 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
29556 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
29557 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
29558 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
29559 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
29560 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
29561 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
29562 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
29563 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
29564 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
29565 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
29566 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
29567 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
29568 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
29569 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
29570 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
29571 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
29572 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
29573 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
29574 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
29575 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
29576 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
29577 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
29578 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
29579 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
29580 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
29581 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
29582 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
29583 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
29584 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
29585 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
29586 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
29587 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
29588 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
29589 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
29590 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
29591 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
29592 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
29593 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
29594 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
29595 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
29596 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
29597 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
29598 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
29599 definitions accordingly.
29601 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
29602 Kuan-Lin Chen <kuanlinchentw@gmail.com>
29605 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
29608 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
29611 * ira-conflicts.c (ira_build_conflicts): Update total conflict
29612 hard regs for inner regno.
29614 2017-01-17 Martin Liska <mliska@suse.cz>
29617 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
29618 assumption and add comment.
29620 2017-01-17 Nathan Sidwell <nathan@acm.org>
29622 * ipa-visibility.c (localize_node): New function, broken out of ...
29623 (function_and_variable_visibility): ... here. Call it.
29625 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
29627 PR middle-end/77445
29628 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
29629 correctly set frequency of oudgoing edge.
29630 (duplicate_thread_path): Fix profile updating.
29632 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29635 * configure.ac: Add GCC_BASE_VER.
29636 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
29637 version from BASE-VER file.
29638 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
29639 (gcc.o): Depend on $(BASEVER).
29640 * common.opt (dumpfullversion): New option.
29641 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
29642 * doc/invoke.texi: Document -dumpfullversion.
29643 * doc/install.texi: Document --with-gcc-major-version-only.
29644 * configure: Regenerated.
29646 2017-01-17 Richard Biener <rguenther@suse.de>
29648 PR tree-optimization/71433
29649 * tree-vrp.c (register_new_assert_for): Merge same asserts
29650 on all incoming edges.
29651 (process_assert_insertions_for): Handle insertions at the
29654 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
29656 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
29657 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
29659 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
29662 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
29665 2017-01-17 Alan Modra <amodra@gmail.com>
29668 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
29669 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
29670 symbolic stack limit when pic.
29672 2017-01-16 Martin Sebor <msebor@redhat.com>
29674 PR tree-optimization/78608
29675 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
29677 2017-01-16 Jeff Law <law@redhat.com>
29680 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
29681 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
29682 for several include directories that may be relative to sysroot.
29683 * config/i386/x-mingw32 (gplus_includedir): Define.
29684 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
29685 (native_system_includedir): Likewise.
29686 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
29687 override if TARGET_SYSTEM_ROOT is defined.
29688 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
29690 PR tree-optimization/79090
29691 PR tree-optimization/33562
29692 PR tree-optimization/61912
29693 PR tree-optimization/77485
29694 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
29695 and computed trims into the dump file.
29697 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
29699 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
29701 2017-01-16 Jakub Jelinek <jakub@redhat.com>
29704 * gimplify.c (gimplify_init_constructor): If want_value and
29705 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
29709 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
29710 sequence. Formatting fixes.
29711 (doloop_optimize): Formatting fixes.
29714 * gcc.c (debug_level_greater_than_spec_func): New function.
29715 (static_spec_functions): Add debug-level-gt spec function.
29716 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
29718 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
29719 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
29720 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
29721 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
29722 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
29723 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
29725 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
29727 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
29728 QImode fixups to general and mask registers only.
29730 2017-01-16 Carl Love <cel@us.ibm.com>
29732 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
29733 for built-in functions
29734 vector signed char vec_nabs (vector signed char)
29735 vector signed short vec_nabs (vector signed short)
29736 vector signed int vec_nabs (vector signed int)
29737 vector signed long long vec_nabs (vector signed long long)
29738 vector float vec_nabs (vector float)
29739 vector double vec_nabs (vector double)
29740 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
29742 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
29743 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
29744 * doc/extend.texi: Update the documentation file for the new built-in
29747 2017-01-16 Martin Sebor <msebor@redhat.com>
29749 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
29752 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29754 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
29755 UNSPEC_VSX__XXSPLTD to require special splat handling.
29757 2017-01-16 David Malcolm <dmalcolm@redhat.com>
29760 * system.h: Poison strndup.
29762 2017-01-16 Alan Modra <amodra@gmail.com>
29765 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
29768 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
29770 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
29772 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
29774 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
29775 call recog here. Assert that INSN_CODE (insn) is non-negative.
29777 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
29780 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
29782 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
29783 in the currently scheduled RTL fragment.
29785 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
29787 PR rtl-optimization/78751
29788 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
29791 2017-01-14 Jeff Law <law@redhat.com>
29793 PR tree-optimization/79090
29794 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
29795 variable length stores.
29796 (compute_trims): Delete dead assignment to *trim_tail.
29797 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
29800 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
29802 PR rtl-optimization/78626
29803 PR rtl-optimization/78727
29804 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
29805 of a block, and split such blocks after everything else is finished.
29807 2017-01-14 Alan Modra <amodra@gmail.com>
29810 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
29811 target legitimate_combined_insn.
29812 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
29813 (rs6000_legitimate_combined_insn): New function.
29814 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
29816 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
29817 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
29818 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
29820 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
29822 * doc/frontends.texi (G++ and GCC): Remove references to Java.
29824 2017-01-13 Jeff Law <law@redhat.com>
29826 PR tree-optimization/33562
29827 PR tree-optimization/61912
29828 PR tree-optimization/77485
29829 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
29831 (delete_dead_assignment): Likewise.
29832 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
29833 statement to delete_dead_call and delete_dead_assignment.
29835 2017-01-13 David Malcolm <dmalcolm@redhat.com>
29838 * substring-locations.c (format_warning_va): Strengthen case 1 so
29839 that both endpoints of the substring must be within the format
29840 range for just the substring to be printed.
29842 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
29844 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
29845 * config/i386/i386.c (ix86_target_string): Add missing options
29846 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
29847 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
29848 flags_other and ix86_target_other to flags2_other. Display unknown
29850 (ix86_valid_target_attribute_inner_p): Add missing options and
29851 reorder options by implied ISAs, as in ix86_target_string.
29853 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
29855 * hash-table.h (hash_table::too_empty_p): New function.
29856 (hash_table::expand): Use it.
29857 (hash_table::traverse): Likewise.
29858 (hash_table::empty_slot): Use sizeof (value_type) instead of
29859 sizeof (PTR) to convert bytes to elements. Shrink the table
29860 if the current size is excessive for the current number of
29863 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
29865 * ira-costs.c (record_reg_classes): Break from the inner loop
29866 early once alt_fail is known to be true. Update outer loop
29867 handling accordingly.
29869 2017-01-13 Jeff Law <law@redhat.com>
29871 * tree-ssa-dse.c (decrement_count): New function.
29872 (increment_start_addr, maybe_trim_memstar_call): Likewise.
29873 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
29874 when we know the partially dead statement is a mem* function.
29876 PR tree-optimization/61912
29877 PR tree-optimization/77485
29878 * tree-ssa-dse.c: Include expr.h.
29879 (maybe_trim_constructor_store): New function.
29880 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
29882 PR tree-optimization/33562
29883 PR tree-optimization/61912
29884 PR tree-optimization/77485
29885 * doc/invoke.texi: Document new dse-max-object-size param.
29886 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
29887 * tree-ssa-dse.c: Include params.h.
29888 (dse_store_status): New enum.
29889 (initialize_ao_ref_for_dse): New, partially extracted from
29891 (valid_ao_ref_for_dse, normalize_ref): New.
29892 (setup_live_bytes_from_ref, compute_trims): Likewise.
29893 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
29894 (maybe_trim_partially_dead_store): Likewise.
29895 (maybe_trim_complex_store): Likewise.
29896 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
29897 Track what bytes live from the original store. Return tri-state
29898 for dead, partially dead or live.
29899 (dse_dom_walker): Add constructor, destructor and new private members.
29900 (delete_dead_call, delete_dead_assignment): New extracted from
29902 (dse_optimize_stmt): Make a member of dse_dom_walker.
29903 Use initialize_ao_ref_for_dse.
29905 PR tree-optimization/33562
29906 PR tree-optimization/61912
29907 PR tree-optimization/77485
29908 * sbitmap.h (bitmap_count_bits): Prototype.
29909 (bitmap_clear_range, bitmap_set_range): Likewise.
29910 * sbitmap.c (bitmap_clear_range): New function.
29911 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
29913 2017-01-13 Martin Liska <mliska@suse.cz>
29916 * function.c (set_cfun): Add new argument force.
29917 * function.h (set_cfun): Likewise.
29918 * ipa-inline-transform.c (inline_call): Use the function when
29919 strict alising from is dropped for function we inline to.
29921 2017-01-13 Richard Biener <rguenther@suse.de>
29923 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
29924 for dumping GIMPLE INTEGER_CSTs.
29926 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29928 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
29929 to 201112L since C++17.
29931 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
29934 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
29935 if -fsanitize=kernel-address is present.
29937 2017-01-13 Richard Biener <rguenther@suse.de>
29939 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
29940 as _Literal ( type ) number in case usual suffixes do not
29941 preserve all information.
29943 2017-01-13 Richard Biener <rguenther@suse.de>
29945 PR tree-optimization/77283
29946 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
29947 and ssa-iterators.h.
29948 (is_feasible_trace): Implement a cost model based on joiner
29951 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
29954 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
29955 char or short to __float128/_Float128 directly.
29957 2017-01-12 Martin Sebor <msebor@redhat.com>
29959 to -Wformat-overflow.
29960 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
29961 (min_bytes_remaining): Same.
29962 (get_string_length): Same.
29963 (format_string): Same.
29964 (format_directive): Same.
29966 (pass_sprintf_length::handle_gimple_call): Same.
29968 2017-01-12 Jakub Jelinek <jakub@redhat.com>
29970 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
29971 info.nowrite calls with no lhs that can't throw. Return bool
29972 whether gsi_remove has been called or not.
29973 (pass_sprintf_length::handle_gimple_call): Return bool whether
29974 try_substitute_return_value called gsi_remove. Formatting fix.
29975 (pass_sprintf_length::execute): Don't use gsi_remove if
29976 handle_gimple_call returned true.
29979 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
29980 be removed due to side-effects, don't remove following barrier nor
29981 turn the successor edge into fallthru edge.
29983 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29986 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
29987 element-reversing loads and stores as not swappable.
29989 2017-01-12 Nathan Sidwell <nathan@acm.org>
29990 Nicolai Stange <nicstange@gmail.com>
29992 * combine.c (try_combine): Don't ignore result of overlap checking
29993 loop. Combine overlap & asm check into single loop.
29995 2017-01-12 Richard Biener <rguenther@suse.de>
29997 * tree-pretty-print.c (dump_generic_node): Provide -gimple
29998 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
30000 2017-01-12 Richard Biener <rguenther@suse.de>
30002 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
30003 and TS_TARGET_OPTION directly derive from TS_BASE.
30004 * tree-core.h (tree_optimization_option): Derive from tree_base.
30005 (tree_target_option): Likewise.
30007 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
30009 * config/i386/i386.c (memory_address_length): Increase len
30010 only when rip_relative_addr_p returns false.
30012 2017-01-11 Julia Koval <julia.koval@intel.com>
30014 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
30015 (OPTION_MASK_ISA_SGX_SET): New.
30016 (ix86_handle_option): Handle OPT_msgx.
30017 * config.gcc: Added sgxintrin.h.
30018 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
30019 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
30020 * config/i386/i386.c (ix86_target_string): Add -msgx.
30022 (ix86_option_override_internal): Handle new options.
30023 (ix86_valid_target_attribute_inner_p): Add sgx.
30024 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
30025 * config/i386/i386.opt: Add msgx.
30026 * config/i386/sgxintrin.h: New file.
30027 * config/i386/x86intrin.h: Add sgxintrin.h.
30029 2017-01-11 Jakub Jelinek <jakub@redhat.com>
30032 * fold-const.c (maybe_nonzero_address): Return 1 for function
30034 (tree_single_nonzero_warnv_p): Don't handle function local objects
30038 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
30041 2017-01-11 David Malcolm <dmalcolm@redhat.com>
30044 * opts.c: Include "spellcheck.h"
30045 (struct string_fragment): New struct.
30046 (struct edit_distance_traits<const string_fragment &>): New
30048 (get_closest_sanitizer_option): New function.
30049 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
30051 2017-01-11 Jakub Jelinek <jakub@redhat.com>
30053 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
30055 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
30056 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
30057 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
30058 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
30059 for initial die_offset if dwarf_split_debug_info.
30060 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
30061 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
30063 (output_skeleton_debug_sections): Formatting fix. Use
30064 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
30065 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
30067 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
30069 * config/arm/cortex-a53.md: Add bypasses for
30070 cortex_a53_r2f_cvt.
30071 (cortex_a53_r2f): Only use for transfers.
30072 (cortex_a53_f2r): Likewise.
30073 (cortex_a53_r2f_cvt): Add reservation for conversions.
30074 (cortex_a53_f2r_cvt): Likewise.
30076 2017-01-11 Tamar Christina <tamar.christina@arm.com>
30078 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
30079 to all inlined functions, change static to extern.
30081 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
30084 * config/arm/arm.c (legitimize_pic_address): Handle reference to
30086 (arm_assemble_integer): Likewise.
30088 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
30090 * config.gcc: Use new awk script to check CPU, FPU and architecture
30091 parameters for --with-... options.
30092 * config/arm/parsecpu.awk: New file
30093 * config/arm/arm-cpus.in: New file.
30094 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
30096 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
30098 * config/arm/t-arm: Update dependency rules.
30099 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
30100 of processing .def files.
30101 * config/arm/genopt.sh: Deleted.
30102 * config/arm/gentune.sh: Deleted.
30103 * config/arm/arm-cores.def: Deleted.
30104 * config/arm/arm-arches.def: Deleted.
30105 * config/arm/arm-fpus.def: Deleted.
30106 * config/arm/arm-tune.md: Regenerated.
30107 * config/arm/arm-tables.opt: Regenerated.
30108 * config/arm/arm-cpu.h: New generated file.
30109 * config/arm/arm-cpu-data.h: New generated file.
30110 * config/arm/arm-cpu-cdata.h: New generated file.
30112 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
30115 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
30117 (input_varpool_node): Unpack dynamically_initialized bit.
30119 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
30121 PR rtl-optimization/79032
30122 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
30123 the alignment of the adjusted memory reference against that of MODE,
30124 instead of the alignment of the original memory reference.
30126 2017-01-11 Martin Jambor <mjambor@suse.cz>
30128 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
30130 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
30131 decorated functions.
30133 2017-01-11 Richard Biener <rguenther@suse.de>
30135 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
30136 set range/nonnull info for PHI results. Do not set it on
30137 stmts marked for removal.
30139 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
30141 * expr.c (store_field): In the bitfield case, fetch the return value
30142 from the registers before applying a single big-endian adjustment.
30143 Always do a final load for a BLKmode value not larger than a word.
30145 2017-01-10 David Malcolm <dmalcolm@redhat.com>
30148 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
30149 that we correctly handle column numbers greater than
30150 LINE_MAP_MAX_COLUMN_NUMBER.
30152 2017-01-10 Martin Sebor <msebor@redhat.com>
30154 PR middle-end/78245
30155 * gimple-ssa-sprintf.c (get_destination_size): Call
30156 {init,fini}object_sizes.
30157 * tree-object-size.c (addr_object_size): Adjust.
30158 (pass_through_call): Adjust.
30159 (pass_object_sizes::execute): Adjust.
30160 * tree-object-size.h (fini_object_sizes): Declare.
30162 2017-01-10 Martin Sebor <msebor@redhat.com>
30164 PR tree-optimization/78775
30165 * builtins.c (get_size_range): Move...
30166 * calls.c: ...to here.
30167 (alloc_max_size): Accept zero argument.
30168 (operand_signed_p): Remove.
30169 (maybe_warn_alloc_args_overflow): Call get_size_range.
30170 * calls.h (get_size_range): Declare.
30172 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
30174 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
30175 from TI's devices.csv file as of September 2016.
30176 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
30178 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
30180 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
30181 * doc/invoke.texi: Likewise.
30182 * doc/md.texi: Likewise.
30183 * doc/objc.texi: Likewise.
30185 2017-01-10 Joshua Conner <joshconner@google.com>
30187 * config/arm/fuchsia-elf.h: New file.
30188 * config/fuchsia.h: New file.
30189 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
30190 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
30192 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
30194 2016-01-10 Richard Biener <rguenther@suse.de>
30196 PR tree-optimization/79034
30197 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
30198 Propagate out degenerate PHIs in the joiner.
30200 2017-01-10 Martin Liska <mliska@suse.cz>
30202 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
30203 (sort_congruence_classes_by_decl_uid): Likewise.
30204 (sort_congruence_class_groups_by_decl_uid): Likewise.
30205 (sem_item_optimizer::merge_classes): Sort class, groups in these
30206 classes and members in the groups by DECL_UID of declarations.
30207 This would make merge operations stable.
30209 2017-01-10 Martin Liska <mliska@suse.cz>
30211 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
30212 usage of m_classes_vec.
30213 (sem_item_optimizer::~sem_item_optimizer): Likewise.
30214 (sem_item_optimizer::get_group_by_hash): Likewise.
30215 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
30216 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
30217 (sem_item_optimizer::verify_classes): Likewise.
30218 (sem_item_optimizer::process_cong_reduction): Likewise.
30219 (sem_item_optimizer::dump_cong_classes): Likewise.
30220 (sem_item_optimizer::merge_classes): Likewise.
30221 * ipa-icf.h (congruence_class_hash): Rename from
30222 congruence_class_group_hash. Remove declaration of m_classes_vec.
30224 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
30226 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
30227 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
30228 * config.gcc: Add avx512vpopcntdqintrin.h.
30229 * config/i386/avx512vpopcntdqintrin.h: New.
30230 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
30231 * config/i386/i386-builtin-types.def: Add new types.
30232 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
30233 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
30234 __builtin_ia32_vpopcountq_v8di_mask): New.
30235 * config/i386/i386-c.c (ix86_target_macros_internal): Define
30236 __AVX512VPOPCNTDQ__.
30237 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
30238 (PTA_AVX512VPOPCNTDQ): Define.
30239 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
30240 TARGET_AVX512VPOPCNTDQ_P): Define.
30241 * config/i386/i386.opt: Add mavx512vpopcntdq.
30242 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
30243 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
30245 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30247 PR middle-end/77484
30248 * predict.def (PRED_CALL): Set to 67.
30250 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
30252 * expr.c (store_field): In the bitfield case, if the value comes from
30253 a function call and is of an aggregate type returned in registers, do
30254 not modify the field mode; extract the value in all cases if the mode
30255 is BLKmode and the size is not larger than a word.
30257 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
30260 * config/i386/cpuid.h: Fix undefined behavior.
30262 2017-01-04 Jeff Law <law@redhat.com>
30264 PR tree-optimization/79007
30265 PR tree-optimization/67955
30266 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
30267 conservative for pt.null when flag_non_call_exceptions is on.
30269 2017-01-09 Jakub Jelinek <jakub@redhat.com>
30271 PR translation/79019
30272 PR translation/79020
30273 * params.def (PARAM_INLINE_MIN_SPEEDUP,
30274 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
30275 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
30277 * config/avr/avr.opt (maccumulate-args): Likewise.
30278 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
30279 * common.opt (freport-bug): Likewise.
30280 * cif-code.def (CIF_FINAL_ERROR): Likewise.
30281 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
30282 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
30283 translatable string.
30284 * config/i386/i386.c (function_value_32): Likewise.
30285 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
30286 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
30288 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
30289 * common/config/msp430/msp430-common.c (msp430_handle_option):
30291 * symtab.c (symtab_node::verify_base): Likewise.
30292 * opts.c (set_debug_level): Likewise.
30293 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
30294 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
30295 missing whitespace to translatable strings.
30296 * config/avr/avr.md (bswapsi2): Fix typo in comment.
30297 * config/sh/superh.h: Likewise.
30298 * config/i386/xopintrin.h: Likewise.
30299 * config/i386/znver1.md: Likewise.
30300 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
30301 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
30302 * double-int.h (struct double_int): Likewise.
30303 * double-int.c (div_and_round_double): Likewise.
30304 * wide-int.cc: Likewise.
30305 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
30306 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
30307 * cfgcleanup.c (crossjumps_occured): Renamed to ...
30308 (crossjumps_occurred): ... this.
30309 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
30312 PR tree-optimization/78899
30313 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
30314 returning bool return struct loop *, NULL for failure and the new
30316 (versionable_outer_loop_p): Don't version outer loop if it has
30317 dont_vectorized bit set.
30318 (tree_if_conversion): When versioning outer loop, ensure
30319 tree_if_conversion is performed also on the inner loop of the
30320 non-vectorizable outer loop copy.
30321 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
30322 LOOP_VECTORIZED in inner loop of the scalar outer loop and
30323 prevent vectorization of it.
30324 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
30325 the outer loop vectorization of the non-scalar version is attempted
30326 before vectorization of the inner loop in scalar version. If
30327 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
30328 vectorization of its inner loop.
30329 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
30330 has 2 inner loops, rename also on edges from bb whose single pred
30331 is outer_loop->header. Fix typo in function comment.
30333 2017-01-09 Martin Sebor <msebor@redhat.com>
30336 * asan.c (asan_emit_stack_protection): Increase local buffer size
30337 to avoid snprintf truncation warning.
30339 2017-01-09 Andrew Pinski <apinski@cavium.com>
30341 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
30342 to reference thunderx2t99 for the tuning structure
30343 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
30345 (thunderx2t99_extra_costs): This.
30346 * config/aarch64/aarch64-tune.md: Regenerate.
30347 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
30348 (vulcan_addrcost_table): This.
30349 (vulcan_regmove_cost): Rename to ...
30350 (thunderx2t99_regmove_cost): This.
30351 (vulcan_vector_cost): Rename to ...
30352 (thunderx2t99_vector_cost): this.
30353 (vulcan_branch_cost): Rename to ...
30354 (thunderx2t99_branch_cost): This.
30355 (vulcan_tunings): Rename to ...
30356 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
30357 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
30359 2017-01-09 Martin Jambor <mjambor@suse.cz>
30363 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
30364 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
30365 (propagate_vr_accross_jump_function): Use the above function for all
30366 value range computations for pass-through jump functions and type
30367 converasion from explicit value range values.
30368 (ipcp_propagate_stage): Do not attempt to deduce types of formal
30369 parameters from TYPE_ARG_TYPES.
30370 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
30371 (ipa_write_node_info): Stream type of the actual argument.
30372 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
30374 2017-01-09 Martin Liska <mliska@suse.cz>
30377 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
30378 (lookup_compiler): Do not show error message with have_E.
30380 2017-01-09 Jakub Jelinek <jakub@redhat.com>
30382 PR tree-optimization/78938
30383 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
30384 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
30385 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
30386 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
30389 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30391 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
30394 2017-01-09 Richard Biener <rguenther@suse.de>
30396 PR tree-optimization/78997
30397 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
30398 name condition properly.
30400 2017-01-09 Richard Biener <rguenther@suse.de>
30403 * dwarf2out.c (is_cxx): New overload with context.
30404 (is_naming_typedef_decl): Use it.
30406 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30408 * invoke.texi (Option Summary): Correct spacing in option lists
30409 and add line breaks to fix over-long lines.
30411 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30413 PR middle-end/17660
30415 * extend.texi (Common Variable Attributes): Add xref to GCC
30416 Internals manual to explain mode attribute keywords.
30418 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30421 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
30422 and Preprocessor Options.
30423 (Options for Linking): Document -pthread here....
30424 (RS/6000 and PowerPC Options): ...not here.
30425 (Solaris 2 Options): ...or here.
30426 * doc/cppopts.texi: Document -pthread.
30428 2017-01-08 Martin Sebor <msebor@redhat.com>
30430 PR middle-end/77708
30431 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
30432 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
30433 New member functions.
30434 (format_directive): Used them.
30436 (pass_sprintf_length::handle_gimple_call): Same.
30437 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
30438 to avoid truncation for any argument.
30439 (extract_affine_mul): Same.
30440 * tree.c (get_file_function_name): Same.
30442 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30444 PR middle-end/77484
30445 * predict.def (PRED_INDIR_CALL): Set to 86.
30447 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30449 PR preprocessor/54124
30450 * doc/cppopts.texi: Reformat -d subtable to list the full name
30451 of the options. Add cross-reference to the docs for the general
30452 compiler -d options.
30453 * doc/invoke.texi (Developer Options): Add cross-reference to the
30454 preprocessor-specific -d option documentation.
30456 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30458 PR preprocessor/13498
30459 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
30460 redudant material, and reflect new command-line options.
30461 (System Headers): Likewise.
30463 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30465 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
30466 -isystem, and -idirafter. Copy-edit.
30467 * doc/cppopts.texi: Copy-edit. Remove contradiction about
30468 default for -ftrack-macro-expansion. Delete obsolete and
30469 badly-formatted implementation details about -fdebug-cpp output.
30470 * doc/cppwarnopts.texi: Copy-edit.
30472 2017-01-07 David Malcolm <dmalcolm@redhat.com>
30475 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
30476 that the transition from a max line width >= 1<<10 to narrower
30477 lines works correctly.
30479 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
30481 * doc/options.texi (PerFunction): New.
30482 * opt-functions.awk (switch_flags): Map both Optimization and
30483 PerFunction to CL_OPTIMIZATION.
30484 * opth-gen.awk: Test for PerFunction flag along with
30486 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
30487 it only when the latter is present. Skip those that don't in
30488 the hash function generator.
30489 * common.opt (fvar-tracking): Mark as PerFunction instead of
30491 (fvar-tracking-assignments): Likewise.
30492 (fvar-tracking-assignments-toggle): Likewise.
30493 (fvar-tracking-uninit): Likewise.
30495 2017-01-07 Jakub Jelinek <jakub@redhat.com>
30497 PR translation/79018
30498 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
30501 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
30504 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
30505 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
30506 TARGET_LONG_JUMP_TABLE_OFFSETS.
30507 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
30508 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
30509 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
30510 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
30511 * config/m68k/m68k.md (tablejump expander): Likewise.
30512 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
30513 TARGET_LONG_JUMP_TABLE_OFFSETS.
30514 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
30515 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
30517 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
30518 David Holsgrove <david.holsgrove@xilinx.com>
30520 * common/config/microblaze/microblaze-common.c
30521 (TARGET_EXCEPT_UNWIND_INFO): Remove.
30522 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
30524 * config/microblaze/microblaze.c (microblaze_must_save_register)
30525 (microblaze_expand_epilogue, microblaze_return_addr): Handle
30527 (microblaze_eh_return): New function.
30528 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
30529 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
30530 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
30531 * config/microblaze/microblaze.md (eh_return): New pattern.
30533 2017-01-06 Jakub Jelinek <jakub@redhat.com>
30535 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
30536 GCC_DIAGNOSTIC_STRINGIFY): Define.
30538 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
30540 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30542 * config/arm/arm.md (<mcrr>): New.
30544 * config/arm/arm.c (arm_arch5te): New.
30545 (arm_option_override): Set arm_arch5te.
30546 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
30548 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
30549 (arm_mcrr_qualifiers): ... this. New.
30550 (MRRC_QUALIFIERS): Define to...
30551 (arm_mrrc_qualifiers): ... this. New.
30552 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
30554 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
30555 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
30556 (MRRCI, mrrc, MRRC): New.
30557 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
30558 VUNSPEC_MRRC2): New.
30560 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30562 * config/arm/arm.md (<mcr>): New.
30564 * config/arm/arm.c (arm_coproc_builtin_available): Add
30565 support for mcr, mrc, mcr2 and mrc2.
30566 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
30567 (arm_mcr_qualifiers): ... this. New.
30568 (MRC_QUALIFIERS): Define to ...
30569 (arm_mrc_qualifiers): ... this. New.
30570 (MCR_QUALIFIERS): Define to ...
30571 (arm_mcr_qualifiers): ... this. New.
30572 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
30574 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
30575 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
30576 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
30577 VUNSPEC_MRC2): New.
30579 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30581 * config/arm/arm.md (*ldc): New.
30585 * config/arm/arm.c (arm_coproc_builtin_available): Add
30586 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
30587 (arm_coproc_ldc_stc_legitimate_address): New.
30588 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
30589 'qualifier_const_pointer'.
30590 (LDC_QUALIFIERS): Define to...
30591 (arm_ldc_qualifiers): ... this. New.
30592 (STC_QUALIFIERS): Define to...
30593 (arm_stc_qualifiers): ... this. New.
30594 * config/arm/arm-protos.h
30595 (arm_coproc_ldc_stc_legitimate_address): New.
30596 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
30597 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
30598 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
30599 stc2, stcl, stc2l): New.
30600 * config/arm/constraints.md (Uz): New.
30601 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
30602 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
30603 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
30604 VUNSPEC_STC2L): New.
30606 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30608 * config/arm/arm.md (<cdp>): New.
30609 * config/arm/arm.c (neon_const_bounds): Rename this ...
30610 (arm_const_bounds): ... this.
30611 (arm_coproc_builtin_available): New.
30612 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
30613 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
30614 (CDP_QUALIFIERS): Define to...
30615 (arm_cdp_qualifiers): ... this. New.
30617 (arm_expand_builtin_args): Add case for 6 arguments.
30618 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
30619 (arm_const_bounds): ... this.
30620 (arm_coproc_builtin_available): New.
30621 * config/arm/arm_acle.h (__arm_cdp): New.
30623 * config/arm/arm_acle_builtins.def (cdp): New.
30625 * config/arm/iterators.md (CDPI,CDP,cdp): New.
30626 * config/arm/neon.md: Rename all 'neon_const_bounds' to
30627 'arm_const_bounds'.
30628 * config/arm/types.md (coproc): New.
30629 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
30630 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
30631 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
30632 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
30634 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30636 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
30637 (UBINOP_QUALIFIERS): New.
30639 (acle_builtin_data): New. Change comment.
30640 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
30641 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
30642 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
30643 arm_acle_builtins.def.
30644 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
30645 (arm_init_acle_builtins): New.
30646 (CRC32_BUILTIN): Remove.
30647 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
30648 crc32cb, crc32ch and crc32cw.
30649 (arm_init_crc32_builtins): Remove.
30650 (arm_init_builtins): Use arm_init_acle_builtins rather
30651 than arm_init_crc32_builtins.
30652 (arm_expand_acle_builtin): New.
30653 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
30654 * config/arm/arm_acle_builtins.def: New.
30656 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30658 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
30659 (arm_builtin_datum): ... this.
30660 (arm_init_neon_builtin): Rename to ...
30661 (arm_init_builtin): ... this. Add a new parameters PREFIX
30662 and USE_SIG_IN_NAME.
30663 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
30664 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
30665 'arm_builtin_datum'.
30666 (arm_init_vfp_builtins): Likewise.
30667 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
30668 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
30669 (arm_expand_neon_args): Rename to ...
30670 (arm_expand_builtin_args): ... this. Rename builtin_arg
30671 enum values and differentiate between ARG_BUILTIN_MEMORY
30672 and ARG_BUILTIN_NEON_MEMORY.
30673 (arm_expand_neon_builtin_1): Rename to ...
30674 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
30675 values, arm_expand_builtin_args and add bool parameter NEON.
30676 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
30677 (arm_expand_vfp_builtin): Likewise.
30678 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
30680 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30682 PR middle-end/77484
30683 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
30684 * predict.c (tree_estimate_probability_bb): Reverse direction of
30685 polymorphic call predictor.
30687 2017-01-06 David Malcolm <dmalcolm@redhat.com>
30689 * passes.c (execute_one_pass): Split out pass-skipping logic into...
30690 (determine_pass_name_match): ...this new function and...
30691 (should_skip_pass_p): ...this new function.
30693 2017-01-06 Nathan Sidwell <nathan@acm.org>
30695 * ipa-visibility.c (function_and_variable_visibility): Reformat
30696 comments and long lines. Remove extrneous if.
30697 * symtab.c (symtab_node::make_decl_local): Fix code format.
30698 (symtab_node::set_section_for_node): Fix comment typo.
30700 2017-01-06 Martin Liska <mliska@suse.cz>
30703 * lra-constraints.c: Rename invariant to lra_invariant.
30704 * predict.c (set_even_probabilities): Initialize e to NULL.
30706 2017-01-05 Martin Sebor <msebor@redhat.com>
30708 PR tree-optimization/78910
30709 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
30710 (format_integer): Correct off-by-one error in the handling
30711 of precision with negative numbers in signed conversions..
30713 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
30715 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
30717 2017-01-05 Jakub Jelinek <jakub@redhat.com>
30719 PR tree-optimization/71016
30720 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
30721 factor_out_conditional_conversion. Formatting fix.
30722 (factor_out_conditional_conversion): Add cond_stmt argument.
30723 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
30724 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
30727 2017-01-05 David Malcolm <dmalcolm@redhat.com>
30729 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
30730 read-rtl-function.o, and selftest-rtl.o.
30731 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
30732 (selftest::aarch64_test_loading_full_dump): New function.
30733 (selftest::aarch64_run_selftests): New function.
30734 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
30735 selftest::aarch64_run_selftests.
30736 * config/i386/i386.c
30737 (selftest::ix86_test_loading_dump_fragment_1): New function.
30738 (selftest::ix86_test_loading_call_insn): New function.
30739 (selftest::ix86_test_loading_full_dump): New function.
30740 (selftest::ix86_test_loading_unspec): New function.
30741 (selftest::ix86_run_selftests): Call the new functions.
30742 * emit-rtl.c (maybe_set_max_label_num): New function.
30743 * emit-rtl.h (maybe_set_max_label_num): New decl.
30744 * function.c (instantiate_decls): Guard call to
30745 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
30746 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
30748 * gensupport.c (gen_reader::gen_reader): Pass "false"
30749 for new "compact" param of rtx_reader.
30750 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
30751 rather than an empty string for NULL strings.
30752 * read-md.c: Potentially include config.h rather than bconfig.h.
30753 Wrap include of errors.h with #ifdef GENERATOR_FILE.
30754 (have_error): New global, copied from errors.c.
30755 (md_reader::read_name): Rename to...
30756 (md_reader::read_name_1): ...this, adding "out_loc" param,
30757 and converting "missing name or number" to returning false, rather
30759 (md_reader::read_name): Reimplement in terms of read_name_1.
30760 (md_reader::read_name_or_nil): New function.
30761 (md_reader::read_string): Handle "(nil)" by returning NULL.
30762 (md_reader::md_reader): Add new param "compact".
30763 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
30764 (md_reader::read_file): New method.
30765 * read-md.h (md_reader::md_reader): Add new param "compact".
30766 (md_reader::read_file): New method.
30767 (md_reader::is_compact): New accessor.
30768 (md_reader::read_name): Convert return type from void to file_location.
30769 (md_reader::read_name_or_nil): New decl.
30770 (md_reader::read_name_1): New decl.
30771 (md_reader::m_compact): New field.
30772 (noop_reader::noop_reader): Pass "false" for new "compact" param
30774 (rtx_reader::rtx_reader): Add new "compact" param.
30775 (rtx_reader::read_rtx_operand): Make virtual and convert return
30776 type from void to rtx.
30777 (rtx_reader::read_until): New decl.
30778 (rtx_reader::handle_any_trailing_information): New virtual function.
30779 (rtx_reader::postprocess): New virtual function.
30780 (rtx_reader::finalize_string): New virtual function.
30781 (rtx_reader::m_in_call_function_usage): New field.
30782 (rtx_reader::m_reuse_rtx_by_id): New field.
30783 * read-rtl-function.c: New file.
30784 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
30785 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
30786 (selftest::verify_three_block_rtl_cfg): New decl.
30787 * read-rtl-function.h: New file.
30788 * read-rtl.c: Potentially include config.h rather than bconfig.h.
30789 For host, include function.h, memmodel.h, and emit-rtl.h.
30790 (one_time_initialization): New function.
30791 (struct compact_insn_name): New struct.
30792 (compact_insn_names): New array.
30793 (find_code): Handle insn codes in compact dumps.
30794 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
30795 (bind_subst_iter_and_attr): Likewise.
30796 (add_condition_to_string): Likewise.
30797 (add_condition_to_rtx): Likewise.
30798 (apply_attribute_uses): Likewise.
30799 (add_current_iterators): Likewise.
30800 (apply_iterators): Likewise.
30801 (initialize_iterators): Guard usage of apply_subst_iterator with
30802 #ifdef GENERATOR_FILE.
30803 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
30804 (md_reader::read_mapping): Likewise.
30805 (add_define_attr_for_define_subst): Likewise.
30806 (add_define_subst_attr): Likewise.
30807 (read_subst_mapping): Likewise.
30808 (check_code_iterator): Likewise.
30809 (rtx_reader::read_rtx): Likewise. Move one-time initialization
30811 (one_time_initialization): New function.
30812 (rtx_reader::read_until): New method.
30813 (read_flags): New function.
30814 (parse_reg_note_name): New function.
30815 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
30816 Handle reuse_rtx ids.
30817 Wrap iterator lookup within #ifdef GENERATOR_FILE.
30818 Add parsing support for RTL dumps, mirroring the special-cases in
30819 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
30820 values, and calling handle_any_trailing_information.
30821 (rtx_reader::read_rtx_operand): Convert return type from void
30822 to rtx, returning return_rtx. Handle case 'e'. Call
30823 finalize_string on XSTR and XTMPL fields.
30824 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
30825 "(nil)" values were omitted. Call the postprocess vfunc on the
30827 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
30828 class ctor. Initialize m_in_call_function_usage. Call
30829 one_time_initialization.
30830 * rtl-tests.c (selftest::test_uncond_jump): Call
30831 set_new_first_and_last_insn.
30832 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
30833 * selftest-rtl.c: New file.
30834 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
30835 (selftest::get_insn_by_uid): New decl.
30836 * selftest-run-tests.c (selftest::run_tests): Call
30837 read_rtl_function_c_tests.
30838 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
30839 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
30842 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
30844 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
30845 operands in a special way. Assert that pos+len <= mode precision.
30847 2017-01-05 Jakub Jelinek <jakub@redhat.com>
30849 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
30850 3 argument Alias with unlimited for the negative form.
30851 (fno-vect-cost-model): Removed.
30853 2017-01-05 Martin Liska <mliska@suse.cz>
30855 * hsa-gen.c (gen_hsa_divmod): New function.
30856 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
30858 2017-01-05 Martin Liska <mliska@suse.cz>
30861 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
30864 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30866 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
30867 small constant length operands.
30869 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30871 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
30872 between loop iterations.
30874 2017-01-05 Martin Liska <mliska@suse.cz>
30877 * gimplify.c (gimplify_decl_expr): Compare to
30878 asan_poisoned_variables instread of checking flags.
30879 (gimplify_target_expr): Likewise.
30880 (gimplify_expr): Likewise.
30881 (gimplify_function_tree): Conditionally initialize
30882 asan_poisoned_variables.
30884 2017-01-04 Jeff Law <law@redhat.com>
30886 PR tree-optimizatin/78812
30887 * rtl.h (contains_mem_rtx_p): Prototype.
30888 * ifcvt.c (containts_mem_rtx_p): Move from here to...
30889 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
30890 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
30891 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
30892 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
30894 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30896 * input.c (assert_char_at_range): Default-initialize actual_range.
30898 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30900 * df-scan.c (df_ref_create_structure): Make regno unsigned,
30901 to match the caller.
30903 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30905 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
30906 insns after final jump in test to emit dummy move.
30908 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30910 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
30911 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
30913 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30915 * multiple_target.c (create_dispatcher_calls): Init e_next.
30916 * tree-ssa-loop-split.c (split_loop): Init border.
30917 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
30920 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
30925 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
30926 (altivec_register_operand): Do not return true if the operand
30927 contains a SUBREG mixing SImode and SFmode.
30928 (vsx_register_operand): Likewise.
30929 (vsx_reg_sfsubreg_ok): New predicate.
30930 (vfloat_operand): Do not return true if the operand contains a
30931 SUBREG mixing SImode and SFmode.
30932 (vint_operand): Likewise.
30933 (vlogical_operand): Likewise.
30934 (gpc_reg_operand): Likewise.
30935 (int_reg_operand): Likewise.
30936 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
30937 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
30938 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
30940 (rs6000_emit_move_si_sf_subreg): New helper function.
30941 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
30942 fixup SUBREGs involving SImode and SFmode.
30943 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
30944 numbers for the new peephole2 optimization.
30945 (peephole2 for SFmode unions): New peephole2 to optimize cases in
30946 the GLIBC math library that do AND/IOR/XOR operations on single
30947 precision floating point.
30948 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
30949 target macros to say whether we need to avoid SUBREGs mixing
30951 (TARGET_ALLOW_SF_SUBREG): Likewise.
30952 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
30953 (UNSPEC_SI_FROM_SF): Likewise.
30954 (iorxor): Change spacing.
30955 (and_ior_xor): New iterator for AND, IOR, and XOR.
30956 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
30957 (movdi_from_sf_zero_ext): Likewise.
30958 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
30959 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
30960 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
30961 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
30962 (fms<mode>4): Likewise.
30963 (fnma<mode>4): Likewise.
30964 (fnms<mode>4): Likewise.
30965 (nfma<mode>4): Likewise.
30966 (nfms<mode>4): Likewise.
30968 2017-01-04 Marek Polacek <polacek@redhat.com>
30971 * doc/invoke.texi: Document -Wpointer-compare.
30973 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30975 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
30978 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
30979 descriptions for -gdwarf-5 and emit them as uleb128 instead of
30982 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
30985 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
30986 documentation of the powerpc_popcntb_ok attribute.
30987 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30988 code to issue warning messages if a requested CPU configuration is
30989 not supported by the binary (assembler and loader) toolchain.
30990 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
30991 made to define a built-in function that has been disabled.
30992 (paired_init_builtins): Add assertion to prevent ICE if attempt is
30993 made to define a built-in function that has been disabled.
30994 (altivec_init_builtins): Add comment explaining why definition
30995 of the DST built-in functions is not preceded by an assertion
30996 check. Add assertions to prevent ICE if attempts are made to
30997 define an altivec predicate or an abs* built-in function that has
30999 (htm_init_builtins): Add comment explaining why definition of the
31000 htm built-in functions is not preceded by an assertion check.
31002 2017-01-04 Jeff Law <law@redhat.com>
31004 PR tree-optimizatin/67955
31005 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
31006 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
31007 the points-to solution does not include pt_null. Use DECL_PT_UID
31010 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
31012 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
31013 Use gen_int_mode instead of gen_lopwart for const_int operands.
31015 2017-01-04 Jakub Jelinek <jakub@redhat.com>
31017 PR tree-optimization/71563
31018 * match.pd: Simplify X << Y into X if Y is known to be 0 or
31019 out of range value - has low bits known to be zero.
31021 2017-01-04 Alan Modra <amodra@gmail.com>
31023 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
31024 * configure: Regenerate.
31025 * config.in: Regenerate.
31027 2017-01-04 Jakub Jelinek <jakub@redhat.com>
31030 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
31031 a substring of the message, but strcmp with the whole message. Ifdef
31032 ENABLE_NLS, translate the message first using dgettext.
31034 2017-01-03 Jeff Law <law@redhat.com>
31036 PR tree-optimizatin/78856
31037 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
31038 (mark_threaded_blocks): Remove code to truncate thread paths that
31039 cross multiple loop headers. Instead invalidate the cached loop
31040 iteration information and handle case of a thread path walking
31041 into an irreducible region.
31043 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
31046 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
31047 assertions. Add support for doing the signbit if the IEEE 128-bit
31048 floating point value is in a GPR.
31049 * config/rs6000/rs6000.md (Fsignbit): Delete.
31050 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
31051 Update the length attribute if the value is in a GPR.
31052 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
31053 the sign or zero extension instruction, since the value is always 0/1.
31054 (signbit<mode>2_dm2): Delete using <Fsignbit>.
31057 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
31058 extracting SImode to a GPR register so that we can generate a
31059 store, limit the vector to be in a traditional Altivec register
31060 for the vextuwrx instruction.
31062 2017-01-03 Ian Lance Taylor <iant@google.com>
31064 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
31066 2017-01-03 Martin Sebor <msebor@redhat.com>
31068 PR tree-optimization/78696
31069 * gimple-ssa-sprintf.c (format_floating): Correct handling of
31070 precision. Use MPFR for %f for greater fidelity. Correct handling
31072 (pass_sprintf_length::compute_format_length): Set width and precision
31073 specified by asrerisk to void_node for vararg functions.
31074 (try_substitute_return_value): Adjust dump output.
31076 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
31078 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
31080 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
31082 * doc/invoke.texi (SPARC options): Document -mlra as the default.
31083 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
31084 -mlra/-mno-lra was passed to the compiler.
31086 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
31088 PR rtl-optimization/65618
31089 * emit-rtl.c (try_split): Move initialization of "before" and
31090 "after" to just before the call to emit_insn_after_setloc.
31092 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
31094 * doc/md.texi (Standard Names): Remove reference to Java frontend.
31096 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
31098 * dwarf2out.c (gen_enumeration_type_die): When
31099 -gno-strict-dwarf, add a DW_AT_encoding attribute.
31101 2017-01-03 Jakub Jelinek <jakub@redhat.com>
31103 PR tree-optimization/78965
31104 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
31105 Change first argument from const call_info & to call_info &. For %n
31106 set info.nowrite to false.
31108 PR middle-end/78901
31109 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
31110 possibly throwing calls.
31112 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
31113 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
31114 and fns handling, rather than in a separate case SSA_NAME.
31116 2017-01-02 Jeff Law <law@redhat.com>
31118 * config/darwin-driver.c (darwin_driver_init): Const-correctness
31119 fixes for first_period and second_period variables.
31121 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
31124 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
31125 (*insvqi_1): New insn pattern.
31126 (*insvqi_1_mem_rex64): Ditto.
31127 (*insvqi_2): Ditto.
31128 (*insvqi_3): Rename from *insvqi.
31130 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
31132 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
31134 * doc/cfg.texi (Edges): Remove reference to Java.
31135 (Maintaining the CFG): Ditto.
31137 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31139 PR middle-end/77674
31140 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
31141 transparent aliases.
31143 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31145 PR middle-end/77484
31146 * predict.def (PRED_CALL): Update hitrate.
31147 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
31148 * predict.c (tree_estimate_probability_bb): Split CALL predictor
31149 into direct/indirect/polymorphic variants.
31151 2017-01-01 Jakub Jelinek <jakub@redhat.com>
31153 Update copyright years.
31155 * gcc.c (process_command): Update copyright notice dates.
31156 * gcov-dump.c (print_version): Ditto.
31157 * gcov.c (print_version): Ditto.
31158 * gcov-tool.c (print_version): Ditto.
31159 * gengtype.c (create_file): Ditto.
31160 * doc/cpp.texi: Bump @copying's copyright year.
31161 * doc/cppinternals.texi: Ditto.
31162 * doc/gcc.texi: Ditto.
31163 * doc/gccint.texi: Ditto.
31164 * doc/gcov.texi: Ditto.
31165 * doc/install.texi: Ditto.
31166 * doc/invoke.texi: Ditto.
31168 Copyright (C) 2017 Free Software Foundation, Inc.
31170 Copying and distribution of this file, with or without modification,
31171 are permitted in any medium without royalty provided the copyright
31172 notice and this notice are preserved.