1 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3 * inchash.h (inchash::hash::add_wide_int): Rename to...
4 (inchash::hash::add_hwi): ...this.
5 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
6 (polymorphic_call_target_hasher::hash): Likewise.
7 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
8 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
9 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
10 * lto-streamer-out.c (hash_tree): Likewise.
11 * optc-save-gen.awk: Likewise.
12 * tree.c (add_expr): Likewise.
14 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
17 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
18 for ordered inequality comparisons even with TARGET_IEEE_FP.
20 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
23 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
24 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
25 Expand with cmp<dwi>_doubleword.
27 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
29 * extend.texi: Add x86 specific to 'nocf_check' attribute.
31 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
32 specific to -fcf-protection option.
34 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
36 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
37 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
38 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
39 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
40 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
41 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
42 (extra_objs): Add cet.o for Linux/x86 targets.
43 (tmake_file): Add i386/t-cet for Linux/x86 targets.
44 * config/i386/cet.c: New file.
45 * config/i386/cetintrin.h: Likewise.
46 * config/i386/t-cet: Likewise.
47 * config/i386/cpuid.h (bit_SHSTK): New.
49 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
50 pass IBT and SHSTK bits.
51 * config/i386/i386-builtin-types.def
52 (VOID_FTYPE_UNSIGNED_PVOID): New.
53 (VOID_FTYPE_UINT64_PVOID): Likewise.
54 * config/i386/i386-builtin.def: Add CET intrinsics.
55 * config/i386/i386-c.c (ix86_target_macros_internal): Add
56 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
57 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
58 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
60 * config/i386/i386.c (rest_of_insert_endbranch): New.
61 (pass_data_insert_endbranch): Likewise.
62 (pass_insert_endbranch): Likewise.
63 (make_pass_insert_endbranch): Likewise.
64 (ix86_notrack_prefixed_insn_p): Likewise.
65 (ix86_target_string): Add -mibt, -mshstk flags.
66 (ix86_option_override_internal): Add flag_cf_protection
68 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
69 (ix86_print_operand): Add 'notrack' prefix output.
70 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
71 (ix86_expand_builtin): Expand CET intrinsics.
72 (x86_output_mi_thunk): Add 'endbranch' instruction.
73 * config/i386/i386.h (TARGET_IBT): New.
74 (TARGET_IBT_P): Likewise.
75 (TARGET_SHSTK): Likewise.
76 (TARGET_SHSTK_P): Likewise.
77 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
78 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
79 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
80 (builtin_setjmp_setup): New pattern.
81 (builtin_longjmp): Likewise.
82 (rdssp<mode>): Likewise.
83 (incssp<mode>): Likewise.
84 (saveprevssp): Likewise.
86 (wrss<mode>): Likewise.
87 (wruss<mode>): Likewise.
90 (nop_endbr): Likewise.
91 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
93 * config/i386/immintrin.h: Include <cetintrin.h>.
94 * config/i386/linux-common.h
95 (file_end_indicate_exec_stack_and_cet): New prototype.
96 (TARGET_ASM_FILE_END): New.
98 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
100 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
102 2017-10-20 Jakub Jelinek <jakub@redhat.com>
104 * config/i386/i386.md (isa): Remove fma_avx512f.
105 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
106 <avx512>_fmadd_<mode>_mask3<round_name>,
107 <avx512>_fmsub_<mode>_mask<round_name>,
108 <avx512>_fmsub_<mode>_mask3<round_name>,
109 <avx512>_fnmadd_<mode>_mask<round_name>,
110 <avx512>_fnmadd_<mode>_mask3<round_name>,
111 <avx512>_fnmsub_<mode>_mask<round_name>,
112 <avx512>_fnmsub_<mode>_mask3<round_name>,
113 <avx512>_fmaddsub_<mode>_mask<round_name>,
114 <avx512>_fmaddsub_<mode>_mask3<round_name>,
115 <avx512>_fmsubadd_<mode>_mask<round_name>,
116 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
117 (*vec_widen_umult_even_v16si<mask_name>,
118 *vec_widen_smult_even_v16si<mask_name>): Likewise.
119 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
121 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
123 * extend.texi: Add 'nocf_check' documentation.
124 * gimple.texi: Add second parameter to
125 gimple_build_call_from_tree.
126 * invoke.texi: Add -fcf-protection documentation.
127 * rtl.texi: Add REG_CALL_NOTRACK documenation.
129 2017-10-20 Richard Biener <rguenther@suse.de>
131 PR tree-optimization/82473
132 * tree-vect-loop.c (vectorizable_reduction): Properly get at
133 the largest input type.
135 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
137 * c-attribs.c (handle_nocf_check_attribute): New function.
138 (c_common_attribute_table): Add 'nocf_check' handling.
139 * gimple-parser.c: Add second argument NULL to
140 gimple_build_call_from_tree.
141 * attrib.c (comp_type_attributes): Check nocf_check attribute.
142 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
144 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
145 * common.opt: Add fcf-protection flag.
146 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
147 * flag-types.h: Add enum cf_protection_level.
148 * gimple.c (gimple_build_call_from_tree): Add second parameter.
149 Add 'nocf_check' attribute propagation to gimple call.
150 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
151 (gimple_build_call_from_tree): Update prototype.
152 (gimple_call_nocf_check_p): New function.
153 (gimple_call_set_nocf_check): Likewise.
154 * gimplify.c: Add second argument to gimple_build_call_from_tree.
155 * ipa-icf.c: Add nocf_check attribute in statement hash.
156 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
157 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
158 * toplev.c (process_options): Add flag_cf_protection handling.
160 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
162 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
164 2017-10-20 Richard Biener <rguenther@suse.de>
166 PR tree-optimization/82603
167 * tree-if-conv.c (predicate_mem_writes): Make sure to only
168 remove false predicated stores.
170 2017-10-20 Richard Biener <rguenther@suse.de>
172 * graphite-isl-ast-to-gimple.c
173 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
174 Remove return value and simplify, dump copied stmt after lhs
176 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
177 Reduce dump verbosity.
178 (gsi_insert_earliest): Likewise.
179 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
180 * graphite.c (print_global_statistics): Adjust dumping.
181 (print_graphite_scop_statistics): Likewise.
182 (print_graphite_statistics): Do not dump loops here.
183 (graphite_transform_loops): But here.
185 2017-10-20 Nicolas Roche <roche@adacore.com>
187 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
188 * configure: Regenerate.
190 2017-10-20 Jakub Jelinek <jakub@redhat.com>
193 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
194 functions when optimizing replace GIMPLE_RETURN stmts with
195 calls to __builtin_unreachable ().
198 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
199 for -fsanitize=thread link of executables.
200 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
204 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
206 (<shift_insn><mode>3<mask_name>): Change the last of the 3
207 define_insns for logical vector shifts to use VI248_AVX512BW
208 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
209 condition, useless isa and prefix attributes. Change the first
210 2 of these define_insns to ...
211 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
212 define_insn for avx512vl.
213 (<shift_insn><mode>3): ... and this, new define_insn without
214 masking for non-avx512vl.
217 * config/i386/sse.md (*andnot<mode>3,
218 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
219 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
220 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
221 not applied use empty suffix even for TARGET_AVX512VL.
222 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
223 is applied, supply evex,evex or evex,evex,evex instead of just
226 2017-10-20 Julia Koval <julia.koval@intel.com>
228 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
229 (OPTION_MASK_ISA_GFNI_UNSET): New.
230 (ix86_handle_option): Handle OPT_mgfni.
231 * config/i386/cpuid.h (bit_GFNI): New.
232 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
233 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
234 * config/i386/i386.c (ix86_target_string): Add -mgfni.
235 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
236 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
237 * config/i386/i386.opt: Add mgfni.
239 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
241 * config/msp430/msp430.c (msp430_option_override): Disable
242 -fdelete-null-pointer-checks.
243 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
245 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
247 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
248 of x87 and SSE instructions.
250 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
252 * asan.c (create_cond_insert_point): Do not update edge count.
253 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
254 (afdo_propagate_circuit): Likewise.
255 (afdo_calculate_branch_prob): Likewise.
256 (afdo_annotate_cfg): Likewise.
257 * basic-block.h (struct edge_def): Remove count.
258 (edge_def::count): New accessor.
259 * bb-reorder.c (rotate_loop): Update.
260 (find_traces_1_round): Update.
261 (connect_traces): Update.
262 (sanitize_hot_paths): Update.
263 * cfg.c (unchecked_make_edge): Update.
264 (make_single_succ_edge): Update.
265 (check_bb_profile): Update.
266 (dump_edge_info): Update.
267 (update_bb_profile_for_threading): Update.
268 (scale_bbs_frequencies_int): Update.
269 (scale_bbs_frequencies_gcov_type): Update.
270 (scale_bbs_frequencies_profile_count): Update.
271 (scale_bbs_frequencies): Update.
272 * cfganal.c (connect_infinite_loops_to_exit): Update.
273 * cfgbuild.c (compute_outgoing_frequencies): Update.
274 (find_many_sub_basic_blocks): Update.
275 * cfgcleanup.c (try_forward_edges): Update.
276 (try_crossjump_to_edge): Update
277 * cfgexpand.c (expand_gimple_cond): Update
278 (expand_gimple_tailcall): Update
279 (construct_exit_block): Update
280 * cfghooks.c (verify_flow_info): Update
281 (redirect_edge_succ_nodup): Update
283 (make_forwarder_block): Update
284 (duplicate_block): Update
285 (account_profile_record): Update
286 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
287 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
288 * cfgloopmanip.c (scale_loop_profile): Update.
290 (lv_adjust_loop_entry_edge): Update.
291 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
292 (force_nonfallthru_and_redirect): Update.
293 (purge_dead_edges): Update.
294 (rtl_flow_call_edges_add): Update.
295 * cgraphunit.c (init_lowered_empty_function): Update.
296 (cgraph_node::expand_thunk): Update.
297 * gimple-pretty-print.c (dump_probability): Update.
298 (dump_edge_probability): Update.
299 * gimple-ssa-isolate-paths.c (isolate_path): Update.
300 * haifa-sched.c (sched_create_recovery_edges): Update.
301 * hsa-gen.c (convert_switch_statements): Update.
302 * ifcvt.c (dead_or_predicable): Update.
303 * ipa-inline-transform.c (inline_transform): Update.
304 * ipa-split.c (split_function): Update.
305 * ipa-utils.c (ipa_merge_profiles): Update.
306 * loop-doloop.c (add_test): Update.
307 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
308 * lto-streamer-in.c (input_cfg): Update.
309 (input_function): Update.
310 * lto-streamer-out.c (output_cfg): Update.
311 * modulo-sched.c (sms_schedule): Update.
312 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
313 * predict.c (maybe_hot_edge_p): Update.
314 (unlikely_executed_edge_p): Update.
315 (probably_never_executed_edge_p): Update.
316 (dump_prediction): Update.
317 (drop_profile): Update.
318 (propagate_unlikely_bbs_forward): Update.
319 (determine_unlikely_bbs): Update.
320 (force_edge_cold): Update.
321 * profile.c (compute_branch_probabilities): Update.
322 * reg-stack.c (better_edge): Update.
323 * shrink-wrap.c (handle_simple_exit): Update.
324 * tracer.c (better_p): Update.
325 * trans-mem.c (expand_transaction): Update.
326 (split_bb_make_tm_edge): Update.
327 * tree-call-cdce.c: Update.
328 * tree-cfg.c (gimple_find_sub_bbs): Update.
329 (gimple_split_edge): Update.
330 (gimple_duplicate_sese_region): Update.
331 (gimple_duplicate_sese_tail): Update.
332 (gimple_flow_call_edges_add): Update.
333 (insert_cond_bb): Update.
334 (execute_fixup_cfg): Update.
335 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
336 * tree-complex.c (expand_complex_div_wide): Update.
337 * tree-eh.c (lower_resx): Update.
338 (unsplit_eh): Update.
339 (cleanup_empty_eh_move_lp): Update.
340 * tree-inline.c (copy_edges_for_bb): Update.
341 (freqs_to_counts): Update.
342 (copy_cfg_body): Update.
343 * tree-ssa-dce.c (remove_dead_stmt): Update.
344 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
345 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
346 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
347 (unloop_loops): Update.
348 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
349 * tree-ssa-loop-split.c (connect_loops): Update.
350 (split_loop): Update.
351 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
352 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
353 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
354 * tree-ssa-reassoc.c (branch_fixup): Update.
355 * tree-ssa-tail-merge.c (replace_block_by): Update.
356 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
357 (compute_path_counts): Update.
358 (update_profile): Update.
359 (recompute_probabilities): Update.
360 (update_joiner_offpath_counts): Update.
361 (estimated_freqs_path): Update.
362 (freqs_to_counts_path): Update.
363 (clear_counts_path): Update.
364 (ssa_fix_duplicate_block_edges): Update.
365 (duplicate_thread_path): Update.
366 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
367 (case_bit_test_cmp): Update.
368 (collect_switch_conv_info): Update.
369 (gen_inbound_check): Update.
370 (do_jump_if_equal): Update.
371 (emit_cmp_and_jump_insns): Update.
372 * tree-tailcall.c (decrease_profile): Update.
373 (eliminate_tail_call): Update.
374 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
375 (vect_do_peeling): Update.
376 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
377 * ubsan.c (ubsan_expand_null_ifn): Update.
378 (ubsan_expand_ptr_ifn): Update.
379 * value-prof.c (gimple_divmod_fixed_value): Update.
380 (gimple_mod_pow2): Update.
381 (gimple_mod_subtract): Update.
383 (gimple_stringop_fixed_value): Update.
385 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
388 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
390 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
392 PR rtl-optimization/82395
393 * ira-color.c (allocno_priority_compare_func): Fix comparison step
394 based on non_spilled_static_chain_regno_p.
396 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
398 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
399 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
400 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
402 2017-10-19 Martin Sebor <msebor@redhat.com>
404 PR tree-optimization/82596
405 * tree.c (array_at_struct_end_p): Handle STRING_CST.
407 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
409 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
410 (get_mem_refs_of_builtin_call): Likewise.
411 * builtins.c (expand_builtin_apply): Adjust call to
412 allocate_dynamic_stack_space.
413 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
414 the third argument to allocate_dynamic_stack_space, otherwise -1.
415 (expand_builtin): Deal with all alloca variants.
416 (is_inexpensive_builtin): Likewise.
417 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
418 * calls.c (special_function_p): Deal with all alloca variants.
419 (initialize_argument_information): Adjust call to
420 allocate_dynamic_stack_space.
421 (expand_call): Likewise.
422 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
423 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
424 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
425 use it for the stack usage computation.
426 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
427 * function.c (gimplify_parameters): Call build_alloca_call_expr.
428 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
429 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
430 (in_loop_p): Remove first argument and useless check.
431 (pass_walloca::execute): Remove useless test and adjust call to above.
432 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
433 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
434 (gimplify_call_expr): Deal with all alloca variants.
435 * hsa-gen.c (gen_hsa_alloca): Likewise.
436 (gen_hsa_insns_for_call): Likewise.
437 * ipa-pure-const.c (special_builtin_state): Likewise.
438 * tree-chkp.c (chkp_build_returned_bound): Likewise.
439 * tree-object-size.c (alloc_object_size): Likewise.
440 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
441 (call_may_clobber_ref_p_1): Likewise.
442 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
443 (ccp_fold_stmt): Likewise.
444 (optimize_stack_restore): Likewise.
445 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
446 (mark_all_reaching_defs_necessary_1): Likewise.
447 (propagate_necessity): Likewise.
448 (eliminate_unnecessary_stmts): Likewise.
449 * tree.c (build_common_builtin_nodes): Build
450 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
451 (build_alloca_call_expr): New function.
452 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
453 (CASE_BUILT_IN_ALLOCA): Likewise.
454 (build_alloca_call_expr): Declare.
455 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
457 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
460 * dwarf2out.c (new_die_raw): New static inline function.
461 (new_die): Use it to create the DIE.
462 (add_AT_external_die_ref): Likewise.
463 (clone_die): Likewise.
464 (clone_as_declaration): Likewise.
465 (dwarf2out_vms_debug_main_pointer): Likewise.
466 (base_type_die): Likewise. Remove early return for corner cases.
467 Do not call add_pubtype on the DIE here.
468 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
469 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
470 typedefs for base types with DW_AT_endianity. Make sure a DIE with
471 native order exists for base types, attach the DIE manually and call
472 add_pubtype on it. Do not equate a reverse order DIE to the type.
474 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
476 * config/arm/arm.c (align_ok_ldrd_strd): New function.
477 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
479 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
481 2017-10-19 Jakub Jelinek <jakub@redhat.com>
483 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
484 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
485 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
486 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
487 * opts.c (sanitizer_opts): Add builtin.
488 * ubsan.c (instrument_builtin): New function.
489 (pass_ubsan::execute): Call it.
490 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
491 * doc/invoke.texi: Document -fsanitize=builtin.
493 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
494 builtins, store max (log2 (align), 0) into uchar field instead of
495 align into uptr field.
496 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
497 store uchar 0 field instead of uptr 0 field.
498 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
499 instead of passing one address of struct with 2 locations pass
500 two addresses of structs with 1 location each.
501 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
502 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
503 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
504 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
505 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
506 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
507 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
508 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
510 2017-10-19 Martin Liska <mliska@suse.cz>
513 * file-find.c (remove_prefix): Remove.
514 * file-find.h (remove_prefix): Likewise.
515 * gcc-ar.c: Remove smartness of lookup.
517 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
519 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
520 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
521 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
523 2017-10-19 Jakub Jelinek <jakub@redhat.com>
526 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
527 (setcc + and to xor + setcc): New peephole2.
529 2017-10-19 Tom de Vries <tom@codesourcery.com>
531 * doc/sourcebuild.texi (Test Directives, Variants of
532 dg-require-support): Add dg-require-stack-size.
534 2017-10-19 Martin Liska <mliska@suse.cz>
537 * gimplify.c (gimplify_decl_expr): Do not instrument variables
538 that have a large alignment.
539 (gimplify_target_expr): Likewise.
541 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
543 PR rtl-optimization/82602
544 * ira.c (rtx_moveable_p): Return false for volatile asm.
546 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
549 * config/i386/i386-modes.def (CCGZ): New CC mode.
550 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
551 * config/i386/predicates.md (ix86_comparison_operator):
553 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
554 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
555 with double-word subtraction.
556 (put_condition_code): Handle CCGZmode.
558 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
560 * wide-int.cc (debug (const wide_int &)): New.
561 (debug (const wide_int *)): New.
562 (debug (const widest_int &)): New.
563 (debug (const widest_int *)): New.
565 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
568 * lra-constraints.c (curr_insn_transform): Use non-input operand
569 instead of output one for matched reload.
571 2017-10-18 Bin Cheng <bin.cheng@arm.com>
573 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
574 (tree-ssa-loop-ivopts.h): New header file.
575 (struct builtin_info): New fields.
576 (classify_builtin_1): Compute and record base and offset parts for
577 memset builtin partition by calling strip_offset.
578 (offset_cmp, fuse_memset_builtins): New functions.
579 (finalize_partitions): Fuse adjacent memset partitions by calling
581 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
582 Expose the interface.
583 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
585 2017-10-18 Bin Cheng <bin.cheng@arm.com>
587 PR tree-optimization/82574
588 * tree-loop-distribution.c (find_single_drs): New parameter. Check
589 that data reference must be executed exactly once per iteration
590 against the outermost loop in nest.
591 (classify_partition): Update call to above function.
593 2017-10-18 Richard Biener <rguenther@suse.de>
595 PR tree-optimization/82591
596 * graphite.c (graphite_transform_loops): Move code gen message
598 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
599 Here. Handle scop_to_isl_ast failing.
600 (scop_to_isl_ast): Limit the number of ISL operations.
602 2017-10-18 Richard Biener <rguenther@suse.de>
604 * graphite-isl-ast-to-gimple.c
605 (translate_isl_ast_to_gimple::set_rename): Simplify.
606 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
607 (graphite_copy_stmts_from_block): ... here.
608 (copy_bb_and_scalar_dependences): Simplify.
609 (add_parameters_to_ivs_params): Canonicalize.
610 (generate_entry_out_of_ssa_copies): Simplify.
611 * graphite-sese-to-poly.c (extract_affine_name): Simplify
612 by passing in ISL dimension.
613 (parameter_index_in_region_1): Rename to ...
614 (parameter_index_in_region): ... this.
615 (extract_affine): Adjust assert, pass down parameter index.
616 (add_param_constraints): Use range-info when available.
617 (build_scop_context): Adjust.
618 * sese.c (new_sese_info): Adjust.
619 (free_sese_info): Likewise.
620 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
621 Remove unused typedefs.
622 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
624 2017-10-18 Martin Liska <mliska@suse.cz>
626 * combine.c (simplify_compare_const): Add gcc_fallthrough.
628 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
630 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
631 (s390_sched_init): Do not reset s390_sched_state if we entered the
632 current basic block via a fallthru edge and all others are unlikely.
634 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
636 * config/s390/s390.c (NUM_SIDES): New variable.
637 (LONGRUNNING_THRESHOLD): New variable.
638 (LATENCY_FACTOR): New variable.
639 (s390_sched_score): Decrease score for long-running instructions on
641 (s390_sched_variable_issue): Perform bookkeeping for long-running
644 2017-10-18 Richard Biener <rguenther@suse.de>
646 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
647 Simplify with removal of the parameter rename map.
648 (set_rename): Likewise.
649 (should_copy_to_new_region): Likewise.
650 (graphite_copy_stmts_from_block): Likewise.
651 (copy_bb_and_scalar_dependences): Remove initialization of
652 unused copied_bb_map.
654 (copy_internal_parameters): Likewise.
655 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
656 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
658 (parameter_index_in_region_1): Rename to ...
659 (assign_parameter_index_in_region): ... this. Assert we have
660 a parameter we handle.
661 (scan_tree_for_params): Adjust.
662 * sese.h (parameter_rename_map_t): Remove.
663 (struct sese_info_t): Remove unused parameter_rename_map and
664 copied_bb_map members.
665 * sese.c (new_sese_info): Adjust.
666 (free_sese_info): Likewise.
668 2017-10-18 Martin Liska <mliska@suse.cz>
671 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
674 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
676 * doc/invoke.texi (ffunction-sections and fdata-sections):
679 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
681 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
682 the use statement can throw internally.
684 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
686 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
687 any RTX present on the RHS of a SET.
688 * compare-elim.c (try_eliminate_compare): Restore comment.
690 2017-10-17 Jakub Jelinek <jakub@redhat.com>
692 * langhooks.h (struct lang_hooks): Document that tree_size langhook
693 may be also called on tcc_type nodes.
694 * langhooks.c (lhd_tree_size): Likewise.
696 2017-10-17 David Malcolm <dmalcolm@redhat.com>
698 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
699 format_warning_at_substring.
700 (maybe_warn): Convert source_range * param to a location_t. Pass
701 UNKNOWN_LOCATION rather than NULL to fmtwarn.
702 (format_directive): Remove code to extract source_ranges and
703 source_range * in favor of just a location_t.
704 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
706 * substring-locations.c (format_warning_va): Convert
707 source_range * param to a location_t.
708 (format_warning_at_substring): Likewise.
709 * substring-locations.h (format_warning_va): Likewise.
710 (format_warning_at_substring): Likewise.
712 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
714 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
716 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
717 and scatter/gather ops.
719 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
720 vec_gather_load and vec_scatter_store.
721 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
722 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
723 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
724 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
725 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
726 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
728 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
730 * reg-stack.c (compare_for_stack_reg): Add bool argument.
731 Detect FTST instruction and handle its register pops. Only pop
732 second operand if can_pop_second_op is true.
733 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
734 set can_pop_second_op to false in the compare_for_stack_reg call.
736 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
737 output_fp_compare for stack register operands.
738 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
739 instructions here. Do not emit stack register pops here. Assert
740 that FCOMPP pops next to top stack register. Rewrite function.
742 2017-10-17 Nathan Sidwell <nathan@acm.org>
745 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
746 use DECL_ASSEMBLER_NAME_RAW.
749 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
752 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
753 Wilco Dijkstra <wilco.dijkstra@arm.com>
755 * builtins.c (expand_builtin_update_setjmp_buf): Add a
756 converstion to Pmode from the buf_addr.
758 2017-10-17 Richard Biener <rguenther@suse.de>
760 * graphite-dependences.c (scop_get_reads_and_writes): Change
761 output parameters to references.
763 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
765 PR 71026/tree-optimization
766 * fold-const.c (distribute_real_division): Removed.
767 (fold_binary_loc): Remove calls to distribute_real_divison.
769 2017-10-17 Richard Biener <rguenther@suse.de>
771 * graphite-scop-detection.c
772 (scop_detection::stmt_has_simple_data_refs_p): Always use
773 the full nest as region.
774 (try_generate_gimple_bb): Likewise.
775 * sese.c (scalar_evolution_in_region): Simplify now that
776 SCEV can handle instantiation in regions.
777 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
778 in the non-loop part of a function if requested.
780 2017-10-17 Richard Biener <rguenther@suse.de>
782 PR tree-optimization/82563
783 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
785 (graphite_regenerate_ast_isl): Call it.
786 * graphite-scop-detection.c (build_scops): Remove entry edge split.
788 2017-10-17 Jakub Jelinek <jakub@redhat.com>
790 PR tree-optimization/82549
791 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
792 Formatting fixes. Instead of calling make_bit_field_ref with negative
795 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
797 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
798 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
799 _mm_maskz_reduce_ss): New.
800 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
801 __builtin_ia32_reducess_mask): Ditto..
802 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
803 * config/i386/sse.md (reduces<mode>): Renamed to ...
804 (reduces<mode><mask_scalar_name>): ... this.
805 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}): Changed
807 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
808 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
810 2017-10-16 David Malcolm <dmalcolm@redhat.com>
812 * Makefile.in (OBJS): Add unique-ptr-tests.o.
813 * selftest-run-tests.c (selftest::run_tests): Call
814 selftest::unique_ptr_tests_cc_tests.
815 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
816 * unique-ptr-tests.cc: New file.
818 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
821 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
823 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
824 (make_hard_regno_born, make_hard_regno_dead): Update
825 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
827 2017-10-16 Jeff Law <law@redhat.com>
829 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
831 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
833 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
835 2017-10-16 Olivier Hainque <hainque@adacore.com>
837 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
838 with_cpu if we were configured for an e500v2 target cpu name.
840 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
842 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
843 * doc/invoke.texi: Document +nodsp as a valid extension for
846 2017-10-16 Martin Liska <mliska@suse.cz>
848 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
849 (test_set_range): Likewise.
850 (test_range_functions): Rename to ...
851 (test_bit_in_range): ... this.
852 (sbitmap_c_tests): Add new test.
854 2017-10-16 Tamar Christina <tamar.christina@arm.com>
856 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32): New.
857 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
858 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
860 2017-10-16 Tamar Christina <tamar.christina@arm.com>
862 * config/aarch64/aarch64-builtins.c
863 (aarch64_types_quadopu_lane_qualifiers): New.
864 (TYPES_QUADOPU_LANE): New.
865 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
866 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
867 (aarch64_<sur>dot_laneq<vsi2qi>): New.
868 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
869 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
870 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
871 (Vdottype, DOTPROD): New.
872 (sur): Add SDOT and UDOT.
874 2017-10-16 Tamar Christina <tamar.christina@arm.com>
876 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
877 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
878 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Add TARGET_DOTPROD.
879 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
880 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75): Enable TARGET_DOTPROD.
881 (cortex-a75.cortex-a55): Likewise.
882 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
884 2017-10-16 Tamar Christina <tamar.christina@arm.com>
886 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
887 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS): New.
888 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane): new.
889 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
890 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
891 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
892 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
893 * config/arm/types.md (neon_dot, neon_dot_q): New.
894 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
896 2017-10-16 Tamar Christina <tamar.christina@arm.com>
898 * config/arm/arm.h (TARGET_DOTPROD): New.
899 * config/arm/arm.c (arm_arch_dotprod): New.
900 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
901 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
902 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
903 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
904 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
905 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
906 * doc/invoke.texi (armv8.2-a): Document dotprod
908 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
910 * i386.c (ix86_vec_cost): New function.
911 (ix86_rtx_costs): Handle vector operations better.
912 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
913 * x86-tune-costs.h: Add new costs to all tables.
915 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
917 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
919 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
920 divsd, sqrtss and sqrtsd
921 * x86-tune-costs.h: Add new entries to all costs.
922 (znver1_cost): Fix to match real instruction latencies.
924 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
925 Michael Collison <michael.collison@arm.com>
927 * compare-elim.c: Include emit-rtl.h.
928 (can_merge_compare_into_arith): New function.
929 (try_validate_parallel): Likewise.
930 (try_merge_compare): Likewise.
931 (try_eliminate_compare): Call the above when no previous clobber
933 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
936 2017-10-14 Jakub Jelinek <jakub@redhat.com>
940 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
941 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
945 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
946 to be any operand_equal_p operands. For & (B - 1) require
947 B to be power of 2. Recognize
948 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
950 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
953 * optabs.c (expand_memory_blockage): Fix call of
954 targetm.have_memory_blockage.
956 2017-10-14 Jakub Jelinek <jakub@redhat.com>
959 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
960 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
961 objects to extra_objs instead of overwriting it.
963 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
965 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
966 Use any_fp_register_operand as operand[3] predicate. Simplify
967 equality test for operands[2] and operands[4] memory location.
968 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
969 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
970 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
971 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
972 any_fp_register_operand as operand[1] predicate. Simplify
973 equality test for operands[0] and operands[3] memory location.
974 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
975 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
976 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
978 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
980 * target-insns.def: Add memory_blockage.
981 * optabs.c (expand_memory_blockage): New function.
982 (expand_asm_memory_barrier): Rename ...
983 (expand_asm_memory_blockage): ... to this.
984 (expand_mem_thread_fence): Call expand_memory_blockage
985 instead of expand_asm_memory_barrier.
986 (expand_mem_singnal_fence): Ditto.
987 (expand_atomic_load): Ditto.
988 (expand_atomic_store): Ditto.
989 * doc/md.texi (Standard Pattern Names For Generation):
990 Document memory_blockage instruction pattern.
992 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
994 * config/rl78/rl78.c (rl78_emit_libcall): New function.
995 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
996 * config/rl78/rl78.md: New define_expand "adddi3".
998 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
1000 * cfghooks.c (verify_flow_info): Disable check that all probabilities
1003 2017-10-13 Jeff Law <law@redhat.com>
1005 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
1007 2017-10-13 Jakub Jelinek <jakub@redhat.com>
1010 * internal-fn.c (expand_mul_overflow): If both operands have
1011 the same highpart of -1 or 0 and the topmost bit of lowpart
1012 is different, overflow is if res <= 0 rather than res < 0.
1014 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
1016 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
1017 TARGET_P9_VECTOR code for unaligned_load case.
1019 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
1021 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
1023 2017-10-13 Nathan Sidwell <nathan@acm.org>
1025 * tree-core.h (tree_contains_struct): Make bool.
1026 * tree.c (tree_contains_struct): Likewise.
1027 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
1028 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
1029 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
1030 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
1031 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
1033 2017-10-13 Richard Biener <rguenther@suse.de>
1035 * graphite-isl-ast-to-gimple.c
1036 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
1037 parameters and dominance check.
1038 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
1039 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
1040 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
1041 Do not update SSA form here or do intermediate IL verification.
1042 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
1043 (graphite_initialize): Remove check on the number of loops in
1044 the function and inline into graphite_transform_loops.
1045 (graphite_finalize): Inline into graphite_transform_loops.
1046 (graphite_transform_loops): Perform SSA update and IL verification
1048 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
1050 2017-10-13 Richard Biener <rguenther@suse.de>
1052 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
1053 graphite_expression_type_precision): Avoid global constructor
1055 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
1056 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
1057 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
1058 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
1059 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
1060 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
1062 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
1065 * config/i386/i386.h (ix86_red_zone_size): New.
1066 * config/i386/i386.md (push peephole2s): Replace
1067 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
1069 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1070 Alan Hayward <alan.hayward@arm.com>
1071 David Sherwood <david.sherwood@arm.com>
1073 * combine.c (can_change_dest_mode): Reject changes in
1074 REGMODE_NATURAL_SIZE.
1076 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1077 Alan Hayward <alan.hayward@arm.com>
1078 David Sherwood <david.sherwood@arm.com>
1080 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
1081 (expand_debug_source_expr): Likewise.
1082 * combine.c (combine_simplify_rtx): Likewise.
1083 * cse.c (fold_rtx): Likewise.
1084 * fwprop.c (canonicalize_address): Likewise.
1085 * targhooks.c (default_shift_truncation_mask): Likewise.
1087 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1088 Alan Hayward <alan.hayward@arm.com>
1089 David Sherwood <david.sherwood@arm.com>
1091 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
1092 (widened_mode): Likewise.
1093 (expand_unop): Likewise.
1094 * ree.c (transform_ifelse): Likewise.
1095 (merge_def_and_ext): Likewise.
1096 (combine_reaching_defs): Likewise.
1097 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1099 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1100 Alan Hayward <alan.hayward@arm.com>
1101 David Sherwood <david.sherwood@arm.com>
1103 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
1104 * combine.c (gen_lowpart_for_combine): Likewise.
1105 * dwarf2out.c (rtl_for_decl_location): Likewise.
1106 * final.c (alter_subreg): Likewise.
1107 * rtlhooks.c (gen_lowpart_general): Likewise.
1108 (gen_lowpart_if_possible): Likewise.
1110 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1111 Alan Hayward <alan.hayward@arm.com>
1112 David Sherwood <david.sherwood@arm.com>
1114 * calls.c (expand_call): Use subreg_lowpart_offset.
1115 * cse.c (cse_insn): Likewise.
1116 * regcprop.c (copy_value): Likewise.
1117 (copyprop_hardreg_forward_1): Likewise.
1119 2017-10-13 Jakub Jelinek <jakub@redhat.com>
1122 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
1123 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
1124 =Q constraints to +Q and into insn condition add check
1125 that operands[0] and operands[1] are equal.
1126 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
1127 =Q constraints to +Q and into insn condition add check
1128 that operands[0] is equal to either operands[1] or operands[2].
1131 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
1132 instead of handling MINUS_EXPR twice (once for each argument),
1133 canonicalize operand order and handle just once, use rtype where
1134 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
1137 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
1138 any values of __C while still being pattern recognizable as a simple
1141 2017-10-13 Richard Biener <rguenther@suse.de>
1143 PR tree-optimization/82451
1145 2017-10-02 Richard Biener <rguenther@suse.de>
1147 PR tree-optimization/82355
1148 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
1149 a mapping for the enclosing loop but avoid generating one for
1151 (copy_bb_and_scalar_dependences): Remove premature codegen
1152 error on PHIs in blocks duplicated into multiple places.
1153 * graphite-scop-detection.c
1154 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
1155 in the region use it as loop and nest to analyze the DR in.
1156 (try_generate_gimple_bb): Likewise.
1157 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
1158 (add_loop_constraints): For blocks in a loop not in the region
1159 create a dimension with a single iteration.
1160 * sese.h (gbb_loop_at_index): Remove assert.
1162 * cfgloop.c (loop_preheader_edge): For the loop tree root
1163 return the single successor of the entry block.
1164 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1165 Reset the SCEV hashtable and niters.
1166 * graphite-scop-detection.c
1167 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
1168 assert that we only have POLYNOMIAL_CHREC that vary in loops
1169 contained in the region.
1170 (scop_detection::graphite_can_represent_expr): Adjust.
1171 (scop_detection::stmt_has_simple_data_refs_p): For loops
1172 not in the region set loop to NULL. The nest is now the
1173 entry edge to the region.
1174 (try_generate_gimple_bb): Likewise.
1175 * sese.c (scalar_evolution_in_region): Adjust for
1176 instantiate_scev change.
1177 * tree-data-ref.h (graphite_find_data_references_in_stmt):
1178 Make nest parameter the edge into the region.
1179 (create_data_ref): Likewise.
1180 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
1181 entry edge into a region and adjust instantiate_scev calls.
1182 (create_data_ref): Likewise.
1183 (graphite_find_data_references_in_stmt): Likewise.
1184 (find_data_references_in_stmt): Pass the loop preheader edge
1185 from the nest argument.
1186 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
1187 parameter the edge into the region.
1188 (instantiate_parameters): Use the loop preheader edge as entry.
1189 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
1191 (get_instantiated_value_entry): Make instantiate_below parameter
1192 the edge into the region.
1193 (instantiate_scev_name): Likewise. Adjust dominance checks,
1194 when we cannot use loop-based instantiation instantiate by
1195 walking use-def chains.
1196 (instantiate_scev_poly): Adjust.
1197 (instantiate_scev_binary): Likewise.
1198 (instantiate_scev_convert): Likewise.
1199 (instantiate_scev_not): Likewise.
1200 (instantiate_array_ref): Remove.
1201 (instantiate_scev_3): Likewise.
1202 (instantiate_scev_2): Likewise.
1203 (instantiate_scev_1): Likewise.
1204 (instantiate_scev_r): Do not blindly handle N-operand trees.
1205 Do not instantiate array-refs. Handle all constants and invariants.
1206 (instantiate_scev): Make instantiate_below parameter
1207 the edge into the region.
1208 (resolve_mixers): Use the loop preheader edge for the region
1209 parameter to instantiate_scev_r.
1210 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
1212 2017-10-13 Richard Biener <rguenther@suse.de>
1214 PR tree-optimization/82525
1215 * graphite-isl-ast-to-gimple.c
1216 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
1218 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
1219 Fail code generation when we cannot represent the isl integer.
1220 (binary_op_to_tree): Elide modulo operations that are no-ops
1221 in the type we code generate. Remove now superfluous code
1224 2017-10-13 Richard Biener <rguenther@suse.de>
1226 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
1227 (scop_detection::harmful_loop_in_region): Remove premature
1228 IV type restriction.
1229 (scop_detection::graphite_can_represent_scev): We can handle
1230 pointer IVs just fine.
1232 2017-10-13 Alan Modra <amodra@gmail.com>
1234 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
1235 "Clobbers and Scratch Registers". Add paragraph on
1236 alternative to clobbers for scratch registers and OpenBLAS
1239 2017-10-13 Alan Modra <amodra@gmail.com>
1241 * doc/extend.texi (Clobbers): Correct vax example. Delete old
1242 example of a memory input for a string of known length. Move
1243 commentary out of table. Add a number of new examples
1244 covering array memory inputs.
1246 2017-10-12 Martin Liska <mliska@suse.cz>
1248 PR tree-optimization/82493
1249 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
1250 (test_range_functions): New function.
1251 (sbitmap_c_tests): Likewise.
1252 * selftest-run-tests.c (selftest::run_tests): Run new tests.
1253 * selftest.h (sbitmap_c_tests): New function.
1255 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
1257 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1259 * config/rs6000/amo.h: Fix spacing issue.
1261 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1264 * config/i386/i386.md (*ashl<mode>3_mask_1,
1265 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
1266 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
1269 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
1271 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
1272 (profile_probability): Set max_probability
1273 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
1275 * profile-count.c (profile_probability::differs_from_p): Do not
1276 rely on max_probaiblity == 10000
1278 2017-10-12 Jeff Law <law@redhat.com>
1280 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
1283 2017-10-12 Martin Sebor <msebor@redhat.com>
1287 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
1288 (handle_alias_pairs): Call it.
1289 * common.opt (-Wattribute-alias): New option.
1290 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
1291 * doc/invoke.texi (-Wattribute-alias): Document.
1293 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
1296 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
1298 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1300 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1301 (make_hard_regno_born, make_hard_regno_dead): Update
1302 bb_killed_pseudos and bb_gen_pseudos.
1304 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
1306 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
1308 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
1310 * config/alpha/alpha.c (alpha_split_conditional_move):
1311 Use std::swap instead of manually swapping.
1312 (alpha_stdarg_optimize_hook): Ditto.
1313 (alpha_canonicalize_comparison): Ditto.
1315 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1317 * tree-loop-distribution.c (struct builtin_info): New struct.
1318 (struct partition): Refactor fields into struct builtin_info.
1319 (partition_free): Free struct builtin_info.
1320 (build_size_arg_loc, build_addr_arg_loc): Delete.
1321 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
1322 information from struct builtin_info.
1323 (find_single_drs): New function refactored from classify_partition.
1324 Also moved builtin validity checks to this function.
1325 (compute_access_range, alloc_builtin): New functions.
1326 (classify_builtin_st, classify_builtin_ldst): New functions.
1327 (classify_partition): Refactor code into functions find_single_drs,
1328 classify_builtin_st and classify_builtin_ldst.
1329 (distribute_loop): Don't do runtime alias check when distributing
1331 (find_seed_stmts_for_distribution): New function.
1332 (pass_loop_distribution::execute): Refactor code finding seed
1333 stmts into above function. Support distribution for the innermost
1334 two-level loop nest. Adjust dump information.
1336 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1338 * tree-loop-distribution.c: Adjust the general comment.
1339 (NUM_PARTITION_THRESHOLD): New macro.
1340 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
1341 (classify_partition): Skip builtin pattern of loop nest's inner loop.
1342 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
1343 in call to build_partition_graph.
1344 (finalize_partitions): New parameter. Make loop distribution more
1345 conservative by fusing more partitions.
1346 (distribute_loop): Don't do runtime alias check in case of loop nest
1348 (find_seed_stmts_for_distribution): New function.
1349 (prepare_perfect_loop_nest): New function.
1350 (pass_loop_distribution::execute): Refactor code finding seed stmts
1351 and loop nest into above functions. Support loop nest distribution.
1352 Adjust dump information accordingly.
1354 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1356 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
1357 and set PTYPE_SEQUENTIAL for merged partition.
1359 2017-10-12 Richard Biener <rguenther@suse.de>
1361 PR tree-optimization/69728
1363 2017-09-19 Richard Biener <rguenther@suse.de>
1365 PR tree-optimization/69728
1366 * graphite-sese-to-poly.c (schedule_error): New global.
1367 (add_loop_schedule): Handle empty domain by failing the
1369 (build_original_schedule): Handle schedule_error.
1371 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
1372 domain by returning an unchanged schedule.
1374 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1376 * genrecog.c (validate_pattern): For VEC_SELECT verify that
1377 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
1379 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
1381 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
1384 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1387 * expr.c (store_field): Don't optimize away bitsize == 0 store
1388 from CALL_EXPR with addressable return type.
1390 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
1392 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
1393 * config/rs6000/rs6000.md (sel): Delete mode attribute.
1394 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
1395 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
1396 TARGET_ISEL instead of TARGET_ISEL<sel>.
1398 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
1400 * config/rs6000/rs6000.c
1401 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
1403 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
1405 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
1407 (reg_or_cint_operand): Fix comment.
1408 (reg_or_zero_operand): New predicate.
1409 * config/rs6000/rs6000-protos.h (output_isel): Delete.
1410 * config/rs6000/rs6000.c (output_isel): Delete.
1411 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
1412 instead of reg_or_cint_operand. Output instruction directly (not via
1414 (isel_unsigned_<mode>): Ditto.
1415 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
1416 gpc_reg_operand. Add an instruction alternative for this. Output
1417 instruction directly.
1418 (*isel_reversed_unsigned_<mode>): Ditto.
1420 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1422 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
1423 (TARGET_CANONICALIZE_COMPARISON): Define.
1425 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
1428 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
1429 Check whether the dest is REG before adding REG_EQUIV note.
1431 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
1434 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1436 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1437 (make_hard_regno_born, make_hard_regno_dead): Update
1438 bb_killed_pseudos and bb_gen_pseudos.
1440 2017-10-11 Nathan Sidwell <nathan@acm.org>
1442 * incpath.h (enum incpath_kind): Name enum, prefix values.
1443 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
1444 * incpath.c (heads, tails): Use INC_MAX.
1445 (add_env_var_paths, add_standard_paths): Use incpath_kind.
1446 (merge_include_chains, split_quote_chain,
1447 register_include_chains): Update incpath_kind names.
1448 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
1449 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
1451 (add_framework_path, darwin_register_objc_includes): Likewise.
1452 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
1454 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1456 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
1457 Do not use float_operator operator predicate.
1458 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
1459 * config/i386/predicates.md (float_operator): Remove predicate.
1461 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1463 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
1464 (*jccxf_i387): Ditto.
1465 (*jcc<mode>_i387): Ditto.
1466 (*jccu<mode>_i387): Ditto.
1467 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
1468 (*jcc_*_i387 splitters): Remove.
1469 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
1470 * config/i386/i386.c (ix86_split_fp_branch): Remove.
1471 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
1474 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
1476 * profile-count.h (slow_safe_scale_64bit): New function.
1477 (safe_scale_64bit): New inline.
1478 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
1479 * profile-count.c: Include wide-int.h
1480 (slow_safe_scale_64bit): New.
1482 2017-10-11 Nathan Sidwell <nathan@acm.org>
1484 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
1485 HAS_DECL_ASSEMBLER_NAME_P.
1486 * gimple-expr.c (gimple_decl_printable_name: Check
1487 HAS_DECL_ASSEMBLER_NAME_P too.
1488 * ipa-utils.h (type_in_anonymous_namespace_p): Check
1489 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
1490 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
1491 * passes.c (rest_of_decl_compilation): Check
1492 HAS_DECL_ASSEMBLER_NAME_P too.
1493 * recog.c (verify_changes): Likewise.
1494 * tree-pretty-print.c (dump_decl_name): Likewise.
1495 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
1497 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
1498 (SET_DECL_ASSEMBLER_NAME): Use it.
1499 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
1500 (COPY_DECL_ASSEMBLER_NAME): Likewise.
1501 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
1503 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
1505 * config.gcc (i386, x86_64): Add extra objects.
1506 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
1507 (ix86_min_insn_size): Declare.
1508 (ix86_issue_rate): Declare.
1509 (ix86_adjust_cost): Declare.
1510 (ia32_multipass_dfa_lookahead): Declare.
1511 (ix86_macro_fusion_p): Declare.
1512 (ix86_macro_fusion_pair_p): Declare.
1513 (ix86_bd_has_dispatch): Declare.
1514 (ix86_bd_do_dispatch): Declare.
1515 (ix86_core2i7_init_hooks): Declare.
1516 (ix86_atom_sched_reorder): Declare.
1517 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
1518 (COSTS_N_BYTES): Move to x86-tune-costs.h.
1519 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
1520 (rip_relative_addr_p): Rename to ...
1521 (ix86_rip_relative_addr_p): ... this one; export.
1522 (memory_address_length): Update.
1523 (ix86_issue_rate): Move to x86-tune-sched.c.
1524 (ix86_flags_dependent): Move to x86-tune-sched.c.
1525 (ix86_agi_dependent): Move to x86-tune-sched.c.
1526 (exact_dependency_1): Move to x86-tune-sched.c.
1527 (exact_store_load_dependency): Move to x86-tune-sched.c.
1528 (ix86_adjust_cost): Move to x86-tune-sched.c.
1529 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
1530 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
1531 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
1532 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
1533 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
1534 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
1535 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
1536 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
1537 (min_insn_size): Rename to ...
1538 (ix86_min_insn_size): ... this one; export.
1539 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
1540 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
1541 (core2i7_first_cycle_multipass_backtrack): Move to
1542 x86-tune-sched-core.c.
1543 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
1544 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
1545 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
1546 (ix86_avoid_jump_mispredicts): Update.
1547 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
1548 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
1549 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
1550 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
1551 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
1552 (MAX_INSN): Move to ix86-tune-sched-bd.c.
1553 (MAX_IMM): Move to ix86-tune-sched-bd.c.
1554 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
1555 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
1556 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
1557 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
1558 (MAX_STORE): Move to ix86-tune-sched-bd.c.
1559 (BIG): Move to ix86-tune-sched-bd.c.
1560 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
1561 (enum insn_path): Move to ix86-tune-sched-bd.c.
1562 (get_mem_group): Move to ix86-tune-sched-bd.c.
1563 (is_cmp): Move to ix86-tune-sched-bd.c.
1564 (dispatch_violation): Move to ix86-tune-sched-bd.c.
1565 (is_branch): Move to ix86-tune-sched-bd.c.
1566 (is_prefetch): Move to ix86-tune-sched-bd.c.
1567 (init_window): Move to ix86-tune-sched-bd.c.
1568 (allocate_window): Move to ix86-tune-sched-bd.c.
1569 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
1570 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
1571 (process_end_window): Move to ix86-tune-sched-bd.c.
1572 (allocate_next_window): Move to ix86-tune-sched-bd.c.
1573 (find_constant): Move to ix86-tune-sched-bd.c.
1574 (get_num_immediates): Move to ix86-tune-sched-bd.c.
1575 (has_immediate): Move to ix86-tune-sched-bd.c.
1576 (get_insn_path): Move to ix86-tune-sched-bd.c.
1577 (get_insn_group): Move to ix86-tune-sched-bd.c.
1578 (count_num_restricted): Move to ix86-tune-sched-bd.c.
1579 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
1580 (add_insn_window): Move to ix86-tune-sched-bd.c.
1581 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
1582 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
1583 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
1584 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
1585 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
1586 (do_dispatch): Move to ix86-tune-sched-bd.c.
1587 (has_dispatch): Move to ix86-tune-sched-bd.c.
1588 * config/i386/t-i386: Add new object files.
1589 * config/i386/x86-tune-costs.h: New file.
1590 * config/i386/x86-tune-sched-atom.c: New file.
1591 * config/i386/x86-tune-sched-bd.c: New file.
1592 * config/i386/x86-tune-sched-core.c: New file.
1593 * config/i386/x86-tune-sched.c: New file.
1595 2017-10-11 Liu Hao <lh_mouse@126.com>
1597 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
1598 the generic version below instead.
1599 (should_colorize): Recognize Windows consoles as terminals
1601 * pretty-print.c [__MINGW32__] (write_all): New function.
1602 [__MINGW32__] (find_esc_head): Likewise.
1603 [__MINGW32__] (find_esc_terminator): Likewise.
1604 [__MINGW32__] (eat_esc_sequence): Likewise.
1605 [__MINGW32__] (mingw_ansi_fputs): New function that handles
1607 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
1610 2017-10-11 Richard Biener <rguenther@suse.de>
1612 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
1613 Properly call analyze_scalar_evolution with the loop of the stmt.
1615 2017-10-11 Richard Biener <rguenther@suse.de>
1617 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
1618 * tree-core.h (tree_base): Add chrec_var union member.
1619 * tree.h (CHREC_VAR): Remove.
1620 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
1621 * tree-chrec.h (build_polynomial_chrec): Adjust.
1622 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
1623 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
1625 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
1627 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
1628 * match.pd: ... here.
1629 ((T) X == (T) Y): Relax condition.
1631 2017-10-11 Bin Cheng <bin.cheng@arm.com>
1633 PR tree-optimization/82472
1634 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
1636 (break_alias_scc_partitions): Update postorder number.
1638 2017-10-11 Martin Liska <mliska@suse.cz>
1641 * opts.c (parse_no_sanitize_attribute): Do not use error_value
1643 * opts.h (parse_no_sanitize_attribute): Remove last argument.
1645 2017-10-11 Martin Liska <mliska@suse.cz>
1647 * print-rtl.c (print_insn): Move declaration of idbuf
1648 to same scope as name.
1650 2017-10-11 Martin Liska <mliska@suse.cz>
1655 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
1658 2017-10-11 Martin Liska <mliska@suse.cz>
1661 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
1664 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
1666 * config.gcc (default_gnu_indirect_function): Default to yes for
1667 arm*-*-linux* with glibc.
1669 2017-10-11 Richard Biener <rguenther@suse.de>
1671 * tree-scalar-evolution.c (get_scalar_evolution): Handle
1672 default-defs and types we do not want to analyze.
1673 (interpret_loop_phi): Replace unreachable code with an assert.
1674 (compute_scalar_evolution_in_loop): Remove and inline ...
1675 (analyze_scalar_evolution_1): ... here, replacing condition with
1676 what makes the intent clearer. Remove handling of cases
1677 get_scalar_evolution now handles.
1679 2017-10-10 Jim Wilson <wilson@tuliptree.org>
1681 PR rtl-optimization/81434
1682 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
1683 comment for main loop. In sched_group_found if, also add checks for
1684 pass and min_cost_group.
1686 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
1688 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
1689 (rs6000_insn_cost): New function.
1690 * config/rs6000/rs6000.md (cost): New attribute.
1692 2017-10-10 Jakub Jelinek <jakub@redhat.com>
1693 H.J. Lu <hongjiu.lu@intel.com>
1697 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
1698 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
1699 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
1700 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
1701 to OPTION_MASK_ISA_AVX512VL - builtins that have both
1702 OPTION_MASK_ISA_MMX and some other bit set require both
1703 mmx and the ISAs without the mmx bit.
1704 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
1705 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
1706 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
1707 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
1708 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
1709 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
1710 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
1711 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
1712 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
1713 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
1714 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
1715 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
1716 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
1717 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
1718 Add OPTION_MASK_ISA_MMX.
1720 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
1722 * config.gcc (armv7*-*-freebsd*): New target.
1723 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
1725 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
1727 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
1728 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
1731 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
1733 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
1734 a template parameter.
1735 (WIDE_INT_REF_FOR): Update accordingly.
1736 * tree.h (wi::int_traits <const_tree>): Delete.
1737 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
1738 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
1739 (wi::tree_to_wide_ref): New typedef.
1740 (wi::to_wide): New function.
1741 * calls.c (get_size_range): Use wi::to_wide when operating on
1743 * cgraph.c (cgraph_node::create_thunk): Likewise.
1744 * config/i386/i386.c (ix86_data_alignment): Likewise.
1745 (ix86_local_alignment): Likewise.
1746 * dbxout.c (stabstr_O): Likewise.
1747 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
1748 * expr.c (const_vector_from_tree): Likewise.
1749 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
1750 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
1751 (fold_negate_expr_1, int_const_binop_1, const_binop)
1752 (fold_convert_const_int_from_real, optimize_bit_field_compare)
1753 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
1754 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
1755 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
1756 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
1757 (fold_not_const, round_up_loc): Likewise.
1758 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
1759 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
1760 (alloca_call_type): Likewise.
1761 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
1762 * godump.c (go_output_typedef): Likewise.
1763 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
1764 * internal-fn.c (get_min_precision): Likewise.
1765 * ipa-cp.c (ipcp_store_vr_results): Likewise.
1766 * ipa-polymorphic-call.c
1767 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1768 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
1769 (ipa_modify_call_arguments): Likewise.
1770 * match.pd: Likewise.
1771 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
1772 * print-tree.c (print_node_brief, print_node): Likewise.
1773 * stmt.c (expand_case): Likewise.
1774 * stor-layout.c (layout_type): Likewise.
1775 * tree-affine.c (tree_to_aff_combination): Likewise.
1776 * tree-cfg.c (group_case_labels_stmt): Likewise.
1777 * tree-data-ref.c (dr_analyze_indices): Likewise.
1778 (prune_runtime_alias_test_list): Likewise.
1779 * tree-dump.c (dequeue_and_dump): Likewise.
1780 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
1781 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
1782 * tree-pretty-print.c (dump_generic_node): Likewise.
1783 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
1784 (simple_iv_with_niters): Likewise.
1785 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1786 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
1787 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
1788 * tree-ssa-loop-niter.c (split_to_var_and_offset)
1789 (refine_value_range_using_guard, number_of_iterations_ne_max)
1790 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
1791 (get_cst_init_from_scev, record_nonwrapping_iv)
1792 (scev_var_range_cant_overflow): Likewise.
1793 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
1794 * tree-ssa-pre.c (compute_avail): Likewise.
1795 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1796 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
1797 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
1798 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
1799 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
1800 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
1801 (dump_case_nodes, try_switch_expansion): Likewise.
1802 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
1803 (vect_do_peeling): Likewise.
1804 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
1805 * tree-vect-stmts.c (vectorizable_load): Likewise.
1806 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
1807 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
1808 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
1809 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
1810 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
1811 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
1812 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
1813 (evrp_dom_walker::before_dom_children): Likewise.
1814 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
1815 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
1816 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
1817 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
1818 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
1819 (get_range_pos_neg): Likewise.
1820 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
1821 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
1822 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
1823 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
1824 * config/avr/avr.c (avr_fold_builtin): Likewise.
1825 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
1826 * config/msp430/msp430.c (msp430_attr): Likewise.
1827 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
1828 * config/powerpcspe/powerpcspe-c.c
1829 (altivec_resolve_overloaded_builtin): Likewise.
1830 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
1831 (rs6000_expand_ternop_builtin): Likewise.
1832 * config/rs6000/rs6000-c.c
1833 (altivec_resolve_overloaded_builtin): Likewise.
1834 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
1835 (rs6000_expand_ternop_builtin): Likewise.
1836 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
1838 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1840 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
1841 when copying loop nest with only one inner loop.
1843 2017-10-10 Richard Biener <rguenther@suse.de>
1845 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
1846 blocks if SCEV is active.
1847 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
1849 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
1850 (scev_initialize): Assert we are not yet initialized.
1852 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1854 * tree-loop-distribution.c (generate_loops_for_partition): Remove
1855 inner loop's exit stmt by making it always exit the loop, otherwise
1856 we would generate an infinite empty loop.
1858 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1860 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
1861 renaming variables in new preheader if it's deleted.
1863 2017-10-10 Bin Cheng <bin.cheng@arm.com>
1865 * tree-loop-distribution.c (struct partition): Remove unused field
1866 loops of the structure.
1867 (partition_alloc, partition_free): Ditto.
1868 (build_rdg_partition_for_vertex): Ditto.
1870 2017-10-09 Jeff Law <law@redhat.com>
1872 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
1873 return type to match prototype and documentation.
1875 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1877 * config/rs6000/rs6000.c (processor_costs): Move to ...
1878 * config/rs6000/rs6000.h: ... here.
1879 (rs6000_cost): Declare.
1881 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
1883 * except.c (setjmp_fn): New global variable.
1884 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
1885 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
1886 if DONT_USE_BUILTIN_SETJMP is defined.
1888 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1890 * target.def (insn_cost): New hook.
1891 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
1892 * doc/tm.texi: Regenerate.
1893 * rtlanal.c (insn_cost): Use the new hook.
1895 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1897 * combine.c (combine_validate_cost): Compute the new insn_cost,
1898 not just pattern_cost.
1899 (try_combine): Adjust comment.
1901 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1903 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
1905 * combine.c (uid_insn_cost): Adjust comment.
1906 (combine_validate_cost): Adjust comment. Use pattern_cost instead
1908 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
1909 * dse.c (find_shift_sequence): Ditto.
1910 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
1911 (bb_valid_for_noce_process_p): Use pattern_cost.
1912 * rtl.h (insn_rtx_cost): Delete.
1913 (pattern_cost): New prototype.
1914 (insn_cost): New prototype.
1915 * rtlanal.c (insn_rtx_cost): Rename to...
1916 (pattern_cost): ... this.
1919 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
1921 * config/i386/i386.md (*jcc_2): Remove insn pattern.
1922 (*jcc<mode>_0_r_i387): Ditto.
1923 (*jccxf_r_i387): Ditto.
1924 (*jcc<mode>_r_i387): Ditto.
1925 (*jccu<mode>_r_i387): Ditto.
1926 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
1927 (*jcc): Rename from *jcc_1.
1929 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1931 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
1932 deferred rescans after the lvx/stvx recombination pre-pass.
1934 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
1936 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
1937 memory operation instruction support.
1938 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
1939 (rs6000-ibm-aix[789]*): Likewise.
1940 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
1941 Document new functions.
1943 2017-10-09 Richard Biener <rguenther@suse.de>
1945 PR tree-optimization/82397
1946 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
1947 equality only for semantically equal trees.
1949 2017-10-09 Richard Biener <rguenther@suse.de>
1951 PR tree-optimization/82449
1952 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
1953 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
1954 allow constant addresses.
1955 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
1958 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1960 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
1963 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1966 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
1969 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1972 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
1974 2017-10-09 Jakub Jelinek <jakub@redhat.com>
1977 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
1978 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
1980 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
1982 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
1983 (WI_BINARY_PREDICATE_RESULT): Likewise.
1984 (wi::binary_traits::operator_result): New type.
1985 (wi::binary_traits::predicate_result): Likewise.
1986 (generic_wide_int::operator~, unary generic_wide_int::operator-)
1987 (generic_wide_int::operator==, generic_wide_int::operator!=)
1988 (generic_wide_int::operator&, generic_wide_int::and_not)
1989 (generic_wide_int::operator|, generic_wide_int::or_not)
1990 (generic_wide_int::operator^, generic_wide_int::operator+
1991 (binary generic_wide_int::operator-, generic_wide_int::operator*):
1993 (operator~, unary operator-, operator==, operator!=, operator&)
1994 (operator|, operator^, operator+, binary operator-, operator*): New
1996 * expr.c (get_inner_reference): Use wi::bit_and_not.
1997 * fold-const.c (fold_binary_loc): Likewise.
1998 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
1999 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
2000 (bit_value_binop): Likewise.
2001 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
2002 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
2003 (extract_range_from_binary_expr_1): Likewise.
2004 (masked_increment): Likewise.
2005 (simplify_bit_ops_using_ranges): Likewise.
2007 2017-10-09 Martin Jambor <mjambor@suse.cz>
2010 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
2011 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
2012 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
2014 (hsa_fixup_mov_insn_type): New function.
2015 (hsa_op_with_type::get_in_type): Use it.
2016 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
2017 immediates in an assert.
2018 (hsa_op_with_type::extend_int_to_32bit): New method.
2019 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
2020 types. Convert to dest type if necessary.
2021 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
2022 (reg_for_gimple_ssa): Pass false as min32int to
2023 hsa_type_for_scalar_tree_type.
2024 (gen_hsa_addr): Fixup type when creating addresable temporary.
2025 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
2026 (gen_hsa_unary_operation): Extend operands and convert to dest type if
2027 necessary. Call hsa_fixup_mov_insn_type.
2028 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
2029 extend operands and convert to dest type if necessary.
2030 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
2031 to dest type if necessary.
2032 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
2033 if conversion nt necessary and size matches.
2034 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
2035 to dest type if necessary.
2036 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
2037 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
2039 (gen_hsa_clrsb): Likewise.
2040 (gen_hsa_ffs): Likewise.
2041 (gen_hsa_divmod): Extend operands and convert to dest type if
2043 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
2045 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
2047 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
2048 Remove empty default arguments. Use a brace block as output
2050 (conditional return): Ditto.
2052 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
2053 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
2055 (group_ending_nop): Ditto.
2056 (doloop_end): Ditto.
2057 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
2058 (splitters for those): Ditto.
2060 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
2062 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
2063 a conditional jump (and the compare for it) so that pc_rtx is the
2065 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
2066 for the deleted and renamed ctr<mode>_internal[234] patterns.
2067 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
2068 Delete second conditional return pattern.
2069 (ctr<mode>_internal2): Delete this second bdnz pattern.
2070 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
2071 (ctr<mode>_internal4): Delete this second bdz pattern.
2073 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
2075 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
2076 (always_initialized_rtx_for_ssa_name_p): New predicate.
2077 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
2078 (finish_out_of_ssa): Free new field of SA.
2079 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
2080 * tree-ssa-coalesce.c: Include tree-ssa.h.
2081 (get_parm_default_def_partitions): Remove extern keyword.
2082 (get_undefined_value_partitions): New function.
2083 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
2084 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
2086 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
2088 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
2090 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
2092 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
2094 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
2095 for targets that preffer 128bit.
2097 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
2099 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
2101 2017-10-08 Olivier Hainque <hainque@adacore.com>
2103 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
2104 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
2106 (thumb_set_return_address): Likewise.
2108 2017-10-08 Olivier Hainque <hainque@adacore.com>
2110 * common/config/arm/arm-common.c (arm_except_unwind_info):
2111 Handle DWARF2_UNWIND_INFO.
2113 2017-10-07 Michael Collison <michael.collison@arm.com>
2115 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
2118 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
2120 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
2121 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
2122 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
2123 defined, force the creation of a new block for a dispatch label.
2125 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
2127 * invoke.texi (Wsuggest-attribute=cold): Document.
2128 * common.opt (Wsuggest-attribute=cold): New
2129 * ipa-pure-const.c (warn_function_cold): New function.
2130 * predict.c (compute_function_frequency): Use it.
2131 * predict.h (warn_function_cold): Declare.
2133 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
2135 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
2138 2017-10-06 Martin Liska <mliska@suse.cz>
2140 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
2141 keyword for member functions.
2142 (struct sanopt_tree_couple): New struct.
2143 (struct sanopt_tree_couple_hash): New function.
2144 (struct sanopt_ctx): Add new hash_map.
2145 (has_dominating_ubsan_ptr_check): New function.
2146 (record_ubsan_ptr_check_stmt): Likewise.
2147 (maybe_optimize_ubsan_ptr_ifn): Likewise.
2148 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
2149 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
2151 2017-10-06 Sudakshina Das <sudi.das@arm.com>
2154 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
2155 aarch64_simd_valid_immediate on CONST_VECTORs.
2156 (aarch64_reg_or_bic_imm): Likewise.
2158 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
2160 PR rtl-optimization/82396
2161 * haifa-sched.c (ready_sort_real): Disable qsort checking.
2163 2017-10-06 Sebastian Pop <sebpop@gmail.com>
2165 * graphite-dependences.c (scop_get_reads): Move code to...
2166 (scop_get_must_writes): Move code to...
2167 (scop_get_may_writes): Move code to...
2168 (scop_get_reads_and_writes): ... here.
2169 (scop_get_dependences): Call scop_get_reads_and_writes.
2171 2017-10-06 Jakub Jelinek <jakub@redhat.com>
2173 PR tree-optimization/82434
2174 * fold-const.h (can_native_encode_type_p,
2175 can_native_encode_string_p): Remove.
2176 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
2177 don't encode anything, just return what would be otherwise returned.
2178 (native_encode_fixed, native_encode_complex, native_encode_vector):
2180 (native_encode_string): Likewise. Inline by hand
2181 can_native_encode_string_p.
2182 (can_native_encode_type_p): Remove.
2183 (can_native_encode_string_p): Remove.
2184 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
2185 STRING_CSTs using can_native_encode_string_p, test all
2186 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
2187 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
2188 argument from native_encode_expr.
2189 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
2190 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
2193 2017-10-06 Richard Biener <rguenther@suse.de>
2195 PR tree-optimization/82397
2196 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
2197 operand_equal_p but rely on data_ref_compare_tree for detecting
2199 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
2200 to match up with dr_group_sort_cmp.
2202 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2205 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
2207 * config/s390/s390-builtin-types.def: Regenerate.
2209 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2212 * config/s390/s390-builtin-types.def: Regenerate.
2213 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
2214 Change flag from B_VXE to B_VX.
2215 (s390_vec_min_dbl): Remove B_VXE flag.
2217 2017-10-06 Richard Biener <rguenther@suse.de>
2219 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
2220 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
2221 translate_isl_ast_to_gimple::is_valid_rename,
2222 translate_isl_ast_to_gimple::get_rename,
2223 translate_isl_ast_to_gimple::get_def_bb_for_const,
2224 translate_isl_ast_to_gimple::get_new_name,
2225 translate_isl_ast_to_gimple::collect_all_ssa_names,
2226 translate_isl_ast_to_gimple::copy_loop_phi_args,
2227 translate_isl_ast_to_gimple::collect_all_ssa_names,
2228 translate_isl_ast_to_gimple::copy_loop_phi_args,
2229 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
2230 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
2231 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
2232 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
2233 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
2234 translate_isl_ast_to_gimple::copy_cond_phi_args,
2235 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
2236 translate_isl_ast_to_gimple::edge_for_new_close_phis,
2237 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
2238 translate_isl_ast_to_gimple::rename_uses,
2239 translate_isl_ast_to_gimple::rename_all_uses): Remove.
2240 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
2241 (set_rename_for_each_def): Likewise.
2242 (graphite_copy_stmts_from_block): Handle debug stmt resetting
2243 here. Handle rewriting SCEV analyzable uses here.
2244 (copy_bb_and_scalar_dependences): Generate code for PHI
2246 (graphite_regenerate_ast_isl): Adjust.
2247 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
2248 (add_write, add_read): New functions.
2249 (build_cross_bb_scalars_def): Use it and simplify.
2250 (build_cross_bb_scalars_use): Likewise.
2251 (graphite_find_cross_bb_scalar_vars): Inline into...
2252 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
2253 simulating out-of-SSA. Compute liveout and add dependencies.
2254 (build_scops): Force an empty entry block.
2255 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
2257 (sese_build_liveouts): Declare.
2258 (sese_trivially_empty_bb_p): Likewise.
2259 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
2260 compute liveout and debug_liveout.
2261 (sese_bad_liveouts_use): Remove.
2262 (sese_reset_debug_liveouts_bb): Likewise.
2263 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
2264 (sese_build_liveouts): Build liveout and debug_liveout and store
2266 (new_sese_info): Adjust.
2267 (free_sese_info): Likewise.
2268 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
2269 do not build liveout here.
2270 (move_sese_in_condition): Adjust region entry.
2271 (scev_analyzable_p): Match up with chrec_apply requirements.
2272 (sese_trivially_empty_bb_p): New.
2273 * tree-into-ssa.c (get_reaching_def): Properly support generating
2274 default-defs for incremental rewrite of anonymous names.
2276 2017-10-06 Richard Biener <rguenther@suse.de>
2278 * graphite-sese-to-poly.c (extract_affine): For casts increasing
2279 precision do not perform modulo reduction.
2281 2017-10-06 Richard Biener <rguenther@suse.de>
2283 PR tree-optimization/82436
2284 * tree-vect-slp.c (vect_supported_load_permutation_p): More
2285 conservatively choose the vectorization factor when checking
2286 whether we can perform the required load permutation.
2287 (vect_transform_slp_perm_load): Assert when we may not fail.
2289 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
2291 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
2292 message for incompatible -msdata=* and -mcall-* options.
2294 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2296 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
2297 rate for post-reload scheduling.
2299 2017-10-05 Tamar Christina <tamar.christina@arm.com>
2301 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
2303 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2305 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
2306 to improve monte carlo in scimark.
2308 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2310 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
2311 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
2312 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
2313 pentium4_cost, nocona_cost): Set reassociation width to 1.
2314 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
2315 width to 2 for fp operations and 1 otherwise.
2316 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
2318 (atom_cost): Set reassociation width to 2.
2319 (slm_cost, generic_cost): Set fp reassociation width
2320 to 2 and 1 otherwise.
2321 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
2322 (core_cost): Set fp reassociation width to 4 and vector to 2.
2323 (ix86_reassociation_width): Rewrite using cost table; special case
2324 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
2325 and TARGET_AVX128_OPTIMAL.
2326 * config/i386/i386.h (processor_costs): Add
2327 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
2328 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
2329 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
2330 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
2331 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
2332 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
2334 2017-10-05 Nathan Sidwell <nathan@acm.org>
2336 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
2338 2017-10-05 Tamar Christina <tamar.christina@arm.com>
2340 * config/arm/arm.c (arm_test_fpu_data): New.
2341 (arm_run_selftests): Call arm_test_fpu_data.
2343 2017-10-04 Nathan Sidwell <nathan@acm.org>
2345 * toplev.c (toplev::main): Remove excess parens on pretty_printer
2347 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
2349 2017-10-04 Sudakshina Das <sudi.das@arm.com>
2351 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
2352 check type for aarch64_simd_valid_immediate.
2353 (aarch64_output_simd_mov_immediate): Update prototype.
2354 (aarch64_simd_valid_immediate): Update prototype.
2355 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
2356 support for ORR-immediate.
2357 (and<mode>3): modified pattern to add support for BIC-immediate.
2358 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
2359 now checks for valid immediate for BIC and ORR based on new enum
2361 (aarch64_output_simd_mov_immediate): Function now used to output
2362 BIC/ORR imm as well based on new enum argument.
2363 * config/aarch64/constraints.md (Do): New vector immediate constraint.
2365 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
2366 (aarch64_reg_or_bic_imm): Likewise.
2368 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2370 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
2371 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
2373 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
2377 PR rtl-optimization/82396
2378 * haifa-sched.c (autopref_multipass_init): Simplify
2380 (autopref_rank_data): Simplify sort order.
2381 * sched-int.h (autopref_multipass_data_): Remove
2382 multi_mem_insn_p, min_offset and max_offset.
2384 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2386 * doc/sourcebuild.texi: Document vect_peeling_profitable.
2388 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2390 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
2393 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2395 * doc/sourcebuild.texi: Document vect_long_mult.
2397 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
2399 PR tree-optimization/82413
2400 * fold-const.c (build_range_check): Use widest_int when comparing
2401 the maximum ETYPE value with HIGH.
2403 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
2405 PR rtl-optimization/82396
2406 * haifa-sched.c (autopref_multipass_init): Simplify
2408 (autopref_rank_data): Simplify sort order.
2409 * sched-int.h (autopref_multipass_data_): Remove
2410 multi_mem_insn_p, min_offset and max_offset.
2412 2017-10-04 Jakub Jelinek <jakub@redhat.com>
2414 PR tree-optimization/82381
2415 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
2416 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
2419 PR tree-optimization/82374
2420 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
2421 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
2422 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
2423 current_function_decl to the new decl.
2425 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
2427 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
2428 helper macro for IEEE float128 hardware built-in functions.
2429 (SQRTF128_ODD): Add built-in functions with the round-to-odd
2431 (TRUNCF128_ODD): Likewise.
2432 (ADDF128_ODD): Likewise.
2433 (SUBF128_ODD): Likewise.
2434 (MULF128_ODD): Likewise.
2435 (DIVF128_ODD): Likewise.
2436 (FMAF128_ODD): Likewise.
2437 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
2438 UNSPEC_TRUNC_ROUND_TO_ODD.
2439 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
2440 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
2441 floating point round to odd instructions.
2442 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
2443 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
2444 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
2445 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
2446 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
2447 (trunc<mode>sf2_hw): Change the truncate with round to odd
2448 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
2449 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
2450 to odd hardware instructions.
2451 (sub<mode>3_odd): Likewise.
2452 (mul<mode>3_odd): Likewise.
2453 (div<mode>3_odd): Likewise.
2454 (sqrt<mode>2_odd): Likewise.
2455 (fma<mode>4_odd): Likewise.
2456 (fms<mode>4_odd): Likewise.
2457 (nfma<mode>4_odd): Likewise.
2458 (nfms<mode>4_odd): Likewise.
2459 (trunc<mode>df2_odd): Change the truncate with round to odd
2460 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
2462 * doc/extend.texi (PowerPC built-in functions): Update documentation
2463 for existing IEEE float128-bit built-in functions. Add built-in
2464 functions that generate the IEEE 128-bit floating point round to
2467 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
2469 PR rtl-optimization/77729
2470 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
2471 to (X&(C1&~C2))|C2 transformations.
2473 2017-10-03 Martin Jambor <mjambor@suse.cz>
2475 PR tree-optimization/82363
2476 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
2477 mismatch, mark lacc written regardless of racc.
2479 2017-10-03 Jakub Jelinek <jakub@redhat.com>
2481 PR tree-optimization/82381
2482 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
2483 stmt_to_insert nor wheather SSA_NAMEs are default defs.
2484 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
2485 fallthrough into reassoc_stmt_dominates_stmt_p.
2488 * combine.c (combine_instructions): Don't combine in unreachable
2491 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
2494 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
2495 function to not use the have_cpu variable. Do not set cpu_index,
2496 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
2498 (rs6000_valid_attribute_p): Remove duplicate initializations of
2499 old_optimize and func_optimize.
2500 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
2501 (rs6000_activate_target_options): Make global.
2502 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
2505 2017-10-02 Jakub Jelinek <jakub@redhat.com>
2507 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
2508 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
2509 Set *poffset to 0 and *psize and *pmax_size to -1 if
2510 *poffset + *psize overflows in HOST_WIDE_INT.
2512 PR tree-optimization/82387
2513 PR tree-optimization/82388
2514 PR tree-optimization/82389
2515 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
2516 instead of live_bytes non-NULL.
2518 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
2521 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
2524 2017-10-02 Richard Biener <rguenther@suse.de>
2526 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
2527 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
2528 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
2530 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
2532 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
2533 requested precision matches the type's.
2534 * calls.c (alloc_max_size): Calculate the new candidate size as
2535 a widest_int and use wi::to_widest when comparing it with the
2536 current candidate size.
2537 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
2538 zero rather than integer_zero_node.
2539 * match.pd: Check for a no-op conversion before using wi::add
2540 rather than after. Use tree_to_uhwi when summing small shift
2541 counts into an unsigned int.
2543 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
2544 Alan Hayward <alan.hayward@arm.com>
2545 David Sherwood <david.sherwood@arm.com>
2548 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
2549 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
2550 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
2551 POINTER_AND_FP_REGS.
2553 2017-10-02 Richard Biener <rguenther@suse.de>
2555 PR tree-optimization/82355
2556 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
2557 a mapping for the enclosing loop but avoid generating one for
2559 (copy_bb_and_scalar_dependences): Remove premature codegen
2560 error on PHIs in blocks duplicated into multiple places.
2561 * graphite-scop-detection.c
2562 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
2563 in the region use it as loop and nest to analyze the DR in.
2564 (try_generate_gimple_bb): Likewise.
2565 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
2566 (add_loop_constraints): For blocks in a loop not in the region
2567 create a dimension with a single iteration.
2568 * sese.h (gbb_loop_at_index): Remove assert.
2570 2017-10-01 Kevin Buettner <kevinb@redhat.com>
2572 * omp-expand.c (adjust_context_scope): New function.
2573 (expand_parallel_call): Call adjust_context_scope.
2575 2017-10-01 Jeff Law <law@redhat.com>
2577 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
2578 dom_opt_dom_walker class with direct access to private members.
2579 Add comments. Call test_for_singularity.
2580 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
2581 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
2582 m_dummy_cond anymore.
2583 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
2585 (pass_dominator:execute): Build the dummy_cond here and pass it
2586 to the dom_opt_dom_walker ctor.
2587 (test_for_singularity): New function.
2589 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
2590 Maya Rashish <coypu@sdf.org>
2592 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
2593 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
2594 (alpha*-*-netbsd*) Use nbsd_tm_file.
2595 (arm*-*-netbsdelf*) Likewise.
2596 (i[34567]86-*-netbsdelf*) Likewise.
2597 (x86_64-*-netbsd*) Likewise.
2598 (mips*-*-netbsd*) Likewise.
2599 (powerpc-*-netbsd*) Likewise.
2600 (sh*-*-netbsd*) Likewise.
2601 (sparc-*-netbsdelf*) Likewise.
2602 (sparc64-*-netbsd*) Likewise.
2603 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
2605 (vax-*-netbsdelf*) Likewise.
2606 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
2607 (UINT_FAST8_TYPE) Likewise.
2608 (INT_FAST16_TYPE) Check CHAR_FAST16.
2609 (UINT_FAST16_TYPE) Likewise.
2611 2017-09-30 Jakub Jelinek <jakub@redhat.com>
2614 * config/i386/i386.md
2615 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
2616 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
2617 *divmodsi4_zext_2): New define_insn_and_split.
2618 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
2619 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
2620 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
2621 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
2622 New define_insn_and_split.
2623 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
2624 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
2625 operands[1] having DImode when mode is SImode.
2627 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
2628 always SImode for DIV and MOD in REG_EQUAL notes.
2630 2017-09-29 Yury Gribov <tetra2005@gmail.com>
2633 * match.pd: Fix handling of NaNs in pattern.
2635 2017-09-29 Jeff Law <law@redhat.com>
2637 * sbitmap.c (bitmap_bit_in_range_p): New function.
2638 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
2639 * tree-ssa-dse.c (live_bytes_read): New function.
2640 (dse_classify_store): Ignore reads of dead bytes.
2642 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
2643 typos and whitespace errors.
2644 * config/i386/predicates.md (address_no_seg_operand): Likewise.
2645 * config/s390/s390.c (s390_emit_prologue): Likewise.
2647 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
2650 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
2651 with a symbol for LRA.
2653 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
2655 PR rtl-optimization/82338
2656 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
2658 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
2660 * genmodes.c (calc_wider_mode): Suppress qsort macro.
2661 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
2662 (qsort_chk): Declare.
2663 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
2664 (qsort_chk): New function.
2666 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2668 PR tree-optimization/82337
2669 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
2670 phi definition if the PHI result appears in an abnormal PHI.
2671 (find_basis_for_base_expr): Don't record a basis if the LHS of the
2672 basis appears in an abnormal PHI.
2674 2017-09-29 Richard Biener <rguenther@suse.de>
2676 * graphite-isl-ast-to-gimple.c
2677 (translate_isl_ast_to_gimple::set_codegen_error): New function.
2678 (binary_op_to_tree): Use it.
2679 (get_rename_from_scev): Likewise.
2680 (copy_loop_phi_nodes): Likewise.
2681 (copy_bb_and_scalar_dependences): Likewise.
2682 (translate_pending_phi_nodes): Likewise.
2684 2017-09-29 Jakub Jelinek <jakub@redhat.com>
2687 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
2688 for movabsq $(i32 << shift), r64.
2690 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
2692 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
2693 index when encoding %esp as %rsp to avoid 0x67 prefix.
2695 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
2697 * config/i386/i386.md (*movsf_internal, *movdf_internal):
2698 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
2700 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
2702 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
2703 Extensions with more than 16 double VFP registers.
2704 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
2705 to_clear_mask and all code related to it. Replace the remaining
2706 entry by a sbitmap and adapt code accordingly.
2708 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
2710 * brig-builtins.def: Change pure attributes to const.
2712 2017-09-28 Joseph Myers <joseph@codesourcery.com>
2714 * config.gcc (default_gnu_indirect_function): Default to yes for
2715 sparc*-*-linux* with glibc.
2717 2017-09-28 Joseph Myers <joseph@codesourcery.com>
2719 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
2720 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
2721 when creating .init_array and .fini_array sections with priority
2724 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
2727 * config/aarch64/aarch64.c
2728 (aarch64_builtin_support_vector_misalignment): Always return false
2729 when misalignment is unknown.
2731 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
2733 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
2734 this function to return false if the definition used by the swap
2735 instruction is artificial, or if the memory address from which the
2736 constant value is loaded is not represented by a base address held
2737 in a register or if the base address register is a frame or stack
2738 pointer. Additionally, return false if the base address of the
2739 loaded constant is a SYMBOL_REF but is not considered to be a
2741 (replace_swapped_load_constant): New function.
2742 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
2743 loaded constant vector with a load of a swapped constant vector.
2745 2017-09-27 Carl Love <cel@us.ibm.com>
2747 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
2748 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
2749 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
2752 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
2754 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
2755 first, always call autopref_rank_data otherwise.
2757 2017-09-27 Richard Biener <rguenther@suse.de>
2759 * graphite-scop-detection.c (find_scop_parameters): Move
2760 loop bound handling ...
2761 (gather_bbs::before_dom_children): ... here, avoiding the need
2762 to build scop_info->loop_nest.
2763 (record_loop_in_sese): Remove.
2764 * sese.h (sese_info_t::loop_nest): Remove.
2765 * sese.c (new_sese_info): Do not allocate loop_nest.
2766 (free_sese_info): Do not free loop_nest.
2768 2017-09-27 Jakub Jelinek <jakub@redhat.com>
2771 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
2772 lhs from calls if the lhs has addressable type.
2774 2017-09-27 Richard Biener <rguenther@suse.de>
2776 * graphite.h (scop::max_alias_set): New member.
2777 * graphite-scop-detection.c: Remove references to non-existing
2778 --param in comments.
2779 (build_alias_sets): Record the maximum alias set used for drs.
2780 (build_scops): Support zero as unlimited for
2781 --param graphite-max-arrays-per-scop.
2782 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
2784 (build_poly_sr_1): ... here. Compute alias set based on the
2785 maximum alias set used for drs rather than
2786 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
2788 2017-09-27 Richard Biener <rguenther@suse.de>
2790 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
2791 --param loop-block-tile-size=0 to disable tiling.
2793 2017-09-27 Richard Biener <rguenther@suse.de>
2795 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
2796 (graphite-max-nb-scop-params): Document special value zero.
2797 * domwalk.h (dom_walker::STOP): New symbolical constant.
2798 (dom_walker::dom_walker): Add optional parameter for bb to
2800 (dom_walker::~dom_walker): Declare.
2801 (dom_walker::before_dom_children): Document STOP return value.
2802 (dom_walker::m_user_bb_to_rpo): New member.
2803 (dom_walker::m_bb_to_rpo): Likewise.
2804 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
2805 mapping here if not provided by the user.
2806 (dom_walker::~dom_walker): Free bb to RPO mapping if not
2807 provided by the user.
2808 (dom_walker::STOP): Define.
2809 (dom_walker::walk): Do not compute bb to RPO mapping here.
2810 Support STOP return value from before_dom_children to stop
2812 * graphite-optimize-isl.c (optimize_isl): If the schedule
2813 is the same still generate code if -fgraphite-identity
2814 or -floop-parallelize-all are given.
2815 * graphite-scop-detection.c: Include cfganal.h.
2816 (gather_bbs::gather_bbs): Get and pass through bb to RPO
2818 (gather_bbs::before_dom_children): Return STOP for BBs
2820 (build_scops): Compute bb to RPO mapping and pass it to
2821 the domwalk. Treat --param graphite-max-nb-scop-params=0
2822 as not limiting the number of params.
2823 * graphite.c (graphite_initialize): Remove limit on the
2824 number of basic-blocks in a function.
2825 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
2826 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
2827 default value of 10.
2829 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2831 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
2832 Adjust code to eliminate needing to do the shift right 32-bits
2833 operation after XSCVDPSPN.
2835 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2837 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
2838 ((X / Y) != 0 -> X >= Y): Likewise.
2840 2017-09-26 Carl Love <cel@us.ibm.com>
2842 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
2843 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
2844 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
2845 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
2846 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
2847 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
2848 definitions and overloading.
2849 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
2850 statement for P9V_BUILTIN_XST_LEN_R.
2851 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
2852 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
2853 define_expand and define_insn for the instructions and builtins.
2854 * doc/extend.texi: Update the built-in documentation file for the new
2856 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
2857 define_insn for the instructions
2859 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
2862 * gcc/config/netbsd-protos.h: New file.
2863 * gcc/config/netbsd.c: New file.
2864 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
2865 * gcc/config/t-netbsd: New file.
2866 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
2867 (tmake_file) Add t-netbsd.
2868 (extra_objs) Add netbsd.o.
2870 2017-09-26 Janus Weil <janus@gcc.gnu.org>
2874 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
2876 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2878 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
2879 sign extension from a vector register to a GPR by doing a 32-bit
2880 direct move and then an EXTSW.
2881 (extendsi<mode>2 splitter): Likewise.
2882 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
2883 right or vector extract after doing XSCVDPSPN. Use
2884 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
2886 (movdi_from_sf_zero_ext): Likewise.
2887 (reload_gpr_from_vsxsf): Likewise.
2888 (p8_mfvsrd_4_disf): Delete, no longer used.
2889 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
2890 then needing to move the SFmode to a GPR to use the XSCVDPSP
2891 instruction instead of FRSP and XSCVDPSPN.
2892 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
2893 it is adjacent to the other XSCVSPDP insns.
2894 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
2895 SFmode to be in traditional Altivec registers.
2896 (vsx_xscvdpspn): Eliminate useless alternative constraint.
2897 (vsx_xscvspdpn): Likewise.
2898 (vsx_xscvspdpn_scalar): Likewise.
2900 2017-09-26 Martin Jambor <mjambor@suse.cz>
2902 * tree-sra.c (compare_access_positions): Put integral types first,
2903 stabilize sorting of integral types, remove conditions putting
2904 non-full-precision integers last.
2905 (sort_and_splice_var_accesses): Disable scalarization if a
2906 non-integert would be represented by a non-full-precision integer.
2908 2017-09-26 Joseph Myers <joseph@codesourcery.com>
2910 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
2911 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
2912 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
2913 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
2914 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
2915 conditionals inside the function instead of around it. Call
2916 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
2917 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
2919 2017-09-26 Richard Biener <rguenther@suse.de>
2921 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
2923 (scop_detection::build_scop_breadth): ... this. Removed.
2924 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
2925 (scop_detection::harmful_stmt_in_bb): Likewise.
2926 (scop_detection::graphite_can_represent_stmt): Likewise.
2927 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
2928 (scop_detection::can_represent_loop): Remove recursion, fold in ...
2929 (scop_detection::can_represent_loop_1): ... this. Removed.
2930 (scop_detection::harmful_loop_in_region): Simplify after inlining
2931 the above and remove more quadraticness.
2932 (build_scops): Adjust.
2933 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
2936 2017-09-26 Jakub Jelinek <jakub@redhat.com>
2939 * config/i386/i386.c (ix86_print_operand_address_as): Only test
2940 REGNO (base) == SP_REG if base is a REG.
2943 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
2944 if it is different SSA_NAME.
2945 (optimize_range_tests_cmp_bitwise): New function.
2946 (optimize_range_tests): Call it.
2948 2017-09-26 Richard Biener <rguenther@suse.de>
2950 PR tree-optimization/82321
2951 * graphite.c (canonicalize_loop_closed_ssa): Properly check
2952 for the def being inside the loop.
2954 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2956 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
2958 * config/s390/s390-builtins.def: Fix constraint on op4.
2960 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2962 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
2963 independent expanders.
2964 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
2965 ("vec_ordered", "vec_unordered"): New expanders.
2967 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2969 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
2972 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2974 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
2975 vec_unpacks_lo_v16qi.
2976 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
2978 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2980 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
2981 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
2982 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
2984 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2986 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
2988 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
2990 ("*vec_slb<mode>"): New insn pattern.
2991 ("vec_shr_<mode>"): New expander.
2992 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
2993 and force the shift count operand to V16QImode.
2994 ("vec_srb<mode>"): Set shift count mode to V16QI.
2996 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2998 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
2999 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
3000 ("vec_widen_smult_hi_<mode>"): New expander definitions.
3002 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
3005 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
3007 2017-09-26 Richard Biener <rguenther@suse.de>
3009 PR tree-optimization/82320
3010 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
3013 2017-09-25 Jeff Law <law@redhat.com>
3015 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
3016 prototype for new argument.
3017 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
3018 mostly extracted from rs6000_emit_allocate_stack.
3019 (rs6000_emit_probe_stack_range_stack_clash): New function.
3020 (rs6000_emit_allocate_stack): Call
3021 rs6000_emit_probe_stack_range_stack_clash as needed.
3022 (rs6000_emit_probe_stack_range): Add additional argument
3023 to call to gen_probe_stack_range{si,di}.
3024 (output_probe_stack_range): New.
3025 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
3026 (output_probe_stack_range_stack_clash): New.
3027 (rs6000_emit_prologue): Emit notes into dump file as requested.
3028 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
3029 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
3030 Add additional operand and pass it to output_probe_stack_range.
3032 2017-09-25 Bin Cheng <bin.cheng@arm.com>
3034 PR tree-optimization/82163
3035 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
3036 (checking_verify_loop_closed_ssa): New parameter.
3037 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
3038 (check_loop_closed_ssa_stmt): Delete.
3039 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
3040 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
3041 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
3044 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
3046 * brig-builtins.def: Treat HSAIL barrier builtins as
3047 setjmp/longjump style functions.
3049 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
3051 * target.def (constant_alignment): New hook.
3052 * defaults.h (CONSTANT_ALIGNMENT): Delete.
3053 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
3054 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
3055 * doc/tm.texi: Regenerate.
3056 * targhooks.h (default_constant_alignment): Declare.
3057 (constant_alignment_word_strings): Likewise.
3058 * targhooks.c (default_constant_alignment): New function.
3059 (constant_alignment_word_strings): Likewise.
3060 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
3061 instead of CONSTANT_ALIGNMENT.
3062 * varasm.c (align_variable, get_variable_align, build_constant_desc)
3063 (force_const_mem): Likewise.
3064 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
3065 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
3066 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
3067 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3068 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
3070 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
3071 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3072 constant_alignment_word_strings.
3073 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
3074 (CONSTANT_ALIGNMENT): Likewise.
3075 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3076 (arm_constant_alignment): New function.
3077 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
3078 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3079 constant_alignment_word_strings.
3080 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
3081 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3082 constant_alignment_word_strings.
3083 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
3084 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3085 (cris_constant_alignment): New function.
3086 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
3087 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3088 (epiphany_constant_alignment): New function.
3089 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
3090 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3091 constant_alignment_word_strings.
3092 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
3093 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3094 constant_alignment_word_strings.
3095 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
3096 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3097 constant_alignment_word_strings.
3098 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
3099 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
3100 * config/i386/i386.c (ix86_constant_alignment): Make static.
3101 Use the same interface as the target hook.
3102 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3103 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
3104 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3105 constant_alignment_word_strings.
3106 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
3107 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
3108 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3109 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
3110 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3111 constant_alignment_word_strings.
3112 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
3113 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3114 constant_alignment_word_strings.
3115 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
3116 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3117 constant_alignment_word_strings.
3118 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
3119 * config/microblaze/microblaze.c (microblaze_constant_alignment):
3121 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3122 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
3123 * config/mips/mips.c (mips_constant_alignment): New function.
3124 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3125 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
3126 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
3127 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3128 (mmix_constant_alignment): Make static. Use the same interface
3130 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
3131 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3132 constant_alignment_word_strings.
3133 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
3134 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3135 constant_alignment_word_strings.
3136 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
3137 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3138 constant_alignment_word_strings.
3139 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
3140 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3141 (rs6000_constant_alignment): New function.
3142 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
3143 * config/riscv/riscv.c (riscv_constant_alignment): New function.
3144 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3145 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
3146 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3147 (rs6000_constant_alignment): New function.
3148 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
3149 * config/s390/s390.c (s390_constant_alignment): New function.
3150 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3151 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
3152 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3153 constant_alignment_word_strings.
3154 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
3155 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3156 (sparc_constant_alignment): New function.
3157 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
3158 * config/spu/spu.c (spu_constant_alignment): New function.
3159 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3160 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
3161 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3162 constant_alignment_word_strings.
3163 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
3164 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3165 constant_alignment_word_strings.
3166 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
3167 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3168 constant_alignment_word_strings.
3169 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
3170 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3171 (visium_constant_alignment): New function.
3172 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
3173 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3174 (xtensa_constant_alignment): New function.
3175 * system.h (CONSTANT_ALIGNMENT): Poison.
3177 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
3179 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
3180 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
3181 (rs6000_builtin_valid_without_lhs): New helper function.
3182 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3183 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
3185 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
3187 * target.h (vec_perm_indices): Use unsigned short rather than
3189 (auto_vec_perm_indices): Likewise.
3190 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
3191 Use unsigned int rather than unsigned char.
3192 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
3194 2017-09-25 Richard Biener <rguenther@suse.de>
3196 * cfgloop.h (sort_sibling_loops): Declare.
3197 * cfgloop.c (sort_sibling_loops_cmp): New helper.
3198 (sort_sibling_loops): New function sorting the sibling loop list
3200 * graphite.c (graphite_transform_loops): Sort sibling loops.
3202 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
3204 * target.def (vec_perm_const_ok): Change sel parameter to
3206 * optabs-query.c (can_vec_perm_p): Update accordingly.
3207 * doc/tm.texi: Regenerate.
3208 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
3209 auto_vec_perm_indices and remove separate nelt field.
3210 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
3211 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
3212 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
3213 (aarch64_expand_vec_perm_const): Update accordingly.
3214 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
3215 to vec_perm_indices.
3216 * config/arm/arm.c (expand_vec_perm_d): Change perm to
3217 auto_vec_perm_indices and remove separate nelt field.
3218 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
3219 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
3220 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
3222 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
3223 to vec_perm_indices.
3224 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
3225 sel to vec_perm_indices.
3226 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
3227 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
3228 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
3230 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
3233 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
3236 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
3237 on the FUNCTION_DECL function context if it has a DIE that is a
3240 2017-09-25 Richard Biener <rguenther@suse.de>
3242 PR tree-optimization/82285
3243 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
3246 2017-09-25 Tom de Vries <tom@codesourcery.com>
3250 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
3253 2017-09-25 Richard Biener <rguenther@suse.de>
3255 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
3256 ISL errors other than isl_error_quota happen. Dump if the
3257 schedule is the same.
3258 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
3259 errors instead of aborting inside ISL.
3261 2017-09-25 Iain Sandoe <iain@codesourcery.com>
3264 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
3265 of libgcc_eh for m64.
3266 * config/i386/darwin64.h: Likewise.
3268 2017-09-25 Richard Biener <rguenther@suse.de>
3271 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
3272 attribute for incomplete types nor twice for complete ones.
3274 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
3277 * config/i386/i386.c (ix86_print_operand_address_as): Encode
3278 %esp as %rsp to avoid 0x67 prefix if there is no index or base
3281 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
3284 * config/i386/i386.opt (mprefer-avx256): Use
3285 ix86_target_flags variable.
3286 * config/i386/i386.c (ix86_target_string): Move
3287 -mprefer-avx256 to flag2_opts.
3289 2017-09-22 Jakub Jelinek <jakub@redhat.com>
3292 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
3293 and x != -1 | y != -1 into (x & y) != -1.
3295 2017-09-22 Steve Ellcey <sellcey@cavium.com>
3297 * config.gcc: Add new case statement to set
3298 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
3299 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
3300 s390x-*-linux* case statements. Added aarch64 to the list of
3301 supported architectures.
3303 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3305 PR tree-optimization/82289
3306 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
3307 STMT_VINFO_RELEVANT_P.
3309 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3310 Alan Hayward <alan.hayward@arm.com>
3311 David Sherwood <david.sherwood@arm.com>
3313 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
3314 for VR_RANGE only; don't allow VR_ANTI_RANGE.
3315 (extract_range_from_binary_expr_1): Don't call
3316 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
3318 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3319 Alan Hayward <alan.hayward@arm.com>
3320 David Sherwood <david.sherwood@arm.com>
3322 * target.def (preferred_vector_alignment): New hook.
3323 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
3325 * doc/tm.texi: Regenerate.
3326 * targhooks.h (default_preferred_vector_alignment): Declare.
3327 * targhooks.c (default_preferred_vector_alignment): New function.
3328 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
3330 (DR_TARGET_ALIGNMENT): New macro.
3331 (aligned_access_p): Update commentary.
3332 (vect_known_alignment_in_bytes): New function.
3333 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
3335 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
3336 Calculate the misalignment based on the target alignment rather than
3338 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
3339 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
3340 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
3341 the target alignment, rather than masking the element misalignment
3342 with the number of elements in a vector. Also use the target
3343 alignment when calculating the maximum number of peels.
3344 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
3345 instead of TYPE_ALIGN_UNIT.
3346 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
3347 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
3348 (vect_create_addr_base_for_vector_ref): Update call accordingly.
3349 (vect_create_data_ref_ptr): Likewise.
3350 (vect_setup_realignment): Realign by ANDing with
3351 -DR_TARGET_MISALIGNMENT.
3352 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
3353 the number of peels based on DR_TARGET_ALIGNMENT.
3354 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
3355 with the guaranteed alignment boundary when deciding whether
3357 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
3358 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
3359 (ensure_base_align): Remove stmt_info parameter. Get the
3360 target base alignment from DR_TARGET_ALIGNMENT.
3361 (vectorizable_store): Update call accordingly. Interpret
3362 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
3364 (vectorizable_load): Likewise.
3366 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3367 Alan Hayward <alan.hayward@arm.com>
3368 David Sherwood <david.sherwood@arm.com>
3370 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
3371 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
3372 (vect_enhance_data_refs_alignment): Likewise.
3374 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
3376 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
3377 error. Only quit immediately if parsing is complete.
3378 (BEGIN): Initialize fatal_err and parse_done.
3379 (begin fpu, end fpu): Check number of arguments.
3380 (begin arch, end arch): Likewise.
3381 (begin cpu, end cpu): Likewise.
3382 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
3383 (optalias): Likewise.
3385 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
3387 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
3388 * config/arm/arm-isa.h: Delete. Move definitions to ...
3389 * arm-cpus.in: ... here. Use new feature and fgroup values.
3390 * config/arm/arm.c (arm_option_override): Use lower case for feature
3392 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
3393 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
3394 * config/arm/parsecpu.awk (END): Add new command 'isa'.
3396 (print_isa_bits_for): New function.
3397 (gen_isa): New function.
3398 (gen_comm_data): Use print_isa_bits_for.
3399 (define feature): New keyword.
3400 (define fgroup): New keyword.
3401 * config/arm/t-arm (TM_H): Remove.
3402 (GTM_H): Add arm-isa.h.
3403 (arm-isa.h): Add rule to generate file.
3404 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
3405 case for feature bit names.
3407 2017-09-22 Richard Biener <rguenther@suse.de>
3409 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
3411 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
3412 print of no dependency loops ...
3413 * graphite.c (graphite_transform_loops): ... here.
3414 (canonicalize_loop_closed_ssa_form): Work from inner to outer
3416 (same_close_phi_node, remove_duplicate_close_phi,
3417 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
3418 (canonicalize_loop_closed_ssa): ... here and simplify.
3419 * graphite-optimize-isl.c: Include tree-vectorizer.h.
3420 (optimize_isl): Use dump_printf_loc to tell when we stopped
3421 optimizing because of an ISL timeout.
3423 2017-09-22 Richard Biener <rguenther@suse.de>
3425 PR tree-optimization/82291
3426 * tree-if-conv.c (predicate_mem_writes): Make sure to
3427 remove writes in blocks predicated with false.
3429 2017-09-22 Richard Biener <rguenther@suse.de>
3431 * sese.c: Include cfganal.h.
3432 (if_region_set_false_region): Remove.
3433 (create_if_region_on_edge): Likewise.
3434 (move_sese_in_condition): Re-implement without destroying
3437 2017-09-22 Richard Biener <rguenther@suse.de>
3439 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
3440 Verify both BBs contain loop PHI nodes before dispatching to
3442 (graphite_regenerate_ast_isl): Do not recompute dominators,
3443 do not verify three times. Restructure for clarity.
3444 * graphite-scop-detection.c (same_close_phi_node,
3445 remove_duplicate_close_phi, make_close_phi_nodes_unique,
3446 defined_in_loop_p, canonicalize_loop_closed_ssa,
3447 canonicalize_loop_closed_ssa_form): Simplify, remove excess
3448 checking and SSA rewrite, move to ...
3449 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
3450 (graphite_initialize): Do not pass in ctx, do not reset the
3451 SCEV cache, compute only dominators.
3452 (graphite_transform_loops): Allocate ISL ctx after
3453 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
3454 Maintain post-dominators only around build_scops.
3455 * sese.c (if_region_set_false_region): Make static. Free
3456 and recompute dominators.
3457 (move_sese_in_condition): Assert we don't get called with
3458 post-dominators computed.
3459 * sese.h (if_region_set_false_region): Remove.
3461 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
3463 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
3464 mode attribute for TARGET_AVX512VL.
3466 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
3468 * config/i386/i386.opt (mprefer-avx256): New option.
3469 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
3471 (ix86_preferred_simd_mode): Return 256-bit AVX modes
3472 for TARGET_PREFER_AVX256.
3473 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
3475 2017-09-21 Jeff Law <law@redhat.com>
3477 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
3478 Fix dump output if the only stack space is for pushed registers.
3480 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3482 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
3485 2017-09-21 Martin Sebor <msebor@redhat.com>
3488 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
3489 code (in C++) or code that triggers warnings.
3491 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
3493 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
3495 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
3497 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
3498 * sched-rgn.c: Ditto.
3499 * sel-sched-ir.c: Ditto.
3501 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
3503 * toplev.h (set_random_seed): Adjust return type.
3504 * toplev.c (init_local_tick): Move eager initialization of random_seed
3505 to get_random_seed. Adjust comment.
3506 (init_random_seed): Inline to get_random_seed, delete.
3507 (get_random_seed): Initialize random_seed lazily.
3508 (set_random_seed): Do not return previous value.
3509 (print_switch_value): Do not call get_random_seed.
3511 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
3513 * cgraph.c (delete_function_version): New, broken out from...
3514 (cgraph_node::delete_function_version): ...here. Rename to
3515 cgraph_node::delete_function_version_by_decl. Update all uses.
3516 (cgraph_node::remove): Call delete_function_version.
3518 2017-09-21 Jakub Jelinek <jakub@redhat.com>
3521 * tree-inline.c (expand_call_inline): Emit clobber stmts for
3522 VAR_DECLs to which addressable non-volatile parameters are mapped
3523 and for id->retvar after the return value assignment. Clear
3524 id->retval and id->retbnd after inlining.
3526 2017-09-21 Richard Biener <rguenther@suse.de>
3528 PR tree-optimization/82276
3529 PR tree-optimization/82244
3530 * tree-vrp.c (build_assert_expr_for): Set
3531 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
3533 (remove_range_assertions): Revert earlier change.
3535 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
3538 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
3540 2017-09-21 Richard Biener <rguenther@suse.de>
3542 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3543 Restore valid IL after code generation errors.
3544 * graphite.c (graphite_transform_loops): Diagnose code
3545 generation issues as MSG_MISSED_OPTIMIZATION and continue
3546 with processing SCOPs.
3548 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3549 Alan Hayward <alan.hayward@arm.com>
3550 David Sherwood <david.sherwood@arm.com>
3552 * calls.c (compute_argument_addresses): Use simplify_gen_binary
3553 rather than choosing between plus_constant and gen_rtx_<CODE>.
3554 * expr.c (emit_push_insn): Likewise.
3555 (expand_expr_real_2): Likewise.
3557 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3558 Alan Hayward <alan.hayward@arm.com>
3559 David Sherwood <david.sherwood@arm.com>
3561 * loop-unroll.c (split_iv): Call copy_rtx on the step.
3563 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3564 Alan Hayward <alan.hayward@arm.com>
3565 David Sherwood <david.sherwood@arm.com>
3567 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
3568 calling tree_to_uhwi.
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 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
3575 INTEGER_CST rather than a negative test for ADDR_EXPR.
3577 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3578 Alan Hayward <alan.hayward@arm.com>
3579 David Sherwood <david.sherwood@arm.com>
3581 * tree-vrp.c (extract_range_from_binary_expr_1): Check
3582 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
3584 2017-09-21 Richard Biener <rguenther@suse.de>
3586 PR tree-optimization/71351
3587 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
3588 graphite_create_new_loop_guard): Remove, fold remaining parts
3590 (translate_isl_ast_node_for): ... here and simplify.
3592 2017-09-21 Jakub Jelinek <jakub@redhat.com>
3595 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
3596 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
3597 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
3598 alternative always use QI mode, for -Os imov (=R,R) alternative
3599 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
3602 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3603 Jeff Law <law@redhat.com>
3605 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
3606 (allocate_stack_space): New function, partially extracted from
3608 (s390_emit_prologue): Track offset to most recent stack probe.
3609 Code to allocate space moved into allocate_stack_space.
3610 Dump actions when no stack is allocated.
3611 (s390_prologue_plus_offset): New function.
3612 (s390_emit_stack_probe): Likewise.
3614 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
3616 * common.opt (Wa, Wl, Wp, g, gz=): Add
3618 (gno-column-info): Remove.
3619 (gcolumn-info): Drop RejectNegative.
3621 (gno-record-gcc-switches): Remove.
3622 (grecord-gcc-switches): Drop RejectNegative.
3623 (gno-split-dwarf): Remove.
3624 (gsplit-dwarf): Drop RejectNegative.
3625 (gno-strict-dwarf): Remove.
3626 (gstrict-dwarf): Drop RejectNegative.
3627 * config/darwin.opt (gfull, gused): Add RejectNegative.
3628 * dwarf2out.c (gen_producer_string): Drop
3629 gno-record-gcc-switches handler.
3630 * optc-gen.awk: Add g to prefixes with negative forms.
3631 * opts-common.c (remapping_prefix_p): New.
3632 (find_opt): Check it.
3633 (generate_canonical_option): Test g prefix.
3634 (option_map): Add -gno- mapping.
3635 (add_misspelling_candidates): Check remapping_prefix_p.
3637 2017-09-20 Jeff Law <law@redhat.com>
3639 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
3640 thinko in stack clash protection support.
3642 * explow.c (compute_stack_clash_protection_loop_data): Use
3643 CONST_INT_P instead of explicit test. Verify object is a
3644 CONST_INT_P before looking at INTVAL.
3645 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
3646 instead of explicit test.
3648 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
3651 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
3652 address instead of to r1 and r11.
3654 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
3656 * config.gcc: Support "knm".
3657 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
3658 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3660 * config/i386/i386.c (m_KNM): Define.
3661 (processor_target_table): Add "knm".
3663 (ix86_option_override_internal): Add "knm".
3664 (ix86_issue_rate): Add PROCESSOR_KNM.
3665 (ix86_adjust_cost): Ditto.
3666 (ia32_multipass_dfa_lookahead): Ditto.
3667 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
3668 (fold_builtin_cpu): Add M_INTEL_KNM.
3669 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
3670 (processor_type): Add PROCESSOR_KNM.
3671 * config/i386/x86-tune.def: Add m_KNM.
3672 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
3674 2017-09-20 Richard Biener <rguenther@suse.de>
3676 PR tree-optimization/80213
3677 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
3678 are allowed in empty BBs as well.
3679 (canonicalize_loop_closed_ssa): Also look for other complex
3681 (scop_detection::get_sese): Include the loop-closed PHI block
3683 (scop_detection::merge_sese): Remove code adding extra blocks.
3684 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
3685 (build_scops): Assert the final returned scop is invalid.
3687 2017-09-20 Richard Biener <rguenther@suse.de>
3689 PR tree-optimization/82264
3690 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
3692 (vn_phi_lookup): Likewise.
3693 (vn_phi_insert): Likewise.
3695 2017-09-20 Jakub Jelinek <jakub@redhat.com>
3697 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
3698 that fits into uhwi or shwi, add DW_AT_const_value regardless
3699 of early_dwarf without going through RTL, using add_AT_unsigned
3702 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
3703 (DEBUG_LTO_ABBREV_SECTION): Likewise.
3704 (DEBUG_LTO_MACINFO_SECTION): Likewise.
3705 (DEBUG_MACRO_SECTION): Likewise.
3706 (DEBUG_LTO_MACRO_SECTION): Likewise.
3707 (DEBUG_STR_DWO_SECTION): Likewise.
3708 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
3709 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
3710 (DEBUG_LTO_DWO_LINE_SECTION): Define.
3711 (DEBUG_LTO_LINE_STR_SECTION): Define.
3712 (init_sections_and_labels): Initialize debug_line_str_section
3713 variable. Initialize debug_loc_section for -gdwarf-5 to
3714 DEBUG_LOCLISTS_SECTION. Formatting fixes.
3716 2017-09-20 Richard Biener <rguenther@suse.de>
3718 * graphite-sese-to-poly.c (extract_affine): Properly handle
3719 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
3721 2017-09-20 Richard Biener <rguenther@suse.de>
3723 PR tree-optimization/81373
3724 * graphite-scop-detection.c (build_cross_bb_scalars_def):
3725 Force SESE live-out defs to be handled even if they are
3728 2017-09-19 Jeff Law <law@redhat.com>
3730 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
3731 nothing for stack adjustments with REG_STACK_CHECK.
3732 * sched-deps.c (parse_add_or_inc): Reject insns with
3733 REG_STACK_CHECK from dependency breaking.
3734 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
3735 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
3736 * reg-notes.def (STACK_CHECK): New note.
3738 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
3739 (ix86_expand_prologue): Dump stack clash info as needed.
3740 Call ix86_adjust_stack_and_probe_stack_clash as needed.
3742 * function.c (dump_stack_clash_frame_info): New function.
3743 * function.h (dump_stack_clash_frame_info): Prototype.
3744 (enum stack_clash_probes): New enum.
3746 * config/alpha/alpha.c (alpha_expand_prologue): Also check
3747 flag_stack_clash_protection.
3748 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
3749 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
3750 (arm_frame_pointer_required): Likewise.
3751 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
3752 (ia64_expand_prologue): Likewise.
3753 * config/mips/mips.c (mips_expand_prologue): Likewise.
3754 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
3755 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
3756 (sparc_flat_expand_prologue): Likewise.
3757 * config/spu/spu.c (spu_expand_prologue): Likewise.
3759 * explow.c: Include "params.h".
3760 (anti_adjust_stack_and_probe_stack_clash): New function.
3761 (get_stack_check_protect): Likewise.
3762 (compute_stack_clash_protection_loop_data): Likewise.
3763 (emit_stack_clash_protection_loop_start): Likewise.
3764 (emit_stack_clash_protection_loop_end): Likewise.
3765 (allocate_dynamic_stack_space): Use get_stack_check_protect.
3766 Use anti_adjust_stack_and_probe_stack_clash.
3767 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
3768 (emit_stack_clash_protection_loop_start): Likewise.
3769 (emit_stack_clash_protection_loop_end): Likewise.
3770 * rtl.h (get_stack_check_protect): Prototype.
3771 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
3772 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
3773 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
3775 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
3777 * doc/tm.texi: Rebuilt.
3778 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
3779 get_stack_check_protect.
3780 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
3781 * config/arm/arm.c (arm_expand_prologue): Likewise.
3782 (arm_frame_pointer_required): Likewise.
3783 * config/i386/i386.c (ix86_expand_prologue): Likewise.
3784 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
3785 * config/mips/mips.c (mips_expand_prologue): Likewise.
3786 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
3787 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
3788 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
3789 (sparc_flat_expand_prologue): Likewise.
3791 * common.opt (-fstack-clash-protection): New option.
3792 * flag-types.h (enum stack_check_type): Note difference between
3793 -fstack-check= and -fstack-clash-protection.
3794 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
3795 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
3796 * toplev.c (process_options): Issue warnings/errors for cases
3797 not handled with -fstack-clash-protection.
3798 * doc/invoke.texi (-fstack-clash-protection): Document new option.
3799 (-fstack-check): Note additional problem with -fstack-check=generic.
3800 Note that -fstack-check is primarily for Ada and refer users
3801 to -fstack-clash-protection for stack-clash-protection.
3802 Document new params for stack clash protection.
3804 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
3806 * config/i386/i386.c (ix86_split_long_move): Do not handle
3807 address used for LEA in a special way.
3809 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
3811 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
3813 2017-09-19 Martin Sebor <msebor@redhat.com>
3816 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
3817 of incompatible types.
3819 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
3821 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
3822 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
3823 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3824 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
3826 2017-09-19 Richard Biener <rguenther@suse.de>
3828 PR tree-optimization/82244
3829 * tree-vrp.c (remove_range_assertions): Do not propagate
3830 a constant to abnormals but replace the assert with a copy.
3832 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
3834 PR rtl-optimization/57878
3835 PR rtl-optimization/68988
3836 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
3837 avoidance test involving non_reload_pseudos. Move frequency test
3838 below the general fragmentation avoidance test.
3840 2017-09-19 Richard Biener <rguenther@suse.de>
3842 PR tree-optimization/69728
3843 * graphite-sese-to-poly.c (schedule_error): New global.
3844 (add_loop_schedule): Handle empty domain by failing the
3846 (build_original_schedule): Handle schedule_error.
3848 2017-09-19 Richard Biener <rguenther@suse.de>
3850 * graphite-scop-detection.c (scop_detection::can_represent_loop):
3851 Do not iterate to sibling loops but only to siblings of inner
3854 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
3857 * config/m68k/m68k.md (moveq feeding equality comparison): Check
3858 that the registers are different.
3860 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
3862 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
3863 to processor_model and "amdfam17h" to arch_names_table.
3864 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
3866 2017-09-18 Jakub Jelinek <jakub@redhat.com>
3869 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
3871 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3872 Alan Hayward <alan.hayward@arm.com>
3873 David Sherwood <david.sherwood@arm.com>
3875 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
3877 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
3879 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
3880 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
3881 vect_schedule_slp_instance.
3882 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
3883 Update call to vect_slp_analyze_node_operations. Simplify return
3885 (vect_slp_analyze_bb_1): Update call accordingly.
3886 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
3887 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
3888 (vect_schedule_slp): Update call accordingly.
3890 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3891 Alan Hayward <alan.hayward@arm.com>
3892 David Sherwood <david.sherwood@arm.com>
3894 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
3895 with types that aren't in fact scalar.
3897 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3899 * tree-vect-slp.c (vect_record_max_nunits): New function,
3901 (vect_build_slp_tree_1): ...here.
3902 (vect_build_slp_tree_2): Call it for phis too.
3904 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3906 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
3907 to vect_get_vec_def_for_operand when getting the mask operand.
3909 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3910 Alan Hayward <alan.hayward@arm.com>
3911 David Sherwood <david.sherwood@arm.com>
3913 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
3916 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3917 Alan Hayward <alan.hayward@arm.com>
3918 David Sherwood <david.sherwood@arm.com>
3920 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
3921 calculation for vector booleans.
3923 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3924 Alan Hayward <alan.hayward@arm.com>
3925 David Sherwood <david.sherwood@arm.com>
3927 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
3929 (vect_transform_stmt): ...here.
3930 (vect_analyze_stmt): Use it instead of calling
3931 vectorizable_live_operation directly.
3933 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
3935 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
3936 non-SIMT targets in acc vector loops.
3938 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
3940 * configure.ac: Add arc and check if assembler supports gdwarf2.
3941 * configure: Regenerate.
3943 2017-09-18 Richard Biener <rguenther@suse.de>
3945 PR tree-optimization/82220
3946 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
3947 epilogue niters from the min_profitable_iters compute.
3949 2017-09-18 Jakub Jelinek <jakub@redhat.com>
3952 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
3953 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
3954 (ix86_init_pic_reg): Revert 2017-09-01 changes.
3956 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
3959 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
3960 switching to a new text section.
3962 2017-09-18 Richard Biener <rguenther@suse.de>
3964 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
3966 (build_alias_set): Reject aliases with no access function.
3968 2017-09-18 Richard Biener <rguenther@suse.de>
3970 PR tree-optimization/79622
3971 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
3973 (build_cross_bb_scalars_use): Likewise.
3975 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
3977 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
3979 2017-09-18 Alan Modra <amodra@gmail.com>
3982 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
3983 stack_pointer_rtx for count 0. Update comments. Break up
3984 large rtl expression.
3986 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
3988 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
3989 Increase to 20 bytes.
3990 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
3991 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
3992 or avx version of the stub.
3994 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
3997 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
3998 compute the minimum stack alignment. Also update preferred stack
3999 boundary for leaf functions.
4001 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
4003 PR tree-optimization/82228
4004 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
4007 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
4009 * common/config/nds32/nds32-common.c
4010 (nds32_option_optimization_table): Refine formatting.
4011 (nds32_option_optimization_table): Use -fsched-pressure and
4012 -fomit-frame-pointer for specific optimization level.
4014 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
4016 * config/nds32/nds32.c: Refine formatting and comments.
4017 * config/nds32/nds32.h: Likewise.
4018 * config/nds32/nds32.md: Likewise.
4019 * config/nds32/nds32-cost.c: Likewise.
4020 * config/nds32/nds32-isr.c: Likewise.
4021 * config/nds32/nds32-md-auxiliary.c: Likewise.
4022 * config/nds32/nds32-multiple.md: Likewise.
4023 * config/nds32/nds32-predicates.c: Likewise.
4025 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
4026 Jakub Jelinek <jakub@redhat.com>
4028 Add support for -std=c++2a.
4029 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
4031 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
4033 2017-09-15 Steve Ellcey <sellcey@cavium.com>
4036 * doc/extend.texi (Common Function Attributes): Add
4037 references to ARM, AArch64, and S/390 specific attributes.
4038 (Function Specific Option Pragmas): Add AArch64 and S/390
4039 to list of back ends that support the target pragma.
4041 2017-09-15 Nathan Sidwell <nathan@acm.org>
4043 * doc/standards.texi: Fix C++17 description. Update URLs for
4046 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
4048 * common.opt (Wcast-align=strict): New warning option.
4049 * doc/invoke.texi: Document -Wcast-align=strict.
4051 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
4053 * cgraph.h (cgraph_thunk_info): Add comments.
4054 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
4055 assert for VIRTUAL_* arguments stricter.
4057 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
4059 PR tree-optimization/71026
4060 * match.pd: Move RDIV patterns from fold-const.c
4061 * fold-const.c (distribute_real_division): Removed.
4062 (fold_binary_loc): Remove calls to distribute_real_divison.
4064 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4066 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
4067 c++1z and gnu++1z as deprecated. Change other references to
4068 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
4069 Change -Wc++1z-compat to -Wc++17-compat.
4070 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
4071 * dwarf2out.c (highest_c_language): Handle C++17.
4072 (gen_compile_unit_die): Likewise.
4074 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4076 PR rtl-optimization/82192
4077 * combine.c (make_extraction): Don't look through non-paradoxical
4078 SUBREGs or TRUNCATE if pos + len is or might be bigger than
4081 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4082 Alan Hayward <alan.hayward@arm.com>
4083 David Sherwood <david.sherwood@arm.com>
4085 * target.def (function_arg_offset): New hook.
4086 * targhooks.h (default_function_arg_offset): Declare.
4087 * targhooks.c (default_function_arg_offset): New function.
4088 * function.c (locate_and_pad_parm): Use
4089 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
4090 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
4091 (TARGET_FUNCTION_ARG_OFFSET): ...this.
4092 * doc/tm.texi: Regenerate.
4093 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
4094 * config/spu/spu.c (spu_function_arg_offset): New function.
4095 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
4096 * system.h (FUNCTION_ARG_OFFSET): Poison.
4098 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4099 Alan Hayard <alan.hayward@arm.com>
4100 David Sherwood <david.sherwood@arm.com>
4102 * target.def (truly_noop_truncation): New hook.
4103 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
4104 than TRULY_NOOP_TRUNCATION.
4105 * hooks.h (hook_bool_uint_uint_true): Declare.
4106 * hooks.c (hook_bool_uint_uint_true): New function.
4107 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
4108 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
4109 * doc/tm.texi: Regenerate.
4110 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
4111 rather than TRULY_NOOP_TRUNCATION in comments.
4112 (simplify_comparison): Likewise.
4113 (record_truncated_value): Likewise.
4114 * expmed.c (extract_bit_field_1): Likewise.
4115 (extract_split_bit_field): Likewise.
4116 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
4117 instead of TRULY_NOOP_TRUNCATION.
4118 * function.c (assign_parm_setup_block): Likewise.
4119 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
4120 * rtlhooks.c: Include target.h.
4121 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
4122 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
4123 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
4124 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
4125 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
4126 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
4127 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
4128 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
4129 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
4130 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
4131 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
4132 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
4133 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
4134 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
4135 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
4136 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
4137 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
4138 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
4139 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
4140 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
4141 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
4142 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
4143 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
4144 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
4145 * config/mips/mips.c (mips_truly_noop_truncation): New function.
4146 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4147 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4148 rather than TRULY_NOOP_TRUNCATION in comments.
4149 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
4150 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
4151 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
4152 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
4153 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
4154 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
4155 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
4156 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
4157 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
4158 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
4159 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
4160 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4161 rather than TRULY_NOOP_TRUNCATION in comments.
4162 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
4163 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
4164 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
4165 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
4166 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
4167 TRULY_NOOP_TRUNCATION condition.
4168 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
4169 (TRULY_NOOP_TRUNCATION): Delete.
4170 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
4171 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
4172 * config/spu/spu.c (spu_truly_noop_truncation): New function.
4173 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4174 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
4175 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
4176 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
4177 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4178 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4179 rather than TRULY_NOOP_TRUNCATION in comments.
4180 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
4181 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
4182 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
4183 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
4184 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
4185 * system.h (TRULY_NOOP_TRUNCATION): Poison.
4187 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
4190 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
4191 (*cmp_ior): Likewise.
4192 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
4193 (*ior_scc_scc_cmp): Likewise.
4194 (*and_scc_scc): Likewise.
4195 (*and_scc_scc_cmp): Likewise.
4197 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4198 Alan Hayard <alan.hayward@arm.com>
4199 David Sherwood <david.sherwood@arm.com>
4201 * target.def (can_change_mode_class): New hook.
4202 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
4203 (hard_regno_nregs): Likewise.
4204 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
4205 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
4206 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
4207 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
4208 (LOAD_EXTEND_OP): Update accordingly.
4209 * doc/tm.texi: Regenerate.
4210 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
4211 CANNOT_CHANGE_MODE_CLASS.
4212 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
4213 (REG_CAN_CHANGE_MODE_P): ...this new macro.
4214 * combine.c (simplify_set): Update accordingly.
4215 * emit-rtl.c (validate_subreg): Likewise.
4216 * recog.c (general_operand): Likewise.
4217 * regcprop.c (mode_change_ok): Likewise.
4218 * reload1.c (choose_reload_regs): Likewise.
4219 (inherit_piecemeal_p): Likewise.
4220 * rtlanal.c (simplify_subreg_regno): Likewise.
4221 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
4222 instead of CANNOT_CHANGE_MODE_CLASS.
4223 (reload_cse_simplify_operands): Likewise.
4224 * reload.c (push_reload): Use targetm.can_change_mode_class
4225 instead of CANNOT_CHANGE_MODE_CLASS.
4226 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
4227 REG_CANNOT_CHANGE_MODE_P.
4228 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4229 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
4230 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4231 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4232 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4233 (arm_can_change_mode_class): New function.
4234 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
4235 than CANNOT_CHANGE_MODE_CLASS in comments.
4236 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4237 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
4238 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
4239 (ix86_can_change_mode_class): ...this new function, inverting the
4240 sense of the return value.
4241 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4242 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4243 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4244 (ia64_can_change_mode_class): New function.
4245 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4246 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
4247 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
4248 (m32c_can_change_mode_class): ...this new function, inverting the
4249 sense of the return value.
4250 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4251 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4252 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
4253 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
4254 (mips_can_change_mode_class): ...this new function, inverting the
4255 sense of the return value.
4256 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4257 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4258 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4259 (msp430_can_change_mode_class): New function.
4260 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4261 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
4262 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4263 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4264 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4265 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
4266 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4267 (pa_cannot_change_mode_class): Replace with...
4268 (pa_can_change_mode_class): ...this new function, inverting the
4269 sense of the return value.
4270 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
4271 than CANNOT_CHANGE_MODE_CLASS in comments.
4272 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4273 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
4274 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4275 (pdp11_cannot_change_mode_class): Replace with...
4276 (pdp11_can_change_mode_class): ...this new function, inverting the
4277 sense of the return value.
4278 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4279 * config/powerpcspe/powerpcspe-protos.h
4280 (rs6000_cannot_change_mode_class_ptr): Delete.
4281 * config/powerpcspe/powerpcspe.c
4282 (rs6000_cannot_change_mode_class_ptr): Delete.
4283 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4284 (rs6000_option_override_internal): Assign to
4285 targetm.can_change_mode_class instead of
4286 rs6000_cannot_change_mode_class_ptr.
4287 (rs6000_cannot_change_mode_class): Replace with...
4288 (rs6000_can_change_mode_class): ...this new function, inverting the
4289 sense of the return value.
4290 (rs6000_debug_cannot_change_mode_class): Replace with...
4291 (rs6000_debug_can_change_mode_class): ...this new function.
4292 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4293 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
4294 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4295 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4296 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
4298 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
4299 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4300 (rs6000_option_override_internal): Assign to
4301 targetm.can_change_mode_class instead of
4302 rs6000_cannot_change_mode_class_ptr.
4303 (rs6000_cannot_change_mode_class): Replace with...
4304 (rs6000_can_change_mode_class): ...this new function, inverting the
4305 sense of the return value.
4306 (rs6000_debug_cannot_change_mode_class): Replace with...
4307 (rs6000_debug_can_change_mode_class): ...this new function.
4308 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4309 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
4310 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
4311 (s390_can_change_mode_class): ...this new function, inverting the
4312 sense of the return value.
4313 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4314 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4315 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
4316 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4317 (sh_cannot_change_mode_class): Replace with...
4318 (sh_can_change_mode_class): ...this new function, inverting the
4319 sense of the return value.
4320 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4321 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4322 (sparc_can_change_mode_class): New function.
4323 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4324 * config/spu/spu.c (spu_can_change_mode_class): New function.
4325 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4326 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4327 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4328 (visium_can_change_mode_class): New function.
4329 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
4331 2017-09-15 Richard Biener <rguenther@suse.de>
4333 PR tree-optimization/82217
4334 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
4335 but not undefined case.
4337 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4340 * postreload.c (reload_cse_simplify_operands): Skip
4341 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
4343 2017-09-15 Richard Biener <rguenther@suse.de>
4345 PR tree-optimization/68823
4346 * graphite-scop-detection.c (build_alias_set): If we have a
4347 possible dependence check whether we can handle them by just
4348 looking at the DRs DR_ACCESS_FNs.
4349 (build_scops): If build_alias_set fails, fail the SCOP.
4351 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
4353 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
4354 to support float128 built-in functions that require the ISA 3.0
4356 (BU_FLOAT128_3_HW): Likewise.
4357 (SQRTF128): Add support for the IEEE 128-bit square root and fma
4359 (FMAF128): Likewise.
4361 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
4362 support for built-in functions that need the ISA 3.0 IEEE 128-bit
4363 floating point instructions.
4364 (rs6000_invalid_builtin): Likewise.
4365 (rs6000_builtin_mask_names): Likewise.
4366 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
4367 (RS6000_BTM_FLOAT128_HW): Likewise.
4368 (RS6000_BTM_COMMON): Likewise.
4369 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
4371 * doc/extend.texi (RS/6000 built-in functions): Document the
4372 IEEE 128-bit floating point square root and fused multiply-add
4375 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
4377 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
4378 reg (r2) isn't in the set of registers defined in the prologue.
4380 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4381 Alan Hayward <alan.hayward@arm.com>
4382 David Sherwood <david.sherwood@arm.com>
4384 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
4385 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
4386 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
4387 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
4388 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
4390 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
4391 max_vectorization_factor.
4392 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
4394 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4395 Alan Hayward <alan.hayward@arm.com>
4396 David Sherwood <david.sherwood@arm.com>
4398 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
4399 (vect_worthwhile_without_simd_p): Declare.
4400 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
4401 (vectorizable_reduction): Use it.
4402 * tree-vect-stmts.c (vectorizable_shift): Likewise.
4403 (vectorizable_operation): Likewise.
4405 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4406 Alan Hayward <alan.hayward@arm.com>
4407 David Sherwood <david.sherwood@arm.com>
4409 * tree-vectorizer.h (vect_get_num_copies): New function.
4410 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
4411 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4412 (vectorizable_induction): Likewise.
4413 (vectorizable_live_operation): Likewise.
4414 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
4415 (vectorizable_bswap): Likewise.
4416 (vectorizable_call): Likewise.
4417 (vectorizable_conversion): Likewise.
4418 (vectorizable_assignment): Likewise.
4419 (vectorizable_shift): Likewise.
4420 (vectorizable_operation): Likewise.
4421 (vectorizable_store): Likewise.
4422 (vectorizable_load): Likewise.
4423 (vectorizable_condition): Likewise.
4424 (vectorizable_comparison): Likewise.
4425 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
4427 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4428 Alan Hayward <alan.hayward@arm.com>
4429 David Sherwood <david.sherwood@arm.com>
4431 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
4432 of vect_init_vector.
4434 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4435 Alan Hayward <alan.hayward@arm.com>
4436 David Sherwood <david.sherwood@arm.com>
4438 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
4439 an inline wrapper that provides a location.
4440 (gimple_build_vector): Likewise.
4441 * gimple-fold.c (gimple_build_vector_from_val): New function.
4442 (gimple_build_vector): Likewise.
4443 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
4444 functions to build the initial value. Always return a gimple value.
4445 (get_initial_defs_for_reduction): Likewise. Only compute
4447 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
4448 vect_init_vector on the results from get_initial_def(s)_for_reduction.
4449 (vectorizable_induction): Use gimple_build_vector rather than
4452 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4453 Alan Hayward <alan.hayward@arm.com>
4454 David Sherwood <david.sherwood@arm.com>
4456 * target.h (vec_perm_indices): New typedef.
4457 (auto_vec_perm_indices): Likewise.
4458 * optabs-query.h: Include target.h
4459 (can_vec_perm_p): Take a vec_perm_indices *.
4460 * optabs-query.c (can_vec_perm_p): Likewise.
4461 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
4462 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
4463 * tree-vect-generic.c (lower_vec_perm): Likewise.
4464 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
4465 (vect_grouped_load_supported): Likewise.
4466 (vect_shift_permute_load_chain): Likewise.
4467 (vect_permute_store_chain): Use auto_vec_perm_indices.
4468 (vect_permute_load_chain): Likewise.
4469 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
4470 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
4471 Update uses of can_vec_perm_p.
4472 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
4473 mode with a number of elements. Take a vec_perm_indices *.
4474 (vect_create_epilog_for_reduction): Update accordingly.
4475 Use auto_vec_perm_indices.
4476 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
4477 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
4478 (vect_transform_slp_perm_load): Likewise.
4479 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
4480 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
4481 (vect_gen_perm_mask_checked): Likewise.
4482 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
4483 (vect_gen_perm_mask_checked): Likewise.
4484 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
4485 (vectorizable_store): Likewise.
4486 (vectorizable_load): Likewise.
4487 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
4488 (vectorizable_bswap): Likewise.
4490 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4491 Alan Hayward <alan.hayward@arm.com>
4492 David Sherwood <david.sherwood@arm.com>
4494 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
4495 * tree.c (build_vector): Likewise.
4496 (build_vector_from_ctor): Update accordingly.
4497 (build_vector_from_val): Likewise.
4498 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
4499 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
4500 * tree-vect-generic.c (add_rshift): Likewise.
4501 (expand_vector_divmod): Likewise.
4502 (optimize_vector_constructor): Likewise.
4503 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
4504 (vect_transform_slp_perm_load): Likewise.
4505 (vect_schedule_slp_instance): Likewise.
4506 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
4507 (vectorizable_call): Likewise.
4508 (vect_gen_perm_mask_any): Likewise. Add elements in order.
4509 * expmed.c (make_tree): Likewise.
4510 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
4511 a vector passed to build_vector.
4512 (fold_convert_const): Likewise.
4513 (exact_inverse): Likewise.
4514 (fold_ternary_loc): Likewise.
4515 (fold_relational_const): Likewise.
4516 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
4517 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
4518 (const_unop): Likewise. Store the reduction accumulator in a
4519 variable rather than an array.
4520 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
4521 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
4522 the new vector, rather than constructing it after the input arrays.
4523 (native_interpret_vector): Use auto_vec<tree> when building
4524 a vector passed to build_vector. Add elements in order.
4525 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
4526 auto_vec<tree> when building a vector passed to build_vector.
4527 (vect_create_epilog_for_reduction): Likewise.
4528 (vectorizable_induction): Likewise.
4529 (get_initial_def_for_reduction): Likewise. Fix indentation of
4531 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
4533 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
4534 passed to build_vector.
4536 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4537 Alan Hayward <alan.hayward@arm.com>
4538 David Sherwood <david.sherwood@arm.com>
4540 * tree-core.h (tree_base::u): Add an "nelts" field.
4541 (tree_vector): Use VECTOR_CST_NELTS as the length.
4542 * tree.c (tree_size): Likewise.
4543 (make_vector): Initialize VECTOR_CST_NELTS.
4544 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
4545 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
4546 TYPE_VECTOR_SUBPARTS.
4547 * expr.c (const_vector_mask_from_tree): Consistently use "units"
4548 as the number of units, setting it from VECTOR_CST_NELTS.
4549 (const_vector_from_tree): Likewise.
4550 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
4551 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
4552 (fold_negate_expr_1): Likewise.
4553 (fold_convert_const): Likewise.
4554 (const_binop): Likewise. Differentiate the number of output and
4556 (const_unop): Likewise.
4557 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
4558 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
4559 in cases that did the opposite.
4561 2017-09-14 Richard Biener <rguenther@suse.de>
4563 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
4566 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
4568 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
4570 2017-09-14 Jakub Jelinek <jakub@redhat.com>
4573 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
4574 if and where to split a bb, except for splitting before debug insn
4575 sequences followed by non-label real insn. Delete debug insns
4576 in between basic blocks.
4578 * combine.c (make_compound_operation_int): Formatting fixes.
4580 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
4581 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4582 * config/netbsd.h (LINK_EH_SPEC): Likewise.
4583 * config/sol2.h (LINK_EH_SPEC): Likewise.
4584 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4585 * config/s390/linux.h (LINK_SPEC): Likewise.
4586 * config/freebsd.h (LINK_EH_SPEC): Likewise.
4587 * config/openbsd.h (LINK_EH_SPEC): Likewise.
4588 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4589 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
4590 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
4591 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4592 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
4593 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
4595 2017-09-13 Jakub Jelinek <jakub@redhat.com>
4597 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
4599 (ENDFILE_LINUX_SPEC): Likewise.
4600 (LINK_EH_SPEC): Likewise.
4601 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
4602 (LINK_OS_LINUX_SPEC32): Likewise.
4603 (LINK_OS_LINUX_SPEC64): Likewise.
4604 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
4605 (LINK_OS_LINUX_SPEC): Likewise.
4607 2017-09-13 Martin Liska <mliska@suse.cz>
4610 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
4611 CASE_HIGH is NULL_TREE.
4613 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
4614 Alan Hayward <alan.hayward@arm.com>
4615 David Sherwood <david.sherwood@arm.com>
4617 * target.def (secondary_memory_needed): New hook.
4618 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
4619 instead of SECONDARY_MEMORY_NEEDED.
4620 (secondary_memory_needed_mode): Likewise.
4621 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
4622 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
4623 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
4624 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
4625 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
4626 * doc/tm.texi: Regenerate.
4627 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
4628 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
4629 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4630 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
4631 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
4632 * config/i386/i386.c (inline_secondary_memory_needed): Put the
4633 mode argument first and change the reg_class arguments to reg_class_t.
4634 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
4635 Make static. Update the call to inline_secondary_memory_needed.
4636 (ix86_register_move_cost): Update the call to
4637 inline_secondary_memory_needed.
4638 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4639 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
4641 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
4642 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4644 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
4645 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
4646 * config/mips/mips.c (mips_secondary_memory_needed): Make static
4647 and match hook interface. Add comment from mips.h.
4648 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4649 * config/mmix/mmix.md (truncdfsf2): Refer to
4650 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4652 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
4653 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
4654 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4655 (pa_secondary_memory_needed): New function.
4656 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
4657 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
4658 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4659 (pdp11_secondary_memory_needed): Make static and match hook interface.
4660 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
4661 * config/powerpcspe/powerpcspe-protos.h
4662 (rs6000_secondary_memory_needed_ptr): Delete.
4663 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
4665 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4666 (rs6000_option_override_internal): Assign to
4667 targetm.secondary_memory_needed rather than
4668 rs6000_secondary_memory_needed_ptr.
4669 (rs6000_secondary_memory_needed): Match hook interface.
4670 (rs6000_debug_secondary_memory_needed): Likewise.
4671 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
4672 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
4673 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
4674 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4675 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
4676 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
4678 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
4679 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4680 (rs6000_option_override_internal): Assign to
4681 targetm.secondary_memory_needed rather than
4682 rs6000_secondary_memory_needed_ptr.
4683 (rs6000_secondary_memory_needed): Match hook interface.
4684 (rs6000_debug_secondary_memory_needed): Likewise.
4685 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
4686 * config/s390/s390.c (s390_secondary_memory_needed): New function.
4687 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4688 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
4689 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4690 (sparc_secondary_memory_needed): New function.
4691 * lra-constraints.c (check_and_process_move): Refer to
4692 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4694 (curr_insn_transform): Likewise.
4695 (process_alt_operands): Use targetm.secondary_memory_needed
4696 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4697 (check_secondary_memory_needed_p): Likewise.
4698 (choose_split_class): Likewise.
4699 * reload.c: Unconditionally include code that was previously
4700 conditional on SECONDARY_MEMORY_NEEDED.
4701 (push_secondary_reload): Use targetm.secondary_memory_needed
4702 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4703 (push_reload): Likewise.
4704 * reload1.c: Unconditionally include code that was previously
4705 conditional on SECONDARY_MEMORY_NEEDED.
4706 (choose_reload_regs): Use targetm.secondary_memory_needed
4707 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4708 (gen_reload): Likewise.
4709 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
4711 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
4712 Alan Hayward <alan.hayward@arm.com>
4713 David Sherwood <david.sherwood@arm.com>
4715 * target.def (secondary_memory_needed_mode): New hook:
4716 * targhooks.c (default_secondary_memory_needed_mode): Declare.
4717 * targhooks.h (default_secondary_memory_needed_mode): New function.
4718 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
4719 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
4720 * doc/tm.texi: Regenerate.
4721 * lra-constraints.c (check_and_process_move): Use
4722 targetm.secondary_memory_needed_mode instead of
4723 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
4724 (curr_insn_transform): Likewise.
4725 * reload.c (get_secondary_mem): Likewise.
4726 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4727 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
4729 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4730 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4731 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
4732 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4733 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
4735 * config/powerpcspe/powerpcspe-protos.h
4736 (rs6000_secondary_memory_needed_mode): Delete.
4737 * config/powerpcspe/powerpcspe.c
4738 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4739 (rs6000_secondary_memory_needed_mode): Make static.
4740 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4741 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
4743 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
4745 (rs6000_secondary_memory_needed_mode): Make static.
4746 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4747 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
4748 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4749 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4750 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
4752 (sparc_secondary_memory_needed_mode): New function.
4753 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
4755 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
4757 * config/aarch64/constraints.md (Umq): New constraint.
4758 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
4760 (mov<mode>): Update condition.
4762 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4764 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
4765 when bitposition is the same.
4767 2017-09-13 Richard Biener <rguenther@suse.de>
4769 * dwarf2out.c (output_die_symbol): Remove.
4770 (output_die): Do not output a DIEs symbol.
4772 2017-09-13 Richard Biener <rguenther@suse.de>
4775 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
4776 default-def to avoid breaking iterator update with the weird
4777 interaction with cgraph_update_edges_for_call_stmt_node.
4779 2017-09-13 Richard Biener <rguenther@suse.de>
4781 * tree-cfg.c (verify_gimple_assign_binary): Add verification
4782 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
4783 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
4784 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
4786 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
4788 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
4789 Disable pc relative literal load irrespective of
4790 TARGET_FIX_ERR_A53_84341 for default.
4792 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
4794 * config/sparc/sparc.c (output_return): Output the source location of
4795 the insn in the delay slot, if any.
4796 (output_sibcall): Likewise.
4798 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
4801 * common.opt (-static-pie): New alias.
4802 (shared): Negate static-pie.
4803 (-no-pie): Update help text.
4805 (static-pie): New option.
4806 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
4807 -static-pie support.
4808 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
4809 (LINK_EH_SPEC): Likewise.
4810 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4811 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
4812 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
4813 * gcc.c (LINK_COMMAND_SPEC): Likewise.
4814 (init_gcc_specs): Likewise.
4815 (init_spec): Likewise.
4816 (display_help): Update help message for -pie.
4817 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
4820 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
4822 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
4823 (movdi_aarch64): Likewise.
4824 (movti_aarch64): Likewise.
4826 2017-09-12 Simon Wright <simon@pushface.org>
4829 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
4830 calculation of the minor version, always output as 0.
4832 2017-09-12 Jakub Jelinek <jakub@redhat.com>
4835 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
4836 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
4837 on it early, rather than manual conversion late. For
4838 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
4839 instead of performing manual conversion.
4841 2017-09-12 Carl Love <cel@us.ibm.com>
4843 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
4844 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
4846 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4847 VMULOSW): Add definitions.
4848 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4849 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4850 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4851 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4852 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4854 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
4856 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
4858 (movti_aarch64): Likewise.
4859 (movdf_aarch64): Likewise.
4860 (movtf_aarch64): Likewise.
4861 (load_pairdi): Likewise.
4862 (store_pairdi): Likewise.
4863 (load_pairdf): Likewise.
4864 (store_pairdf): Likewise.
4865 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
4866 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
4867 (ldr_got_small_<mode>): Likewise.
4868 (ldr_got_small_28k_<mode>): Likewise.
4869 (ldr_got_tiny): Likewise.
4870 * config/aarch64/iterators.md (ldst_sz): New.
4871 (ldpstp_sz): Likewise.
4872 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
4874 (thunderx_load): Split load_8 to load_16.
4875 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
4877 (thunderx2t99_storepair_basic): Split store_8 to store_16.
4878 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
4879 (xgene1_store_pair): Split store_8 to store_16.
4880 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
4881 (falkor_st_0_st_sd): Split store_8 to store_16.
4883 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
4885 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
4886 and store1/2/3/4 to store_4/8/12/16.
4887 * config/aarch64/aarch64.md: Update for rename.
4888 * config/arm/arm.md: Likewise.: Likewise.
4889 * config/arm/arm.c: Likewise.
4890 * config/arm/thumb1.md: Likewise.
4891 * config/arm/thumb2.md: Likewise.
4892 * config/arm/vfp.md: Likewise.
4893 * config/arm/arm-generic.md: Likewise.
4894 * config/arm/arm1020e.md: Likewise.
4895 * config/arm/arm1026ejs.md: Likewise.
4896 * config/arm/arm1136jfs.md: Likewise.
4897 * config/arm/arm926ejs.md: Likewise.
4898 * config/arm/cortex-a15.md: Likewise.
4899 * config/arm/cortex-a17.md: Likewise.
4900 * config/arm/cortex-a5.md: Likewise.
4901 * config/arm/cortex-a53.md: Likewise.
4902 * config/arm/cortex-a57.md: Likewise.
4903 * config/arm/cortex-a7.md: Likewise.
4904 * config/arm/cortex-a8.md: Likewise.
4905 * config/arm/cortex-a9.md: Likewise.
4906 * config/arm/cortex-m4.md: Likewise.
4907 * config/arm/cortex-m7.md: Likewise.
4908 * config/arm/cortex-r4.md: Likewise.
4909 * config/arm/exynos-m1.md: Likewise.
4910 * config/arm/fa526.md: Likewise.
4911 * config/arm/fa606te.md: Likewise.
4912 * config/arm/fa626te.md: Likewise.
4913 * config/arm/fa726te.md: Likewise.
4914 * config/arm/fmp626.md: Likewise.
4915 * config/arm/iwmmxt.md: Likewise.
4916 * config/arm/ldmstm.md: Likewise.
4917 * config/arm/marvell-pj4.md: Likewise.
4918 * config/arm/xgene1.md: Likewise.
4919 * config/aarch64/thunderx.md: Likewise.
4920 * config/aarch64/thunderx2t99.md: Likewise.
4921 * config/aarch64/falkor.md: Likewise.
4923 2017-09-12 Martin Liska <mliska@suse.cz>
4925 * attribs.c (private_lookup_attribute): New function.
4926 * attribs.h (private_lookup_attribute): Declared here.
4927 (lookup_attribute): Called from this place.
4929 2017-09-12 Richard Biener <rguenther@suse.de>
4931 PR tree-optimization/82157
4932 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
4933 stmts with side-effects.
4935 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4936 Alan Hayward <alan.hayward@arm.com>
4937 David Sherwood <david.sherwood@arm.com>
4939 * target.def (hard_regno_nregs): New hook.
4940 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
4941 * targhooks.h (default_hard_regno_nregs): Declare.
4942 * targhooks.c (default_hard_regno_nregs): New function.
4943 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
4944 (TARGET_HARD_REGNO_NREGS): ...this hook.
4945 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
4946 (CLASS_MAX_NREGS): Likewise.
4947 * doc/tm.texi: Regenerate.
4948 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
4949 instead of HARD_REGNO_NREGS.
4950 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
4951 HARD_REGNO_NREGS in the comment.
4952 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
4953 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
4954 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
4955 Return an unsigned int.
4956 (TARGET_HARD_REGNO_NREGS): Redefine.
4957 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
4958 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
4959 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
4960 (arc_hard_regno_nregs): New function.
4961 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
4962 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
4963 (arm_hard_regno_nregs): New function.
4964 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
4965 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
4966 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
4967 (TARGET_HARD_REGNO_NREGS): Redefine.
4968 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
4969 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
4970 (HARD_REGNO_NREGS): Delete.
4971 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
4972 (cr16_hard_regno_nregs): New function.
4973 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
4974 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
4975 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
4976 (cris_hard_regno_nregs): New function.
4977 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
4978 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
4979 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
4980 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
4981 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
4982 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
4983 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
4984 (frv_hard_regno_nregs): Make static. Take and return an
4986 (frv_class_max_nregs): Remove outdated copy of documentation.
4987 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
4988 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
4989 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
4990 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
4991 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
4992 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
4993 (TARGET_HARD_REGNO_NREGS): Redefine.
4994 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
4995 (CLASS_MAX_NREGS): Update comment.
4996 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
4997 (ia64_hard_regno_nregs): New function.
4998 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
4999 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
5000 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
5001 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
5002 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
5004 (m32c_hard_regno_nregs): Likewise. Make static.
5005 (TARGET_HARD_REGNO_NREGS): Redefine.
5006 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
5007 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
5008 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
5009 (m68k_hard_regno_nregs): New function.
5010 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
5011 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
5012 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
5013 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
5014 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
5015 Take and return an unsigned int.
5016 (TARGET_HARD_REGNO_NREGS): Redefine.
5017 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
5018 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
5019 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
5020 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
5021 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
5022 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
5023 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
5024 (msp430_hard_regno_nregs): Make static. Take and return an
5026 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
5027 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
5028 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
5029 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
5030 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
5031 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
5032 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
5033 (TARGET_HARD_REGNO_NREGS): Redefine.
5034 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
5035 (PA_HARD_REGNO_NREGS): ...this.
5036 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
5037 (PA_HARD_REGNO_NREGS): ...this.
5038 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
5039 (pa_hard_regno_nregs): New function.
5040 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
5041 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
5042 (pdp11_hard_regno_nregs): New function.
5043 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
5044 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
5045 (rs6000_hard_regno_nregs_hook): New function.
5046 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
5047 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
5048 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
5049 Take and return an unsigned int. Move earlier in file.
5050 (TARGET_HARD_REGNO_NREGS): Redefine.
5051 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
5052 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
5053 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
5054 (rl78_hard_regno_nregs): Make static. Take and return an
5056 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
5057 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
5058 (rs6000_hard_regno_nregs_hook): New function.
5059 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
5060 * config/rx/rx.c (rx_hard_regno_nregs): New function.
5061 (TARGET_HARD_REGNO_NREGS): Redefine.
5062 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
5063 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
5064 instead of HARD_REGNO_NREGS.
5065 (s390_hard_regno_nregs): New function.
5066 (s390_hard_regno_mode_ok): Add comment from s390.h.
5067 (TARGET_HARD_REGNO_NREGS): Redefine.
5068 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
5069 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
5070 (sh_hard_regno_nregs): New function.
5071 (sh_pass_in_reg_p): Use it.
5072 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
5073 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
5074 (sparc_hard_regno_nregs): New function.
5075 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
5076 * config/spu/spu.c (spu_hard_regno_nregs): New function.
5077 (spu_function_arg_advance): Use it, supplying a valid register number.
5078 (TARGET_HARD_REGNO_NREGS): Redefine.
5079 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
5080 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
5081 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
5082 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
5083 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
5084 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
5085 (CLASS_MAX_NREGS): Remove copy of old documentation.
5086 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
5087 (visium_hard_regno_nregs): New function.
5088 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
5089 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
5090 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
5091 xtensa_hard_regno_nregs): New function.
5092 * system.h (HARD_REGNO_NREGS): Poison.
5094 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5096 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
5097 hard_regno_nregs instead of HARD_REGNO_NREGS.
5098 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
5099 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
5100 (c6x_expand_epilogue): Likewise.
5101 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
5102 (frv_read_iacc_argument): Likewise.
5103 * config/sh/sh.c: Include regs.h.
5104 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
5105 (regs_used): Likewise.
5106 (output_stack_adjust): Likewise.
5107 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
5108 * expmed.c: Include regs.h.
5109 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
5110 * ree.c: Include regs.h.
5111 (combine_reaching_defs): Use hard_regno_nregs instead of
5113 (add_removable_extension): Likewise.
5115 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5117 * regs.h (hard_regno_nregs): Turn into a function.
5118 (end_hard_regno): Update accordingly.
5119 * caller-save.c (setup_save_areas): Likewise.
5120 (save_call_clobbered_regs): Likewise.
5121 (replace_reg_with_saved_mem): Likewise.
5122 (insert_restore): Likewise.
5123 (insert_save): Likewise.
5124 * combine.c (can_change_dest_mode): Likewise.
5125 (move_deaths): Likewise.
5126 (distribute_notes): Likewise.
5127 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
5128 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
5129 (rs6000_split_multireg_move): Likewise.
5130 (rs6000_register_move_cost): Likewise.
5131 (rs6000_memory_move_cost): Likewise.
5132 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
5133 (rs6000_split_multireg_move): Likewise.
5134 (rs6000_register_move_cost): Likewise.
5135 (rs6000_memory_move_cost): Likewise.
5136 * cselib.c (cselib_reset_table): Likewise.
5137 (cselib_lookup_1): Likewise.
5138 * emit-rtl.c (set_mode_and_regno): Likewise.
5139 * function.c (aggregate_value_p): Likewise.
5140 * ira-color.c (setup_profitable_hard_regs): Likewise.
5141 (check_hard_reg_p): Likewise.
5142 (calculate_saved_nregs): Likewise.
5143 (assign_hard_reg): Likewise.
5144 (improve_allocation): Likewise.
5145 (calculate_spill_cost): Likewise.
5146 * ira-emit.c (modify_move_list): Likewise.
5147 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
5148 (ira_hard_reg_in_set_p): Likewise.
5149 * ira.c (setup_reg_mode_hard_regset): Likewise.
5150 (clarify_prohibited_class_mode_regs): Likewise.
5151 (check_allocation): Likewise.
5152 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5153 (lra_setup_reg_renumber): Likewise.
5154 (setup_try_hard_regno_pseudos): Likewise.
5155 (spill_for): Likewise.
5156 (assign_hard_regno): Likewise.
5157 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
5158 * lra-constraints.c (in_class_p): Likewise.
5159 (lra_constraint_offset): Likewise.
5160 (simplify_operand_subreg): Likewise.
5161 (lra_constraints): Likewise.
5162 (split_reg): Likewise.
5163 (split_if_necessary): Likewise.
5164 (invariant_p): Likewise.
5165 (inherit_in_ebb): Likewise.
5166 * lra-lives.c (process_bb_lives): Likewise.
5167 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
5168 (get_hard_regs): Likewise.
5169 (do_remat): Likewise.
5170 * lra-spills.c (assign_spill_hard_regs): Likewise.
5171 * mode-switching.c (create_pre_exit): Likewise.
5172 * postreload.c (reload_combine_recognize_pattern): Likewise.
5173 * recog.c (peep2_find_free_register): Likewise.
5174 * regcprop.c (kill_value_regno): Likewise.
5175 (set_value_regno): Likewise.
5176 (copy_value): Likewise.
5177 (maybe_mode_change): Likewise.
5178 (find_oldest_value_reg): Likewise.
5179 (copyprop_hardreg_forward_1): Likewise.
5180 * regrename.c (check_new_reg_p): Likewise.
5181 (regrename_do_replace): Likewise.
5182 * reload.c (push_reload): Likewise.
5183 (combine_reloads): Likewise.
5184 (find_dummy_reload): Likewise.
5185 (operands_match_p): Likewise.
5186 (find_reloads): Likewise.
5187 (find_equiv_reg): Likewise.
5188 (reload_adjust_reg_for_mode): Likewise.
5189 * reload1.c (count_pseudo): Likewise.
5190 (count_spilled_pseudo): Likewise.
5191 (find_reg): Likewise.
5192 (clear_reload_reg_in_use): Likewise.
5193 (free_for_value_p): Likewise.
5194 (allocate_reload_reg): Likewise.
5195 (choose_reload_regs): Likewise.
5196 (reload_adjust_reg_for_temp): Likewise.
5197 (emit_reload_insns): Likewise.
5198 (delete_output_reload): Likewise.
5199 * rtlanal.c (subreg_get_info): Likewise.
5200 * sched-deps.c (sched_analyze_reg): Likewise.
5201 * sel-sched.c (init_regs_for_mode): Likewise.
5202 (mark_unavailable_hard_regs): Likewise.
5203 (choose_best_reg_1): Likewise.
5204 (verify_target_availability): Likewise.
5205 * valtrack.c (dead_debug_insert_temp): Likewise.
5206 * var-tracking.c (track_loc_p): Likewise.
5207 (emit_note_insn_var_location): Likewise.
5208 * varasm.c (make_decl_rtl): Likewise.
5209 * reginfo.c (choose_hard_reg_mode): Likewise.
5210 (init_reg_modes_target): Refer directly to
5211 this_target_regs->x_hard_regno_nregs.
5213 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5215 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
5216 instead of hard_regno_nregs.
5218 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5220 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
5221 end_hard_regno instead of hard_regno_nregs.
5222 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
5223 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
5224 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
5225 * ira-color.c (improve_allocation): Likewise.
5226 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5227 * lra-lives.c (mark_regno_live): Likewise.
5228 (mark_regno_dead): Likewise.
5229 * lra-remat.c (operand_to_remat): Likewise.
5230 * lra.c (collect_non_operand_hard_regs): Likewise.
5231 * postreload.c (reload_combine_note_store): Likewise.
5232 (move2add_valid_value_p): Likewise.
5233 * reload.c (regno_clobbered_p): Likewise.
5235 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5237 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
5239 * config/v850/v850.c (v850_reorg): Likewise.
5240 * reload.c (refers_to_regno_for_reload_p): Likewise.
5241 (find_equiv_reg): Likewise.
5242 * reload1.c (reload_reg_reaches_end_p): Likewise.
5244 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5246 * caller-save.c (add_used_regs): Use REG_NREGS instead of
5248 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
5249 * config/arm/arm.c (output_move_neon): Likewise.
5250 (arm_attr_length_move_neon): Likewise.
5251 (neon_split_vcombine): Likewise.
5252 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
5253 (c6x_mark_reg_written): Likewise.
5254 (c6x_dwarf_register_span): Likewise.
5255 * config/i386/i386.c (ix86_save_reg): Likewise.
5256 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
5257 (rws_access_reg): Likewise.
5258 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
5259 * mode-switching.c (create_pre_exit): Likewise.
5260 * ree.c (combine_reaching_defs): Likewise.
5261 (add_removable_extension): Likewise.
5262 * regcprop.c (find_oldest_value_reg): Likewise.
5263 (copyprop_hardreg_forward_1): Likewise.
5264 * reload.c (reload_inner_reg_of_subreg): Likewise.
5265 (push_reload): Likewise.
5266 (combine_reloads): Likewise.
5267 (find_dummy_reload): Likewise.
5268 (reload_adjust_reg_for_mode): Likewise.
5269 * reload1.c (find_reload_regs): Likewise.
5270 (forget_old_reloads_1): Likewise.
5271 (reload_reg_free_for_value_p): Likewise.
5272 (reload_adjust_reg_for_temp): Likewise.
5273 (emit_reload_insns): Likewise.
5274 (delete_output_reload): Likewise.
5275 * sel-sched.c (choose_best_reg_1): Likewise.
5276 (choose_best_pseudo_reg): Likewise.
5278 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5279 Alan Hayward <alan.hayward@arm.com>
5280 David Sherwood <david.sherwood@arm.com>
5282 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
5283 * target.def (slow_unaligned_access): New hook.
5284 * targhooks.h (default_slow_unaligned_access): Declare.
5285 * targhooks.c (default_slow_unaligned_access): New function.
5286 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
5287 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
5288 * doc/tm.texi: Regenerate.
5289 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
5290 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
5291 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
5293 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
5294 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
5296 (rs6000_slow_unaligned_access): New function.
5297 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
5298 (expand_block_compare): Likewise.
5299 (expand_strn_compare): Likewise.
5300 (rs6000_rtx_costs): Likewise.
5301 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
5302 (riscv_slow_unaligned_access): Likewise.
5303 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
5304 (riscv_slow_unaligned_access_p): ...this and make static.
5305 (riscv_option_override): Update accordingly.
5306 (riscv_slow_unaligned_access): New function.
5307 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
5308 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
5309 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
5310 (rs6000_slow_unaligned_access): New function.
5311 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
5312 (rs6000_rtx_costs): Likewise.
5313 * config/rs6000/rs6000-string.c (expand_block_compare)
5314 (expand_strn_compare): Use targetm.slow_unaligned_access instead
5315 of SLOW_UNALIGNED_ACCESS.
5316 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
5317 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
5318 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
5319 of SLOW_UNALIGNED_ACCESS.
5320 * expmed.c (simple_mem_bitfield_p): Likewise.
5321 * expr.c (alignment_for_piecewise_move): Likewise.
5322 (emit_group_load_1): Likewise.
5323 (emit_group_store): Likewise.
5324 (copy_blkmode_from_reg): Likewise.
5325 (emit_push_insn): Likewise.
5326 (expand_assignment): Likewise.
5327 (store_field): Likewise.
5328 (expand_expr_real_1): Likewise.
5329 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
5330 * lra-constraints.c (simplify_operand_subreg): Likewise.
5331 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
5332 * gimple-ssa-store-merging.c: Likewise in block comment at start
5334 * tree-ssa-strlen.c: Include target.h.
5335 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
5336 of SLOW_UNALIGNED_ACCESS.
5337 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
5339 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5341 PR rtl-optimization/82185
5342 * expmed.c (emit_store_flag_int): Only test tem if it has been
5345 2017-09-12 Richard Biener <rguenther@suse.de>
5348 * match.pd ((FTYPE) N CMP CST): Fix typo.
5350 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
5352 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
5354 (mips_near_type_p): Add 'short_call' attribute as a synonym
5356 * doc/extend.texi (short_call): Document new function attribute.
5358 2017-09-12 Jakub Jelinek <jakub@redhat.com>
5361 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
5362 assertion check that in the condition.
5363 (get_atomic_generic_size): Likewise. Before testing if parameter
5364 has pointer type, if it has array type, call for C++
5365 default_conversion to perform array-to-pointer conversion.
5367 2017-09-12 Richard Biener <rguenther@suse.de>
5369 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
5370 for operations we cannot scalarize.
5372 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
5374 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
5375 vectors heap vectors. Clean up comments.
5376 Make visited_bbs a reference.
5377 (profitable_jump_thread_path): Make GC
5378 vectors heap vectors. Clean up comments.
5380 (convert_and_register_jump_thread_path): Make GC vectors heap
5382 (check_subpath_and_update_thread_path): Same. Clean up comments.
5383 Make visited_bbs a reference.
5384 (handle_phi): Abstract common code to to
5385 register_jump_thread_path_if_profitable.
5386 Rename VAR_BB to DEF_BB.
5388 Make GC vectors heap vectors.
5389 Make visited_bbs a reference.
5390 (handle_assignment): Same.
5391 (register_jump_thread_path_if_profitable): New.
5392 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
5394 Make GC vectors heap vectors. Clean up comments.
5395 Make visited_bbs a reference.
5396 (find_jump_threads_backwards): Make visited_bbs live in the stack.
5397 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
5400 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
5403 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
5404 words of E_DImode object are reachable by xtensa_uimm8x4 access.
5406 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
5408 Revert r251800 and r251799.
5410 2017-09-11 Martin Jambor <mjambor@suse.cz>
5413 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
5414 arguments in advance.
5415 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
5416 use it to find predecessor edges.
5417 (naive_outof_ssa): Collect vector of predecessors.
5419 2017-09-08 Jason Merrill <jason@redhat.com>
5421 PR c++/70029 - ICE with ref-qualifier and -flto
5422 * langhooks.h (struct lang_hooks_for_types): Add
5423 copy_lang_qualifiers.
5424 * attribs.c (build_type_attribute_qual_variant): Use it.
5425 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
5427 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
5428 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
5430 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
5433 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
5434 (*mulsi3_sp64): New instruction.
5435 (mulsi3): New expander.
5437 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
5439 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
5441 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5443 * sancov.c: Include memmodel.h.
5445 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
5448 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
5451 2017-09-07 Carl Love <cel@us.ibm.com>
5453 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
5454 the sldi instruction.
5456 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
5458 * sancov.c: Include tm_p.h.
5460 2017-09-07 Jakub Jelinek <jakub@redhat.com>
5463 * output.h (switch_to_other_text_partition): New declaration.
5464 * varasm.c (switch_to_other_text_partition): New function.
5465 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
5466 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
5467 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
5468 to the other text partition before emitting LCL label and switch back
5469 after emitting the word after it.
5471 2017-09-07 Richard Biener <rguenther@suse.de>
5473 * passes.def (pass_split_crit_edges): Remove instance before PRE.
5474 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
5475 critical edges here, after loop init.
5476 (pass_data_pre): Remove PROP_no_crit_edges flags.
5477 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
5478 for valueization of call args to avoid leaking VN_TOP.
5479 (visit_use): Assert we do not visit default defs.
5480 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
5481 Use error_mark_node to more easily detect leaking VN_TOP.
5482 All default-defs are varying, not VN_TOP. Mark them visited.
5483 (run_scc_vn): Make code match comment.
5485 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5487 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
5488 OPTION_MASK_FLOAT128_KEYWORD.
5489 (POWERPC_MASKS): Likewise.
5490 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
5491 support for the -mfloat128-type option, and make -mfloat128
5492 default on PowerPC Linux systems. Define or undefine
5493 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
5494 Define __float128 to be __ieee128 if IEEE 128-bit support is
5495 enabled, or undefine it.
5496 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
5497 Delete defining __FLOAT128_TYPE__.
5498 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
5499 -mfloat128-type option and make -mfloat128 default on PowerPC
5501 (TARGET_FLOAT128_TYPE): Likewise.
5502 (-mfloat128-type): Likewise.
5503 * config/rs6000/rs6000.c (rs6000_option_override_internal):
5504 Delete the -mfloat128-type option and make -mfloat128 default on
5505 PowerPC Linux systems. Always use __ieee128 to be the keyword for
5506 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
5507 128-bit floating point is enabled. Change tests from using
5508 -mfloat128-type to -mfloat128.
5509 (rs6000_mangle_type): Use the correct mangling for the __float128
5510 type even if normal long double is restricted to 64-bits.
5511 (floatn_mode): Enable the _Float128 type by default on VSX Linux
5513 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
5514 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
5515 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
5517 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
5518 documentation for -mfloat128.
5520 2017-09-06 Olivier Hainque <hainque@adacore.com>
5522 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
5524 2017-09-06 Wish Wu <wishwu007@gmail.com>
5525 Jakub Jelinek <jakub@redhat.com>
5527 * asan.c (initialize_sanitizer_builtins): Add
5528 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
5529 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
5530 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
5531 BT_FN_VOID_UINT64_PTR variables.
5532 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
5533 (BT_FN_VOID_UINT16_UINT16): Likewise.
5534 (BT_FN_VOID_UINT32_UINT32): Likewise.
5535 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
5536 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
5537 (BT_FN_VOID_UINT64_PTR): Likewise.
5538 * common.opt (flag_sanitize_coverage): New variable.
5539 (fsanitize-coverage=trace-pc): Remove.
5540 (fsanitize-coverage=): Add.
5541 * flag-types.h (enum sanitize_coverage_code): New enum.
5542 * fold-const.c (fold_range_test): Disable non-short-circuit
5543 optimization if flag_sanitize_coverage.
5544 (fold_truth_andor): Likewise.
5545 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
5546 * opts.c (COVERAGE_SANITIZER_OPT): Define.
5547 (coverage_sanitizer_opts): New array.
5548 (get_closest_sanitizer_option): Add OPTS argument, handle also
5549 OPT_fsanitize_coverage_.
5550 (parse_sanitizer_options): Adjusted to also handle
5551 OPT_fsanitize_coverage_.
5552 (common_handle_option): Add OPT_fsanitize_coverage_.
5553 * sancov.c (instrument_comparison, instrument_switch): New function.
5554 (sancov_pass): Add trace-cmp support.
5555 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
5556 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
5557 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
5558 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
5559 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
5560 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
5561 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
5562 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
5563 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
5564 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
5566 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
5568 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
5569 error. Only quit immediately if parsing is complete.
5570 (BEGIN): Initialize fatal_err and parse_done.
5571 (begin fpu, end fpu): Check number of arguments.
5572 (begin arch, end arch): Likewise.
5573 (begin cpu, end cpu): Likewise.
5574 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
5575 (optalias): Likewise.
5577 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
5579 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
5580 * config/arm/arm-isa.h: Delete. Move definitions to ...
5581 * arm-cpus.in: ... here. Use new feature and fgroup values.
5582 * config/arm/arm.c (arm_option_override): Use lower case for feature
5584 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
5585 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
5586 * config/arm/parsecpu.awk (END): Add new command 'isa'.
5588 (print_isa_bits_for): New function.
5589 (gen_isa): New function.
5590 (gen_comm_data): Use print_isa_bits_for.
5591 (define feature): New keyword.
5592 (define fgroup): New keyword.
5593 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
5594 (arm-isa.h): Add rule to generate file.
5595 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
5596 case for feature bit names.
5598 2017-09-06 Richard Biener <rguenther@suse.de>
5600 * tree-ssa-pre.c (NECESSARY): Remove.
5601 (create_expression_by_pieces): Do not touch pass-local flags.
5602 (insert_into_preds_of_block): Likewise.
5603 (do_pre_regular_insertion): Likewise.
5604 (eliminate_insert): Likewise.
5605 (eliminate_dom_walker::before_dom_children): Likewise.
5606 (fini_eliminate): Do not look at inserted_exprs.
5607 (mark_operand_necessary): Remove.
5608 (remove_dead_inserted_code): Replace with simple work-list
5609 algorithm based on inserted_exprs and SSA uses.
5610 (pass_pre::execute): Re-order fini_eliminate and
5611 remove_dead_inserted_code.
5613 2017-09-06 Olivier Hainque <hainque@adacore.com>
5615 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
5616 for VxWorks 7. Adjust surrounding comments.
5618 2017-09-06 Richard Biener <rguenther@suse.de>
5620 * gimple-ssa-strength-reduction.c
5621 (find_candidates_dom_walker::before_dom_children): Also allow
5624 2017-09-06 Richard Biener <rguenther@suse.de>
5626 PR tree-optimization/82108
5627 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
5628 for gap in the non-permutation SLP case.
5630 2017-09-06 Martin Jambor <mjambor@suse.cz>
5632 PR tree-optimization/82078
5633 * tree-sra.c (sort_and_splice_var_accesses): Move call to
5634 add_access_to_work_queue...
5635 (build_accesses_from_assign): ...here.
5636 (propagate_all_subaccesses): Make sure racc is the group
5637 representative, if there is one.
5639 2017-09-06 Jakub Jelinek <jakub@redhat.com>
5642 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
5645 2017-09-06 Richard Biener <rguenther@suse.de>
5647 * gimple-ssa-strength-reduction.c
5648 (find_candidates_dom_walker::before_doom_children): Use a
5649 type and not a mode check.
5651 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
5654 * config/arm/predicates.md (arm_general_adddi_operand): Create new
5656 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
5658 2017-09-05 Jeff Law <law@redhat.com>
5660 PR tree-optimization/64910
5661 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
5662 cases where we have 3 or more operands.
5664 2017-09-05 Jakub Jelinek <jakub@redhat.com>
5667 * omp-low.c (lower_omp_for): Recompute tree invariant if
5668 gimple_omp_for_initial/final is ADDR_EXPR.
5671 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
5672 into gimple val before gimplification fo the COND_EXPR.
5674 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
5676 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
5677 REGION_COPY argument.
5678 (thread_through_all_blocks): Remove unused argument to
5679 duplicate_thread_path.
5681 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5682 Alan Hayward <alan.hayward@arm.com>
5683 David Sherwood <david.sherwood@arm.com>
5685 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
5686 Take a scalar_mode rather than a machine_mode.
5687 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
5688 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
5689 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
5690 (aarch64_gen_adjusted_ldpstp): Likewise.
5691 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
5693 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5694 Alan Hayward <alan.hayward@arm.com>
5695 David Sherwood <david.sherwood@arm.com>
5697 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
5698 Take a scalar_int_mode instead of a machine_mode.
5699 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
5700 (aarch64_output_scalar_simd_mov_immediate): Likewise.
5701 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
5702 (aarch64_simd_attr_length_rglist): Delete.
5703 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
5704 a scalar_int_mode instead of a machine_mode.
5705 (aarch64_add_offset): Likewise.
5706 (aarch64_internal_mov_immediate): Likewise
5707 (aarch64_add_constant_internal): Likewise.
5708 (aarch64_add_constant): Likewise.
5709 (aarch64_movw_imm): Likewise.
5710 (aarch64_rtx_arith_op_extract_p): Likewise.
5711 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
5712 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
5713 Remove assert that the mode isn't a vector.
5714 (aarch64_output_scalar_simd_mov_immediate): Likewise.
5715 (aarch64_expand_mov_immediate): Update calls after above changes.
5716 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
5717 (aarch64_and_bitmask_imm): Check for scalar integer modes.
5718 (aarch64_move_imm): Likewise.
5719 (aarch64_can_const_movi_rtx_p): Likewise.
5720 (aarch64_strip_extend): Likewise.
5721 (aarch64_extr_rtx_p): Likewise.
5722 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
5723 a CONST_INT when the mode parameter is VOIDmode.
5724 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
5726 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5728 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
5729 * stor-layout.c (bitwise_mode_for_mode): Likewise.
5730 (bitwise_type_for_mode): Update accordingly.
5732 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5734 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
5735 * stor-layout.c (mode_for_size_tree): Likewise.
5736 (mode_for_array): Update accordingly.
5737 (layout_decl): Likewise.
5738 (compute_record_mode): Likewise. Only set the mode once.
5740 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5742 * target.def (get_mask_mode): Change return type to opt_mode.
5744 * doc/tm.texi: Regenerate.
5745 * targhooks.h (default_get_mask_mode): Return an opt_mode.
5746 * targhooks.c (default_get_mask_mode): Likewise.
5747 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
5748 * optabs-query.c (can_vec_mask_load_store_p): Update use of
5749 targetm.get_mask_mode.
5750 * tree.c (build_truth_vector_type): Likewise.
5752 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5754 * machmode.h (mode_for_vector): Return an opt_mode.
5755 * stor-layout.c (mode_for_vector): Likewise.
5756 (mode_for_int_vector): Update accordingly.
5757 (layout_type): Likewise.
5758 * config/i386/i386.c (emit_memmov): Likewise.
5759 (ix86_expand_set_or_movmem): Likewise.
5760 (ix86_expand_vector_init): Likewise.
5761 (ix86_get_mask_mode): Likewise.
5762 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
5764 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
5765 * expmed.c (extract_bit_field_1): Likewise.
5766 * expr.c (expand_expr_real_2): Likewise.
5767 * optabs-query.c (can_vec_perm_p): Likewise.
5768 (can_vec_mask_load_store_p): Likewise.
5769 * optabs.c (expand_vec_perm): Likewise.
5770 * targhooks.c (default_get_mask_mode): Likewise.
5771 * tree-vect-stmts.c (vectorizable_store): Likewise.
5772 (vectorizable_load): Likewise.
5773 (get_vectype_for_scalar_type_and_size): Likewise.
5775 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5777 * machmode.h (mode_for_int_vector): New function.
5778 * stor-layout.c (mode_for_int_vector): Likewise.
5779 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
5780 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
5781 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
5782 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
5783 (s390_expand_vcond): Likewise.
5785 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5787 * machmode.h (opt_machine_mode): New type.
5788 (opt_mode<T>): Allow construction from anything that can be
5790 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
5791 (mode_for_size): Return an opt_machine_mode.
5792 * stor-layout.c (mode_for_size): Likewise.
5793 (mode_for_size_tree): Update call accordingly.
5794 (bitwise_mode_for_mode): Likewise.
5795 (make_fract_type): Likewise.
5796 (make_accum_type): Likewise.
5797 * caller-save.c (replace_reg_with_saved_mem): Update call
5799 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5800 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5801 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5802 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
5803 * expmed.c (extract_bit_field_1): Likewise.
5804 * reload.c (get_secondary_mem): Likewise.
5805 * varasm.c (assemble_integer): Likewise.
5806 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
5809 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5811 * machmode.h (decimal_float_mode_for_size): New function.
5812 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
5813 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
5814 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
5815 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
5816 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
5817 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
5819 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5821 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
5822 (get_builtin_sync_mode): Likewise.
5823 (expand_ifn_atomic_compare_exchange): Likewise.
5824 (expand_builtin_atomic_clear): Likewise.
5825 (expand_builtin_atomic_test_and_set): Likewise.
5826 (fold_builtin_atomic_always_lock_free): Likewise.
5827 * calls.c (compute_argument_addresses): Likewise.
5828 (emit_library_call_value_1): Likewise.
5829 (store_one_arg): Likewise.
5830 * combine.c (combine_instructions): Likewise.
5831 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
5832 * config/arm/arm.c (arm_function_value): Likewise.
5833 (aapcs_allocate_return_reg): Likewise.
5834 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
5835 * config/i386/i386.c (construct_container): Likewise.
5836 (ix86_gimplify_va_arg): Likewise.
5837 (ix86_expand_sse_cmp): Likewise.
5838 (emit_memmov): Likewise.
5839 (emit_memset): Likewise.
5840 (expand_small_movmem_or_setmem): Likewise.
5841 (ix86_expand_pextr): Likewise.
5842 (ix86_expand_pinsr): Likewise.
5843 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
5844 * config/microblaze/microblaze.c (microblaze_block_move_straight):
5846 * config/mips/mips.c (mips_function_value_1) Likewise.
5847 (mips_block_move_straight): Likewise.
5848 (mips_expand_ins_as_unaligned_store): Likewise.
5849 * config/powerpcspe/powerpcspe.c
5850 (rs6000_darwin64_record_arg_advance_flush): Likewise.
5851 (rs6000_darwin64_record_arg_flush): Likewise.
5852 * config/rs6000/rs6000.c
5853 (rs6000_darwin64_record_arg_advance_flush): Likewise.
5854 (rs6000_darwin64_record_arg_flush): Likewise.
5855 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
5856 (sparc_function_value_1): Likewise.
5857 * config/spu/spu.c (adjust_operand): Likewise.
5858 (spu_emit_branch_or_set): Likewise.
5859 (arith_immediate_p): Likewise.
5860 * emit-rtl.c (gen_lowpart_common): Likewise.
5861 * expr.c (expand_expr_real_1): Likewise.
5862 * function.c (assign_parm_setup_block): Likewise.
5863 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
5864 * reload1.c (alter_reg): Likewise.
5865 * stor-layout.c (mode_for_vector): Likewise.
5866 (layout_type): Likewise.
5868 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5870 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
5871 (spu_convert_move): Likewise.
5872 * lower-subreg.c (resolve_simple_move): Likewise.
5874 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5877 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
5878 define_insn to a define_expand.
5879 (altivec_vsum2sws_direct): New define_insn.
5880 (altivec_vsumsws): Convert from a define_insn to a define_expand.
5882 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
5884 * config/arm/arm.c (arm_option_params_internal): Improve setting of
5887 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
5892 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
5893 realign stack if stack alignment needed is less than incoming
5896 2017-09-05 Marek Polacek <polacek@redhat.com>
5899 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
5902 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
5904 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
5906 2017-09-05 Richard Biener <rguenther@suse.de>
5908 PR tree-optimization/82084
5909 * fold-const.c (can_native_encode_string_p): Handle wide characters.
5911 2017-09-05 Richard Biener <rguenther@suse.de>
5913 PR tree-optimization/82102
5914 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
5916 2017-09-05 Martin Liska <mliska@suse.cz>
5918 PR tree-optimization/82032
5919 * tree-cfg.c (generate_range_test): New function.
5920 * tree-cfg.h (generate_range_test): Declared here.
5921 * tree-cfgcleanup.c (convert_single_case_switch): New function.
5922 (cleanup_control_expr_graph): Use it.
5923 * tree-switch-conversion.c (try_switch_expansion): Remove
5925 (emit_case_nodes): Use generate_range_test.
5927 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
5930 * config/i386/i386.md (*<btsc><mode>_mask): Add
5931 TARGET_USE_BT to insn constraint.
5932 (*btr<mode>_mask): Ditto.
5934 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
5936 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
5937 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
5939 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
5942 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
5943 TARGET_NEON and TARGET_IWMMXT.
5944 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
5945 TARGET_NEON and TARGET_IWMMXT.
5946 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
5948 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
5950 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
5951 (ix86_rewrite_tls_address): Ditto.
5952 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
5953 (ix86_rewrite_tls_address_1): Ditto.
5954 (ix86_rewrite_tls_address): Ditto.
5955 * config/i386/predicates.md (tls_address_pattern): New predicate.
5956 * config/i386/i386.md (TLS address splitter): New splitter.
5958 2017-09-04 Richard Biener <rguenther@suse.de>
5960 PR tree-optimization/82084
5961 * fold-const.h (can_native_encode_string_p): Declare.
5962 * fold-const.c (can_native_encode_string_p): Factor out from ...
5963 (native_encode_string): ... here.
5964 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
5965 vectorizing stores from constants we later cannot handle.
5967 2017-09-04 Marek Polacek <polacek@redhat.com>
5970 * doc/invoke.texi: Update -Wtautological-compare documentation.
5972 2017-09-04 Jeff Law <law@redhat.com>
5974 PR tree-optimization/64910
5975 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
5976 swap the first and last operand if the last is a constant.
5978 2017-09-04 Marek Polacek <polacek@redhat.com>
5981 * convert.c (do_narrow): When sanitizing signed integer overflows,
5982 bail out for signed types.
5983 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
5985 2017-09-04 Richard Biener <rguenther@suse.de>
5987 PR tree-optimization/82060
5988 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
5989 Move devirtualization after stmt folding and before EH/AB/noreturn
5990 cleanup to get the stmt refs canonicalized. Use a bool instead
5991 of gimple_modified_p since that doesn't work for NOPs. Schedule
5992 NOPs generated by folding for removal.
5994 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5995 Alan Hayward <alan.hayward@arm.com>
5996 David Sherwood <david.sherwood@arm.com>
5998 * coretypes.h (pad_direction): New enum.
5999 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
6000 (FUNCTION_ARG_PADDING): Likewise.
6001 * target.def (function_arg_padding): New hook.
6002 * targhooks.h (default_function_arg_padding): Declare.
6003 * targhooks.c (default_function_arg_padding): New function.
6004 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
6005 (TARGET_FUNCTION_ARG_PADDING): ...this.
6006 * doc/tm.texi: Regenerate.
6007 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
6008 instead of direction.
6009 (compute_argument_addresses): Likewise.
6010 (load_register_parameters): Likewise.
6011 (emit_library_call_value_1): Likewise.
6012 (store_one_arg): Use targetm.calls.function_arg_padding instead
6013 of FUNCTION_ARG_PADDING.
6014 (must_pass_in_stack_var_size_or_pad): Likewise.
6015 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
6016 (emit_group_store): Likewise.
6017 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
6018 instead of FUNCTION_ARG_PADDING.
6019 (emit_push_insn): Likewise, and propagate enum change throughout
6021 * function.h (direction): Delete.
6022 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
6024 * function.c (assign_parm_find_stack_rtl): Likewise.
6025 (assign_parm_setup_block_p): Likewise.
6026 (assign_parm_setup_block): Likewise.
6027 (gimplify_parameters): Likewise.
6028 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
6029 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
6031 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
6032 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
6033 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
6034 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
6035 (aarch64_function_arg_padding): ...this new function.
6036 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
6037 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6038 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
6039 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
6040 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
6041 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6042 (arm_pad_arg_upward): Replace with...
6043 (arm_function_arg_padding): ...this new function.
6044 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
6046 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
6047 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
6048 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6049 (ia64_hpux_function_arg_padding): Replace with...
6050 (ia64_function_arg_padding): ...this new function. Use pad_direction
6051 instead of direction. Check for TARGET_HPUX.
6052 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
6053 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6054 (iq2000_function_arg_padding): New function.
6055 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
6056 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
6057 (mips_function_arg_padding): ...this new function.
6058 (mips_pad_reg_upward): Update accordingly.
6059 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6060 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
6061 targetm.calls.function_arg_padding.
6062 (FUNCTION_ARG_PADDING): Delete.
6063 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
6064 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
6065 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
6066 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
6067 (nios2_block_reg_padding): Return pad_direction instead of direction.
6068 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
6069 instead of direction.
6070 (nios2_function_arg_padding): Likewise. Make static.
6071 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6072 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
6073 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
6074 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
6075 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6076 (pa_function_arg_padding): Make static. Return pad_direction instead
6078 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
6079 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
6080 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
6081 instead of direction. Use targetm.calls.function_arg_padding.
6082 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
6083 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
6084 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
6085 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
6086 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
6088 (function_arg_padding): Rename to...
6089 (rs6000_function_arg_padding): ...this. Make static. Return
6090 pad_direction instead of direction.
6091 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
6092 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
6093 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
6094 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
6095 instead of direction. Use targetm.calls.function_arg_padding.
6096 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
6097 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
6098 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
6099 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
6100 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
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/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
6106 * config/s390/s390.c (s390_function_arg_padding): New function.
6107 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6108 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
6109 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
6110 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6111 (function_arg_padding): Rename to...
6112 (sparc_function_arg_padding): ...this. Make static. Return
6113 pad_direction instead of direction.
6114 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
6115 * config/spu/spu.c (spu_function_arg_padding): New function.
6116 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6117 * system.h (FUNCTION_ARG_PADDING): Poison.
6119 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6120 Alan Hayward <alan.hayward@arm.com>
6121 David Sherwood <david.sherwood@arm.com>
6123 * target.def (modes_tieable_p): New hook.
6124 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
6125 (TARGET_MODES_TIEABLE_P): ...this.
6126 * doc/tm.texi.in: Regenerate.
6127 * hooks.h (hook_bool_mode_mode_true): Declare.
6128 * hooks.c (hook_bool_mode_mode_true): New function.
6129 * combine.c (subst): Use targetm.modes_tieable_p instead of
6131 * dse.c (find_shift_sequence): Likewise.
6132 * expmed.c (extract_low_bits): Likewise.
6133 * lower-subreg.c: Include target.h.
6134 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
6136 * rtlanal.c (rtx_cost): Likewise.
6137 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
6138 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
6139 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
6140 (TARGET_MODES_TIEABLE_P): Redefine.
6141 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
6142 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
6143 (TARGET_MODES_TIEABLE_P): Redefine.
6144 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
6145 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
6146 (arc_modes_tieable_p): New function.
6147 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
6148 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
6149 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
6150 (arm_modes_tieable_p): Make static.
6151 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
6152 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
6153 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
6154 (TARGET_MODES_TIEABLE_P): Redefine.
6155 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
6156 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
6157 (TARGET_MODES_TIEABLE_P): Redefine.
6158 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
6159 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
6160 (cr16_modes_tieable_p): New function.
6161 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
6162 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
6163 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
6164 (TRULY_NOOP_TRUNCATION): Update comment.
6165 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
6166 (TRULY_NOOP_TRUNCATION): Update comment.
6167 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
6168 (frv_modes_tieable_p): New function.
6169 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
6170 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
6171 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
6172 (TARGET_MODES_TIEABLE_P): Redefine.
6173 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
6174 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
6175 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
6176 (TARGET_MODES_TIEABLE_P): Redefine.
6177 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
6178 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
6179 (ia64_modes_tieable_p): New function.
6180 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
6181 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
6182 (iq2000_modes_tieable_p): New function.
6183 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
6184 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
6185 (lm32_modes_tieable_p): New function.
6186 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
6187 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
6188 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
6189 (TARGET_MODES_TIEABLE_P): Redefine.
6190 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
6191 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
6192 (m32r_modes_tieable_p): New function.
6193 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
6194 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
6195 (m68k_modes_tieable_p): New function.
6196 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
6197 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
6198 (mcore_modes_tieable_p): New function.
6199 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
6200 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
6202 (TARGET_MODES_TIEABLE_P): Redefine.
6203 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
6204 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
6205 * config/mips/mips.c (mips_modes_tieable_p): Make static.
6206 (TARGET_MODES_TIEABLE_P): Redefine.
6207 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
6208 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
6209 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
6210 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
6211 (mn10300_modes_tieable_p): ...this and make static.
6212 (TARGET_MODES_TIEABLE_P): Redefine.
6213 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
6214 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
6215 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
6216 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
6217 (msp430_modes_tieable_p): Make static.
6218 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
6219 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
6220 (TARGET_MODES_TIEABLE_P): Redefine.
6221 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
6222 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
6223 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
6224 (TARGET_MODES_TIEABLE_P): Redefine.
6225 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
6226 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
6227 * config/pa/pa.c (pa_modes_tieable_p): Make static.
6228 (TARGET_MODES_TIEABLE_P): Redefine.
6229 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
6230 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
6231 (pdp11_modes_tieable_p): New function.
6232 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
6233 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
6234 (rs6000_modes_tieable_p): New function.
6235 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
6236 * config/powerpcspe/powerpcspe.md: Update comment.
6237 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
6238 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
6239 (TARGET_MODES_TIEABLE_P): Redefine.
6240 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
6241 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
6242 (rl78_modes_tieable_p): New function.
6243 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
6244 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
6245 (rs6000_modes_tieable_p): New function.
6246 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
6247 * config/rs6000/rs6000.md: Update comment.
6248 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
6249 * config/rx/rx.c (rx_modes_tieable_p): New function.
6250 (TARGET_MODES_TIEABLE_P): Redefine.
6251 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
6252 * config/s390/s390.c (s390_modes_tieable_p): New function.
6253 (TARGET_MODES_TIEABLE_P): Redefine.
6254 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
6255 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
6256 (sh_modes_tieable_p): New function.
6257 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
6258 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
6259 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
6260 (sparc_modes_tieable_p): Make static.
6261 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
6262 * config/spu/spu.c (spu_modes_tieable_p): New function.
6263 (TARGET_MODES_TIEABLE_P): Redefine.
6264 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
6265 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
6266 (TARGET_MODES_TIEABLE_P): Redefine.
6267 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
6268 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
6269 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
6270 * config/v850/v850.c (v850_modes_tieable_p): New function.
6271 (TARGET_MODES_TIEABLE_P): Redefine.
6272 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
6273 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
6274 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
6275 (visium_modes_tieable_p): New function.
6276 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
6277 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
6278 (xtensa_modes_tieable_p): New function.
6279 * system.h (MODES_TIEABLE_P): Poison.
6281 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6282 Alan Hayward <alan.hayward@arm.com>
6283 David Sherwood <david.sherwood@arm.com>
6285 * target.def (hard_regno_mode_ok): New hook.
6286 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
6287 (TARGET_HARD_REGNO_MODE_OK): ...this.
6288 * doc/tm.texi.in: Regenerate.
6289 * hooks.h (hook_bool_uint_mode_true): Declare.
6290 * hooks.c (hook_bool_uint_mode_true): New function.
6291 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
6293 * genpreds.c (write_insn_preds_c): Add an include of target.h.
6294 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
6295 instead of HARD_REGNO_MODE_OK.
6296 * caller-save.c: Include target.h.
6297 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
6299 * combine.c (can_combine_p): Likewise.
6300 (combinable_i3pat): Likewise.
6301 (can_change_dest_mode): Likewise.
6302 * expr.c (init_expr_target): Likewise.
6303 (convert_move): Likewise.
6304 (convert_modes): Likewise.
6305 * ira.c (setup_prohibited_class_mode_regs): Likewise.
6306 (setup_prohibited_mode_move_regs): Likewise.
6307 * ira.h (target_ira): Likewise.
6308 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6309 * lra-constraints.c (process_alt_operands): Likewise.
6310 (split_reg): Likewise.
6311 * recog.c (peep2_find_free_register): Likewise.
6312 * ree.c (combine_reaching_defs): Likewise.
6313 * regcprop.c (maybe_mode_change): Likewise.
6314 * reginfo.c (init_reg_sets_1): Likewise.
6315 (choose_hard_reg_mode): Likewise.
6316 (simplifiable_subregs): Likewise.
6317 * regrename.c (check_new_reg_p): Likewise.
6318 * reload.c (find_valid_class): Likewise.
6319 (find_valid_class_1): Likewise.
6320 (reload_inner_reg_of_subreg): Likewise.
6321 (push_reload): Likewise.
6322 (combine_reloads): Likewise.
6323 (find_dummy_reload): Likewise.
6324 (find_reloads): Likewise.
6325 * reload1.c (find_reg): Likewise.
6326 (set_reload_reg): Likewise.
6327 (allocate_reload_reg): Likewise.
6328 (choose_reload_regs): Likewise.
6329 (reload_adjust_reg_for_temp): Likewise.
6330 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
6331 (simplify_subreg_regno): Likewise.
6332 * sel-sched.c (init_regs_for_mode): Likewise.
6333 * varasm.c (make_decl_rtl): Likewise.
6334 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
6335 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
6337 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
6338 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
6339 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6340 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
6341 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
6342 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6343 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
6344 (arc_mode_class): Delete.
6345 (HARD_REGNO_MODE_OK): Delete.
6346 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6347 (arc_hard_regno_mode_ok): Rename old array to...
6348 (arc_hard_regno_mode_ok_modes): ...this.
6349 (arc_conditional_register_usage): Update accordingly.
6350 (arc_mode_class): Make static.
6351 (arc_hard_regno_mode_ok): New function.
6352 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
6353 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
6354 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6355 (arm_hard_regno_mode_ok): Make static.
6356 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
6358 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
6359 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
6360 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
6362 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6363 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
6364 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
6365 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
6366 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
6367 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6368 * config/bfin/predicates.md (valid_reg_operand): Use
6369 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
6370 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
6371 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
6372 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6373 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
6374 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
6375 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6376 (cr16_hard_regno_mode_ok): Make static and return a bool.
6377 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
6378 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6379 (cris_hard_regno_mode_ok): New function.
6380 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
6381 (epiphany_mode_class): Delete.
6382 (HARD_REGNO_MODE_OK): Delete.
6383 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
6384 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6385 (hard_regno_mode_ok): Rename to...
6386 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
6387 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
6388 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
6390 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
6391 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
6392 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6393 (frv_hard_regno_mode_ok): Make static and return a bool.
6394 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
6396 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
6397 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
6398 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
6399 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
6401 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6402 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
6403 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
6404 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
6406 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6407 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
6408 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6409 (ia64_hard_regno_mode_ok): New function.
6410 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
6411 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6412 (iq2000_hard_regno_mode_ok): New function.
6413 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
6414 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6415 (lm32_hard_regno_mode_ok): New function.
6416 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
6417 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
6418 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
6419 instead of HARD_REGNO_MODE_OK.
6420 (m32c_hard_regno_ok): Rename to...
6421 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
6422 (m32c_cannot_change_mode_class): Update accordingly.
6423 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6424 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
6425 (m32r_mode_class): Delete.
6426 (HARD_REGNO_MODE_OK): Delete.
6427 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6428 (m32r_hard_regno_mode_ok): Rename to...
6429 (m32r_hard_regno_modes): ...this.
6430 (m32r_mode_class): Make static.
6431 (m32r_hard_regno_mode_ok): New function.
6432 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
6433 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
6434 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6435 (m68k_hard_regno_mode_ok): Make static.
6436 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
6437 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6438 (mcore_hard_regno_mode_ok): New function.
6439 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
6440 (HARD_REGNO_MODE_OK): Delete.
6441 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
6443 (microblaze_hard_regno_mode_ok_p): ...this and make static.
6444 (microblaze_hard_regno_mode_ok): New function.
6445 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6446 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
6447 (mips_hard_regno_mode_ok): Delete.
6448 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
6449 (mips_hard_regno_mode_ok_p): ...this and make static.
6450 (mips_hard_regno_mode_ok_p): Rename to...
6451 (mips_hard_regno_mode_ok_uncached): ...this.
6452 (mips_hard_regno_mode_ok): New function.
6453 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
6454 of HARD_REGNO_MODE_OK.
6455 (mips_option_override): Update after above name changes.
6456 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6457 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
6458 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
6459 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
6460 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
6461 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6462 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
6463 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
6464 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
6465 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6466 (msp430_hard_regno_mode_ok): Make static and return a bool.
6467 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
6468 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
6469 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
6471 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6472 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
6473 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
6474 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
6475 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
6476 (PA_HARD_REGNO_MODE_OK): ...this
6477 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
6478 (PA_HARD_REGNO_MODE_OK): ...this.
6479 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6480 (pa_hard_regno_mode_ok): New function.
6481 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
6482 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6483 (pdp11_hard_regno_mode_ok): New function.
6484 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
6485 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
6487 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
6489 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6490 (rs6000_hard_regno_mode_ok): Rename to...
6491 (rs6000_hard_regno_mode_ok_uncached): ...this.
6492 (rs6000_init_hard_regno_mode_ok): Update accordingly.
6493 (rs6000_hard_regno_mode_ok): New function.
6494 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
6495 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
6496 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
6497 (riscv_hard_regno_mode_ok): ...this and make static.
6498 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6499 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
6500 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
6501 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6502 (rl78_hard_regno_mode_ok): Make static and return bool.
6503 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
6504 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
6506 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
6507 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6508 (rs6000_hard_regno_mode_ok): Rename to...
6509 (rs6000_hard_regno_mode_ok_uncached): ...this.
6510 (rs6000_init_hard_regno_mode_ok): Update accordingly.
6511 (rs6000_hard_regno_mode_ok): New function.
6512 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
6513 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
6514 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6515 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
6516 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
6517 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
6518 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6519 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
6520 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
6521 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6522 (sh_hard_regno_mode_ok): Make static.
6523 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
6524 instead of HARD_REGNO_MODE_OK.
6525 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
6526 (sparc_mode_class): Delete.
6527 (HARD_REGNO_MODE_OK): Delete.
6528 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6529 (hard_regno_mode_classes): Make static.
6530 (sparc_mode_class): Likewise.
6531 (sparc_hard_regno_mode_ok): New function.
6532 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
6533 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
6534 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
6536 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6537 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
6538 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
6539 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
6540 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
6541 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6542 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
6543 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
6544 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6545 (visium_hard_regno_mode_ok): New function.
6546 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
6547 instead of HARD_REGNO_MODE_OK.
6548 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
6549 (HARD_REGNO_MODE_OK): Delete.
6550 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
6551 (xtensa_hard_regno_mode_ok_p): ...this and make static.
6552 (xtensa_option_override): Update accordingly.
6553 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6554 (xtensa_hard_regno_mode_ok): New function.
6555 * system.h (HARD_REGNO_MODE_OK): Poison.
6557 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6558 Alan Hayward <alan.hayward@arm.com>
6559 David Sherwood <david.sherwood@arm.com>
6561 * target.def (hard_regno_call_part_clobbered): New hook.
6562 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
6563 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
6564 * doc/tm.texi: Regenerate.
6565 * hooks.h (hook_bool_uint_mode_false): Declare.
6566 * hooks.c (hook_bool_uint_mode_false): New function.
6567 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6568 * cselib.c (cselib_process_insn): Use
6569 targetm.hard_regno_call_part_clobbered instead of
6570 HARD_REGNO_CALL_PART_CLOBBERED.
6571 * ira-conflicts.c (ira_build_conflicts): Likewise.
6572 * ira-costs.c (ira_tune_allocno_costs): Likewise.
6573 * lra-constraints.c (need_for_call_save_p): Likewise.
6574 * lra-lives.c: Include target.h.
6575 (check_pseudos_live_through_calls): Use
6576 targetm.hard_regno_call_part_clobbered instead of
6577 HARD_REGNO_CALL_PART_CLOBBERED.
6578 * regcprop.c: Include target.h.
6579 (copyprop_hardreg_forward_1): Use
6580 targetm.hard_regno_call_part_clobbered instead of
6581 HARD_REGNO_CALL_PART_CLOBBERED.
6582 * reginfo.c (choose_hard_reg_mode): Likewise.
6583 * regrename.c (check_new_reg_p): Likewise.
6584 * reload.c (find_equiv_reg): Likewise.
6585 * reload1.c (emit_reload_insns): Likewise.
6586 * sched-deps.c (deps_analyze_insn): Likewise.
6587 * sel-sched.c (init_regs_for_mode): Likewise.
6588 (mark_unavailable_hard_regs): Likewise.
6589 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
6590 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6591 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
6593 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6594 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6595 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
6597 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
6599 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6600 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6601 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
6603 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6604 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6605 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
6607 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6608 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
6610 * config/powerpcspe/powerpcspe.c
6611 (rs6000_hard_regno_call_part_clobbered): New function.
6612 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6613 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6614 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
6616 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6617 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6618 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
6620 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6621 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6622 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
6624 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6625 Alan Hayward <alan.hayward@arm.com>
6626 David Sherwood <david.sherwood@arm.com>
6628 * rtl.h (subreg_memory_offset): Declare.
6629 * emit-rtl.c (subreg_memory_offset): New function.
6630 * expmed.c (store_bit_field_1): Use it.
6631 * expr.c (undefined_operand_subword_p): Likewise.
6632 * simplify-rtx.c (simplify_subreg): Likewise.
6634 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
6636 PR rtl-optimization/57448
6639 * optabs.c (expand_atomic_load): Place compiler memory barriers if
6640 using atomic_load pattern.
6641 (expand_atomic_store): Likewise.
6643 2017-09-04 Jakub Jelinek <jakub@redhat.com>
6646 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
6647 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
6648 handling. Use replace_call_with_value with NULL instead of
6649 gsi_replace, unlink_stmt_vdef and release_defs.
6651 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
6654 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
6656 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6659 * rtl.h (emit_library_call_value_1): Declare.
6660 (emit_library_call): Replace declaration with a series of overloads.
6661 Remove the parameter count argument.
6662 (emit_library_call_value): Likewise.
6663 * calls.c (emit_library_call_value_1): Make global. Replace varargs
6664 with an "rtx_mode_t *".
6665 (emit_library_call_value): Delete.
6666 (emit_library_call): Likewise.
6667 * asan.c (asan_emit_stack_protection): Update calls accordingly.
6668 (asan_emit_allocas_unpoison): Likewise.
6669 * builtins.c (expand_builtin_powi): Likewise.
6670 (expand_asan_emit_allocas_unpoison): Likewise.
6671 * cfgexpand.c (expand_main_function): Likewise.
6672 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
6673 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
6674 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
6675 * config/arm/arm.c (arm_trampoline_init): Likewise.
6676 (arm_call_tls_get_addr): Likewise.
6677 (arm_expand_divmod_libfunc): Likewise.
6678 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
6679 (smulsi3_highpart): Likewise.
6680 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
6681 (c6x_expand_compare): Likewise.
6682 (c6x_expand_movmem): Likewise.
6683 * config/frv/frv.c (frv_trampoline_init): Likewise.
6684 * config/i386/i386.c (ix86_trampoline_init): Likewise.
6685 (ix86_expand_divmod_libfunc): Likewise.
6686 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
6687 (ia64_expand_compare): Likewise.
6688 (ia64_profile_hook): Likewise.
6689 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
6690 (nonlocal_goto): Likewise.
6691 (restore_stack_nonlocal): Likewise.
6692 * config/m32r/m32r.c (block_move_call): Likewise.
6693 (m32r_trampoline_init): Likewise.
6694 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
6695 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
6696 (m68k_call_m68k_read_tp): Likewise.
6697 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
6698 (microblaze_expand_divide): Likewise.
6699 * config/mips/mips.h (mips_args): Likewise.
6700 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
6701 (MIPS_ICACHE_SYNC): Likewise.
6702 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
6703 (nios2_trampoline_init): Likewise.
6704 * config/pa/pa.c (hppa_tls_call): Likewise.
6705 (pa_trampoline_init): Likewise.
6706 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
6707 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
6708 (expand_strn_compare): Likewise.
6709 (rs6000_generate_compare): Likewise.
6710 (rs6000_expand_float128_convert): Likewise.
6711 (output_profile_hook): Likewise.
6712 (rs6000_trampoline_init): Likewise.
6713 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
6714 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
6715 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
6716 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
6717 (rs6000_generate_compare): Likewise.
6718 (rs6000_expand_float128_convert): Likewise.
6719 (output_profile_hook): Likewise.
6720 (rs6000_trampoline_init): Likewise.
6721 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
6722 * config/sh/sh.c (sh_trampoline_init): Likewise.
6723 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
6724 (sparc_emit_float_lib_cmp): Likewise.
6725 (sparc32_initialize_trampoline): Likewise.
6726 (sparc64_initialize_trampoline): Likewise.
6727 (sparc_profile_hook): Likewise.
6728 * config/spu/spu.c (ea_load_store): Likewise.
6729 * config/spu/spu.md (floatunssidf2): Likewise.
6730 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
6731 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
6732 * config/visium/visium.c (expand_block_move_4): Likewise.
6733 (expand_block_move_2): Likewise.
6734 (expand_block_move_1): Likewise.
6735 (expand_block_set_4): Likewise.
6736 (expand_block_set_2): Likewise.
6737 (expand_block_set_1): Likewise.
6738 (visium_trampoline_init): Likewise.
6739 (visium_profile_hook): Likewise.
6740 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
6741 (xtensa_setup_frame_addresses): Likewise.
6742 (xtensa_trampoline_init): Likewise.
6743 * except.c (sjlj_emit_function_enter): Likewise.
6744 (sjlj_emit_function_exit): Likewise.
6745 * explow.c (allocate_dynamic_stack_space): Likewise.
6746 (probe_stack_range): Likewise.
6747 * expr.c (convert_mode_scalar): Likewise.
6748 * optabs.c (expand_binop): Likewise.
6749 (expand_twoval_binop_libfunc): Likewise.
6750 (expand_unop): Likewise.
6751 (prepare_cmp_insn): Likewise.
6752 (prepare_float_lib_cmp): Likewise.
6753 (expand_float): Likewise.
6754 (expand_fix): Likewise.
6755 (expand_fixed_convert): Likewise.
6756 (maybe_emit_sync_lock_test_and_set): Likewise.
6757 (expand_atomic_compare_and_swap): Likewise.
6758 (expand_mem_thread_fence): Likewise.
6759 (expand_atomic_fetch_op): Likewise.
6761 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
6763 * doc/generic.texi (OpenACC): Adjust URL.
6764 * doc/invoke.texi (C Dialect Options): Ditto.
6766 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
6768 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
6769 predicate for operand 1. Add (m,<S>) constraint.
6770 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
6771 Prevent memory operand 1 with register operand 2.
6773 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
6775 PR rtl-optimization/82024
6776 * combine.c (try_combine): If the combination result is a PARALLEL,
6777 and we only need to retain the SET in there that would be placed
6778 at I2, check that we can place that at I3 instead, before doing so.
6780 2017-09-01 Jakub Jelinek <jakub@redhat.com>
6783 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
6785 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
6786 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
6789 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
6790 Jeff Law <law@redhat.com>
6792 * varasm.c (bss_initializer_p): Do not put constants into .bss
6793 (categorize_decl_for_section): Handle bss_initializer_p returning
6794 false when DECL_INITIAL is NULL.
6796 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6799 * config/s390/s390.c (s390_can_inline_p): New function.
6801 2017-09-01 Jeff Law <law@redhat.com>
6803 PR tree-optimization/82052
6804 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
6805 Always initialize the returned slot after a hash table miss
6806 when INSERT is true.
6808 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
6810 * config/s390/s390.md (mem_signal_fence): Remove.
6811 * doc/md.texi (mem_signal_fence): Remove.
6812 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
6814 * target-insns.def (mem_signal_fence): Remove.
6816 2017-09-01 Jakub Jelinek <jakub@redhat.com>
6819 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
6822 * asan.c (create_odr_indicator): Strip name encoding from assembler
6823 name before appending it after __odr_asan_.
6825 2017-09-01 Martin Liska <mliska@suse.cz>
6827 PR tree-optimization/82059
6828 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
6829 frequency only when an edge is redirected.
6831 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6833 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
6834 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
6835 (arc_conditional_register_usage): Remove ARC600 lp_count
6837 (arc_file_start): Emit Tag_ARC_CPU_variation.
6838 (arc_can_use_doloop_p): New conditions to use ZOLs.
6839 (hwloop_fail): New function.
6840 (hwloop_optimize): Likewise.
6841 (hwloop_pattern_reg): Likewise.
6842 (arc_doloop_hooks): New struct, to be used with reorg_loops.
6843 (arc_reorg_loops): New function, calls reorg_loops.
6844 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
6845 (arc600_corereg_hazard): Remove ZOL checking, case handled by
6847 (arc_loop_hazard): Remove function, functionality moved into
6849 (arc_hazard): Remove arc_loop_hazard call.
6850 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
6851 into hwloop_optimize.
6852 (arc_label_align): Remove ZOL handling.
6853 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
6854 * config/arc/arc.md (doloop_begin): Remove pattern.
6855 (doloop_begin_i): Likewise.
6856 (doloop_end_i): Likewise.
6857 (doloop_fallback): Likewise.
6858 (doloop_fallback_m): Likewise.
6859 (doloop_end): Reimplement expand.
6860 (arc_lp): New pattern for LP instruction.
6861 (loop_end): New pattern.
6862 (loop_fail): Likewise.
6863 (decrement_and_branch_until_zero): Likewise.
6864 * config/arc/arc.opt (mlpc-width): New option.
6865 * doc/invoke.texi (mlpc-width): Document option.
6867 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6869 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
6870 (arc_ccfsm_advance): Fix checking for delay slots.
6871 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
6873 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6875 * config/arc/arc.md (movqi_insn): Add stores to save constant long
6877 (movhi_insn): Update store instruction constraint which are saving
6878 6-bit short immediates.
6879 (movsi_insn): Consider also short scaled load operations.
6880 (zero_extendhisi2_i): Use Usd constraint instead of T.
6881 (extendhisi2_i): Add q constraint.
6882 (arc_clzsi2): Add type and length attributes.
6883 (arc_ctzsi2): Likewise.
6884 * config/arc/constraints.md (Usc): Update constraint, the
6885 assembler can parse two relocations for a single instruction.
6887 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6889 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
6890 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
6892 2017-08-31 Olivier Hainque <hainque@adacore.com>
6894 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
6895 match as powerpc-wrs-vxworks*.
6897 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
6899 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
6900 register constraint for by-element operand.
6901 (aarch64_mls_elt_merge<mode>): Likewise.
6903 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6905 * config/arc/arc.c (arc_can_follow_jump): Check for short
6908 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6910 * config.gcc: Use g.opt for arc.
6911 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
6912 functionality moved to ...
6913 (legitimate_scaled_address_p): New function, ...here.
6914 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
6915 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
6916 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
6918 (arc_override_options): Handle G option.
6919 (arc_output_pic_addr_const): Correct function definition.
6920 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
6921 (arc_decl_anon_ns_mem_p): Delete.
6922 (arc_in_small_data_p): Overhaul this function to take into
6923 consideration the value given via G option.
6924 (arc_rewrite_small_data_1): Renamed and corrected old
6925 arc_rewrite_small_data function.
6926 (arc_rewrite_small_data): New function.
6927 (small_data_pattern): Don't use pic_offset_table_rtx.
6928 * config/arc/arc.h (CC1_SPEC): Recognize G option.
6929 * config/arc/simdext.md (movmisalignv2hi): Use
6930 prepare_move_operands function.
6932 (movmisalign*): Likewise.
6933 * doc/invoke.texi (ARC options): Document -G option.
6935 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6937 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
6939 * config/arc/arc.c (arc_print_operand): Output scalled address for
6940 sdata whenever is possible.
6941 (arc_in_small_data_p): Allow sdata for 64bit datum when double
6942 load/stores are available.
6943 (compact_sda_memory_operand): Check for the alignment required by
6944 code density instructions.
6945 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
6947 * config/arc/constraints.md (Usd): Update constraint.
6948 (Us0): New constraint.
6949 (Usc): Update constraint.
6951 2017-08-31 Richard Biener <rguenther@suse.de>
6954 * dwarf2out.c (dwarf2out_early_global_decl): Process each
6957 2017-08-31 Tamar Christina <tamar.christina@arm.com>
6959 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
6960 Resize type_signature.
6962 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6963 Alan Hayward <alan.hayward@arm.com>
6964 David Sherwood <david.sherwood@arm.com>
6966 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
6967 subregs whose inner modes can be stored in GPRs.
6968 (aarch64_classify_index): Likewise.
6970 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6971 Alan Hayward <alan.hayward@arm.com>
6972 David Sherwood <david.sherwood@arm.com>
6974 * config/aarch64/iterators.md (V_cmp_result): Rename to...
6975 (V_INT_EQUIV): ...this.
6976 (v_cmp_result): Rename to...
6977 (v_int_equiv): ...this.
6978 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
6979 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
6980 (copysign<mode>3): Likewise.
6981 (aarch64_simd_bsl<mode>_internal): Likewise.
6982 (aarch64_simd_bsl<mode>): Likewise.
6983 (vec_cmp<mode><mode>): Likewise.
6984 (vcond<mode><mode>): Likewise.
6985 (vcond<v_cmp_mixed><mode>): Likewise.
6986 (vcondu<mode><v_cmp_mixed>): Likewise.
6987 (aarch64_cm<optab><mode>): Likewise.
6988 (aarch64_cmtst<mode>): Likewise.
6989 (aarch64_fac<optab><mode>): Likewise.
6990 (vec_perm_const<mode>): Likewise.
6991 (vcond_mask_<mode><v_cmp_result>): Rename to...
6992 (vcond_mask_<mode><v_int_equiv>): ...this.
6993 (vec_cmp<mode><v_cmp_result>): Rename to...
6994 (vec_cmp<mode><v_int_equiv>): ...this.
6996 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6997 Alan Hayward <alan.hayward@arm.com>
6998 David Sherwood <david.sherwood@arm.com>
7000 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
7002 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
7003 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
7004 (UNSPEC_LD4_DREG): New unspecs.
7005 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
7006 (aarch64_ld2<mode>_dreg_be): Replace with...
7007 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
7009 (aarch64_ld3<mode>_dreg_le)
7010 (aarch64_ld3<mode>_dreg_be): Replace with...
7011 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
7013 (aarch64_ld4<mode>_dreg_le)
7014 (aarch64_ld4<mode>_dreg_be): Replace with...
7015 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
7018 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7020 PR tree-optimization/81987
7021 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
7022 insert an initializer in a location not dominated by the stride
7025 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
7027 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
7028 on the entire header of the finally block in the fallthru case.
7030 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
7032 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
7034 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
7036 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
7037 rs6000_emit_move_from_cr and call renamed function.
7038 (rs6000_emit_prologue): Call renamed functions.
7039 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
7040 movesi_from_cr, remove volatile CRs.
7042 2017-08-30 Jon Beniston <jon@beniston.com>
7043 Richard Biener <rguenther@suse.de>
7045 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
7046 of VECTOR_MODE_P check.
7047 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
7048 element vector types.
7050 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7052 * df.h (df_read_modify_subreg_p): Remove in favor of...
7053 * rtl.h (read_modify_subreg_p): ...this new function. Take a
7054 const_rtx instead of an rtx.
7055 * cprop.c (local_cprop_find_used_regs): Update accordingly.
7056 * df-problems.c (df_word_lr_mark_ref): Likewise.
7057 * ira-lives.c (mark_pseudo_reg_live): Likewise.
7058 (mark_pseudo_reg_dead): Likewise.
7059 (mark_ref_dead): Likewise.
7060 * reginfo.c (init_subregs_of_mode): Likewise.
7061 * sched-deps.c (sched_analyze_1): Likewise.
7062 * df-scan.c (df_def_record_1): Likewise.
7063 (df_uses_record): Likewise.
7064 (df_read_modify_subreg_p): Remove in favor of...
7065 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
7066 const_rtx instead of an rtx.
7068 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7069 Alan Hayward <alan.hayward@arm.com>
7070 David Sherwood <david.sherwood@arm.com>
7072 * rtl.h (partial_subreg_p): New function.
7073 * caller-save.c (save_call_clobbered_regs): Use it.
7074 * calls.c (expand_call): Likewise.
7075 * combine.c (combinable_i3pat): Likewise.
7076 (simplify_set): Likewise.
7077 (make_extraction): Likewise.
7078 (make_compound_operation_int): Likewise.
7079 (gen_lowpart_or_truncate): Likewise.
7080 (force_to_mode): Likewise.
7081 (make_field_assignment): Likewise.
7082 (reg_truncated_to_mode): Likewise.
7083 (record_truncated_value): Likewise.
7084 (move_deaths): Likewise.
7085 * cse.c (record_jump_cond): Likewise.
7086 (cse_insn): Likewise.
7087 * cselib.c (cselib_lookup_1): Likewise.
7088 * expmed.c (extract_bit_field_using_extv): Likewise.
7089 * function.c (assign_parm_setup_reg): Likewise.
7090 * ifcvt.c (noce_convert_multiple_sets): Likewise.
7091 * ira-build.c (create_insn_allocnos): Likewise.
7092 * lra-coalesce.c (merge_pseudos): Likewise.
7093 * lra-constraints.c (match_reload): Likewise.
7094 (simplify_operand_subreg): Likewise.
7095 (curr_insn_transform): Likewise.
7096 * lra-lives.c (process_bb_lives): Likewise.
7097 * lra.c (new_insn_reg): Likewise.
7098 (lra_substitute_pseudo): Likewise.
7099 * regcprop.c (mode_change_ok): Likewise.
7100 (maybe_mode_change): Likewise.
7101 (copyprop_hardreg_forward_1): Likewise.
7102 * reload.c (push_reload): Likewise.
7103 (find_reloads): Likewise.
7104 (find_reloads_subreg_address): Likewise.
7105 * reload1.c (alter_reg): Likewise.
7106 (eliminate_regs_1): Likewise.
7107 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7109 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
7111 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
7112 back to if statements, including unpack.
7114 2017-08-30 Martin Liska <mliska@suse.cz>
7117 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
7119 (func_checker::compare_asm_inputs_outputs): ... this function.
7120 (func_checker::compare_gimple_asm): Use the function to compare
7121 also ASM constrains.
7122 * ipa-icf-gimple.h: Rename the function.
7124 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7125 Alan Hayward <alan.hayward@arm.com>
7126 David Sherwood <david.sherwood@arm.com>
7128 * coretypes.h (complex_mode): New type.
7129 * gdbhooks.py (build_pretty_printer): Handle it.
7130 * machmode.h (complex_mode): New class.
7131 (complex_mode::includes_p): New function.
7132 (is_complex_int_mode): Likewise.
7133 (is_complex_float_mode): Likewise.
7134 * genmodes.c (get_mode_class): Handle complex mode classes.
7135 * function.c (expand_function_end): Use is_complex_int_mode.
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 (scalar_mode_pod): New typedef.
7142 * gdbhooks.py (build_pretty_printer): Handle it.
7143 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
7144 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
7145 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
7146 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
7149 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7150 Alan Hayward <alan.hayward@arm.com>
7151 David Sherwood <david.sherwood@arm.com>
7153 * machmode.h (mode_for_vector): Take a scalar_mode instead
7155 * stor-layout.c (mode_for_vector): Likewise.
7156 * explow.c (promote_mode): Use as_a <scalar_mode>.
7157 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
7159 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7160 Alan Hayward <alan.hayward@arm.com>
7161 David Sherwood <david.sherwood@arm.com>
7163 * target.def (preferred_simd_mode): Take a scalar_mode
7164 instead of a machine_mode.
7165 * targhooks.h (default_preferred_simd_mode): Likewise.
7166 * targhooks.c (default_preferred_simd_mode): Likewise.
7167 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
7168 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
7169 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
7170 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
7171 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
7172 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
7173 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
7174 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
7176 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
7177 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
7178 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
7179 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
7180 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
7181 * doc/tm.texi: Regenerate.
7182 * optabs-query.c (can_vec_mask_load_store_p): Return false for
7185 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7186 Alan Hayward <alan.hayward@arm.com>
7187 David Sherwood <david.sherwood@arm.com>
7189 * target.def (scalar_mode_supported_p): Take a scalar_mode
7190 instead of a machine_mode.
7191 * targhooks.h (default_scalar_mode_supported_p): Likewise.
7192 * targhooks.c (default_scalar_mode_supported_p): Likewise.
7193 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
7194 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
7195 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
7196 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
7197 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
7198 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
7199 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
7200 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
7201 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
7202 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
7203 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
7204 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
7206 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
7207 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
7208 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
7209 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
7210 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
7212 * doc/tm.texi: Regenerate.
7214 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7215 Alan Hayward <alan.hayward@arm.com>
7216 David Sherwood <david.sherwood@arm.com>
7218 * coretypes.h (opt_scalar_mode): New typedef.
7219 * gdbhooks.py (build_pretty_printers): Handle it.
7220 * machmode.h (mode_iterator::get_2xwider): Add overload for
7222 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
7224 * expr.c (convert_mode_scalar): Likewise.
7225 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7226 * optabs.c (expand_float): Likewise.
7227 (expand_fix): Likewise.
7228 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7231 Alan Hayward <alan.hayward@arm.com>
7232 David Sherwood <david.sherwood@arm.com>
7234 * optabs.c (expand_float): Explicitly check for scalars before
7235 using a branching expansion.
7236 (expand_fix): Likewise.
7238 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7239 Alan Hayward <alan.hayward@arm.com>
7240 David Sherwood <david.sherwood@arm.com>
7242 * expr.c (convert_mode): Split scalar handling out into...
7243 (convert_mode_scalar): ...this new function. Treat the modes
7246 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7247 Alan Hayward <alan.hayward@arm.com>
7248 David Sherwood <david.sherwood@arm.com>
7250 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
7252 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
7254 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7255 Alan Hayward <alan.hayward@arm.com>
7256 David Sherwood <david.sherwood@arm.com>
7258 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
7259 rather than a machine_mode.
7260 (fixed_from_string): Likewise.
7261 (fixed_convert): Likewise.
7262 (fixed_convert_from_int): Likewise.
7263 (fixed_convert_from_real): Likewise.
7264 (real_convert_from_fixed): Likewise.
7265 * fixed-value.c (fixed_from_double_int): Likewise.
7266 (fixed_from_string): Likewise.
7267 (fixed_convert): Likewise.
7268 (fixed_convert_from_int): Likewise.
7269 (fixed_convert_from_real): Likewise.
7270 (real_convert_from_fixed): Likewise.
7271 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
7273 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7274 Alan Hayward <alan.hayward@arm.com>
7275 David Sherwood <david.sherwood@arm.com>
7277 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
7278 of separate mode class checks. Do not allow vector modes here.
7279 (immed_wide_int_const): Use as_a <scalar_mode>.
7280 * explow.c (trunc_int_for_mode): Likewise.
7281 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
7282 (wi::shwi): Likewise.
7283 (wi::min_value): Likewise.
7284 (wi::max_value): Likewise.
7285 * dwarf2out.c (loc_descriptor): Likewise.
7286 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
7289 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7290 Alan Hayward <alan.hayward@arm.com>
7291 David Sherwood <david.sherwood@arm.com>
7293 * tree.h (SCALAR_TYPE_MODE): New macro.
7294 * expr.c (expand_expr_addr_expr_1): Use it.
7295 (expand_expr_real_2): Likewise.
7296 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
7297 (fold_convert_const_fixed_from_int): Likewise.
7298 (fold_convert_const_fixed_from_real): Likewise.
7299 (native_encode_fixed): Likewise
7300 (native_encode_complex): Likewise
7301 (native_encode_vector): Likewise.
7302 (native_interpret_fixed): Likewise.
7303 (native_interpret_real): Likewise.
7304 (native_interpret_complex): Likewise.
7305 (native_interpret_vector): Likewise.
7306 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
7307 (simd_clone_adjust_argument_types): Likewise.
7308 (simd_clone_init_simd_arrays): Likewise.
7309 (simd_clone_adjust): Likewise.
7310 * stor-layout.c (layout_type): Likewise.
7311 * tree.c (build_minus_one_cst): Likewise.
7312 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
7313 * tree-inline.c (estimate_move_cost): Likewise.
7314 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
7315 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
7316 (vectorizable_reduction): Likewise.
7317 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
7318 (vect_recog_mixed_size_cond_pattern): Likewise.
7319 (check_bool_pattern): Likewise.
7320 (adjust_bool_pattern): Likewise.
7321 (search_type_for_mask_1): Likewise.
7322 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
7323 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7324 (vectorizable_load): Likewise.
7325 (vectorizable_store): Likewise.
7326 * ubsan.c (ubsan_encode_value): Likewise.
7327 * varasm.c (output_constant): Likewise.
7329 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7330 Alan Hayward <alan.hayward@arm.com>
7331 David Sherwood <david.sherwood@arm.com>
7333 * coretypes.h (scalar_mode): New class.
7334 * machmode.h (scalar_mode): Likewise.
7335 (scalar_mode::includes_p): New function.
7336 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
7337 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
7338 * genmodes.c (get_mode_class): Handle remaining scalar modes.
7339 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
7340 * expmed.c (store_bit_field_1): Likewise.
7341 (extract_bit_field_1): Likewise.
7342 * expr.c (write_complex_part): Likewise.
7343 (read_complex_part): Likewise.
7344 (emit_move_complex_push): Likewise.
7345 (expand_expr_real_2): Likewise.
7346 * function.c (assign_parm_setup_reg): Likewise.
7347 (assign_parms_unsplit_complex): Likewise.
7348 * optabs.c (expand_binop): Likewise.
7349 * rtlanal.c (subreg_get_info): Likewise.
7350 * simplify-rtx.c (simplify_immed_subreg): Likewise.
7351 * varasm.c (output_constant_pool_2): Likewise.
7353 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7354 Alan Hayward <alan.hayward@arm.com>
7355 David Sherwood <david.sherwood@arm.com>
7357 * expmed.c (extract_high_half): Use scalar_int_mode and remove
7359 (expmed_mult_highpart_optab): Likewise.
7360 (expmed_mult_highpart): Likewise.
7362 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7363 Alan Hayward <alan.hayward@arm.com>
7364 David Sherwood <david.sherwood@arm.com>
7366 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
7367 instead of a machine_mode.
7368 (builtin_memset_read_str): Likewise.
7369 * builtins.c (c_readstr): Likewise.
7370 (builtin_memcpy_read_str): Likewise.
7371 (builtin_strncpy_read_str): Likewise.
7372 (builtin_memset_read_str): Likewise.
7373 (builtin_memset_gen_str): Likewise.
7374 (expand_builtin_signbit): Use scalar_int_mode for local variables.
7375 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
7376 instead of a machine_mode.
7377 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
7379 (make_extraction): Likewise.
7380 (try_widen_shift_mode): Take and return scalar_int_modes instead
7382 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
7383 a scalar_int_mode instead of a machine_mode.
7384 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
7385 (avr_addr_space_pointer_mode): Likewise.
7386 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
7387 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
7388 (msp430_unwind_word_mode): Likewise.
7389 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
7390 (spu_addr_space_pointer_mode): Likewise.
7391 (spu_addr_space_address_mode): Likewise.
7392 (spu_libgcc_cmp_return_mode): Likewise.
7393 (spu_libgcc_shift_count_mode): Likewise.
7394 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
7395 (rl78_addr_space_pointer_mode): Likewise.
7396 (fl78_unwind_word_mode): Likewise.
7397 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
7399 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
7400 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
7401 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
7402 (mips_valid_pointer_mode): Likewise.
7403 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
7404 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
7405 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
7407 (ft32_addr_space_address_mode): Likewise.
7408 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
7409 scalar_int_mode instead of a machine_mode.
7410 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
7412 (m32c_addr_space_address_mode): Likewise.
7413 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
7414 (rs6000_eh_return_filter_mode): Likewise.
7415 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
7416 (rs6000_eh_return_filter_mode): Likewise.
7417 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
7418 (s390_libgcc_shift_count_mode): Likewise.
7419 (s390_unwind_word_mode): Likewise.
7420 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
7422 * target.def (mode_rep_extended): Likewise.
7423 (valid_pointer_mode): Likewise.
7424 (addr_space.valid_pointer_mode): Likewise.
7425 (eh_return_filter_mode): Return a scalar_int_mode rather than
7427 (libgcc_cmp_return_mode): Likewise.
7428 (libgcc_shift_count_mode): Likewise.
7429 (unwind_word_mode): Likewise.
7430 (addr_space.pointer_mode): Likewise.
7431 (addr_space.address_mode): Likewise.
7432 * doc/tm.texi: Regenerate.
7433 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
7435 (do_jump): Use scalar_int_mode for local variables.
7436 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
7437 rather than a machine_mode.
7438 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
7439 (scompare_loc_descriptor_wide): Likewise.
7440 (scompare_loc_descriptor_narrow): Likewise.
7441 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
7443 * except.c (sjlj_emit_dispatch_table): Likewise.
7444 (expand_builtin_eh_copy_values): Likewise.
7445 * explow.c (convert_memory_address_addr_space_1): Likewise.
7446 Take a scalar_int_mode rather than a machine_mode.
7447 (convert_memory_address_addr_space): Take a scalar_int_mode rather
7448 than a machine_mode.
7449 (memory_address_addr_space): Use scalar_int_mode for local variables.
7450 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
7451 rather than a machine_mode.
7452 * expmed.c (mask_rtx): Likewise.
7453 (init_expmed_one_conv): Likewise.
7454 (expand_mult_highpart_adjust): Likewise.
7455 (extract_high_half): Likewise.
7456 (expmed_mult_highpart_optab): Likewise.
7457 (expmed_mult_highpart): Likewise.
7458 (expand_smod_pow2): Likewise.
7459 (expand_sdiv_pow2): Likewise.
7460 (emit_store_flag_int): Likewise.
7461 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
7463 (extract_low_bits): Likewise.
7464 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
7466 * expr.c (pieces_addr::adjust): Likewise.
7467 (can_store_by_pieces): Likewise.
7468 (store_by_pieces): Likewise.
7469 (clear_by_pieces_1): Likewise.
7470 (expand_expr_addr_expr_1): Likewise.
7471 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
7472 (expand_expr_real_1): Likewise.
7473 (try_casesi): Likewise.
7474 * final.c (shorten_branches): Likewise.
7475 * fold-const.c (fold_convert_const_int_from_fixed): Change the
7476 type of "mode" to machine_mode.
7477 * internal-fn.c (expand_arith_overflow_result_store): Take a
7478 scalar_int_mode rather than a machine_mode.
7479 (expand_mul_overflow): Use scalar_int_mode for local variables.
7480 * loop-doloop.c (doloop_modify): Likewise.
7481 (doloop_optimize): Likewise.
7482 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
7483 than a machine_mode.
7484 (expand_doubleword_shift_condmove): Likewise.
7485 (expand_doubleword_shift): Likewise.
7486 (expand_doubleword_clz): Likewise.
7487 (expand_doubleword_popcount): Likewise.
7488 (expand_doubleword_parity): Likewise.
7489 (expand_absneg_bit): Use scalar_int_mode for local variables.
7490 (prepare_float_lib_cmp): Likewise.
7491 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
7492 rather than a machine_mode.
7493 (convert_memory_address_addr_space): Likewise.
7494 (get_mode_bounds): Likewise.
7495 (get_address_mode): Return a scalar_int_mode rather than a
7497 * rtlanal.c (get_address_mode): Likewise.
7498 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
7499 than a machine_mode.
7500 * targhooks.c (default_mode_rep_extended): Likewise.
7501 (default_valid_pointer_mode): Likewise.
7502 (default_addr_space_valid_pointer_mode): Likewise.
7503 (default_eh_return_filter_mode): Return a scalar_int_mode rather
7504 than a machine_mode.
7505 (default_libgcc_cmp_return_mode): Likewise.
7506 (default_libgcc_shift_count_mode): Likewise.
7507 (default_unwind_word_mode): Likewise.
7508 (default_addr_space_pointer_mode): Likewise.
7509 (default_addr_space_address_mode): Likewise.
7510 * targhooks.h (default_eh_return_filter_mode): Likewise.
7511 (default_libgcc_cmp_return_mode): Likewise.
7512 (default_libgcc_shift_count_mode): Likewise.
7513 (default_unwind_word_mode): Likewise.
7514 (default_addr_space_pointer_mode): Likewise.
7515 (default_addr_space_address_mode): Likewise.
7516 (default_mode_rep_extended): Take a scalar_int_mode rather than
7518 (default_valid_pointer_mode): Likewise.
7519 (default_addr_space_valid_pointer_mode): Likewise.
7520 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
7522 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
7523 rather than a machine_mode.
7524 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
7525 for local variables.
7526 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7527 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
7528 than a machine_mode.
7530 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7531 Alan Hayward <alan.hayward@arm.com>
7532 David Sherwood <david.sherwood@arm.com>
7534 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
7535 the mode argument to scalar_int_mode.
7536 (do_jump_by_parts_zero_rtx): Likewise.
7537 (do_jump_by_parts_equality_rtx): Likewise.
7538 (do_jump_by_parts_greater): Take a mode argument.
7539 (do_jump_by_parts_equality): Likewise.
7540 (do_jump_1): Update calls accordingly.
7542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7543 Alan Hayward <alan.hayward@arm.com>
7544 David Sherwood <david.sherwood@arm.com>
7546 * is-a.h (safe_dyn_cast): New function.
7547 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
7548 (jump_table_for_label): Likewise.
7549 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
7550 instead of an rtx_insn *.
7551 (shorten_branches): Use dyn_cast instead of LABEL_P and
7552 JUMP_TABLE_DATA_P. Use jump_table_for_label and
7553 rtx_jump_table_data::get_data_mode.
7554 (final_scan_insn): Likewise.
7556 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7557 Alan Hayward <alan.hayward@arm.com>
7558 David Sherwood <david.sherwood@arm.com>
7560 * combine.c (try_combine): Use is_a <scalar_int_mode> when
7561 trying to combine a full-register integer set with a subreg
7564 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7565 Alan Hayward <alan.hayward@arm.com>
7566 David Sherwood <david.sherwood@arm.com>
7568 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
7569 that is always either address_mode or pointer_mode.
7571 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7572 Alan Hayward <alan.hayward@arm.com>
7573 David Sherwood <david.sherwood@arm.com>
7575 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
7576 when the two are known to be equal.
7578 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7579 Alan Hayward <alan.hayward@arm.com>
7580 David Sherwood <david.sherwood@arm.com>
7582 * simplify-rtx.c (simplify_const_unary_operation): Use
7583 is_a <scalar_int_mode> instead of checking for a nonzero
7584 precision. Forcibly convert op_mode to a scalar_int_mode
7585 in that case. More clearly differentiate the operand and
7586 result modes and use the former when deciding what the value
7587 of a count-bits operation should be. Use is_int_mode instead
7588 of checking for a MODE_INT. Remove redundant check for whether
7589 this mode has a zero precision.
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 * optabs.c (widen_leading): Change the type of the mode argument
7596 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
7597 (widen_bswap): Likewise.
7598 (expand_parity): Likewise.
7599 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
7600 (expand_ffs): Likewise.
7601 (epand_unop): Check for scalar integer modes before calling the
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 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
7610 (expand_expr_real_1): Update call accordingly.
7612 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7613 Alan Hayward <alan.hayward@arm.com>
7614 David Sherwood <david.sherwood@arm.com>
7616 * expmed.c (store_bit_field_using_insv): Add op0_mode and
7617 value_mode arguments. Use scalar_int_mode internally.
7618 (store_bit_field_1): Rename the new integer mode from imode
7619 to op0_mode and use it instead of GET_MODE (op0). Update calls
7620 to store_split_bit_field, store_bit_field_using_insv and
7621 store_fixed_bit_field.
7622 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
7623 Use scalar_int_mode internally. Use a bit count rather than a mode
7624 when calculating the largest bit size for get_best_mode.
7625 Update calls to store_split_bit_field and store_fixed_bit_field_1.
7626 (store_fixed_bit_field_1): Add mode and value_mode arguments.
7627 Remove assertion that OP0 has a scalar integer mode.
7628 (store_split_bit_field): Add op0_mode and value_mode arguments.
7629 Update calls to extract_fixed_bit_field.
7630 (extract_bit_field_using_extv): Add an op0_mode argument.
7631 Use scalar_int_mode internally.
7632 (extract_bit_field_1): Rename the new integer mode from imode to
7633 op0_mode and use it instead of GET_MODE (op0). Update calls to
7634 extract_split_bit_field, extract_bit_field_using_extv and
7635 extract_fixed_bit_field.
7636 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
7637 to extract_split_bit_field and extract_fixed_bit_field_1.
7638 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
7639 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
7641 (extract_split_bit_field): Add an op0_mode argument. Update call
7642 to extract_fixed_bit_field.
7644 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7645 Alan Hayward <alan.hayward@arm.com>
7646 David Sherwood <david.sherwood@arm.com>
7648 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
7649 * explow.c (hard_function_value): Likewise.
7650 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
7651 convert_to_mode call outside the loop.
7652 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
7653 for the mode iterator. Require the mode specified by max_pieces
7655 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
7657 (copy_blkmode_to_reg): Likewise.
7658 (set_storage_via_setmem): Likewise.
7659 * optabs.c (prepare_cmp_insn): Likewise.
7660 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7661 * stor-layout.c (finish_bitfield_representative): Likewise.
7663 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7664 Alan Hayward <alan.hayward@arm.com>
7665 David Sherwood <david.sherwood@arm.com>
7667 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
7668 * expr.c (convert_move): Use them.
7669 (convert_modes): Likewise.
7670 (store_expr_with_bounds): Likewise.
7672 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7673 Alan Hayward <alan.hayward@arm.com>
7674 David Sherwood <david.sherwood@arm.com>
7676 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
7677 parameter for the mode of "x". Remove the "known_x", "known_mode"
7678 and "known_ret" arguments. Change the type of the mode argument
7680 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
7681 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
7682 (reg_num_sign_bit_copies_for_combine): Likewise.
7683 * rtlanal.c (nonzero_bits1): Likewise.
7684 (num_sign_bit_copies1): Likewise.
7685 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
7686 (reg_num_sign_bit_copies_general): Likewise.
7687 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
7688 (reg_nonzero_bits_general): Likewise.
7690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7691 Alan Hayward <alan.hayward@arm.com>
7692 David Sherwood <david.sherwood@arm.com>
7694 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
7695 than in subroutines. Return 1 for non-integer modes.
7696 (cached_num_sign_bit_copies): Change the type of the mode parameter
7698 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
7699 classes. Handle CONST_INT_P first and then check whether X also
7700 has a scalar integer mode. Check the same thing for inner registers
7701 of a SUBREG and for values that are being extended or truncated.
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 (nonzero_bits): Handle VOIDmode here rather than
7708 in subroutines. Return the mode mask for non-integer modes.
7709 (cached_nonzero_bits): Change the type of the mode parameter
7711 (nonzero_bits1): Likewise. Remove early exit for other mode
7712 classes. Handle CONST_INT_P first and then check whether X
7713 also has a scalar integer mode.
7715 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7716 Alan Hayward <alan.hayward@arm.com>
7717 David Sherwood <david.sherwood@arm.com>
7719 * expr.c (widest_int_mode_for_size): Make the comment match the code.
7720 Return a scalar_int_mode and assert that the size is greater than
7722 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
7723 (op_by_pieces_d::op_by_pieces_d): Likewise.
7724 (op_by_pieces_d::run): Likewise.
7725 (can_store_by_pieces): Likewise.
7727 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7728 Alan Hayward <alan.hayward@arm.com>
7729 David Sherwood <david.sherwood@arm.com>
7731 * combine.c (extract_left_shift): Add a mode argument and update
7733 (make_compound_operation_int): Change the type of the mode parameter
7734 to scalar_int_mode and update the call to extract_left_shift.
7736 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7737 Alan Hayward <alan.hayward@arm.com>
7738 David Sherwood <david.sherwood@arm.com>
7740 * combine.c (simplify_and_const_int): Change the type of the mode
7741 parameter to scalar_int_mode.
7742 (simplify_and_const_int_1): Likewise. Update recursive call.
7744 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7745 Alan Hayward <alan.hayward@arm.com>
7746 David Sherwood <david.sherwood@arm.com>
7748 * combine.c (simplify_compare_const): Check that the mode is a
7749 scalar_int_mode (rather than VOIDmode) before testing its
7751 (simplify_comparison): Move COMPARISON_P handling out of the
7752 loop and restrict the latter part of the loop to scalar_int_modes.
7753 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
7754 and when considering SUBREG_REGs. Use is_int_mode instead of
7755 checking GET_MODE_CLASS against MODE_INT.
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 (try_widen_shift_mode): Move check for equal modes to...
7762 (simplify_shift_const_1): ...here. Use scalar_int_mode for
7763 shift_unit_mode and for modes involved in scalar shifts.
7765 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7766 Alan Hayward <alan.hayward@arm.com>
7767 David Sherwood <david.sherwood@arm.com>
7769 * combine.c (force_int_to_mode): New function, split out from...
7770 (force_to_mode): ...here. Keep xmode up-to-date and use it
7771 instead of GET_MODE (x).
7773 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7774 Alan Hayward <alan.hayward@arm.com>
7775 David Sherwood <david.sherwood@arm.com>
7777 * optabs-query.h (extraction_insn::struct_mode): Change type to
7778 opt_scalar_int_mode and update comment.
7779 (extraction_insn::field_mode): Change type to scalar_int_mode.
7780 (extraction_insn::pos_mode): Likewise.
7781 * combine.c (make_extraction): Update accordingly.
7782 * optabs-query.c (get_traditional_extraction_insn): Likewise.
7783 (get_optab_extraction_insn): Likewise.
7784 * recog.c (simplify_while_replacing): Likewise.
7785 * expmed.c (narrow_bit_field_mem): Change the type of the mode
7786 parameter to opt_scalar_int_mode.
7788 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7789 Alan Hayward <alan.hayward@arm.com>
7790 David Sherwood <david.sherwood@arm.com>
7792 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
7793 to a scalar_int_mode instead of a machine_mode.
7794 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
7795 (get_best_mode): Return a boolean and use a pointer argument to store
7796 the selected mode. Replace the limit mode parameter with a bit limit.
7797 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
7798 for the values returned by bit_field_mode_iterator::next_mode.
7799 (store_bit_field): Update call to get_best_mode.
7800 (store_fixed_bit_field): Likewise.
7801 (extract_fixed_bit_field): Likewise.
7802 * expr.c (optimize_bitfield_assignment_op): Likewise.
7803 * fold-const.c (optimize_bit_field_compare): Likewise.
7804 (fold_truth_andor_1): Likewise.
7805 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
7806 Update for new type of m_mode.
7807 (get_best_mode): As above.
7809 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7810 Alan Hayward <alan.hayward@arm.com>
7811 David Sherwood <david.sherwood@arm.com>
7813 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
7814 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
7815 (store_bit_field): Check is_a <scalar_int_mode> before calling
7816 strict_volatile_bitfield_p.
7817 (extract_bit_field): Likewise.
7819 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7820 Alan Hayward <alan.hayward@arm.com>
7821 David Sherwood <david.sherwood@arm.com>
7823 * target.def (cstore_mode): Return a scalar_int_mode.
7824 * doc/tm.texi: Regenerate.
7825 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
7826 * targhooks.h (default_cstore_mode): Likewise.
7827 * targhooks.c (default_cstore_mode): Likewise, using a forced
7829 * expmed.c (emit_cstore): Expect the target of the cstore to be
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 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
7838 (niter_desc): Likewise mode.
7839 (iv_analyze): Add a mode parameter.
7841 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
7842 and change its type to scalar_int_mode.
7843 * loop-iv.c: Update commentary at head of file.
7844 (iv_constant): Pass the mode paraeter before the rtx it describes
7845 and change its type to scalar_int_mode. Remove VOIDmode handling.
7846 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
7847 (iv_extend): Likewise.
7848 (shorten_into_mode): Likewise.
7849 (iv_add): Use scalar_int_mode.
7850 (iv_mult): Likewise.
7851 (iv_shift): Likewise.
7852 (canonicalize_iv_subregs): Likewise.
7853 (get_biv_step_1): Pass the outer_mode parameter before the rtx
7854 it describes and change its mode to scalar_int_mode. Also change
7855 the type of the returned inner_mode to scalar_int_mode.
7856 (get_biv_step): Likewise, turning outer_mode from a pointer
7857 into a direct parameter. Update call to get_biv_step_1.
7858 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
7859 iv_constant and get_biv_step.
7860 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
7861 and change its type to scalar_int_mode. Don't initialise iv->mode
7862 to VOIDmode and remove later checks for its still being VOIDmode.
7863 Update calls to iv_analyze_op and iv_analyze_expr. Check
7864 is_a <scalar_int_mode> when changing the mode under consideration.
7865 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
7866 Update call to iv_analyze_expr.
7867 (iv_analyze_op): Add a mode parameter. Reject subregs whose
7868 inner register is not also a scalar_int_mode. Update call to
7870 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
7871 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
7872 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
7873 separate mode class checks. Update calls to iv_analyze. Remove
7874 fix-up of VOIDmodes after iv_analyze_biv.
7875 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
7876 don't have a scalar_int_mode. Update call to biv_p.
7878 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7879 Alan Hayward <alan.hayward@arm.com>
7880 David Sherwood <david.sherwood@arm.com>
7882 * cfgexpand.c (convert_debug_memory_address): Use
7883 as_a <scalar_int_mode>.
7884 * combine.c (expand_compound_operation): Likewise.
7885 (make_extraction): Likewise.
7886 (change_zero_ext): Likewise.
7887 (simplify_comparison): Likewise.
7888 * cse.c (cse_insn): Likewise.
7889 * dwarf2out.c (minmax_loc_descriptor): Likewise.
7890 (mem_loc_descriptor): Likewise.
7891 (loc_descriptor): Likewise.
7892 * expmed.c (init_expmed_one_mode): Likewise.
7893 (synth_mult): Likewise.
7894 (emit_store_flag_1): Likewise.
7895 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
7896 of a comparison with size.
7897 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
7898 (reduce_to_bit_field_precision): Likewise.
7899 * function.c (expand_function_end): Likewise.
7900 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
7901 * loop-doloop.c (doloop_modify): Likewise.
7902 * optabs.c (expand_binop): Likewise.
7903 (expand_unop): Likewise.
7904 (expand_copysign_absneg): Likewise.
7905 (prepare_cmp_insn): Likewise.
7906 (maybe_legitimize_operand): Likewise.
7907 * recog.c (const_scalar_int_operand): Likewise.
7908 * rtlanal.c (get_address_mode): Likewise.
7909 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7910 (simplify_cond_clz_ctz): Likewise.
7911 * tree-nested.c (get_nl_goto_field): Likewise.
7912 * tree.c (build_vector_type_for_mode): Likewise.
7913 * var-tracking.c (use_narrower_mode): Likewise.
7915 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7916 Alan Hayward <alan.hayward@arm.com>
7917 David Sherwood <david.sherwood@arm.com>
7919 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
7920 * builtins.c (expand_builtin_signbit): Use it.
7921 * cfgexpand.c (expand_debug_expr): Likewise.
7922 * dojump.c (do_jump): Likewise.
7923 (do_compare_and_jump): Likewise.
7924 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
7925 * expmed.c (make_tree): Likewise.
7926 * expr.c (expand_expr_real_2): Likewise.
7927 (expand_expr_real_1): Likewise.
7928 (try_casesi): Likewise.
7929 * fold-const-call.c (fold_const_call_ss): Likewise.
7930 * fold-const.c (unextend): Likewise.
7931 (extract_muldiv_1): Likewise.
7932 (fold_single_bit_test): Likewise.
7933 (native_encode_int): Likewise.
7934 (native_encode_string): Likewise.
7935 (native_interpret_int): Likewise.
7936 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
7937 * internal-fn.c (expand_addsub_overflow): Likewise.
7938 (expand_neg_overflow): Likewise.
7939 (expand_mul_overflow): Likewise.
7940 (expand_arith_overflow): Likewise.
7941 * match.pd: Likewise.
7942 * stor-layout.c (layout_type): Likewise.
7943 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
7944 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
7945 * tree-ssanames.c (get_range_info): Likewise.
7946 * tree-switch-conversion.c (array_value_type) Likewise.
7947 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
7948 (vect_recog_divmod_pattern): Likewise.
7949 (vect_recog_mixed_size_cond_pattern): Likewise.
7950 * tree-vrp.c (extract_range_basic): Likewise.
7951 (simplify_float_conversion_using_ranges): Likewise.
7952 * tree.c (int_fits_type_p): Likewise.
7953 * ubsan.c (instrument_bool_enum_load): Likewise.
7954 * varasm.c (mergeable_string_section): Likewise.
7955 (narrowing_initializer_constant_valid_p): Likewise.
7956 (output_constant): Likewise.
7958 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7959 Alan Hayward <alan.hayward@arm.com>
7960 David Sherwood <david.sherwood@arm.com>
7962 * machmode.h (NARROWEST_INT_MODE): New macro.
7963 * expr.c (alignment_for_piecewise_move): Use it instead of
7964 GET_CLASS_NARROWEST_MODE (MODE_INT).
7965 (push_block): Likewise.
7966 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
7968 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7970 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7971 Alan Hayward <alan.hayward@arm.com>
7972 David Sherwood <david.sherwood@arm.com>
7974 * postreload.c (move2add_valid_value_p): Change the type of the
7975 mode parameter to scalar_int_mode.
7976 (move2add_use_add2_insn): Add a mode parameter and use it instead
7978 (move2add_use_add3_insn): Likewise.
7979 (reload_cse_move2add): Update accordingly.
7981 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7982 Alan Hayward <alan.hayward@arm.com>
7983 David Sherwood <david.sherwood@arm.com>
7985 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
7987 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
7988 * optabs.c (expand_unop): Likewise.
7990 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7991 Alan Hayward <alan.hayward@arm.com>
7992 David Sherwood <david.sherwood@arm.com>
7994 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
7995 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
7996 (popcount_loc_descriptor): Likewise.
7997 (bswap_loc_descriptor): Likewise.
7998 (rotate_loc_descriptor): Likewise.
7999 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
8000 calling the functions above.
8002 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8003 Alan Hayward <alan.hayward@arm.com>
8004 David Sherwood <david.sherwood@arm.com>
8006 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
8008 (try_combine): Likewise.
8009 (simplify_if_then_else): Likewise.
8010 * cse.c (cse_insn): Likewise.
8011 * dwarf2out.c (mem_loc_descriptor): Likewise.
8012 * emit-rtl.c (gen_lowpart_common): Likewise.
8013 * simplify-rtx.c (simplify_truncation): Likewise.
8014 (simplify_binary_operation_1): Likewise.
8015 (simplify_const_relational_operation): Likewise.
8016 (simplify_ternary_operation): Likewise.
8017 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
8019 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8020 Alan Hayward <alan.hayward@arm.com>
8021 David Sherwood <david.sherwood@arm.com>
8023 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
8024 * reload.c (push_reload): Likewise.
8025 (find_reloads): Likewise.
8027 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8028 Alan Hayward <alan.hayward@arm.com>
8029 David Sherwood <david.sherwood@arm.com>
8031 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
8032 (make_compound_operation_int): Likewise.
8033 (change_zero_ext): Likewise.
8034 * expr.c (convert_move): Likewise.
8035 (convert_modes): Likewise.
8036 * fwprop.c (forward_propagate_subreg): Likewise.
8037 * loop-iv.c (get_biv_step_1): Likewise.
8038 * optabs.c (widen_operand): Likewise.
8039 * postreload.c (move2add_valid_value_p): Likewise.
8040 * recog.c (simplify_while_replacing): Likewise.
8041 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8042 (simplify_binary_operation_1): Likewise. Remove redundant
8043 mode equality check.
8045 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8046 Alan Hayward <alan.hayward@arm.com>
8047 David Sherwood <david.sherwood@arm.com>
8049 * combine.c (combine_simplify_rtx): Add checks for
8050 is_a <scalar_int_mode>.
8051 (simplify_if_then_else): Likewise.
8052 (make_field_assignment): Likewise.
8053 (simplify_comparison): Likewise.
8054 * ifcvt.c (noce_try_bitop): Likewise.
8055 * loop-invariant.c (canonicalize_address_mult): Likewise.
8056 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
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 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
8063 is_a <scalar_int_mode> instead of != BLKmode.
8065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8066 Alan Hayward <alan.hayward@arm.com>
8067 David Sherwood <david.sherwood@arm.com>
8069 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
8070 instead of != VOIDmode.
8071 * combine.c (if_then_else_cond): Likewise.
8072 (change_zero_ext): Likewise.
8073 * dwarf2out.c (mem_loc_descriptor): Likewise.
8074 (loc_descriptor): Likewise.
8075 * rtlanal.c (canonicalize_condition): Likewise.
8076 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
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 * simplify-rtx.c (simplify_binary_operation_1): Use
8083 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
8085 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8086 Alan Hayward <alan.hayward@arm.com>
8087 David Sherwood <david.sherwood@arm.com>
8089 * wide-int.h (int_traits<unsigned char>) New class.
8090 (int_traits<unsigned short>) Likewise.
8091 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
8092 Use GET_MODE_UNIT_PRECISION and remove redundant test for
8094 * combine.c (set_nonzero_bits_and_sign_copies): Use
8095 is_a <scalar_int_mode>.
8096 (find_split_point): Likewise.
8097 (combine_simplify_rtx): Likewise.
8098 (simplify_logical): Likewise.
8099 (expand_compound_operation): Likewise.
8100 (expand_field_assignment): Likewise.
8101 (make_compound_operation): Likewise.
8102 (extended_count): Likewise.
8103 (change_zero_ext): Likewise.
8104 (simplify_comparison): Likewise.
8105 * dwarf2out.c (scompare_loc_descriptor): Likewise.
8106 (ucompare_loc_descriptor): Likewise.
8107 (minmax_loc_descriptor): Likewise.
8108 (mem_loc_descriptor): Likewise.
8109 (loc_descriptor): Likewise.
8110 * expmed.c (init_expmed_one_mode): Likewise.
8111 * lra-constraints.c (lra_constraint_offset): Likewise.
8112 * optabs.c (prepare_libcall_arg): Likewise.
8113 * postreload.c (move2add_note_store): Likewise.
8114 * reload.c (operands_match_p): Likewise.
8115 * rtl.h (load_extend_op): Likewise.
8116 * rtlhooks.c (gen_lowpart_general): Likewise.
8117 * simplify-rtx.c (simplify_truncation): Likewise.
8118 (simplify_unary_operation_1): Likewise.
8119 (simplify_binary_operation_1): Likewise.
8120 (simplify_const_binary_operation): Likewise.
8121 (simplify_const_relational_operation): Likewise.
8122 (simplify_subreg): Likewise.
8123 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8124 * var-tracking.c (adjust_mems): Likewise.
8125 (prepare_call_arguments): Likewise.
8127 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8128 Alan Hayward <alan.hayward@arm.com>
8129 David Sherwood <david.sherwood@arm.com>
8131 * machmode.h (is_int_mode): New fuction.
8132 * combine.c (find_split_point): Use it.
8133 (combine_simplify_rtx): Likewise.
8134 (simplify_if_then_else): Likewise.
8135 (simplify_set): Likewise.
8136 (simplify_shift_const_1): Likewise.
8137 (simplify_comparison): Likewise.
8138 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
8139 * cse.c (notreg_cost): Likewise.
8140 (cse_insn): Likewise.
8141 * cselib.c (cselib_lookup_1): Likewise.
8142 * dojump.c (do_jump_1): Likewise.
8143 (do_compare_rtx_and_jump): Likewise.
8144 * dse.c (get_call_args): Likewise.
8145 * dwarf2out.c (rtl_for_decl_init): Likewise.
8146 (native_encode_initializer): Likewise.
8147 * expmed.c (emit_store_flag_1): Likewise.
8148 (emit_store_flag): Likewise.
8149 * expr.c (convert_modes): Likewise.
8150 (store_field): Likewise.
8151 (expand_expr_real_1): Likewise.
8152 * fold-const.c (fold_read_from_constant_string): Likewise.
8153 * gimple-ssa-sprintf.c (get_format_string): Likewise.
8154 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
8155 * optabs.c (expand_binop): Likewise.
8156 (expand_unop): Likewise.
8157 (expand_abs_nojump): Likewise.
8158 (expand_one_cmpl_abs_nojump): Likewise.
8159 * simplify-rtx.c (mode_signbit_p): Likewise.
8160 (val_signbit_p): Likewise.
8161 (val_signbit_known_set_p): Likewise.
8162 (val_signbit_known_clear_p): Likewise.
8163 (simplify_relational_operation_1): Likewise.
8164 * tree.c (vector_type_mode): Likewise.
8166 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8167 Alan Hayward <alan.hayward@arm.com>
8168 David Sherwood <david.sherwood@arm.com>
8170 * machmode.h (smallest_mode_for_size): Fix formatting.
8171 (smallest_int_mode_for_size): New function.
8172 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
8173 instead of smallest_mode_for_size.
8174 * combine.c (make_extraction): Likewise.
8175 * config/arc/arc.c (arc_expand_movmem): Likewise.
8176 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
8177 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8178 * config/s390/s390.c (s390_expand_insv): Likewise.
8179 * config/sparc/sparc.c (assign_int_registers): Likewise.
8180 * config/spu/spu.c (spu_function_value): Likewise.
8181 (spu_function_arg): Likewise.
8182 * coverage.c (get_gcov_type): Likewise.
8183 (get_gcov_unsigned_t): Likewise.
8184 * dse.c (find_shift_sequence): Likewise.
8185 * expmed.c (store_bit_field_1): Likewise.
8186 * expr.c (convert_move): Likewise.
8187 (store_field): Likewise.
8188 * internal-fn.c (expand_arith_overflow): Likewise.
8189 * optabs-query.c (get_best_extraction_insn): Likewise.
8190 * optabs.c (expand_twoval_binop_libfunc): Likewise.
8191 * stor-layout.c (layout_type): Likewise.
8192 (initialize_sizetypes): Likewise.
8193 * targhooks.c (default_get_mask_mode): Likewise.
8194 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
8196 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8197 Alan Hayward <alan.hayward@arm.com>
8198 David Sherwood <david.sherwood@arm.com>
8200 * machmode.h (opt_mode::else_blk): New function.
8201 (int_mode_for_mode): Declare.
8202 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
8203 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
8205 * cfgexpand.c (expand_debug_expr): Likewise.
8206 * combine.c (gen_lowpart_or_truncate): Likewise.
8207 (gen_lowpart_for_combine): Likewise.
8208 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
8209 * config/avr/avr.c (avr_to_int_mode): Likewise.
8210 (avr_out_plus_1): Likewise.
8211 (avr_out_plus): Likewise.
8212 (avr_out_round): Likewise.
8213 * config/i386/i386.c (ix86_split_to_parts): Likewise.
8214 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8215 (s390_expand_vcond): Likewise.
8216 * config/spu/spu.c (spu_split_immediate): Likewise.
8217 (spu_expand_mov): Likewise.
8218 * dse.c (get_stored_val): Likewise.
8219 * expmed.c (store_bit_field_1): Likewise.
8220 (convert_extracted_bit_field): Use int_mode_for_mode instead of
8222 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
8223 (extract_low_bits): Likewise.
8224 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
8225 handling rather than repeating the check.
8226 (emit_group_store): Likewise.
8227 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
8228 * optabs.c (expand_absneg_bit): Likewise.
8229 (expand_copysign_absneg): Likewise.
8230 (expand_copysign_bit): Likewise.
8231 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
8232 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
8233 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
8234 * var-tracking.c (prepare_call_arguments): Likewise.
8235 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
8236 int_mode_for_mode instead of mode_for_size.
8237 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8239 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8240 Alan Hayward <alan.hayward@arm.com>
8241 David Sherwood <david.sherwood@arm.com>
8243 * machmode.h (int_mode_for_size): New function.
8244 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
8245 instead of mode_for_size.
8246 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
8248 * combine.c (expand_field_assignment): Use int_mode_for_size
8249 instead of mode_for_size.
8250 (make_extraction): Likewise.
8251 (simplify_shift_const_1): Likewise.
8252 (simplify_comparison): Likewise.
8253 * dojump.c (do_jump): Likewise.
8254 * dwarf2out.c (mem_loc_descriptor): Likewise.
8255 * emit-rtl.c (init_derived_machine_modes): Likewise.
8256 * expmed.c (flip_storage_order): Likewise.
8257 (convert_extracted_bit_field): Likewise.
8258 * expr.c (copy_blkmode_from_reg): Likewise.
8259 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
8260 * internal-fn.c (expand_mul_overflow): Likewise.
8261 * lower-subreg.c (simple_move): Likewise.
8262 * optabs-libfuncs.c (init_optabs): Likewise.
8263 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8264 * tree.c (vector_type_mode): Likewise.
8265 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
8266 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
8267 * tree-vect-generic.c (expand_vector_parallel): Likewise.
8268 * tree-vect-stmts.c (vectorizable_load): Likewise.
8269 (vectorizable_store): Likewise.
8271 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8272 Alan Hayward <alan.hayward@arm.com>
8273 David Sherwood <david.sherwood@arm.com>
8275 * coretypes.h (pod_mode): New type.
8276 (scalar_int_mode_pod): New typedef.
8277 * machmode.h (pod_mode): New class.
8278 (int_n_data_t::m): Change type to scalar_int_mode_pod.
8279 * genmodes.c (emit_mode_int_n): Update accordingly.
8280 * lower-subreg.h (target_lower_subreg): Change type to
8281 scalar_int_mode_pod.
8282 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
8283 scalar_int_mode_pod.
8285 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8286 Alan Hayward <alan.hayward@arm.com>
8287 David Sherwood <david.sherwood@arm.com>
8289 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
8290 machine_mode to scalar_int_mode.
8291 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
8292 (rs6000_option_override_internal): Remove cast to int.
8293 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
8294 machine_mode to scalar_int_mode.
8295 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
8296 (rs6000_option_override_internal): Remove cast to int.
8297 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
8298 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
8300 * config/s390/s390.c (s390_expand_builtin): Likewise.
8301 * coretypes.h (scalar_int_mode): New type.
8302 (opt_scalar_int_mode): New typedef.
8303 * machmode.h (scalar_int_mode): New class.
8304 (scalar_int_mode::includes_p): New function.
8305 (byte_mode): Change type to scalar_int_mode.
8306 (word_mode): Likewise.
8307 (ptr_mode): Likewise.
8308 * emit-rtl.c (byte_mode): Likewise.
8309 (word_mode): Likewise.
8310 (ptr_mode): Likewise.
8311 (init_derived_machine_modes): Update accordingly.
8312 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
8313 and MODE_PARTIAL_INT.
8314 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
8315 opt_scalar_int_mode.
8317 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8318 Alan Hayward <alan.hayward@arm.com>
8319 David Sherwood <david.sherwood@arm.com>
8321 * target.def (libgcc_floating_mode_supported_p): Take a
8323 * doc/tm.texi: Regenerate.
8324 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
8326 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
8327 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
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 (default_floatn_mode): Return an opt_scalar_float_mode.
8335 * doc/tm.texi: Regenerate.
8336 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
8337 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
8338 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
8339 * targhooks.h (default_floatn_mode): Likewise.
8340 * targhooks.c (default_floatn_mode): Likewise.
8341 * tree.c (build_common_tree_nodes): Update accordingly.
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 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
8348 (mode_iterator::iterate_p): Likewise.
8349 (mode_iterator::get_wider): Likewise.
8350 * expr.c (init_expr_target): Use opt_scalar_float_mode.
8352 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8353 Alan Hayward <alan.hayward@arm.com>
8354 David Sherwood <david.sherwood@arm.com>
8356 * coretypes.h (opt_scalar_float_mode): New typedef.
8357 * machmode.h (float_mode_for_size): New function.
8358 * emit-rtl.c (double_mode): Delete.
8359 (init_emit_once): Use float_mode_for_size.
8360 * stor-layout.c (layout_type): Likewise.
8361 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
8363 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8364 Alan Hayward <alan.hayward@arm.com>
8365 David Sherwood <david.sherwood@arm.com>
8367 * output.h (assemble_real): Take a scalar_float_mode.
8368 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
8369 * config/arm/arm.md (consttable_4): Likewise.
8370 (consttable_8): Likewise.
8371 (consttable_16): Likewise.
8372 * config/mips/mips.md (consttable_float): Likewise.
8373 * config/s390/s390.c (s390_output_pool_entry): Likewise.
8374 * varasm.c (assemble_real): Take a scalar_float_mode.
8375 (output_constant_pool_2): Update accordingly.
8376 (output_constant): Likewise.
8378 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8379 Alan Hayward <alan.hayward@arm.com>
8380 David Sherwood <david.sherwood@arm.com>
8382 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
8383 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
8384 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
8385 (native_encode_real): Likewise.
8386 (native_interpret_real): Likewise.
8387 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
8388 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8390 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8391 Alan Hayward <alan.hayward@arm.com>
8392 David Sherwood <david.sherwood@arm.com>
8394 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
8395 <scalar_float_mode>. Simplify.
8396 (gen_extend_conv_libfunc): Likewise.
8398 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8399 Alan Hayward <alan.hayward@arm.com>
8400 David Sherwood <david.sherwood@arm.com>
8402 * coretypes.h (scalar_float_mode): New type.
8403 * machmode.h (mode_traits::from_int): Use machine_mode if
8404 USE_ENUM_MODES is defined.
8405 (is_a): New function.
8407 (dyn_cast): Likewise.
8408 (scalar_float_mode): New class.
8409 (scalar_float_mode::includes_p): New function.
8410 (is_float_mode): Likewise.
8411 * gdbhooks.py (MachineModePrinter): New class.
8412 (build_pretty_printer): Use it for scalar_float_mode.
8413 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
8414 (format_helper::format_helper): Turn into a template.
8415 * genmodes.c (get_mode_class): New function.
8416 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
8417 or machine_mode if none.
8418 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
8419 as_a <scalar_float_mode>.
8420 * dwarf2out.c (mem_loc_descriptor): Likewise.
8421 (insert_float): Likewise.
8422 (add_const_value_attribute): Likewise.
8423 * simplify-rtx.c (simplify_immed_subreg): Likewise.
8424 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
8425 (expand_unop): Update accordingly.
8426 (expand_abs_nojump): Likewise.
8427 (expand_copysign_absneg): Take a scalar_float_mode.
8428 (expand_copysign_bit): Likewise.
8429 (expand_copysign): Update accordingly.
8431 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8432 Alan Hayward <alan.hayward@arm.com>
8433 David Sherwood <david.sherwood@arm.com>
8435 * coretypes.h (opt_mode): New class.
8436 * machmode.h (opt_mode): Likewise.
8437 (opt_mode::else_void): New function.
8438 (opt_mode::require): Likewise.
8439 (opt_mode::exists): Likewise.
8440 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
8441 (GET_MODE_2XWIDER_MODE): Likewise.
8442 (mode_iterator::get_wider): Update accordingly.
8443 (mode_iterator::get_2xwider): Likewise.
8444 (mode_iterator::get_known_wider): Likewise, turning into a template.
8445 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
8446 forcing a wider mode to exist.
8447 * config/cr16/cr16.h (LONG_REG_P): Likewise.
8448 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8449 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
8450 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
8451 * lower-subreg.c (init_lower_subreg): Likewise.
8452 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
8453 on the final iteration.
8454 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
8455 a wider mode exists before asking for a move pattern.
8456 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
8457 forcing a wider mode to exist.
8458 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
8459 returning false if no such mode exists.
8460 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
8461 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
8462 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
8463 Avoid checking for a MODE_INT if we already know the mode is not a
8465 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
8466 forcing a wider mode to exist.
8467 (expmed_mult_highpart_optab): Likewise.
8468 (expmed_mult_highpart): Likewise.
8469 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
8471 * lto-streamer-in.c (lto_input_mode_table): Likewise.
8472 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
8473 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8474 * internal-fn.c (expand_mul_overflow): Update use of
8475 GET_MODE_2XWIDER_MODE.
8476 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8477 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
8478 GET_MODE_WIDER_MODE.
8479 (convert_plusminus_to_widen): Likewise.
8480 * tree-switch-conversion.c (array_value_type): Likewise.
8481 * var-tracking.c (emit_note_insn_var_location): Likewise.
8482 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8483 Return false inside rather than outside the loop if no wider mode
8485 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
8486 and GET_MODE_2XWIDER_MODE
8487 (can_compare_p): Use else_void.
8488 * gdbhooks.py (OptMachineModePrinter): New class.
8489 (build_pretty_printer): Use it for opt_mode.
8491 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8492 Alan Hayward <alan.hayward@arm.com>
8493 David Sherwood <david.sherwood@arm.com>
8495 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
8496 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
8498 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8499 Alan Hayward <alan.hayward@arm.com>
8500 David Sherwood <david.sherwood@arm.com>
8502 * machmode.h (mode_traits): New structure.
8503 (get_narrowest_mode): New function.
8504 (mode_iterator::start): Likewise.
8505 (mode_iterator::iterate_p): Likewise.
8506 (mode_iterator::get_wider): Likewise.
8507 (mode_iterator::get_known_wider): Likewise.
8508 (mode_iterator::get_2xwider): Likewise.
8509 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
8510 (FOR_EACH_MODE): Likewise.
8511 (FOR_EACH_MODE_FROM): Likewise.
8512 (FOR_EACH_MODE_UNTIL): Likewise.
8513 (FOR_EACH_WIDER_MODE): Likewise.
8514 (FOR_EACH_2XWIDER_MODE): Likewise.
8515 * builtins.c (expand_builtin_strlen): Use new mode iterators.
8516 * combine.c (simplify_comparison): Likewise
8517 * config/i386/i386.c (type_natural_mode): Likewise.
8518 * cse.c (cse_insn): Likewise.
8519 * dse.c (find_shift_sequence): Likewise.
8520 * emit-rtl.c (init_derived_machine_modes): Likewise.
8521 (init_emit_once): Likewise.
8522 * explow.c (hard_function_value): Likewise.
8523 * expmed.c (extract_fixed_bit_field_1): Likewise.
8524 (extract_bit_field_1): Likewise.
8525 (expand_divmod): Likewise.
8526 (emit_store_flag_1): Likewise.
8527 * expr.c (init_expr_target): Likewise.
8528 (convert_move): Likewise.
8529 (alignment_for_piecewise_move): Likewise.
8530 (widest_int_mode_for_size): Likewise.
8531 (emit_block_move_via_movmem): Likewise.
8532 (copy_blkmode_to_reg): Likewise.
8533 (set_storage_via_setmem): Likewise.
8534 (compress_float_constant): Likewise.
8535 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8536 * optabs-query.c (get_best_extraction_insn): Likewise.
8537 * optabs.c (expand_binop): Likewise.
8538 (expand_twoval_unop): Likewise.
8539 (expand_twoval_binop): Likewise.
8540 (widen_leading): Likewise.
8541 (widen_bswap): Likewise.
8542 (expand_parity): Likewise.
8543 (expand_unop): Likewise.
8544 (prepare_cmp_insn): Likewise.
8545 (prepare_float_lib_cmp): Likewise.
8546 (expand_float): Likewise.
8547 (expand_fix): Likewise.
8548 (expand_sfix_optab): Likewise.
8549 * postreload.c (move2add_use_add2_insn): Likewise.
8550 * reg-stack.c (reg_to_stack): Likewise.
8551 * reginfo.c (choose_hard_reg_mode): Likewise.
8552 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8553 * stor-layout.c (mode_for_size): Likewise.
8554 (smallest_mode_for_size): Likewise.
8555 (mode_for_vector): Likewise.
8556 (finish_bitfield_representative): Likewise.
8557 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
8558 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
8559 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8560 * var-tracking.c (prepare_call_arguments): Likewise.
8562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8563 Alan Hayward <alan.hayward@arm.com>
8564 David Sherwood <david.sherwood@arm.com>
8566 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
8567 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
8568 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
8569 * machmode.h (mode_size): Move earlier in file.
8570 (mode_precision): Likewise.
8571 (mode_inner): Likewise.
8572 (mode_nunits): Likewise.
8573 (mode_unit_size): Likewise.
8574 (unit_unit_precision): Likewise.
8575 (mode_wider): Likewise.
8576 (mode_2xwider): Likewise.
8577 (machine_mode): New class.
8578 (mode_to_bytes): New function.
8579 (mode_to_bits): Likewise.
8580 (mode_to_precision): Likewise.
8581 (mode_to_inner): Likewise.
8582 (mode_to_unit_size): Likewise.
8583 (mode_to_unit_precision): Likewise.
8584 (mode_to_nunits): Likewise.
8585 (GET_MODE_SIZE): Use mode_to_bytes.
8586 (GET_MODE_BITSIZE): Use mode_to_bits.
8587 (GET_MODE_PRECISION): Use mode_to_precision.
8588 (GET_MODE_INNER): Use mode_to_inner.
8589 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
8590 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
8591 (GET_MODE_NUNITS): Use mode_to_nunits.
8592 * system.h (ALWAYS_INLINE): New macro.
8593 * config/powerpcspe/powerpcspe-c.c
8594 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
8595 int for arg1_mode and arg2_mode.
8597 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8598 Alan Hayward <alan.hayward@arm.com>
8599 David Sherwood <david.sherwood@arm.com>
8601 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
8602 Prefix mode names with E_ in case statements.
8603 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8604 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
8605 (aarch64_split_simd_move): Likewise.
8606 (aarch64_gen_storewb_pair): Likewise.
8607 (aarch64_gen_loadwb_pair): Likewise.
8608 (aarch64_gen_store_pair): Likewise.
8609 (aarch64_gen_load_pair): Likewise.
8610 (aarch64_get_condition_code_1): Likewise.
8611 (aarch64_constant_pool_reload_icode): Likewise.
8612 (get_rsqrte_type): Likewise.
8613 (get_rsqrts_type): Likewise.
8614 (get_recpe_type): Likewise.
8615 (get_recps_type): Likewise.
8616 (aarch64_gimplify_va_arg_expr): Likewise.
8617 (aarch64_simd_container_mode): Likewise.
8618 (aarch64_emit_load_exclusive): Likewise.
8619 (aarch64_emit_store_exclusive): Likewise.
8620 (aarch64_expand_compare_and_swap): Likewise.
8621 (aarch64_gen_atomic_cas): Likewise.
8622 (aarch64_emit_bic): Likewise.
8623 (aarch64_emit_atomic_swap): Likewise.
8624 (aarch64_emit_atomic_load_op): Likewise.
8625 (aarch64_evpc_trn): Likewise.
8626 (aarch64_evpc_uzp): Likewise.
8627 (aarch64_evpc_zip): Likewise.
8628 (aarch64_evpc_ext): Likewise.
8629 (aarch64_evpc_rev): Likewise.
8630 (aarch64_evpc_dup): Likewise.
8631 (aarch64_gen_ccmp_first): Likewise.
8632 (aarch64_gen_ccmp_next): Likewise.
8633 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
8634 (alpha_emit_xfloating_libcall): Likewise.
8635 (emit_insxl): Likewise.
8636 (alpha_arg_type): Likewise.
8637 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
8638 (arc_preferred_simd_mode): Likewise.
8639 (arc_secondary_reload): Likewise.
8640 (get_arc_condition_code): Likewise.
8641 (arc_print_operand): Likewise.
8642 (arc_legitimate_constant_p): Likewise.
8643 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8644 * config/arc/arc.md (casesi_load): Likewise.
8645 (casesi_compact_jump): Likewise.
8646 * config/arc/predicates.md (proper_comparison_operator): Likewise.
8647 (cc_use_register): Likewise.
8648 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8649 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
8650 (arm_init_iwmmxt_builtins): Likewise.
8651 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
8652 (neon_expand_vector_init): Likewise.
8653 (arm_attr_length_move_neon): Likewise.
8654 (maybe_get_arm_condition_code): Likewise.
8655 (arm_emit_vector_const): Likewise.
8656 (arm_preferred_simd_mode): Likewise.
8657 (arm_output_iwmmxt_tinsr): Likewise.
8658 (thumb1_output_casesi): Likewise.
8659 (thumb2_output_casesi): Likewise.
8660 (arm_emit_load_exclusive): Likewise.
8661 (arm_emit_store_exclusive): Likewise.
8662 (arm_expand_compare_and_swap): Likewise.
8663 (arm_evpc_neon_vuzp): Likewise.
8664 (arm_evpc_neon_vzip): Likewise.
8665 (arm_evpc_neon_vrev): Likewise.
8666 (arm_evpc_neon_vtrn): Likewise.
8667 (arm_evpc_neon_vext): Likewise.
8668 (arm_validize_comparison): Likewise.
8669 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
8670 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
8671 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
8672 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
8673 (c6x_preferred_simd_mode): Likewise.
8674 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
8675 (epiphany_rtx_costs): Likewise.
8676 * config/epiphany/predicates.md (proper_comparison_operator):
8678 * config/frv/frv.c (condexec_memory_operand): Likewise.
8679 (frv_emit_move): Likewise.
8680 (output_move_single): Likewise.
8681 (output_condmove_single): Likewise.
8682 (frv_hard_regno_mode_ok): Likewise.
8683 (frv_matching_accg_mode): Likewise.
8684 * config/h8300/h8300.c (split_adds_subs): Likewise.
8685 (h8300_rtx_costs): Likewise.
8686 (h8300_print_operand): Likewise.
8687 (compute_mov_length): Likewise.
8688 (output_logical_op): Likewise.
8689 (compute_logical_op_length): Likewise.
8690 (compute_logical_op_cc): Likewise.
8691 (h8300_shift_needs_scratch_p): Likewise.
8692 (output_a_shift): Likewise.
8693 (compute_a_shift_length): Likewise.
8694 (compute_a_shift_cc): Likewise.
8695 (expand_a_rotate): Likewise.
8696 (output_a_rotate): Likewise.
8697 * config/i386/i386.c (classify_argument): Likewise.
8698 (function_arg_advance_32): Likewise.
8699 (function_arg_32): Likewise.
8700 (function_arg_64): Likewise.
8701 (function_value_64): Likewise.
8702 (ix86_gimplify_va_arg): Likewise.
8703 (ix86_legitimate_constant_p): Likewise.
8704 (put_condition_code): Likewise.
8705 (split_double_mode): Likewise.
8706 (ix86_avx256_split_vector_move_misalign): Likewise.
8707 (ix86_expand_vector_logical_operator): Likewise.
8708 (ix86_split_idivmod): Likewise.
8709 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
8710 (ix86_build_const_vector): Likewise.
8711 (ix86_build_signbit_mask): Likewise.
8712 (ix86_match_ccmode): Likewise.
8713 (ix86_cc_modes_compatible): Likewise.
8714 (ix86_expand_branch): Likewise.
8715 (ix86_expand_sse_cmp): Likewise.
8716 (ix86_expand_sse_movcc): Likewise.
8717 (ix86_expand_int_sse_cmp): Likewise.
8718 (ix86_expand_vec_perm_vpermi2): Likewise.
8719 (ix86_expand_vec_perm): Likewise.
8720 (ix86_expand_sse_unpack): Likewise.
8721 (ix86_expand_int_addcc): Likewise.
8722 (ix86_split_to_parts): Likewise.
8723 (ix86_vectorize_builtin_gather): Likewise.
8724 (ix86_vectorize_builtin_scatter): Likewise.
8725 (avx_vpermilp_parallel): Likewise.
8726 (inline_memory_move_cost): Likewise.
8727 (ix86_tieable_integer_mode_p): Likewise.
8728 (x86_maybe_negate_const_int): Likewise.
8729 (ix86_expand_vector_init_duplicate): Likewise.
8730 (ix86_expand_vector_init_one_nonzero): Likewise.
8731 (ix86_expand_vector_init_one_var): Likewise.
8732 (ix86_expand_vector_init_concat): Likewise.
8733 (ix86_expand_vector_init_interleave): Likewise.
8734 (ix86_expand_vector_init_general): Likewise.
8735 (ix86_expand_vector_set): Likewise.
8736 (ix86_expand_vector_extract): Likewise.
8737 (emit_reduc_half): Likewise.
8738 (ix86_emit_i387_round): Likewise.
8739 (ix86_mangle_type): Likewise.
8740 (ix86_expand_round_sse4): Likewise.
8741 (expand_vec_perm_blend): Likewise.
8742 (canonicalize_vector_int_perm): Likewise.
8743 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
8744 (expand_vec_perm_1): Likewise.
8745 (expand_vec_perm_interleave3): Likewise.
8746 (expand_vec_perm_even_odd_pack): Likewise.
8747 (expand_vec_perm_even_odd_1): Likewise.
8748 (expand_vec_perm_broadcast_1): Likewise.
8749 (ix86_vectorize_vec_perm_const_ok): Likewise.
8750 (ix86_expand_vecop_qihi): Likewise.
8751 (ix86_expand_mul_widen_hilo): Likewise.
8752 (ix86_expand_sse2_abs): Likewise.
8753 (ix86_expand_pextr): Likewise.
8754 (ix86_expand_pinsr): Likewise.
8755 (ix86_preferred_simd_mode): Likewise.
8756 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
8757 * config/i386/sse.md (*andnot<mode>3): Likewise.
8758 (<mask_codefor><code><mode>3<mask_name>): Likewise.
8759 (*<code><mode>3): Likewise.
8760 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
8761 (ia64_expand_atomic_op): Likewise.
8762 (ia64_arg_type): Likewise.
8763 (ia64_mode_to_int): Likewise.
8764 (ia64_scalar_mode_supported_p): Likewise.
8765 (ia64_vector_mode_supported_p): Likewise.
8766 (expand_vec_perm_broadcast): Likewise.
8767 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
8768 (iq2000_function_arg_advance): Likewise.
8769 (iq2000_function_arg): Likewise.
8770 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
8771 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
8772 (m68k_libcall_value): Likewise.
8773 (m68k_function_value): Likewise.
8774 (sched_attr_op_type): Likewise.
8775 * config/mcore/mcore.c (mcore_output_move): Likewise.
8776 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
8778 (microblaze_function_arg): Likewise.
8779 * config/mips/mips.c (mips16_build_call_stub): Likewise.
8780 (mips_print_operand): Likewise.
8781 (mips_mode_ok_for_mov_fmt_p): Likewise.
8782 (mips_vector_mode_supported_p): Likewise.
8783 (mips_preferred_simd_mode): Likewise.
8784 (mips_expand_vpc_loongson_even_odd): Likewise.
8785 (mips_expand_vec_unpack): Likewise.
8786 (mips_expand_vi_broadcast): Likewise.
8787 (mips_expand_vector_init): Likewise.
8788 (mips_expand_vec_reduc): Likewise.
8789 (mips_expand_msa_cmp): Likewise.
8790 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
8791 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
8792 (cc_flags_for_mode): Likewise.
8793 * config/msp430/msp430.c (msp430_print_operand): Likewise.
8794 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
8795 (nds32_output_casesi_pc_relative): Likewise.
8796 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8797 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
8798 (nvptx_gen_unpack): Likewise.
8799 (nvptx_gen_pack): Likewise.
8800 (nvptx_gen_shuffle): Likewise.
8801 (nvptx_gen_wcast): Likewise.
8802 (nvptx_preferred_simd_mode): Likewise.
8803 * config/pa/pa.c (pa_secondary_reload): Likewise.
8804 * config/pa/predicates.md (base14_operand): Likewise.
8805 * config/powerpcspe/powerpcspe-c.c
8806 (altivec_resolve_overloaded_builtin): Likewise.
8807 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
8809 (rs6000_preferred_simd_mode): Likewise.
8810 (output_vec_const_move): Likewise.
8811 (rs6000_expand_vector_extract): Likewise.
8812 (rs6000_split_vec_extract_var): Likewise.
8813 (reg_offset_addressing_ok_p): Likewise.
8814 (rs6000_legitimate_offset_address_p): Likewise.
8815 (rs6000_legitimize_address): Likewise.
8816 (rs6000_emit_set_const): Likewise.
8817 (rs6000_const_vec): Likewise.
8818 (rs6000_emit_move): Likewise.
8819 (spe_build_register_parallel): Likewise.
8820 (rs6000_darwin64_record_arg_recurse): Likewise.
8821 (swap_selector_for_mode): Likewise.
8822 (spe_init_builtins): Likewise.
8823 (paired_init_builtins): Likewise.
8824 (altivec_init_builtins): Likewise.
8825 (do_load_for_compare): Likewise.
8826 (rs6000_generate_compare): Likewise.
8827 (rs6000_expand_float128_convert): Likewise.
8828 (emit_load_locked): Likewise.
8829 (emit_store_conditional): Likewise.
8830 (rs6000_output_function_epilogue): Likewise.
8831 (rs6000_handle_altivec_attribute): Likewise.
8832 (rs6000_function_value): Likewise.
8833 (emit_fusion_gpr_load): Likewise.
8834 (emit_fusion_p9_load): Likewise.
8835 (emit_fusion_p9_store): Likewise.
8836 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
8837 (fusion_gpr_mem_load): Likewise.
8838 (fusion_addis_mem_combo_load): Likewise.
8839 (fusion_addis_mem_combo_store): Likewise.
8840 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
8841 (fusion_gpr_mem_load): Likewise.
8842 (fusion_addis_mem_combo_load): Likewise.
8843 (fusion_addis_mem_combo_store): Likewise.
8844 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8846 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
8847 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
8848 (rs6000_preferred_simd_mode): Likewise.
8849 (output_vec_const_move): Likewise.
8850 (rs6000_expand_vector_extract): Likewise.
8851 (rs6000_split_vec_extract_var): Likewise.
8852 (reg_offset_addressing_ok_p): Likewise.
8853 (rs6000_legitimate_offset_address_p): Likewise.
8854 (rs6000_legitimize_address): Likewise.
8855 (rs6000_emit_set_const): Likewise.
8856 (rs6000_const_vec): Likewise.
8857 (rs6000_emit_move): Likewise.
8858 (rs6000_darwin64_record_arg_recurse): Likewise.
8859 (swap_selector_for_mode): Likewise.
8860 (paired_init_builtins): Likewise.
8861 (altivec_init_builtins): Likewise.
8862 (rs6000_expand_float128_convert): Likewise.
8863 (emit_load_locked): Likewise.
8864 (emit_store_conditional): Likewise.
8865 (rs6000_output_function_epilogue): Likewise.
8866 (rs6000_handle_altivec_attribute): Likewise.
8867 (rs6000_function_value): Likewise.
8868 (emit_fusion_gpr_load): Likewise.
8869 (emit_fusion_p9_load): Likewise.
8870 (emit_fusion_p9_store): Likewise.
8871 * config/rx/rx.c (rx_gen_move_template): Likewise.
8872 (flags_from_mode): Likewise.
8873 * config/s390/predicates.md (s390_alc_comparison): Likewise.
8874 (s390_slb_comparison): Likewise.
8875 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
8876 (s390_vector_mode_supported_p): Likewise.
8877 (s390_cc_modes_compatible): Likewise.
8878 (s390_match_ccmode_set): Likewise.
8879 (s390_canonicalize_comparison): Likewise.
8880 (s390_emit_compare_and_swap): Likewise.
8881 (s390_branch_condition_mask): Likewise.
8882 (s390_rtx_costs): Likewise.
8883 (s390_secondary_reload): Likewise.
8884 (__SECONDARY_RELOAD_CASE): Likewise.
8885 (s390_expand_cs): Likewise.
8886 (s390_preferred_simd_mode): Likewise.
8887 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
8888 * config/sh/sh.c (sh_print_operand): Likewise.
8889 (dump_table): Likewise.
8890 (sh_secondary_reload): Likewise.
8891 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8892 * config/sh/sh.md (casesi_worker_1): Likewise.
8893 (casesi_worker_2): Likewise.
8894 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
8895 (fcc_comparison_operator): Likewise.
8896 * config/sparc/sparc.c (sparc_expand_move): Likewise.
8897 (emit_soft_tfmode_cvt): Likewise.
8898 (sparc_preferred_simd_mode): Likewise.
8899 (output_cbranch): Likewise.
8900 (sparc_print_operand): Likewise.
8901 (sparc_expand_vec_perm_bmask): Likewise.
8902 (vector_init_bshuffle): Likewise.
8903 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
8904 (spu_vector_mode_supported_p): Likewise.
8905 (spu_expand_insv): Likewise.
8906 (spu_emit_branch_or_set): Likewise.
8907 (spu_handle_vector_attribute): Likewise.
8908 (spu_builtin_splats): Likewise.
8909 (spu_builtin_extract): Likewise.
8910 (spu_builtin_promote): Likewise.
8911 (spu_expand_sign_extend): Likewise.
8912 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
8913 (tilegx_simd_int): Likewise.
8914 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
8915 (tilepro_simd_int): Likewise.
8916 * config/v850/v850.c (const_double_split): Likewise.
8917 (v850_print_operand): Likewise.
8918 (ep_memory_offset): Likewise.
8919 * config/vax/vax.c (vax_rtx_costs): Likewise.
8920 (vax_output_int_move): Likewise.
8921 (vax_output_int_add): Likewise.
8922 (vax_output_int_subtract): Likewise.
8923 * config/visium/predicates.md (visium_branch_operator): Likewise.
8924 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
8925 (visium_print_operand_address): Likewise.
8926 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8927 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
8928 (xtensa_expand_conditional_branch): Likewise.
8929 (xtensa_copy_incoming_a7): Likewise.
8930 (xtensa_output_literal): Likewise.
8931 * dfp.c (decimal_real_maxval): Likewise.
8932 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
8934 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8935 Alan Hayward <alan.hayward@arm.com>
8936 David Sherwood <david.sherwood@arm.com>
8938 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
8939 (mode_nunits_inline): Likewise.
8940 (mode_inner_inline): Likewise.
8941 (mode_unit_size_inline): Likewise.
8942 (mode_unit_precision_inline): Likewise.
8943 (emit_insn_modes_h): Likewise. Also emit a #define of the
8945 (emit_mode_wider): Add an E_ prefix to mode names.
8946 (emit_mode_complex): Likewise.
8947 (emit_mode_inner): Likewise.
8948 (emit_mode_adjustments): Likewise.
8949 (emit_mode_int_n): Likewise.
8950 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
8951 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
8952 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
8953 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
8954 (CRC32_BUILTIN, ENTRY): Likewise.
8955 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
8956 (aarch64_pop_regs): Likewise.
8957 (aarch64_process_components): Likewise.
8958 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
8959 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
8960 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
8961 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
8962 * config/arm/arm.c (arm_init_libfuncs): Likewise.
8963 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
8965 * config/i386/i386-builtin.def (pcmpestr): Likewise.
8966 (pcmpistr): Likewise.
8967 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
8968 * config/mmix/mmix.c (mmix_output_condition): Likewise.
8969 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
8971 * config/rl78/rl78.c (mduc_regs): Likewise.
8972 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8973 (htm_expand_builtin): Likewise.
8974 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
8975 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
8976 * config/xtensa/xtensa.c (print_operand): Likewise.
8977 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
8978 (NUM_MODE_VECTOR_INT): Likewise.
8979 * genoutput.c (null_operand): Likewise.
8980 (output_operand_data): Likewise.
8981 * genrecog.c (print_parameter_value): Likewise.
8982 * lra.c (debug_operand_data): Likewise.
8984 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8986 * dwarf2out.c (scompare_loc_descriptor_wide)
8987 (scompare_loc_descriptor_narrow): New functions, split out from...
8988 (scompare_loc_descriptor): ...here.
8989 * expmed.c (emit_store_flag_int): New function, split out from...
8990 (emit_store_flag): ...here.
8992 2017-08-30 Richard Biener <rguenther@suse.de>
8994 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
8995 (dwarf2out_early_finish): Move setting of AT_pubnames from
8996 early debug output to early finish.
8998 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9000 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
9001 and -mdata-region to the assembler.
9003 2017-08-30 Richard Biener <rguenther@suse.de>
9005 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
9007 (gen_subprogram_die): Add DW_AT_object_pointer only early.
9008 (dwarf2out_early_global_decl): Only generate a DIE for the
9009 abstract origin if it doesn't already exist or is a declaration DIE.
9010 (resolve_addr): Do not add the linkage name twice when
9011 generating a stub DIE for the DW_TAG_GNU_call_site target.
9013 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9015 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9016 Use machine_mode rather than int for arg1_mode.
9018 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
9021 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
9022 that the second argument of the built-in functions to unpack
9023 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
9024 switch statement instead a lot of if statements.
9025 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
9026 Allow 64-bit values to be in Altivec registers as well as
9027 traditional floating point registers.
9028 (pack<mode>, FMOVE128_VSX iterator): Likewise.
9030 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
9032 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
9033 MAX_REGS_PER_ADDRESS == 1.
9035 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
9037 * config/i386/i386.opt (flag_fentry): Do not init to -1.
9038 * config/i386/i386.c (ix86_option_override_internal): Simplify
9039 setting of opts->x_flag_entry.
9041 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9042 Jakub Jelinek <jakub@redhat.com>
9043 Richard Biener <rguenther@suse.de>
9045 PR tree-optimization/81503
9046 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
9047 folded constant fits in the target type; reorder tests for clarity.
9049 2017-08-29 Martin Liska <mliska@suse.cz>
9051 * passes.def: Include pass_lower_switch.
9052 * stmt.c (dump_case_nodes): Remove and move to
9053 tree-switch-conversion.
9054 (case_values_threshold): Likewise.
9055 (expand_switch_as_decision_tree_p): Likewise.
9056 (emit_case_decision_tree): Likewise.
9057 (expand_case): Likewise.
9058 (balance_case_nodes): Likewise.
9059 (node_has_low_bound): Likewise.
9060 (node_has_high_bound): Likewise.
9061 (node_is_bounded): Likewise.
9062 (emit_case_nodes): Likewise.
9063 (struct simple_case_node): New struct.
9064 (add_case_node): Remove.
9065 (emit_case_dispatch_table): Use vector instead of case_list.
9066 (reset_out_edges_aux): Remove.
9067 (compute_cases_per_edge): Likewise.
9068 (expand_case): Build list of simple_case_node.
9069 (expand_sjlj_dispatch_table): Use it.
9070 * tree-switch-conversion.c (struct case_node): Moved from
9071 stmt.c and adjusted.
9072 (emit_case_nodes): Likewise.
9073 (node_has_low_bound): Likewise.
9074 (node_has_high_bound): Likewise.
9075 (node_is_bounded): Likewise.
9076 (case_values_threshold): Likewise.
9077 (reset_out_edges_aux): Likewise.
9078 (compute_cases_per_edge): Likewise.
9079 (add_case_node): Likewise.
9080 (dump_case_nodes): Likewise.
9081 (balance_case_nodes): Likewise.
9082 (expand_switch_as_decision_tree_p): Likewise.
9083 (emit_jump): Likewise.
9084 (emit_case_decision_tree): Likewise.
9085 (try_switch_expansion): Likewise.
9086 (do_jump_if_equal): Likewise.
9087 (emit_cmp_and_jump_insns): Likewise.
9088 (fix_phi_operands_for_edge): New function.
9089 (record_phi_operand_mapping): Likewise.
9090 (class pass_lower_switch): New pass.
9091 (pass_lower_switch::execute): New function.
9092 (make_pass_lower_switch): Likewise.
9093 (conditional_probability):
9094 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
9095 * tree-pass.h: Add make_pass_lower_switch.
9097 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9100 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
9103 2017-08-29 Richard Biener <rguenther@suse.de>
9105 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
9106 we do not add a DW_AT_inline attribute twice.
9107 (gen_subprogram_die): Remove code setting DW_AT_inline on
9108 DECL_ABSTRACT_P nodes.
9110 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
9112 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
9113 calls to internal functions.
9114 (gimplify_modify_expr): Likewise.
9115 * tree-call-cdce.c (use_internal_fn): Likewise.
9116 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
9117 (convert_to_divmod): Set the nothrow flag.
9118 * tree-if-conv.c (predicate_mem_writes): Likewise.
9119 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
9120 (vectorizable_call): Likewise.
9121 (vectorizable_store): Likewise.
9122 (vectorizable_load): Likewise.
9123 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
9124 (vect_recog_mask_conversion_pattern): Likewise.
9126 2017-08-29 Martin Liska <mliska@suse.cz>
9129 * gcc.c (driver_handle_option): Add new argument.
9130 * opts-common.c (handle_option): Pass
9131 target_option_override_hook.
9132 * opts-global.c (lang_handle_option): Add new option.
9133 (set_default_handlers): Add new argument.
9134 (decode_options): Likewise.
9135 * opts.c (target_handle_option): Likewise.
9136 (common_handle_option): Call target_option_override_hook.
9137 * opts.h (struct cl_option_handler_func): Add hook for
9138 target option override.
9139 (struct cl_option_handlers): Likewise.
9140 (set_default_handlers): Add new argument.
9141 (decode_options): Likewise.
9142 (common_handle_option): Likewise.
9143 (target_handle_option): Likewise.
9144 * toplev.c (toplev::main): Pass targetm.target_option.override
9147 2017-08-29 Richard Biener <rguenther@suse.de>
9148 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9150 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
9151 life to the active subtree.
9153 2017-08-28 Jeff Law <law@redhat.com>
9155 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
9156 derive_equivalences.
9157 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
9159 (edge_info::derive_equivalences): New private member function
9161 * tree-ssa-dom.c (class edge_info): Changed from a struct
9162 to a class. Add ctor/dtor, methods and data members.
9163 (edge_info::edge_info): Renamed from allocate_edge_info.
9164 Initialize additional members.
9165 (edge_info::~edge_info): New.
9166 (free_dom_edge_info): Delete the edge info.
9167 (record_edge_info): Use new class & associated member functions.
9168 Tighten forms for testing for edge equivalences.
9169 (record_temporary_equivalences): Iterate over the simple
9170 equivalences rather than assuming there's only one per edge.
9171 (cprop_into_successor_phis): Iterate over the simple
9172 equivalences rather than assuming there's only one per edge.
9173 (optimize_stmt): Use operand_equal_p rather than pointer
9174 equality for mini-DSE code.
9176 2017-08-28 Nathan Sidwell <nathan@acm.org>
9178 * gcc.c (execute): Fold SIGPIPE handling into switch
9179 statement. Adjust internal error message.
9181 2017-08-28 Richard Biener <rguenther@suse.de>
9184 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
9185 Do nothing for removed DIEs.
9187 2017-08-28 Richard Biener <rguenther@suse.de>
9189 PR tree-optimization/81977
9190 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
9193 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
9196 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
9197 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
9198 using targetm.gen_mem_thread_fence.
9200 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
9203 * config/i386/i386.md (*<btsc><mode>): Change operand 2
9204 predicate to register_operand. Reorder operands.
9205 (*btr<mode>): Ditto.
9206 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
9207 (*btr<mode>_mask): Ditto.
9209 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
9211 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
9212 * config/rs6000/xmmintrin.h: New file.
9213 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
9215 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9218 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
9219 parameter and_insn and return it.
9220 (recombine_lvx_pattern): Insert a copy to ensure availability of
9221 the base register of the copied masking operation at the point of
9222 the instruction replacement.
9223 (recombine_stvx_pattern): Likewise.
9225 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
9227 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
9228 undocumented switches.
9229 (-mpower9-dform-vector): Likewise.
9230 (-mpower9-dform): Likewise.
9231 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
9232 comments to delete references to -mpower9-dform* switches.
9233 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
9234 Delete reference to -mpower9-dform* switches, test for
9235 -mpower9-vector instead.
9236 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
9237 (OTHER_P9_VECTOR_MASKS): Likewise.
9238 (POWERPC_MASKS): Likewise.
9239 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
9240 tests against -mpower9-dform* to -mpower9-vector. Delete code
9241 that checked for -mpower9-dform* consistancy with other options.
9242 Add test for -mpower9-misc to enable other power9 switches.
9243 (rs6000_init_hard_regno_mode_ok): Likewise.
9244 (rs6000_option_override_internal): Likewise.
9245 (rs6000_emit_prologue): Likewise.
9246 (rs6000_emit_epilogue): Likewise.
9247 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
9248 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
9249 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
9251 (emit_fusion_p9_store): Likewise.
9252 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
9253 resetting these macros if the assembler does not support ISA 3.0
9255 (TARGET_P9_DFORM_VECTOR): Likewise.
9256 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
9257 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
9259 2017-08-25 Alan Modra <amodra@gmail.com>
9261 PR rtl-optimization/81747
9262 * cse.c (cse_extended_basic_block): Don't attempt to record
9263 equivalences for degenerate conditional branches that branch
9264 to their fall-through.
9266 2017-08-24 Martin Sebor <msebor@redhat.com>
9269 * gimple-fold.c (size_must_be_zero_p): New function.
9270 (gimple_fold_builtin_memory_op): Call it.
9272 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
9274 * config/rs6000/mm_malloc.h: New file.
9276 2017-08-24 Bin Cheng <bin.cheng@arm.com>
9278 PR tree-optimization/81913
9279 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
9280 analysis when either IVs in condition can wrap.
9282 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
9284 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
9285 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
9287 2017-08-24 Richard Biener <rguenther@suse.de>
9290 * targhooks.c (default_target_can_inline_p): Properly
9291 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
9292 is present and always compare.
9293 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
9294 infer -mfpmath=sse from TARGET_SSE_P.
9295 (ix86_can_inline_p): Properly use target_option_default_node when
9296 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
9298 2017-08-24 Richard Biener <rguenther@suse.de>
9301 * dwarf2out.c (output_die): Handle flag_generate_offload like
9303 (output_comp_unit): Likewise.
9304 (gen_array_type_die): Likewise.
9305 (dwarf2out_early_finish): Likewise.
9306 (note_variable_value_in_expr): Likewise.
9307 (dwarf2out_finish): Likewise. Adjust assert.
9308 * cgraphunit.c (symbol_table::compile): Move setting of
9309 flag_generate_offload earlier ...
9310 (symbol_table::finalize_compilation_unit): ... here, before
9311 early debug finalization.
9313 2017-08-24 Richard Biener <rguenther@suse.de>
9315 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
9316 and ipa-fnsummary.h.
9317 (ix86_can_inline_p): When ix86_fpmath flags do not match
9318 check whether the callee uses FP math at all.
9320 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
9323 * tree-ssanames.c (get_nonzero_bits): Use element_precision
9324 instead of TYPE_PRECISION.
9326 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
9327 Alan Hayward <alan.hayward@arm.com>
9328 David Sherwood <david.sherwood@arm.com>
9330 * combine.c (make_extraction): Use subreg_offset_from_lsb.
9332 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
9334 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
9336 (ix86_frame::stack_realign_allocate): New field.
9337 (struct machine_frame_state): Modify comments.
9338 (machine_frame_state::sp_realigned_fp_end): New field.
9339 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
9341 (sp_valid_at): Add assertion to assure no attempt to access invalid
9342 offset of a realigned stack.
9343 (fp_valid_at): Likewise.
9344 (choose_baseaddr): Modify comments.
9345 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
9346 ix86_expand_prologue.
9347 (ix86_expand_prologue): Modify stack realignment and allocation.
9348 (ix86_expand_epilogue): Modify comments.
9349 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
9350 avx2_runtime, avx512f, and avx512f_runtime.
9352 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
9354 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
9355 (mstackrealign): Do not init to -1.
9356 * config/i386/i386.c (ix86_option_override_internal):
9357 Check opts_set, not opts when setting default value of
9358 opts->x_ix86_force_align_arg_pointer.
9360 2017-08-23 Richard Biener <rguenther@suse.de>
9362 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
9363 lang_hooks.decl_printable_name.
9364 * print-rtl-function.c (print_rtx_function): Likewise.
9365 * tree-pretty-print.c (dump_function_header): Likewise.
9367 2017-08-23 Richard Biener <rguenther@suse.de>
9370 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
9371 -g0 at compile-time.
9373 2017-08-23 Tamar Christina <tamar.christina@arm.com>
9376 * doc/sourcebuild.texi (Other hardware attributes):
9379 2017-08-23 Tamar Christina <tamar.christina@arm.com>
9382 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
9383 Fix single-use check.
9385 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
9387 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
9389 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
9391 * doc/install.texi: Modify to add more details on running selected
9394 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
9396 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
9397 is combined with -mabi=ms.
9398 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
9401 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9403 PR tree-optimization/81488
9404 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
9405 and cached_basis fields.
9406 (MAX_SPREAD): New constant.
9407 (alloc_cand_and_find_basis): Initialize new fields.
9408 (clear_visited): New function.
9409 (create_phi_basis_1): Rename from create_phi_basis, set visited
9410 and cached_basis fields.
9411 (create_phi_basis): New wrapper function.
9412 (phi_add_costs_1): Rename from phi_add_costs, add spread
9413 parameter, set visited field, short-circuit when limits reached.
9414 (phi_add_costs): New wrapper function.
9415 (record_phi_increments_1): Rename from record_phi_increments, set
9417 (record_phi_increments): New wrapper function.
9418 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
9419 (phi_incr_cost): New wrapper function.
9420 (all_phi_incrs_profitable_1): Rename from
9421 all_phi_incrs_profitable, set visited field.
9422 (all_phi_incrs_profitable): New wrapper function.
9424 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
9425 Alan Hayward <alan.hayward@arm.com>
9426 David Sherwood <david.sherwood@arm.com>
9428 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
9429 that takes the outer and inner modes.
9430 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
9431 comparison as the canonical test for a paradoxical subreg.
9432 * combine.c (simplify_set): Use paradoxical_subreg_p.
9433 (make_extraction): Likewise.
9434 (force_to_mode): Likewise.
9435 (rtx_equal_for_field_assignment_p): Likewise.
9436 (gen_lowpart_for_combine): Likewise.
9437 (simplify_comparison): Likewise.
9438 * cse.c (equiv_constant): Likewise.
9439 * expmed.c (store_bit_field_1): Likewise.
9440 * final.c (alter_subreg): Likewise.
9441 * fwprop.c (propagate_rtx): Likewise.
9442 (forward_propagate_subreg): Likewise.
9443 * ira-conflicts.c (ira_build_conflicts): Likewise.
9444 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
9445 * lra-constraints.c (curr_insn_transform): Likewise.
9446 (split_reg): Likewise.
9447 * lra-eliminations.c (move_plus_up): Likewise.
9448 (lra_eliminate_regs_1): Likewise.
9449 * recog.c (general_operand): Likewise.
9450 * ree.c (combine_reaching_defs): Likewise.
9451 * reload.c (push_reload): Likewise.
9452 (find_reloads): Likewise.
9453 * reload1.c (elimination_effects): Likewise.
9454 (compute_reload_subreg_offset): Likewise.
9455 (choose_reload_regs): Likewise.
9456 * rtlanal.c (subreg_lsb_1): Likewise.
9457 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9458 (simplify_subreg): Likewise.
9459 * var-tracking.c (track_loc_p): Likewise.
9460 * emit-rtl.c (byte_lowpart_offset): Likewise.
9461 (paradoxical_subreg_p): Delete out-of-line definition.
9463 2017-08-22 Jeff Law <law@redhat.com>
9465 PR tree-optimization/81741
9466 PR tree-optimization/71947
9467 * tree-ssa-dom.c: Include tree-inline.h.
9468 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
9469 equivalences if one is more expensive to compute than the other.
9470 * tree-ssa-scopedtables.h (class const_or_copies): Make
9471 record_const_or_copy_raw method private.
9472 (class avail_exprs_stack): New method simplify_binary_operation.
9473 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
9474 avail_exprs_stack::simplify_binary_operation as needed.
9475 (avail_exprs_stack::simplify_binary_operation): New function.
9477 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
9479 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
9480 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
9481 (DOT_SYMBOLS): Likewise.
9482 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
9483 (RELOCATABLE_NEEDS_FIXUP): Likewise.
9484 (RS6000_ABI_NAME): Likewise.
9485 (TARGET_CMODEL): Likewise.
9486 (TOC_SECTION_ASM_OP): Likewise.
9487 (SET_CMODEL): New macro.
9488 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
9490 2017-08-22 Richard Biener <rguenther@suse.de>
9492 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
9493 to allow for free-lang-data replacements similar to verify_type_variant.
9495 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
9496 Alan Hayward <alan.hayward@arm.com>
9497 David Sherwood <david.sherwood@arm.com>
9499 * config/aarch64/aarch64.md (casesi): Use DImode rather than
9500 VOIDmode for the LABEL_REF.
9502 2017-08-22 Richard Biener <rguenther@suse.de>
9504 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
9506 2017-08-22 Richard Biener <rguenther@suse.de>
9508 * common.opt (feliminate-dwarf2-dups): Ignore.
9509 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
9510 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
9511 same_die_p_wrap, compute_section_prefix,
9512 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
9513 (comdat_symbol_id, comdat_symbol_number): Likewise.
9514 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
9516 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
9517 (output_die): Mark unreachable path unreachable.
9518 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
9519 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
9520 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
9521 (dwarf2out_early_finish): Likewise.
9523 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
9525 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
9527 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
9530 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
9531 not VAR_P. Filter attribute warnings with OPT_Wattributes.
9532 (avr_attribute_table) <io, io_low, address>: Initialize
9533 .decl_required with true.
9535 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
9537 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
9538 undocumented debugging options.
9539 (-mvsx-scalar-double): Likewise.
9540 (-mallow-df-permute): Likewise.
9541 (-mvectorize-builtins): Likewise.
9542 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9543 (rs6000_builtin_vectorized_function): Likewise.
9544 (rs6000_builtin_md_vectorized_function): Likewise.
9545 (rs6000_opt_vars): Likewise.
9547 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
9550 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
9551 (*btrq_imm): Rename from *btrq.
9552 (*btcq_imm): Rename from *btcq.
9553 (btsc): New code attribute.
9554 (*<btsc><mode>): New insn pattern.
9555 (*btr<mode>): Ditto.
9556 (*<btsc><mode>_mask): New insn_and_split pattern.
9557 (*btr<mode>_mask): Ditto.
9559 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9560 Alan Hayward <alan.hayward@arm.com>
9561 David Sherwood <david.sherwood@arm.com>
9563 * function.c (pad_below): Simplify padding calculation.
9565 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9566 Alan Hayward <alan.hayward@arm.com>
9567 David Sherwood <david.sherwood@arm.com>
9569 * target.def (function_prologue): Remove frame size argument.
9570 (function_epilogue): Likewise.
9571 * doc/tm.texi: Regenerate.
9572 * final.c (final_start_function): Update call to function_prologue.
9573 (final_end_function): Update call to function_epilogue.
9574 (default_function_pro_epilogue): Remove frame size argument.
9575 * output.h (default_function_pro_epilogue): Likewise.
9576 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
9577 (arm_output_function_prologue): Likewise.
9578 * config/frv/frv.c (frv_function_prologue): Likewise.
9579 (frv_function_epilogue): Likewise.
9580 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
9581 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
9582 (ia64_output_function_epilogue): Likewise.
9583 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
9584 (m32r_output_function_epilogue): Likewise.
9585 * config/microblaze/microblaze.c (microblaze_function_prologue)
9586 (microblaze_function_epilogue): Likewise.
9587 * config/mips/mips.c (mips_output_function_prologue): Likewise.
9588 (mips_output_function_epilogue): Likewise.
9589 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
9590 (mmix_target_asm_function_epilogue): Likewise.
9591 * config/msp430/msp430.c (msp430_start_function): Likewise.
9592 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
9593 (nds32_asm_function_epilogue): Likewise.
9594 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
9595 * config/pa/pa.c (pa_output_function_prologue): Likewise.
9596 (pa_output_function_epilogue): Likewise.
9597 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
9598 (rs6000_output_function_epilogue): Likewise.
9599 * config/rl78/rl78.c (rl78_start_function): Likewise.
9600 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
9601 (rs6000_output_function_epilogue): Likewise.
9602 * config/rx/rx.c (rx_output_function_prologue): Likewise.
9603 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
9604 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
9605 (sparc_asm_function_epilogue): Likewise.
9607 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9609 * tree.h (type_has_mode_precision_p): New function.
9610 * convert.c (convert_to_integer_1): Use it.
9611 * expr.c (expand_expr_real_2): Likewise.
9612 (expand_expr_real_1): Likewise.
9613 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
9614 * match.pd: Likewise.
9615 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
9616 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
9617 * tree-tailcall.c (process_assignment): Likewise.
9618 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9619 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
9620 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
9621 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9622 (vectorizable_assignment): Likewise.
9623 (vectorizable_shift): Likewise.
9624 (vectorizable_operation): Likewise.
9625 * tree-vrp.c (register_edge_assert_for_2): Likewise.
9627 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
9629 * match.pd: Add pow (C, x) simplification.
9631 2017-08-21 Richard Biener <rguenther@suse.de>
9633 PR tree-optimization/81900
9634 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
9635 for blocks with abnormal predecessors.
9636 (compute_antic): Do not set visited flag prematurely.
9638 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
9641 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
9643 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9645 * stor-layout.h (vector_type_mode): Move to...
9646 * tree.h (vector_type_mode): ...here.
9647 * stor-layout.c (vector_type_mode): Move to...
9648 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
9650 2017-08-21 Richard Biener <rguenther@suse.de>
9652 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
9653 register_external_die hooks.
9654 (debug_false_tree_charstarstar_uhwistar): Declare.
9655 (debug_nothing_tree_charstar_uhwi): Likewise.
9656 * debug.c (do_nothing_debug_hooks): Adjust.
9657 (debug_false_tree_charstarstar_uhwistar): New do nothing.
9658 (debug_nothing_tree_charstar_uhwi): Likewise.
9659 * dbxout.c (dbx_debug_hooks): Adjust.
9660 (xcoff_debug_hooks): Likewise.
9661 * sdbout.c (sdb_debug_hooks): Likewise.
9662 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
9663 * dwarf2out.c (macinfo_label_base): New global.
9664 (dwarf2out_register_external_die): New function for the
9665 register_external_die hook.
9666 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
9667 (dwarf2_debug_hooks): Use them.
9668 (dwarf2_lineno_debug_hooks): Adjust.
9669 (struct die_struct): Add with_offset flag.
9670 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
9671 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
9672 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
9673 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
9674 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
9675 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
9676 defining section names for the early LTO debug variants.
9677 (reset_indirect_string): New helper.
9678 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
9679 (print_dw_val): Add support for offsetted symbol references.
9680 (get_ultimate_context): Split out from is_cxx.
9681 (is_cxx): Use get_ultimate_context.
9682 (is_fortran): Add decl overload.
9683 (compute_comp_unit_symbol): Split out worker from
9684 compute_section_prefix.
9685 (compute_section_prefix): Call compute_comp_unit_symbol and
9686 set comdat_type_p here.
9687 (output_die): Skip DIE symbol output for the LTO added one.
9688 Handle DIE symbol references with offset.
9689 (output_comp_unit): Guard section name mangling properly.
9690 For LTO debug sections emit a symbol at the section beginning
9691 which we use to refer to its DIEs.
9692 (add_abstract_origin_attribute): For DIEs registered via
9693 dwarf2out_register_external_die directly refer to the early
9694 DIE rather than indirectly through the shadow one we created.
9695 Remove obsolete call to dwarf2out_abstract_function for
9696 non-function/block origins.
9697 (gen_array_type_die): When generating early LTO debug do
9698 not emit DW_AT_string_length.
9699 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
9700 late when in LTO. As suggested place a gcc_unreachable for
9701 the DECL_ABSTRACT_P case.
9702 (gen_subprogram_die): Avoid another specification DIE
9703 for early built declarations/definitions for the late LTO case.
9704 (gen_variable_die): Add type references for late duplicated VLA dies
9706 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
9707 we have the abstract instance already.
9708 (process_scope_var): Adjust decl DIE contexts in LTO which
9709 first puts them in limbo.
9710 (gen_decl_die): Do not generate type DIEs late apart from
9711 types for VLAs or for decls we do not yet have a DIE. Do not
9712 call dwarf2out_abstract_function late.
9713 (dwarf2out_early_global_decl): Make sure to create DIEs
9714 for abstract instances of a decl first.
9715 (dwarf2out_late_global_decl): Adjust comment.
9716 (output_macinfo_op): With multiple macro sections use
9717 macinfo_label_base to distinguish labels.
9718 (output_macinfo): Likewise. Update macinfo_label_base.
9719 Pass in the line info label.
9720 (note_variable_value_in_expr): When generating LTO resolve
9721 all variable values here by generating DIEs as needed.
9722 (init_sections_and_labels): Add early LTO debug flag parameter
9723 and generate different sections and names if set. Add generation
9724 counter for the labels so we can have multiple of them.
9725 (reset_dies): Helper to allow DIEs to be output multiple times.
9726 (dwarf2out_finish): When outputting DIEs to the fat part of an
9727 LTO object first reset DIEs.
9728 (dwarf2out_early_finish): Output early DIEs when generating LTO.
9729 (modified_type_die): Check for decl_ultimate_origin being self
9731 (gen_type_die_with_usage): Likewise.
9732 (gen_typedef_die): Allow decl_ultimate_origin being self.
9733 (set_decl_abstract_flags): Remove.
9734 (set_block_abstract_flags): Likewise.
9735 (dwarf2out_abstract_function): Treat the early generated DIEs
9736 as the abstract copy and only add DW_AT_inline and
9737 DW_AT_artificial here and call set_decl_origin_self.
9738 If the DIE has an abstract origin don't do anything.
9739 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
9740 if we have none yet (Go fails to build one, PR78628).
9741 (variably_modified_type_p): Prevent endless recursion for Ada
9742 cyclic pointer types.
9743 * lto-streamer-in.c: Include debug.h.
9744 (dref_queue): New global.
9745 (lto_read_tree_1): Stream in DIE references.
9746 (lto_input_tree): Register DIE references.
9747 (input_function): Stream DECL_DEBUG_ARGS.
9748 * lto-streamer-out.c: Include debug.h.
9749 (lto_write_tree_1): Output DIE references.
9750 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
9751 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
9752 (output_function): Stream DECL_DEBUG_ARGS.
9753 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
9754 Stream DECL_ABSTRACT_ORIGIN.
9755 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
9756 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
9757 DECL_CONTEXT for file-scope decls.
9758 * lto-streamer.h (struct dref_entry): Declare.
9759 (dref_queue): Likewise.
9760 * cfgexpand.c (pass_expand::execute): Do not call the
9761 outlining_inline_function hook here.
9762 * lto-wrapper.c (debug_obj): New global.
9763 (tool_cleanup): Unlink it if required.
9764 (debug_objcopy): New function.
9765 (run_gcc): Handle early debug sections in the IL files by
9766 extracting them to separate files, partially linkin them and
9767 feeding the result back as result to the linker.
9768 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
9769 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
9770 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
9771 sections into a separate segment.
9772 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
9774 (darwin_asm_dwarf_section): Likewise.
9775 (darwin_asm_output_dwarf_offset): Likewise.
9776 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
9778 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9779 Alan Hayward <alan.hayward@arm.com>
9780 David Sherwood <david.sherwood@arm.com>
9782 * read-md.h (md_reader::record_potential_iterator_use): Replace
9783 pointer argument with an rtx and an index.
9784 * read-rtl.c (iterator_group::apply_iterator): Likewise.
9785 (apply_mode_iterator): Likewise.
9786 (apply_code_iterator): Likewise.
9787 (apply_int_iterator): Likewise.
9788 (apply_subst_iterator): Likewise.
9789 (record_iterator_use): Likewise.
9790 (record_attribute_use): Likewise.
9791 (md_reader::record_potential_iterator_use): Likewise. Update calls
9792 to record_iterator_use and apply_iterator.
9793 (iterator_use): Replace ptr with x and index.
9794 (attribute_use): Likewise.
9795 (apply_attribute_uses): Update calls to apply_iterator.
9796 (apply_iterators): Likewise. Update initialization of iterator_use.
9797 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
9798 and record_potential_iterator_use.
9799 (rtx_reader::read_rtx_operand): Likewise.
9801 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9802 Alan Hayward <alan.hayward@arm.com>
9803 David Sherwood <david.sherwood@arm.com>
9805 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
9808 2017-08-21 Richard Biener <rguenther@suse.de>
9811 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
9812 at struct end conservatively when comparing common bases.
9814 2017-08-21 Richard Biener <rguenther@suse.de>
9816 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
9817 (mem_ref_in_stmt): Remove.
9818 (determine_max_movement): Use ref index to get at the reference.
9819 (invariantness_dom_walker::before_dom_children): Deal with
9820 lim data already initialized.
9821 (gather_mem_refs_stmt): Initialize lim data and record ref index.
9823 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
9825 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
9826 (TARGET_ISA_ROUND): Ditto.
9827 (TARGET_ROUND): Ditto.
9828 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
9829 * config/i386/i386.md: Ditto.
9830 * config/i386/sse.md: Ditto.
9831 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
9832 with OPTION_MASK_ISA_SSE4_1.
9834 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
9837 * doc/extend.texi (x86 Built-in Functions): Correct the name of
9838 __builtin_ia32_lzcnt_u16.
9840 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
9843 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
9844 (rs6000_set_current_function): Rewrite function to use it.
9846 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
9849 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
9850 and TYPE_WARN_IF_NOT_ALIGN.
9851 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
9852 (handle_warn_if_not_align): New.
9853 (place_union_field): Call handle_warn_if_not_align.
9854 (place_field): Call handle_warn_if_not_align.
9855 Copy TYPE_WARN_IF_NOT_ALIGN.
9856 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
9857 (layout_type): Likewise.
9858 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
9860 (tree_decl_common): Add warn_if_not_align.
9861 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
9862 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
9863 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
9864 (DECL_WARN_IF_NOT_ALIGN): Likewise.
9865 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
9866 * doc/extend.texi: Document warn_if_not_aligned attribute.
9867 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
9869 2017-08-17 Martin Liska <mliska@suse.cz>
9872 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
9873 (get_data_dependence): Use it as pointer type.
9874 (distribute_loop): Likewise.
9876 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9878 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
9879 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
9880 (p8_vmrgow_<mode>_direct): New define_insn.
9881 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
9882 handle endianness for vmrgew and vmrgow permute patterns.
9884 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
9886 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
9887 * config/rs6000/rs6000-cpus.def: Remove comment.
9888 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
9889 (POWERPC_MASKS): Likewise.
9890 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
9891 use of TARGET_VSX_TIMODE.
9892 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
9893 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
9894 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
9895 (rs6000_option_override_internal): Remove dead code.
9896 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
9897 (rs6000_legitimize_reload_address): Likewise.
9898 (rs6000_legitimate_address_p): Likewise.
9899 (rs6000_opt_masks): Delete "vsx-timode".
9900 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
9901 from function comment.
9902 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
9903 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
9904 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
9906 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
9907 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
9908 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
9911 (define_peephole2): Likewise.
9913 2017-08-17 Martin Sebor <msebor@redhat.com>
9916 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
9917 past the end of an array.
9918 (test_pp_format): Add test cases.
9920 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
9922 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
9923 missing ECF_NOTHROW flags.
9925 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
9928 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
9929 operands residing in integer registers.
9930 (*vsx_le_perm_load_<mode>): Likewise.
9931 (*vsx_le_perm_store_<mode>): Likewise.
9932 (define_peephole2): Add peepholes to optimize the above.
9934 2017-08-17 Marek Polacek <polacek@redhat.com>
9937 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
9938 to mimic what shorten_compare did. Change the return type to bool.
9939 (fold_cond_expr_with_comparison): Update call to
9940 operand_equal_for_comparison_p.
9941 (fold_ternary_loc): Likewise.
9943 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
9945 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
9947 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
9949 2017-08-17 Richard Biener <rguenther@suse.de>
9951 * tree-ssa-structalias.c (solve_graph): When propagating
9952 to successors update the graphs succ edges and avoid duplicate work.
9954 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
9957 * config/i386/i386.c (ix86_option_override_internal): Save target
9958 specific options after ix86_stack_protector_guard_reg was changed.
9960 2017-08-17 Richard Biener <rguenther@suse.de>
9962 PR tree-optimization/81827
9963 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
9964 (new_var_info): Initialize it conservatively.
9965 (get_call_vi): Mark register vars.
9966 (new_scalar_tmp_constraint_exp): Likewise.
9967 (handle_rhs_call): Likewise.
9968 (handle_const_call): Likewise.
9969 (create_function_info_for): Likewise.
9970 (solve_constraints): Sort varinfos to separate register from
9971 non-register vars to pack points-to solution bitmaps during
9974 2017-08-17 Marek Polacek <polacek@redhat.com>
9976 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
9978 2017-08-17 Richard Biener <rguenther@suse.de>
9980 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
9981 to true when overflow is undefined and we saturated the result.
9983 2017-08-17 Alan Modra <amodra@gmail.com>
9986 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
9987 Don't use store multiple if only one reg needs saving.
9988 (interesting_frame_related_regno): New function.
9989 (rs6000_frame_related): Don't emit frame info for regs that
9991 (rs6000_emit_epilogue): Likewise.
9993 2017-08-16 Nathan Sidwell <nathan@acm.org>
9995 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
9996 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
9997 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
9998 * tree.c (free_lang_data_in_type): Use else-if chain. Always
9999 clear TYPE_LANG_1. Remove obsolete member-function stripping.
10000 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
10001 (verify_type): Adjust for TYPE_BINFO move.
10002 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
10003 process TYPE_BINFO directly.
10004 (hash_tree): Likewise.
10005 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
10007 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
10010 2017-08-16 David Malcolm <dmalcolm@redhat.com>
10012 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
10014 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
10017 * config/i386/i386.md (*anddi_1_btr): Change predicates of
10018 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
10019 Add ix86_binary_operator_ok to insn constraint.
10020 (*iordi_1_bts): Ditto.
10021 (*xordi_1_btc): Ditto.
10022 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
10023 Update corresponding peephole2 pattern.
10027 2017-08-16 Bin Cheng <bin.cheng@arm.com>
10029 PR tree-optimization/81832
10030 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
10031 copy loop header which has IFN_LOOP_DIST_ALIAS call.
10033 2017-08-16 Marek Polacek <polacek@redhat.com>
10036 * fold-const.c (fold_indirect_ref_1): Restore original behavior
10037 regarding size_zero_node.
10039 2017-08-16 Martin Liska <mliska@suse.cz>
10042 * config.gcc: Respect previously set extra_objs in case
10045 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
10047 PR tree-optimization/81835
10048 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
10049 the phi SSA_NAME. Check that the condition in a COND_EXPR does
10050 not depend on the phi.
10052 2017-08-16 Alan Modra <amodra@gmail.com>
10054 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
10057 2017-08-16 Alan Modra <amodra@gmail.com>
10059 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
10060 (save_reg_p): ..into this. Update all callers.
10061 (first_reg_to_save): Simplify.
10063 2017-08-16 Alan Modra <amodra@gmail.com>
10065 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
10068 2017-08-15 Joseph Myers <joseph@codesourcery.com>
10072 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
10073 constant count if that count is less than 32.
10075 2017-08-15 Nathan Sidwell <nathan@acm.org>
10077 * gcc.c (execute): Emit friendlier message if inferior is killed
10078 by an external cause.
10080 2017-08-15 Richard Biener <rguenther@suse.de>
10082 PR tree-optimization/81790
10083 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
10084 CONSTRUCTORs from simplifying and VN.
10086 2017-08-14 Martin Sebor <msebor@redhat.com>
10088 * builtin-attrs.def: Add comments.
10090 2017-08-14 Martin Sebor <msebor@redhat.com>
10093 * doc/extend.texi (attribute nonstring): Document new attribute.
10095 2017-08-14 Martin Sebor <msebor@redhat.com>
10098 * tree-diagnostic.c (default_tree_printer): Handle %G.
10099 * gimple-pretty-print.h (percent_G_format): Declare new function.
10100 * gimple-pretty-print.c (percent_G_format): Define.
10101 * tree-pretty-print.c (percent_K_format): Add argument.
10103 2017-08-14 Martin Sebor <msebor@redhat.com>
10105 PR translation/79998
10106 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
10107 Remove a stray space.
10109 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
10112 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
10113 (*iordi_1_bts): Ditto.
10114 (*xordi_1_btc): Ditto.
10116 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10119 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
10120 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10122 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
10123 quoted strings, and make more translator-friendly.
10124 (darwin_rs6000_override_options): Likewise.
10125 (rs6000_option_override_internal): Likewise.
10126 (rs6000_return_in_memory): Fix overlong line.
10127 (init_cmulative_args): Use quoted strings, and make more
10128 translator-friendly.
10129 (rs6000_pass_by_reference): Fix overlong line.
10130 (def_builtin): Use quoted strings.
10131 (altivec_expand_predicate_builtin): Use quoted strings, and make
10132 more translator-friendly.
10133 (htm_expand_builtin): Use quoted strings.
10134 (cpu_expand_builtin): Use quoted strings, and make more
10135 translator-friendly.
10136 (altivec_expand_builtin): Likewise.
10137 (paired_expand_predicate_builtin): Likewise.
10138 (rs6000_invalid_builtin): Likewise.
10139 (builtin_function_type): Use quoted strings.
10140 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
10141 more translator-friendly.
10142 (rs6000_trampoline_init): Likewise.
10143 (rs6000_handle_altivec_attribute): Likewise.
10144 (rs6000_inner_target_options): Use quoted strings.
10145 (rs6000_disable_incompatible_switches): Likewise.
10146 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
10147 strings, and make more translator-friendly.
10148 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
10150 2017-08-14 Bin Cheng <bin.cheng@arm.com>
10152 PR tree-optimization/81799
10153 * tree-loop-distribution.c (version_loop_by_alias_check): Force
10154 cond_expr to simple gimple operand.
10156 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
10158 PR middle-end/46932
10159 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
10161 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
10165 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
10166 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
10167 TARGET_GASISR_PROLOGUES.
10168 * config/avr/avr.c (avr_option_override): Same.
10169 (avr_pass_pre_proep::execute): Same.
10171 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
10174 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
10175 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
10177 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
10179 * config/i386/i386.md (*load_tp_<mode>): Redefine as
10180 define_insn_and_split. Split to a memory load from 0 in
10181 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
10182 using PTR mode iterator.
10183 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
10184 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
10185 (*add_tp_<mode>): Redefine as define_insn_and_split.
10186 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
10187 address space. Merge with *add_tp_x32 using PTR mode iterator.
10188 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
10189 Split to an add with a memory load from 0 in
10190 DEFAULT_TLS_SEG_REG address space.
10192 2017-08-12 Andrew Pinski <apinski@cavium.com>
10194 * config/aarch64/aarch64-option-extensions.def (rdma):
10195 Fix feature string to what Linux prints out in /proc/cpuinfo.
10197 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
10200 * dwarf2out.c (modified_type_die): For C typedef types that have
10201 an ultimate origin, process the ultimate origin instead of the
10203 (gen_typedef_die): Assert that input DECLs have no ultimate
10205 (gen_type_die_with_usage): For typedef variants that have an
10206 ultimate origin, just call gen_decl_die on the original DECL.
10207 (process_scope_var): Avoid creating DIEs for local typedefs and
10208 concrete static variables.
10210 2017-08-12 Alan Modra <amodra@gmail.com>
10214 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
10215 match gnu-user.h startfile.
10216 (ENDFILE_LINUX_SPEC): Similarly.
10218 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
10221 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
10223 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
10225 2017-08-11 Tamar Christina <tamar.christina@arm.com>
10226 * config/aarch64/aarch64.md (mov<mode>): Change.
10227 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
10228 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
10229 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
10231 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
10233 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
10234 for storage order barriers.
10236 2017-08-11 Martin Liska <mliska@suse.cz>
10239 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
10240 variables of void type.
10242 2017-08-11 Martin Liska <mliska@suse.cz>
10244 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
10245 TARGET_SUPPORTS_ALIASES.
10246 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
10247 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
10248 (optimize_weakref): Likewise.
10249 * symtab.c (symtab_node::noninterposable_alias): Likewise.
10250 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
10251 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
10253 2017-08-11 Martin Liska <mliska@suse.cz>
10256 * config/i386/i386.c (make_resolver_func): Do complete
10257 refactoring of the function.
10259 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
10262 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
10263 * config/i386/i386.c (ix86_stack_protect_guard): Use
10264 ix86_stack_protect_guard_symbol_str to generate varible declaration.
10265 * doc/invoke.texi (x86 Options): Document
10266 -mstack-protector-guard-symbol= option.
10268 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
10270 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
10271 * config/i386/i386.c (ix86_split_stack_guard): New function.
10272 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
10273 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
10274 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
10275 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
10276 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
10277 (split_stack_space_check): Call ix86_split_stack_guard.
10279 2017-08-10 Martin Sebor <msebor@redhat.com>
10281 * print-tree.c (print_node): Print location using the established
10283 Replace spaces with colons.
10284 (debug_raw, debug): Ditto.
10286 2017-08-10 Martin Sebor <msebor@redhat.com>
10289 * pretty-print.c (pp_format): Correct the handling of %s precision.
10291 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
10294 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
10296 (ix86_finalize_stack_frame_flags): This. Also clear
10297 frame_pointer_needed if -fno-omit-frame-pointer is used without
10299 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
10300 with ix86_finalize_stack_frame_flags.
10301 (ix86_expand_epilogue): Likewise.
10302 (ix86_expand_split_stack_prologue): Likewise.
10303 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
10305 2017-08-10 Martin Liska <mliska@suse.cz>
10308 * c-attribs.c (handle_target_attribute):
10309 Report warning for an empty string argument of target attribute.
10311 2017-08-09 Jakub Jelinek <jakub@redhat.com>
10314 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
10316 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
10317 or DECL_NONLOCAL labels.
10318 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
10319 or DECL_NONLOCAL labels here.
10321 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
10323 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
10324 to indicate when early gimple folding has been disabled.
10325 (rs6000_gimple_fold_builtin): Add debug content.
10326 (rs6000_invalid_builtin): Fix whitespace.
10327 (rs6000_expand_builtin): Fix whitespace.
10328 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
10330 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
10333 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
10334 SAVE_MULTIPLE if not all the registers that saves, should be saved.
10336 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
10338 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
10339 (qdf24xx): Likewise.
10340 * config/aarch64/aarch64.md: Include falkor.md.
10341 * config/aarch64/falkor.md: New.
10343 2017-08-09 Marek Polacek <polacek@redhat.com>
10346 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
10347 * diagnostic.c (emit_diagnostic): Add a comment.
10348 (emit_diagnostic_valist): New function.
10350 2017-08-09 Marek Polacek <polacek@redhat.com>
10353 * input.c (make_location): New overload.
10354 * input.h (make_location): Declare.
10356 2017-08-08 Alan Modra <amodra@gmail.com>
10357 H.J. Lu <hongjiu.lu@intel.com>
10360 * gcc.c (NO_PIE_SPEC): Delete.
10361 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
10363 (LINK_PIE_SPEC): ..to here.
10364 (LINK_COMMAND_SPEC): Support -no-pie.
10365 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
10366 chain of crtbegin*.o selection, update for PIE_SPEC changes and
10368 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
10369 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
10370 (ENDFILE_CRTEND_SPEC): Similarly.
10372 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
10375 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
10376 (mstack-protector-guard-offset=): Ditto.
10377 * config/i386/i386.c (ix86_option_override): Handle
10378 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
10380 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
10381 ix86_stack_protect_guard_offset variables.
10382 (TARGET_STACK_PROTECT_GUARD): Always define.
10383 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
10384 and -mstack-protector-guard-offset= options.
10386 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10388 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
10389 boundary case for the last candidate.
10391 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10393 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
10394 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
10396 2017-08-08 Tamar Christina <tamar.christina@arm.com>
10398 PR middle-end/19706
10399 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
10400 * config/aarch64/aarch64-builtins.c
10401 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
10402 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
10403 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
10405 2017-08-08 Tamar Christina <tamar.christina@arm.com>
10406 Andrew Pinski <pinskia@gmail.com>
10408 PR middle-end/19706
10409 * internal-fn.def (XORSIGN): New.
10410 * optabs.def (xorsign_optab): New.
10411 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
10412 (convert_expand_mult_copysign): New.
10413 (pass_optimize_widening_mul::execute): Call
10414 convert_expand_mult_copysign.
10416 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10418 PR tree-optimization/81354
10419 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
10420 Insert on edges rather than explicitly creating landing pads.
10421 (analyze_candidates_and_replace): Commit edge inserts.
10423 2017-08-08 Richard Biener <rguenther@suse.de>
10425 PR middle-end/81719
10426 * tree-ssa-loop-niter.c: Include tree-dfa.h.
10427 (expand_simple_operations): Also look through ADDR_EXPRs with
10428 MEM_REF bases treating them as POINTER_PLUS_EXPR.
10430 2017-08-08 Richard Biener <rguenther@suse.de>
10432 PR tree-optimization/81723
10433 * tree-vect-slp.c (struct bst_traits): New hash traits.
10434 (bst_fail): New global.
10435 (vect_build_slp_tree_2): New worker, split out from ...
10436 (vect_build_slp_tree): ... this now wrapping it with using
10437 bst_fail set to cache SLP tree build fails. Properly handle
10439 (vect_analyze_slp_instance): Allocate and free bst_fail.
10441 2017-08-08 Martin Liska <mliska@suse.cz>
10444 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
10445 LABEL_DECLs that can be from a different function.
10447 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10449 PR tree-optimization/81744
10450 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
10451 loop's number of iterations.
10453 2017-08-08 Martin Liska <mliska@suse.cz>
10455 * asan.c: Include header files.
10456 * attribs.c (build_decl_attribute_variant): New function moved
10458 (build_type_attribute_qual_variant): Likewise.
10459 (cmp_attrib_identifiers): Likewise.
10460 (simple_cst_list_equal): Likewise.
10461 (omp_declare_simd_clauses_equal): Likewise.
10462 (attribute_value_equal): Likewise.
10463 (comp_type_attributes): Likewise.
10464 (build_type_attribute_variant): Likewise.
10465 (lookup_ident_attribute): Likewise.
10466 (remove_attribute): Likewise.
10467 (merge_attributes): Likewise.
10468 (merge_type_attributes): Likewise.
10469 (merge_decl_attributes): Likewise.
10470 (merge_dllimport_decl_attributes): Likewise.
10471 (handle_dll_attribute): Likewise.
10472 (attribute_list_equal): Likewise.
10473 (attribute_list_contained): Likewise.
10474 * attribs.h (lookup_attribute): New function moved from tree.[ch].
10475 (lookup_attribute_by_prefix): Likewise.
10476 * bb-reorder.c: Include header files.
10477 * builtins.c: Likewise.
10478 * calls.c: Likewise.
10479 * cfgexpand.c: Likewise.
10480 * cgraph.c: Likewise.
10481 * cgraphunit.c: Likewise.
10482 * convert.c: Likewise.
10483 * dwarf2out.c: Likewise.
10484 * final.c: Likewise.
10485 * fold-const.c: Likewise.
10486 * function.c: Likewise.
10487 * gimple-expr.c: Likewise.
10488 * gimple-fold.c: Likewise.
10489 * gimple-pretty-print.c: Likewise.
10490 * gimple.c: Likewise.
10491 * gimplify.c: Likewise.
10492 * hsa-common.c: Likewise.
10493 * hsa-gen.c: Likewise.
10494 * internal-fn.c: Likewise.
10495 * ipa-chkp.c: Likewise.
10496 * ipa-cp.c: Likewise.
10497 * ipa-devirt.c: Likewise.
10498 * ipa-fnsummary.c: Likewise.
10499 * ipa-inline.c: Likewise.
10500 * ipa-visibility.c: Likewise.
10502 * lto-cgraph.c: Likewise.
10503 * omp-expand.c: Likewise.
10504 * omp-general.c: Likewise.
10505 * omp-low.c: Likewise.
10506 * omp-offload.c: Likewise.
10507 * omp-simd-clone.c: Likewise.
10508 * opts-global.c: Likewise.
10509 * passes.c: Likewise.
10510 * predict.c: Likewise.
10511 * sancov.c: Likewise.
10512 * sanopt.c: Likewise.
10513 * symtab.c: Likewise.
10514 * toplev.c: Likewise.
10515 * trans-mem.c: Likewise.
10516 * tree-chkp.c: Likewise.
10517 * tree-eh.c: Likewise.
10518 * tree-into-ssa.c: Likewise.
10519 * tree-object-size.c: Likewise.
10520 * tree-parloops.c: Likewise.
10521 * tree-profile.c: Likewise.
10522 * tree-ssa-ccp.c: Likewise.
10523 * tree-ssa-live.c: Likewise.
10524 * tree-ssa-loop.c: Likewise.
10525 * tree-ssa-sccvn.c: Likewise.
10526 * tree-ssa-structalias.c: Likewise.
10527 * tree-ssa.c: Likewise.
10528 * tree-streamer-in.c: Likewise.
10529 * tree-vectorizer.c: Likewise.
10530 * tree-vrp.c: Likewise.
10531 * tsan.c: Likewise.
10532 * ubsan.c: Likewise.
10533 * varasm.c: Likewise.
10534 * varpool.c: Likewise.
10535 * tree.c: Remove functions moved to attribs.[ch].
10536 * tree.h: Likewise.
10537 * config/aarch64/aarch64.c: Add attrs.h header file.
10538 * config/alpha/alpha.c: Likewise.
10539 * config/arc/arc.c: Likewise.
10540 * config/arm/arm.c: Likewise.
10541 * config/avr/avr.c: Likewise.
10542 * config/bfin/bfin.c: Likewise.
10543 * config/c6x/c6x.c: Likewise.
10544 * config/cr16/cr16.c: Likewise.
10545 * config/cris/cris.c: Likewise.
10546 * config/darwin.c: Likewise.
10547 * config/epiphany/epiphany.c: Likewise.
10548 * config/fr30/fr30.c: Likewise.
10549 * config/frv/frv.c: Likewise.
10550 * config/ft32/ft32.c: Likewise.
10551 * config/h8300/h8300.c: Likewise.
10552 * config/i386/winnt.c: Likewise.
10553 * config/ia64/ia64.c: Likewise.
10554 * config/iq2000/iq2000.c: Likewise.
10555 * config/lm32/lm32.c: Likewise.
10556 * config/m32c/m32c.c: Likewise.
10557 * config/m32r/m32r.c: Likewise.
10558 * config/m68k/m68k.c: Likewise.
10559 * config/mcore/mcore.c: Likewise.
10560 * config/microblaze/microblaze.c: Likewise.
10561 * config/mips/mips.c: Likewise.
10562 * config/mmix/mmix.c: Likewise.
10563 * config/mn10300/mn10300.c: Likewise.
10564 * config/moxie/moxie.c: Likewise.
10565 * config/msp430/msp430.c: Likewise.
10566 * config/nds32/nds32-isr.c: Likewise.
10567 * config/nds32/nds32.c: Likewise.
10568 * config/nios2/nios2.c: Likewise.
10569 * config/nvptx/nvptx.c: Likewise.
10570 * config/pa/pa.c: Likewise.
10571 * config/pdp11/pdp11.c: Likewise.
10572 * config/powerpcspe/powerpcspe.c: Likewise.
10573 * config/riscv/riscv.c: Likewise.
10574 * config/rl78/rl78.c: Likewise.
10575 * config/rx/rx.c: Likewise.
10576 * config/s390/s390.c: Likewise.
10577 * config/sh/sh.c: Likewise.
10578 * config/sol2.c: Likewise.
10579 * config/sparc/sparc.c: Likewise.
10580 * config/spu/spu.c: Likewise.
10581 * config/stormy16/stormy16.c: Likewise.
10582 * config/tilegx/tilegx.c: Likewise.
10583 * config/tilepro/tilepro.c: Likewise.
10584 * config/v850/v850.c: Likewise.
10585 * config/vax/vax.c: Likewise.
10586 * config/visium/visium.c: Likewise.
10587 * config/xtensa/xtensa.c: Likewise.
10589 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10592 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
10593 constraints since the -mupper-regs-* switches have been
10595 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
10596 into a vector from a double word element that was extracted from
10597 another vector, and eliminate extra XXPERMDI instructions.
10598 (vsx_concat_<mode>_2): Likewise.
10599 (vsx_concat_<mode>_3): Likewise.
10600 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
10601 concat to allow optimizing inserts from previous extracts.
10603 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
10605 * config/i386/i386.c (ix86_stack_protect_guard): Generate
10606 memory reference to a SSP offset in TLS address space.
10607 (ix86_print_operand) <case '@'>: Remove.
10608 (ix86_print_operand_punct_valid_p): Remove '@' code.
10609 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
10610 UNSPEC_SP_TLS_TEST.
10611 (stack_tls_protect_set_<mode>): Remove.
10612 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
10613 (stack_tls_protect_test_<mode>): Remove.
10614 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
10616 2017-08-07 Olivier Hainque <hainque@adacore.com>
10619 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
10621 2017-08-07 Douglas Rupp <rupp@adacore.com>
10623 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
10624 variable was referenced as multidir in command.
10626 2017-08-07 Jakub Jelinek <jakub@redhat.com>
10629 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
10632 2017-08-07 Martin Liska <mliska@suse.cz>
10634 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
10635 * config/rl78/rl78.c: Add include of attribs.h.
10636 * config/sh/sh.c: Likewise.
10637 * config/v850/v850.c: Likewise.
10639 2017-08-07 Tom de Vries <tom@codesourcery.com>
10641 PR middle-end/78266
10642 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
10644 2017-08-07 Martin Liska <mliska@suse.cz>
10646 * config/mips/mips.c: Include attribs.h.
10648 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
10651 * doc/invoke.texi: Document change in behvaior for -Ofast for
10654 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
10656 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
10658 (aarch64_pop_regs): Likewise.
10659 (aarch64_gen_load_pair): Likewise.
10660 (aarch64_save_callee_saves): Likewise.
10661 (aarch64_restore_callee_saves): Likewise.
10663 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10665 * config/i386/i386.c: Revert the last change.
10667 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10670 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
10672 (ix86_finalize_stack_frame_flags): This. Also clear
10673 frame_pointer_needed if -fno-omit-frame-pointer is used without
10675 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
10676 with ix86_finalize_stack_frame_flags.
10677 (ix86_expand_epilogue): Likewise.
10678 (ix86_expand_split_stack_prologue): Likewise.
10680 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10683 * config/i386/i386.c (get_builtin_code_for_version): Set priority
10684 to P_AES for Westmere.
10686 2017-08-07 Jonathan Yong <10walls@gmail.com>
10688 * config/i386/mingw.opt (fset-stack-executable): Removed.
10689 * config/i386/cygming.opt (fset-stack-executable): Moved
10691 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
10693 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
10695 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
10697 2017-08-07 Marek Polacek <polacek@redhat.com>
10699 PR middle-end/81737
10700 * fold-const.c (fold_indirect_ref_1): Check type_domain.
10702 2017-08-07 Martin Liska <mliska@suse.cz>
10704 * attribs.h (canonicalize_attr_name): New function.
10705 (cmp_attribs): Move from c-format.c and adjusted.
10706 (is_attribute_p): Moved from tree.h.
10707 * tree-inline.c: Add new includes.
10708 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
10709 (private_is_attribute_p): Remove.
10710 (private_lookup_attribute): Likewise.
10711 (private_lookup_attribute_by_prefix): Simplify.
10712 (remove_attribute): Use is_attribute_p.
10713 * tree.h: Remove removed declarations.
10715 2017-08-07 Jakub Jelinek <jakub@redhat.com>
10717 PR middle-end/81698
10718 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
10719 instead of computing it in the function. Formatting fix.
10720 (expand_case): Don't rely on default_edge being the first edge,
10721 clear it if removing it, pass default_edge to
10722 emit_case_dispatch_table.
10723 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
10726 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
10728 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
10729 insn in the function, emit NOP after the insn.
10731 2017-08-06 Tom de Vries <tom@codesourcery.com>
10733 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
10736 2017-08-06 Tom de Vries <tom@codesourcery.com>
10738 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
10741 2017-08-04 Yury Gribov <tetra2005@gmail.com>
10743 PR tree-optimization/57371
10744 * match.pd: New pattern.
10746 2017-08-04 Marek Polacek <polacek@redhat.com>
10748 PR middle-end/81695
10749 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
10750 perform the computation in offset_int.
10752 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10754 PR tree-optimization/81136
10755 * tree-vectorizer.h: Include tree-hash-traits.h.
10756 (vec_base_alignments): New typedef.
10757 (vec_info): Add a base_alignments field.
10758 (vect_record_base_alignments): Declare.
10759 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
10761 (DR_IS_CONDITIONAL_IN_STMT): New macro.
10762 (create_data_ref): Add an is_conditional_in_stmt argument.
10763 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
10764 the is_conditional_in_stmt field.
10765 (data_ref_loc): Add an is_conditional_in_stmt field.
10766 (get_references_in_stmt): Set the is_conditional_in_stmt field.
10767 (find_data_references_in_stmt): Update call to create_data_ref.
10768 (graphite_find_data_references_in_stmt): Likewise.
10769 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
10770 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
10771 (vect_record_base_alignment): New function.
10772 (vect_record_base_alignments): Likewise.
10773 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
10774 for nested statements even if we fail to compute a misalignment.
10775 Use pooled base alignments for unconditional references.
10776 (vect_find_same_alignment_drs): Compare base addresses instead
10778 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
10779 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
10781 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10783 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
10784 Add an explicit name for the enum. Use auto_vec for slp_instances
10785 and grouped_stores.
10786 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
10788 (_bb_vec_info): Add a constructor and destructor.
10789 (vinfo_for_stmt): Return NULL for uids of -1 as well.
10790 (destroy_loop_vec_info): Delete.
10791 (vect_destroy_datarefs): Likewise.
10792 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
10793 (vec_info::vec_info): New function.
10794 (vec_info::~vec_info): Likewise.
10795 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
10796 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
10797 destroy_loop_vec_info.
10798 * tree-vect-loop.c (new_loop_vec_info): Replace with...
10799 (_loop_vec_info::_loop_vec_info): ...this.
10800 (destroy_loop_vec_info): Replace with...
10801 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
10802 the stmt_vec_infos. Leave handling of vec_info information to its
10803 destructor. Remove explicit vector releases.
10804 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
10805 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
10806 * tree-vect-slp.c (new_bb_vec_info): Replace with...
10807 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
10808 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
10809 (destroy_bb_vec_info): Replace with...
10810 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
10811 information to its destructor.
10812 (vect_slp_analyze_bb_1): Use new and delete instead of
10813 new_bb_vec_info and destroy_bb_vec_info.
10814 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
10817 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
10819 * tree-data-ref.h (subscript): Add access_fn field.
10820 (data_dependence_relation): Add could_be_independent_p.
10821 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
10822 (same_access_functions): Move to tree-data-ref.c.
10823 * tree-data-ref.c (ref_contains_union_access_p): New function.
10824 (access_fn_component_p): Likewise.
10825 (access_fn_components_comparable_p): Likewise.
10826 (dr_analyze_indices): Add a reference to access_fn_component_p.
10827 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
10829 (constant_access_functions): Likewise.
10830 (add_other_self_distances): Likewise.
10831 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
10832 (initialize_data_dependence_relation): Use XCNEW and remove
10833 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
10834 of access functions that have the same type. Allow the
10835 subsequence to end with different bases in some circumstances.
10836 Record the chosen access functions in SUB_ACCESS_FN.
10837 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
10838 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
10839 (subscript_dependence_tester_1): Likewise dra and drb.
10840 (build_classic_dist_vector): Update calls accordingly.
10841 (subscript_dependence_tester): Likewise.
10842 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
10843 DDR_COULD_BE_INDEPENDENT_P.
10844 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
10845 comp_alias_ddrs instead of may_alias_ddrs.
10846 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
10848 (vect_analyze_data_ref_dependence): Use it if
10849 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
10850 distance vectors if that fails.
10851 (dependence_distance_ge_vf): New function.
10852 (vect_prune_runtime_alias_test_list): Use it. Don't clear
10853 LOOP_VINFO_MAY_ALIAS_DDRS.
10855 2017-08-04 Richard Biener <rguenther@suse.de>
10857 PR middle-end/81705
10858 * fold-const.c (fold_binary_loc): Properly restrict
10859 minus_var0 && minus_var1 case when associating undefined overflow
10862 2017-08-04 Tom de Vries <tom@codesourcery.com>
10864 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
10866 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10868 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10869 Don't start diagnostic messages with a capital letter.
10870 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10872 (rs6000_invalid_builtin): Likewise.
10873 (rs6000_trampoline_init): Likewise.
10875 2017-08-03 Jakub Jelinek <jakub@redhat.com>
10878 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
10879 after setting changeable df flags.
10881 2017-08-03 Richard Biener <rguenther@suse.de>
10883 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
10884 up if the use is in USE - X.
10886 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
10888 * toplev.c (dumpfile.h): New include.
10889 (internal_error_reentered): New static function. Use it...
10890 (internal_error_function): ...here to handle reentered internal_error.
10892 2017-08-03 Richard Biener <rguenther@suse.de>
10894 PR middle-end/81148
10895 * fold-const.c (split_tree): Add minus_var and minus_con
10896 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
10897 here but always use minus_*.
10898 (associate_trees): Assert we never associate with MINUS_EXPR
10899 and NULL first operand. Do not recurse for PLUS_EXPR operands
10900 when associating as MINUS_EXPR either.
10901 (fold_binary_loc): Track minus_var and minus_con.
10903 2017-08-03 Tom de Vries <tom@codesourcery.com>
10906 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
10907 ACCEL_COMPILER, apply finish_options on
10908 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
10910 2017-08-03 Tom de Vries <tom@codesourcery.com>
10913 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
10914 function_entry_patch_area_size > 0.
10916 2017-08-03 Jakub Jelinek <jakub@redhat.com>
10919 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
10920 instead of 10??LU, perform unit multiplication in wide_int,
10921 don't change alloc_object_size_limit if the limit is larger
10924 PR tree-optimization/81655
10925 PR tree-optimization/81588
10926 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
10927 the case when ranges[i].low and high are 1 for unsigned type with
10930 PR middle-end/81052
10931 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
10932 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
10934 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10936 * tree-vrp.h: Add include guard.
10938 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
10941 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
10942 (ud2): New insn pattern.
10943 * config/i386/i386.c (ix86_expand_epilogue):
10944 For naked functions, generate ud2 instead of trap insn.
10946 2017-08-02 Marek Polacek <polacek@redhat.com>
10949 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
10951 2017-08-02 Tom de Vries <tom@codesourcery.com>
10952 Cesar Philippidis <cesar@codesourcery.com>
10954 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
10955 Add missing edge probabilities.
10957 2017-08-02 Tamar Christina <tamar.christina@arm.com>
10959 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
10960 Correct endianness.
10962 2017-08-02 Jakub Jelinek <jakub@redhat.com>
10964 PR middle-end/79499
10965 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
10966 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
10967 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
10968 prologue_seq sequences - if any.
10970 2017-08-02 Richard Biener <rguenther@suse.de>
10972 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
10973 via vectors if supported, integer extracts via punning if supported
10974 or otherwise vector extracts.
10976 2017-08-02 Richard Biener <rguenther@suse.de>
10978 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
10980 (bitmap_insert_into_set): ... this.
10982 2017-08-02 Richard Biener <rguenther@suse.de>
10984 PR tree-optimization/81633
10986 2015-08-17 Alan Hayward <alan.hayward@arm.com>
10988 PR tree-optimization/71752
10989 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
10991 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
10993 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
10994 (machine_function::call_ms2sysv_pad_out): Remove field.
10995 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
10996 (ix86_compute_frame_layout): Likewise.
10998 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
11001 * config/i386/i386.c (ix86_set_func_type): Disallow naked
11002 attribute with interrupt attribute.
11004 2017-08-01 Andrew Pinski <apinski@cavium.com>
11006 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
11007 BIT_INSERT_EXPR's operand 1
11008 to see if the types precision matches.
11010 2017-08-01 Martin Liska <mliska@suse.cz>
11012 PR middle-end/70140
11013 * builtins.c (expand_builtin_memcpy_args): Remove.
11014 (expand_builtin_memcpy): Call newly added function
11015 expand_builtin_memory_copy_args.
11016 (expand_builtin_memcpy_with_bounds): Likewise.
11017 (expand_builtin_mempcpy): Remove last argument.
11018 (expand_builtin_mempcpy_with_bounds): Likewise.
11019 (expand_builtin_memory_copy_args): New function created from
11020 expand_builtin_mempcpy_args with small modifications.
11021 (expand_builtin_mempcpy_args): Remove.
11022 (expand_builtin_stpcpy): Remove unused argument.
11023 (expand_builtin): Likewise.
11024 (expand_builtin_with_bounds): Likewise.
11026 2017-08-01 Martin Liska <mliska@suse.cz>
11029 Make mempcpy more optimal (PR middle-end/70140).
11031 2017-08-01 Jakub Jelinek <jakub@redhat.com>
11034 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
11035 __builtin_vec_cmpne verify both arguments are compatible vectors
11036 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
11037 verify arg1_type is a pointer or array type. For __builtin_vec_st,
11038 move computation of aligned to after checking the argument types.
11042 * config/rs6000/vsx.md (vextract_fp_from_shorth,
11043 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
11046 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
11047 Doug Rupp <rupp@adacore.com>
11048 Olivier Hainque <hainque@adacore.com>
11050 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
11051 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
11053 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
11054 for TARGET_OS_CPP_BUILTIN.
11055 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
11056 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
11058 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
11059 passing required abi options to the assembler for EABI configurations.
11060 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
11061 of .text.hot and .text.unlikely sections for kernel modules when
11062 using ARM style exceptions.
11063 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
11064 options. Add EXTRA_CC1_SPEC.
11065 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
11067 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
11069 (ARM_TARGET2_DWARF_FORMAT): Define.
11070 * config/arm/t-vxworks: Adjust multilib control to removal of the
11071 Diab command line options.
11073 2017-08-01 Martin Liska <mliska@suse.cz>
11075 PR gcov-profile/81561
11076 * gcov.c (unblock): Make unblocking safe as we need to preserve
11077 index correspondence of blocks and block_lists.
11079 2017-08-01 Richard Biener <rguenther@suse.de>
11081 PR tree-optimization/81181
11082 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
11083 (compute_antic): ... end of iteration here.
11085 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
11087 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
11088 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
11089 (ftree-slp-vectorize): Likewise.
11090 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
11091 can no longer be set independent of flag_tree_loop_vectorize.
11092 * omp-general.c (emp_max_vf): Likewise.
11093 * opts.c (enable_fdo_optimizations): Remove references to
11094 flag_tree_vectorize, these are now implicit.
11095 (common_handle_option): Remove handling for OPT_ftree_vectorize,
11096 and leave it for the options machinery.
11098 2017-08-01 Martin Liska <mliska@suse.cz>
11100 PR middle-end/70140
11101 * builtins.c (expand_builtin_memcpy_args): Remove.
11102 (expand_builtin_memcpy): Call newly added function
11103 expand_builtin_memory_copy_args.
11104 (expand_builtin_memcpy_with_bounds): Likewise.
11105 (expand_builtin_mempcpy): Remove last argument.
11106 (expand_builtin_mempcpy_with_bounds): Likewise.
11107 (expand_builtin_memory_copy_args): New function created from
11108 expand_builtin_mempcpy_args with small modifications.
11109 (expand_builtin_mempcpy_args): Remove.
11110 (expand_builtin_stpcpy): Remove unused argument.
11111 (expand_builtin): Likewise.
11112 (expand_builtin_with_bounds): Likewise.
11114 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
11117 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
11118 print "ds:" only for immediates in generic address space.
11120 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
11123 * config/i386/i386.c (ix86_funciton_naked): New prototype.
11124 (ix86_function_ok_for_sibcall): Return false for naked functions.
11126 2017-08-01 Richard Biener <rguenther@suse.de>
11128 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
11129 (compute_antic): Seed worklist with exit block predecessors.
11130 * cfganal.c (dfs_find_deadend): For a cycle return the source
11131 of the edge closing it.
11133 2017-08-01 Tamar Christina <tamar.christina@arm.com>
11135 * config/aarch64/aarch64.c
11136 (aarch64_can_const_movi_rtx_p): Move 0 check.
11138 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11140 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
11141 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
11143 * match.pd: Ditto in address comparison pattern.
11145 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11147 PR tree-optimization/81627
11148 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
11149 closed ssa form for store-store chain.
11151 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11153 PR tree-optimization/81620
11154 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
11155 for store-store chain.
11157 2017-08-01 Jakub Jelinek <jakub@redhat.com>
11159 PR tree-optimization/81588
11160 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
11161 ranges[i].in_p, invert comparison code ccode. For >/>=,
11162 swap rhs1 and rhs2 and comparison code unconditionally,
11163 for </<= don't do that. Don't swap rhs1/rhs2 again if
11164 ranges[i].in_p, instead invert comparison code ccode if
11165 opcode or oe->rank is BIT_IOR_EXPR.
11168 * optabs.def (vec_extract_optab, vec_init_optab): Change from
11169 a direct optab to conversion optab.
11170 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
11171 with GET_MODE_INNER as last argument instead of optab_handler.
11172 * expmed.c (extract_bit_field_1): Likewise. Use vector from
11173 vector extraction if possible and optab is available.
11174 * expr.c (store_constructor): Use convert_optab_handler instead
11175 of optab_handler. Use vector initialization from smaller
11176 vectors if possible and optab is available.
11177 * tree-vect-stmts.c (vectorizable_load): Likewise.
11178 * doc/md.texi (vec_extract, vec_init): Document that the optabs
11179 now have two modes.
11180 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
11181 of vec_init from half-sized vectors with the same element mode.
11182 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
11183 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
11184 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
11185 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
11186 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
11187 after mode in gen_vec_extract* calls.
11188 (vec_extract<mode>): Renamed to ...
11189 (vec_extract<mode><ssescalarmodelower>): ... this.
11190 (vec_extract<mode><ssehalfvecmodelower>): New expander.
11191 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
11192 element mode after mode in gen_vec_init* calls.
11193 (VEC_INIT_HALF_MODE): New mode iterator.
11194 (vec_init<mode>): Renamed to ...
11195 (vec_init<mode><ssescalarmodelower>): ... this.
11196 (vec_init<mode><ssehalfvecmodelower>): New expander.
11197 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
11198 (vec_extractv2sfsf): ... this.
11199 (vec_initv2sf): Renamed to ...
11200 (vec_initv2sfsf): ... this.
11201 (vec_extractv2si): Renamed to ...
11202 (vec_extractv2sisi): ... this.
11203 (vec_initv2si): Renamed to ...
11204 (vec_initv2sisi): ... this.
11205 (vec_extractv4hi): Renamed to ...
11206 (vec_extractv4hihi): ... this.
11207 (vec_initv4hi): Renamed to ...
11208 (vec_initv4hihi): ... this.
11209 (vec_extractv8qi): Renamed to ...
11210 (vec_extractv8qiqi): ... this.
11211 (vec_initv8qi): Renamed to ...
11212 (vec_initv8qiqi): ... this.
11213 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
11214 (vec_init<mode>): Renamed to ...
11215 (vec_init<mode><VEC_base_l>): ... this.
11216 (vec_extract<mode>): Renamed to ...
11217 (vec_extract<mode><VEC_base_l>): ... this.
11218 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
11219 (vec_initv2sfsf): ... this.
11220 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
11221 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
11222 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
11223 element mode after mode in gen_vec_init* calls.
11224 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
11225 (vec_init<mode><Vel>): ... this.
11226 (vec_extract<mode>): Renamed to ...
11227 (vec_extract<mode><Vel>): ... this.
11228 * config/aarch64/iterators.md (Vel): New mode attribute.
11229 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
11230 Add element mode after mode in gen_vec_extract* calls.
11231 * config/s390/vector.md (non_vec_l): New mode attribute.
11232 (vec_extract<mode>): Renamed to ...
11233 (vec_extract<mode><non_vec_l>): ... this.
11234 (vec_init<mode>): Renamed to ...
11235 (vec_init<mode><non_vec_l>): ... this.
11236 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
11237 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
11239 * config/arm/iterators.md (V_elem_l): New mode attribute.
11240 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
11241 (vec_extract<mode><V_elem_l>): ... this.
11242 (vec_extractv2di): Renamed to ...
11243 (vec_extractv2didi): ... this.
11244 (vec_init<mode>): Renamed to ...
11245 (vec_init<mode><V_elem_l>): ... this.
11246 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
11247 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
11248 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
11249 Add element mode after gen_vec_extract* calls.
11250 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
11251 (vec_init<mode><unitmode>): ... this.
11252 (vec_extract<mode>): Renamed to ...
11253 (vec_extract<mode><unitmode>): ... this.
11254 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
11255 (vec_init<mode><unitmode>): ... this.
11256 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
11257 (vec_initv2sfsf): ... this.
11258 (vec_extractv2sf): Renamed to ...
11259 (vec_extractv2sfsf): ... this.
11260 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
11261 Add element mode after gen_vec_extract* calls.
11262 * config/mips/mips.md (unitmode): New mode iterator.
11263 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
11264 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
11265 * config/spu/spu.md (inner_l): New mode attribute.
11266 (vec_init<mode>): Renamed to ...
11267 (vec_init<mode><inner_l>): ... this.
11268 (vec_extract<mode>): Renamed to ...
11269 (vec_extract<mode><inner_l>): ... this.
11270 * config/sparc/sparc.md (veltmode): New mode iterator.
11271 (vec_init<VMALL:mode>): Renamed to ...
11272 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
11273 * config/ia64/vect.md (vec_initv2si): Renamed to ...
11274 (vec_initv2sisi): ... this.
11275 (vec_initv2sf): Renamed to ...
11276 (vec_initv2sfsf): ... this.
11277 (vec_extractv2sf): Renamed to ...
11278 (vec_extractv2sfsf): ... this.
11279 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
11280 (vec_init<mode>): Renamed to ...
11281 (vec_init<mode><VEC_base_l>): ... this.
11282 (vec_extract<mode>): Renamed to ...
11283 (vec_extract<mode><VEC_base_l>): ... this.
11284 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
11285 (vec_initv2sfsf): ... this.
11286 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
11287 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
11288 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
11289 gen_vec_init* calls.
11291 2017-08-01 Richard Biener <rguenther@suse.de>
11293 PR tree-optimization/81297
11294 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
11295 TREE_OVERFLOW from INTEGER_CSTs.
11297 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
11299 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
11301 2017-07-31 Carl Love <cel@us.ibm.com>
11303 * config/rs6000/rs6000-c: Add support for built-in functions
11304 vector signed char vec_xl_be (signed long long, signed char *);
11305 vector unsigned char vec_xl_be (signed long long, unsigned char *);
11306 vector signed int vec_xl_be (signed long long, signed int *);
11307 vector unsigned int vec_xl_be (signed long long, unsigned int *);
11308 vector signed long long vec_xl_be (signed long long, signed long long *);
11309 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
11310 vector signed short vec_xl_be (signed long long, signed short *);
11311 vector unsigned short vec_xl_be (signed long long, unsigned short *);
11312 vector double vec_xl_be (signed long long, double *);
11313 vector float vec_xl_be (signed long long, float *);
11314 * config/rs6000/altivec.h (vec_xl_be): Add #define.
11315 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
11316 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
11318 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
11319 (altivec_expand_builtin): Add switch statement to call
11320 altivec_expand_xl_be for each builtin.
11321 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
11322 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
11323 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
11324 __builtin_vsx_le_be_v16qi.
11325 * doc/extend.texi: Update the built-in documentation file for the
11326 new built-in functions.
11328 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
11331 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
11333 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
11335 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11337 * config.gcc: Add z14.
11338 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
11339 CPU model numbers for z13s and z14.
11340 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
11342 * config/s390/s390-opts.h (enum processor_type): Rename
11343 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
11344 * config/s390/s390.c (processor_table): Add field for CPU name to
11345 be passed to Binutils.
11346 (s390_asm_output_machine_for_arch): Use the new field in
11347 processor_table for Binutils.
11348 (s390_expand_builtin): Replace arch12 with z14.
11349 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
11350 (s390_get_sched_attrmask): Likewise.
11351 (s390_get_unit_mask): Likewise.
11352 * config/s390/s390.opt: Add z14 to processor_type enum.
11354 2017-07-31 Martin Jambor <mjambor@suse.cz>
11357 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
11358 regardless of optimization level.
11360 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
11361 Martin Liska <mliska@suse.cz>
11363 * predict.def: Remove old comment and adjust probability.
11364 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
11365 PREDICT statements.
11367 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
11370 * config/i386/i386.c (ix86_function_naked): New function.
11371 (ix86_can_use_return_insn_p): Return false for naked functions.
11372 (ix86_expand_prologue): Skip prologue for naked functions.
11373 (ix86_expand_epilogue): Skip epilogue for naked functions
11374 and emit trap instruction.
11375 (ix86_warn_func_return): New function.
11376 (ix86_attribute_table): Add "naked" attribute specification.
11377 (TARGET_WARN_FUNC_RETURN): Define.
11378 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
11380 2017-07-31 Martin Liska <mliska@suse.cz>
11382 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
11383 (dump_gimple_bb_header): Always dump BB info.
11384 (pp_cfg_jump): Do not append info about BB when dumping a jump.
11386 2017-07-31 Martin Liska <mliska@suse.cz>
11389 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
11390 also with current_function_decl non-null equality.
11392 2017-07-31 Jakub Jelinek <jakub@redhat.com>
11395 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
11396 change type to the element type, instead add eltype variable and
11397 use it where we are interested in the element type.
11399 PR tree-optimization/81603
11400 * ipa-polymorphic-call.c
11401 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
11402 offset arithmetic in offset_int, bail out if the resulting bit offset
11403 doesn't fit into shwi.
11405 2017-07-31 Martin Liska <mliska@suse.cz>
11407 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
11408 (gimplify_save_expr): Fix comment.
11410 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
11413 * config/i386/i386.c (ix86_function_arg): Update arguments for
11415 (ix86_compute_frame_layout): Set the initial stack offset to
11416 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
11417 INCOMING_FRAME_SP_OFFSET.
11418 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
11419 stack before exception handler returns.
11420 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
11421 the 'ERROR_CODE' for exception handler.
11423 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
11425 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
11426 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
11427 (ASM_OUTPUT_REG_POP): Ditto.
11428 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
11429 instead of asm_fprintf to output pure string.
11431 2017-07-29 Jakub Jelinek <jakub@redhat.com>
11433 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
11434 to imported_module_or_decl hook.
11435 (debug_nothing_tree_tree_tree_bool): Remove.
11436 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
11437 * debug.c (do_nothing_debug_hooks): Use
11438 debug_nothing_tree_tree_tree_bool_bool instead of
11439 debug_nothing_tree_tree_tree_bool.
11440 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
11441 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
11442 * sdbout.c (sdb_debug_hooks): Likewise.
11443 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
11444 (gen_namespace_die): Add DW_AT_export_symbols attribute if
11446 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
11447 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
11448 attribute, don't add anything.
11450 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11452 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
11453 (fold_build2_stat_loc): Likewise.
11454 (fold_build3_stat_loc): Likewise.
11455 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
11456 (fold_build1_loc): Remove macro.
11457 (fold_build2_loc): Likewise.
11458 (fold_build3_loc): Likewise.
11460 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11462 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
11463 (gimple_build_debug_bind_source_stat): Likewise.
11464 * gimple.h (gimple_build_debug_bind): Remove macro.
11465 (gimple_build_debug_bind_source): Likewise.
11467 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11469 * bitmap.c (bitmap_alloc): Adjust.
11470 (bitmap_gc_alloc): Likewise.
11471 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
11473 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11475 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
11476 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
11477 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
11478 (bitmap_gc_alloc_stat): Likewise.
11479 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
11481 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11483 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
11484 * rtl.h (shallow_copy_rtx): Remove macro.
11486 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11488 * emit-rtl.c (gen_raw_REG): Adjust.
11489 * gengenrtl.c (gendef): Likewise.
11490 * rtl.c (rtx_alloc_stat): Remove _stat from name.
11491 * rtl.h (rtx_alloc): Remove macro.
11493 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11495 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
11496 (build_tree_list_stat): Likewise.
11497 * tree.h (build_tree_list): Remove macro.
11498 (build_tree_list_vec): Likewise.
11500 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11502 * tree.c (make_vector_stat): Remove _stat from name.
11503 (build_vector_stat): Likewise.
11504 * tree.h (make_vector_stat): Remove macro.
11505 (build_vector_stat): Likewise.
11507 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11509 * tree.h (build_var_debug_value): Remove prototype.
11511 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11513 * tree.c (tree_cons_stat): Remove _stat from name.
11514 * tree.h (tree_cons): Remove macro.
11516 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11518 * tree.c (build_vl_exp_stat): Remove _stat from name.
11519 * tree.h (build_vl_exp): Remove macro.
11521 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11523 * tree.c (build_decl_stat): Remove _stat from name.
11524 * tree.h (build_decl): Remove macro.
11526 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11528 * gimple.c (gimple_build_with_ops_stat): Adjust.
11529 (gimple_alloc_stat): Remove _stat from name.
11530 * gimple.h (gimple_alloc): Remove macro.
11532 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11534 * tree.c (make_tree_vec_stat): Remove _stat from name.
11535 (grow_tree_vec_stat): Likewise.
11536 * tree.h (make_tree_vec_stat): Adjust prototype.
11537 (grow_tree_vec_stat): Likewise.
11538 (make_tree_vec): Remove macro.
11539 (grow_tree_vec): Likewise.
11541 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11543 * fold-const.c (fold_build1_stat_loc): Adjust.
11544 (fold_build2_stat_loc): Likewise.
11545 (fold_build3_stat_loc): Likewise.
11546 * tree.c (build0_stat): Remove _stat from name.
11547 (build1_stat): Likewise.
11548 (build2_stat): Likewise.
11549 (build3_stat): Likewise.
11550 (build4_stat): Likewise.
11551 (build5_stat): Likewise.
11552 * tree.h (build1_loc): Remove macro, and rename _stat function
11554 (build2_loc): Likewise.
11555 (build3_loc): Likewise.
11556 (build4_loc): Likewise.
11557 (build5_loc): Likewise.
11559 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11561 * tree.c (make_int_cst_stat): Remove _stat from name.
11562 * tree.h (make_int_cst_stat): Adjust prototype.
11563 (make_int_cst): Remove macro.
11565 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11567 * tree.c (make_tre_binfo_stat): Remove _stat from name.
11568 * tree.h (make_tree_binfo_stat): Adjust prototype.
11569 (make_tree_binfo): Remove.
11571 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11573 * tree.c (copy_node_stat): Rename to copy_node.
11574 (build_distinct_type_copy): Adjust.
11575 * tree.h (copy_node_stat): Adjust prototype.
11576 (copy_node): Remove macro.
11578 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11580 * tree.c (make_node_stat): rename to make_node.
11581 (build_tree_list_stat): Adjust.
11582 (build0_stat): Likewise.
11583 (build2_stat): Likewise.
11584 (build3_stat): Likewise.
11585 (build4_stat): Likewise.
11586 (build5_stat): Likewise.
11587 (build_decl_stat): Likewise.
11588 * tree.h (make_node_stat): Adjust prototype.
11589 (make_node): remove macro.
11591 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
11593 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
11594 (PPC_FEATURE2_SCV): Likewise.
11595 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
11597 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11599 * config/aarch64/aarch64.c
11600 (aarch64_internal_mov_immediate): Add new special pattern.
11601 * config/aarch64/aarch64.md (*movdi_aarch64):
11602 Add reg/32bit const mov case.
11604 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11605 Richard Sandiford <richard.sandiford@linaro.org>
11607 * config/aarch64/aarch64.md (mov<mode>): Generalize.
11608 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
11609 Add integer and movi cases.
11610 (movi-split-hf-df-sf split, fp16): New.
11611 (enabled): Added TARGET_FP_F16INST.
11612 * config/aarch64/iterators.md (GPF_HF): New.
11613 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
11615 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11617 * config/aarch64/aarch64.c
11618 (aarch64_simd_container_mode): Add prototype.
11619 (aarch64_expand_mov_immediate): Add HI support.
11620 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
11621 (aarch64_can_const_movi_rtx_p): New.
11622 (aarch64_preferred_reload_class):
11623 Remove restrictions of using FP registers for certain SIMD operations.
11624 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
11625 (aarch64_valid_floating_const): Add integer move validation.
11626 (aarch64_simd_imm_scalar_p): Remove.
11627 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
11628 (aarch64_legitimate_constant_p): Expand list of supported cases.
11629 * config/aarch64/aarch64-protos.h
11630 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
11631 (aarch64_reinterpret_float_as_int): New.
11632 (aarch64_simd_imm_scalar_p): Remove.
11633 * config/aarch64/constraints.md (Uvi): New.
11634 (Dd): Split into Ds and new Dd.
11635 * config/aarch64/aarch64.md (*movsi_aarch64):
11637 (*movdi_aarch64): Add SIMD mov case.
11639 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11641 * tree-predcom.c: (struct chain): Handle store-store chain in which
11642 stores for elimination only store loop invariant values.
11643 (execute_pred_commoning_chain): Ditto.
11644 (prepare_initializers_chain_store_elim): Ditto.
11645 (prepare_finalizers): Ditto.
11646 (is_inv_store_elimination_chain): New function.
11647 (initialize_root_vars_store_elim_1): New function.
11649 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11651 * tree-predcom.c: Revise general description of the pass.
11652 (enum chain_type): New enum type for store elimination.
11653 (struct chain): New field supporting store elimination.
11654 (struct component): Ditto.
11655 (dump_chain): Dump store-stores chain.
11656 (release_chain): Release resources.
11657 (split_data_refs_to_components): Compute and create component
11658 contains only stores for elimination.
11659 (get_chain_last_ref_at): New function.
11660 (make_invariant_chain): Initialization.
11661 (make_rooted_chain): Specify chain type in parameter and record it.
11662 (add_looparound_copies): Skip for store-stores chain.
11663 (determine_roots_comp): Compute type of chain and pass it to
11665 (initialize_root_vars_store_elim_2): New function.
11666 (finalize_eliminated_stores): New function.
11667 (remove_stmt): Handle store for elimination.
11668 (execute_pred_commoning_chain): Execute predictive commoning on
11669 store-store chains.
11670 (determine_unroll_factor): Skip unroll for store-stores chain.
11671 (prepare_initializers_chain_store_elim): New function.
11672 (prepare_initializers_chain): Hanlde store-store chain.
11673 (prepare_finalizers_chain, prepare_finalizers): New function.
11674 (tree_predictive_commoning_loop): Return integer value indicating
11675 if loop is unrolled or lcssa form is corrupted.
11676 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
11678 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11680 * tree-predcom.c (initialize_root): Delete.
11681 (execute_pred_commoning_chain): Initialize root vars and replace
11682 reference of non-combined chain directly, rather than call above
11685 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11687 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
11688 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
11690 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11692 * tree-predcom.c (struct chain): New field init_seq.
11693 (release_chain): Release init_seq.
11694 (prepare_initializers_chain): Record intialization stmts in above
11696 (insert_init_seqs): New function.
11697 (tree_predictive_commoning_loop): Call insert_init_seqs.
11699 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11701 * tree-predcom.c (determine_roots_comp): Skip trivial components.
11703 2017-07-28 Richard Biener <rguenther@suse.de>
11705 * match.pd: Remove superfluous :c.
11706 * genmatch.c (simplify::id): Add member.
11707 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
11709 (current_id): New global.
11710 (dt_node::parent): Move from ...
11711 (dt_operand::parent): ... here. Add for_id member.
11712 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
11713 (decision_tree::find_node): Relax order requirement when
11714 merging DT_TRUE nodes to ones inbetween the current simplify
11715 and the one we try to merge with. Add diagnostic whenever
11716 we need to enforce pattern order by not merging.
11717 (decision_tree::insert): Set current_id.
11718 (decision_tree::print_node): Dump parent node and for_id.
11719 (parser::last_id): Add member.
11720 (parser::push_simplify): Assign unique id.
11721 (parser::parser): Initialize last_id.
11723 2017-07-28 Martin Liska <mliska@suse.cz>
11726 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
11727 gimple_build_debug_bind.
11729 2017-07-28 Richard Biener <rguenther@suse.de>
11731 PR tree-optimization/81502
11732 * match.pd: Add pattern combining BIT_INSERT_EXPR with
11734 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
11736 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
11737 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
11738 for BIT_FIELD_REF args.
11739 * fold-const.c (make_bit_field_ref): Likewise.
11740 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
11742 2017-07-28 Jakub Jelinek <jakub@redhat.com>
11745 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
11746 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
11747 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
11748 Or it into SANITIZER_UNDEFINED.
11749 * ubsan.c: Include gimple-fold.h and varasm.h.
11750 (ubsan_expand_ptr_ifn): New function.
11751 (instrument_pointer_overflow): New function.
11752 (maybe_instrument_pointer_overflow): New function.
11753 (instrument_object_size): Formatting fix.
11754 (pass_ubsan::execute): Call instrument_pointer_overflow
11755 and maybe_instrument_pointer_overflow.
11756 * internal-fn.c (expand_UBSAN_PTR): New function.
11757 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
11758 * sanitizer.def (__ubsan_handle_pointer_overflow,
11759 __ubsan_handle_pointer_overflow_abort): New builtins.
11760 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
11761 * internal-fn.def (UBSAN_PTR): New internal function.
11762 * opts.c (sanitizer_opts): Add pointer-overflow.
11763 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
11764 * fold-const.c (build_range_check): Compute pointer range check in
11765 integral type if pointer arithmetics would be needed. Formatting
11768 2017-07-28 Martin Liska <mliska@suse.cz>
11771 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
11772 parameters that are of a variable-length.
11774 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
11776 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
11778 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
11779 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
11780 (CRT_CALL_STATIC_FUNCTION): Likewise.
11781 (ASM_DEFAULT_SPEC): New define.
11782 (ASM_SPEC32): Likewise.
11783 (ASM_SPEC64): Likewise.
11784 (ASM_SPEC_COMMON): Likewise.
11785 (ASM_SPEC): Likewise.
11786 (INVALID_64BIT): Likewise.
11787 (LINK_OS_DEFAULT_SPEC): Likewise.
11788 (LINK_OS_SPEC32): Likewise.
11789 (LINK_OS_SPEC64): Likewise.
11790 (POWERPC_LINUX): Likewise.
11791 (PTRDIFF_TYPE): Likewise.
11792 (RESTORE_FP_PREFIX): Likewise.
11793 (RESTORE_FP_SUFFIX): Likewise.
11794 (SAVE_FP_PREFIX): Likewise.
11795 (SAVE_FP_SUFFIX): Likewise.
11796 (SIZE_TYPE): Likewise.
11797 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
11798 (TARGET_64BIT): Likewise.
11799 (TARGET_64BIT): Likewise.
11800 (TARGET_AIX): Likewise.
11801 (WCHAR_TYPE_SIZE): Likewise.
11802 (WCHAR_TYPE): Undefine.
11803 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
11804 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
11805 (CPP_OS_RTEMS_SPEC): Delete.
11806 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
11807 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
11809 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
11811 2017-07-28 Jakub Jelinek <jakub@redhat.com>
11813 PR tree-optimization/81578
11814 * tree-parloops.c (build_new_reduction): Bail out if
11815 reduction_code isn't one of the standard OpenMP reductions.
11816 Move the details printing after that decision.
11818 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11820 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
11821 related to reload_in_progress.
11822 (splat_input_operand): Likewise.
11823 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
11825 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
11827 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
11828 (TARGET_INSTANTIATE_DECLS): Likewise.
11829 (legitimate_indexed_address_p): Delete reload_in_progress code.
11830 (rs6000_debug_legitimate_address_p): Likewise.
11831 (rs6000_eliminate_indexed_memrefs): Likewise.
11832 (rs6000_emit_le_vsx_store): Likewise.
11833 (rs6000_emit_move_si_sf_subreg): Likewise.
11834 (rs6000_emit_move): Likewise.
11835 (register_to_reg_type): Likewise.
11836 (rs6000_pre_atomic_barrier): Likewise.
11837 (rs6000_machopic_legitimize_pic_address): Likewise.
11838 (rs6000_allocate_stack_temp): Likewise.
11839 (rs6000_address_for_fpconvert): Likewise.
11840 (rs6000_address_for_altivec): Likewise.
11841 (rs6000_secondary_memory_needed_rtx): Delete function.
11842 (rs6000_check_sdmode): Likewise.
11843 (rs6000_alloc_sdmode_stack_slot): Likewise.
11844 (rs6000_instantiate_decls): Likewise.
11845 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
11846 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
11847 Delete reload_in_progress.
11848 (*vec_reload_and_plus_<mptrsize>): Likewise.
11849 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
11850 (vsx_div_v2di): Likewise.
11851 (vsx_udiv_v2di): Likewise.
11853 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11855 * config/rs6000/rs6000.opt (mlra): Replace with stub.
11856 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
11857 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
11858 (rs6000_debug_reg_global): Delete print of LRA status.
11859 (rs6000_option_override_internal): Delete dead LRA related code.
11860 (rs6000_lra_p): Delete function.
11861 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
11863 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
11865 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
11866 * config/riscv/rtems.h: New file.
11868 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11869 Sudakshina Das <sudi.das@arm.com>
11871 * config/aarch64/aarch64.md
11872 (define_split for and<mode>3nr_compare): Move
11873 non aarch64_logical_operand to a register.
11874 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
11875 register immediate operand to a register.
11876 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
11878 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11880 PR middle-end/81564
11881 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
11883 2017-07-27 Richard Biener <rguenther@suse.de>
11885 PR tree-optimization/81573
11886 PR tree-optimization/81494
11887 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
11888 multi defuse cycle case.
11890 2017-07-27 Richard Biener <rguenther@suse.de>
11892 PR tree-optimization/81571
11893 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
11896 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
11898 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
11899 earlier and only if MASK_FPU is set. Adjust formatting.
11901 2017-07-27 Martin Liska <mliska@suse.cz>
11903 * opt-functions.awk: Add validation of value of Init.
11904 * optc-gen.awk: Pass new argument.
11906 2017-07-27 Martin Liska <mliska@suse.cz>
11908 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
11909 Fix wrong condition.
11911 2017-07-27 Martin Liska <mliska@suse.cz>
11913 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
11914 BBs and edges seen by autoFDO.
11916 2017-07-27 Richard Biener <rguenther@suse.de>
11918 PR tree-optimization/81502
11919 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
11920 with incompatible but same sized type.
11921 (execute_update_addresses_taken): Likewise.
11923 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
11925 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
11926 flag_tree_loop_vectorize rather than flag_tree_vectorize.
11928 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11931 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
11932 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
11933 Change s_operand to memory_operand.
11935 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
11937 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
11938 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
11939 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
11940 Emit instructions rather than returning an expression. Handle TFmode
11941 and KFmode by casting to TImode.
11942 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
11943 (rs6000_emit_le_vsx_store): Likewise.
11944 * config/rs6000/vsx.md (VSX_TI): New iterator.
11945 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
11946 (*vsx_le_undo_permute_<mode>): Likewise.
11947 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
11948 emit the split sequence.
11949 (*vsx_le_perm_store_<mode>): Likewise.
11951 2017-07-27 Jakub Jelinek <jakub@redhat.com>
11953 PR tree-optimization/81555
11954 PR tree-optimization/81556
11955 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
11956 if true, force CHANGED for the recursive invocation.
11957 (reassociate_bb): Remember original length of ops array, pass
11958 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
11960 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
11961 attributes for noipa attribute. For naked attribute use
11962 lookup_attribute first before lookup_attribute_spec.
11963 * final.c (rest_of_handle_final): Disable IPA RA for functions with
11965 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
11966 for functions with noipa attribute.
11967 (cgraph_externally_visible_p): Return true for functions with noipa
11969 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
11970 for functions with noipa attribute.
11971 * doc/extend.texi: Document noipa function attribute.
11972 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
11973 also for functions with noipa attribute.
11974 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
11976 2017-07-26 Andrew Pinski <apinski@cavium.com>
11978 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
11979 vec_unalign_load_cost and vec_unalign_store_cost.
11981 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11983 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
11984 -mvsx-small-integer option.
11985 (ISA_3_0_MASKS_IEEE): Likewise.
11986 (OTHER_VSX_VECTOR_MASKS): Likewise.
11987 (POWERPC_MASKS): Likewise.
11988 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
11989 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
11990 code, only testing for DImode being allowed in non-VSX floating
11992 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
11993 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
11995 (rs6000_option_override_internal): Delete -mvsx-small-integer.
11996 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
11997 TARGET_P8_VECTOR test.
11998 (rs6000_secondary_reload_simple_move): Likewise.
11999 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
12000 since TARGET_P9_VECTOR was already tested.
12001 (rs6000_opt_masks): Remove -mvsx-small-integer.
12002 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
12003 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
12005 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
12006 test for TARGET_VEXTRACTUB was used, and that uses
12008 (p9 extract splitter): Likewise.
12009 (vsx_extract_<mode>_di_p9): Likewise.
12010 (vsx_extract_<mode>_store_p9): Likewise.
12011 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
12012 for TARGET_P9_VECTOR was used. Delete code that is now dead with
12013 the elimination of TARGET_VSX_SMALL_INTEGER.
12014 (vsx_extract_<mode>_p8): Likewise.
12015 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
12016 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
12017 (vsx_set_<mode>_p9): Likewise.
12018 (vsx_set_v4sf_p9): Likewise.
12019 (vsx_set_v4sf_p9_zero): Likewise.
12020 (vsx_insert_extract_v4sf_p9): Likewise.
12021 (vsx_insert_extract_v4sf_p9_2): Likewise.
12022 * config/rs6000/rs6000.md (sign extend splitter): Change
12023 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
12024 (floatsi<mode>2_lfiwax_mem): Likewise.
12025 (floatunssi<mode>2_lfiwzx_mem): Likewise.
12026 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
12027 since a test for TARGET_P9_VECTOR was used.
12028 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12029 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
12030 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12031 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
12032 TARGET_P8_VECTOR test.
12033 (fix_trunc<mode>si2_stfiwx): Likewise.
12034 (fix_trunc<mode>si2_internal): Likewise.
12035 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
12036 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
12038 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
12039 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
12040 TARGET_P8_VECTOR test.
12041 (fixuns_trunc<mode>si2_stfiwx): Likewise.
12042 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
12043 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
12045 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
12046 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
12047 since a test for TARGET_P9_VECTOR was used.
12048 (splitter for loading small constants): Likewise.
12050 2017-07-26 Andrew Pinski <apinski@cavium.com>
12052 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
12055 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
12058 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
12059 (fp_valid_at): Likewise.
12061 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
12063 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
12064 (qdf24xx_addrcost_table): Likewise.
12065 (cortexa57_tunings): Update to use generic_branch_cost.
12066 (cortexa72_tunings): Likewise.
12067 (cortexa73_tunings): Likewise.
12068 (qdf24xx_tunings): Likewise.
12070 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
12072 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
12073 (thunderx2t99_branch_cost): Likewise.
12074 (cortexa35_tunings): Update to use generic_branch_cost.
12075 (cortexa53_tunings): Likewise.
12076 (cortexa57_tunings): Likewise.
12077 (cortexa72_tunings): Likewise.
12078 (cortexa73_tunings): Likewise.
12079 (thunderx2t99_tunings): Likewise.
12081 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
12083 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
12084 (sparc_option_override): Honour MASK_FSMULD.
12085 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
12086 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
12087 * config/sparc/sparc.opt (mfsmuld): New option.
12088 * doc/invoke.texi (mfsmuld): Document option.
12090 2017-07-26 Marek Polacek <polacek@redhat.com>
12092 PR middle-end/70992
12093 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
12095 2017-07-26 Richard Biener <rguenther@suse.de>
12097 * gimple-match-head.c (do_valueize): Return OP if valueize
12099 (get_def): New helper to get at the def stmt of a SSA name
12100 if valueize allows.
12101 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
12102 do_valueize to get at the def stmt.
12103 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
12105 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
12107 PR middle-end/46932
12108 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
12110 2017-07-26 Martin Liska <mliska@suse.cz>
12113 * function.c (expand_function_start): Make expansion of
12114 nonlocal_goto_save_area after parm_birth_insn.
12116 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
12118 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
12119 from all CPU target flags enable members.
12121 2017-07-26 Richard Biener <rguenther@suse.de>
12123 * genmatch.c (dt_simplify::gen): Make iterator vars const.
12124 (decision_tree::gen): Make 'type' const.
12125 (write_predicate): Likewise.
12127 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
12129 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
12130 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
12131 (rs6000_option_override_internal): Likewise.
12132 (rs6000_expand_vector_set): Likewise.
12133 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
12134 (TARGET_UPPER_REGS_SF): Likewise.
12135 (TARGET_UPPER_REGS_DI): Likewise.
12136 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
12137 (TARGET_DIRECT_MOVE_64BIT): Likewise.
12138 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
12139 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12140 (Splitters for DI constants in Altivec registers): Likewise.
12141 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
12142 (vsx_set_v4sf_p9): Likewise.
12143 (vsx_set_v4sf_p9_zero): Likewise.
12144 (vsx_insert_extract_v4sf_p9): Likewise.
12145 (vsx_insert_extract_v4sf_p9_2): Likewise.
12147 2017-07-25 Carl Love <cel@us.ibm.com>
12149 * doc/extend.texi: Update the built-in documentation file for the
12150 existing built-in functions
12151 vector signed char vec_cnttz (vector signed char);
12152 vector unsigned char vec_cnttz (vector unsigned char);
12153 vector signed short vec_cnttz (vector signed short);
12154 vector unsigned short vec_cnttz (vector unsigned short);
12155 vector signed int vec_cnttz (vector signed int);
12156 vector unsigned int vec_cnttz (vector unsigned int);
12157 vector signed long long vec_cnttz (vector signed long long);
12158 vector unsigned long long vec_cnttz (vector unsigned long long);
12160 2017-07-25 Andrew Pinski <apinski@cavium.com>
12162 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
12163 accesses where the use is for the first operand of a BIT_INSERT.
12165 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
12168 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
12169 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
12171 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
12173 * config/i386/gstabs.h: Delete.
12174 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
12176 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
12178 * config/i386/i386.c (ix86_decompose_address): Do not check for
12179 register RTX when looking at index_reg or base_reg.
12180 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
12182 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
12184 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
12185 to update EH info here.
12187 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12189 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
12191 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12193 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
12195 2017-07-25 Torsten Duwe <duwe@suse.de>
12197 * common.opt: Introduce -fpatchable-function-entry
12198 command line option, and its variables function_entry_patch_area_size
12199 and function_entry_patch_area_start.
12200 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
12201 including a two-value parser.
12202 * target.def (print_patchable_function_entry): New target hook.
12203 * targhooks.h (default_print_patchable_function_entry): New function.
12204 * targhooks.c (default_print_patchable_function_entry): Likewise.
12205 * toplev.c (process_options): Switch off IPA-RA if
12206 patchable function entries are being generated.
12207 * varasm.c (assemble_start_function): Look at the
12208 patchable-function-entry command line switch and current
12209 function attributes and maybe generate NOP instructions by
12210 calling the print_patchable_function_entry hook.
12211 * doc/extend.texi: Document patchable_function_entry attribute.
12212 * doc/invoke.texi: Document -fpatchable_function_entry
12213 command line option.
12214 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
12216 * doc/tm.texi: Re-generate.
12218 2017-07-25 Jakub Jelinek <jakub@redhat.com>
12221 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
12222 TARGET_AVX512DQ rather than TARGET_AVX512BW.
12224 2017-07-25 Tamar Christina <tamar.christina@arm.com>
12226 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
12228 2017-07-25 Richard Biener <rguenther@suse.de>
12230 PR tree-optimization/81455
12231 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
12232 not walk in cycles when looking for guards.
12234 2017-07-25 Richard Biener <rguenther@suse.de>
12236 PR tree-optimization/81529
12237 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
12238 when optimizing backedge uses.
12240 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
12242 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
12244 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
12245 to dl_section_ref. On AIX, append an expression to subtract
12246 the size of the section length to dl_section_ref.
12248 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
12250 * configure.ac: If any of the config.* scripts fail, exit 1.
12251 * configure: Regenerate.
12253 2017-07-25 Richard Biener <rguenther@suse.de>
12255 PR middle-end/81546
12256 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
12257 of immediate uses, be more verbose on errors.
12259 2017-07-25 Richard Biener <rguenther@suse.de>
12261 PR tree-optimization/81510
12262 * tree-vect-loop.c (vect_is_simple_reduction): When the
12263 reduction stmt is not inside the loop bail out.
12265 2017-07-25 Richard Biener <rguenther@suse.de>
12267 PR tree-optimization/81303
12268 * tree-vect-loop-manip.c (vect_loop_versioning): Build
12269 profitability check against LOOP_VINFO_NITERSM1.
12271 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12273 * domwalk.c (cmp_bb_postorder): Simplify.
12274 (sort_bbs_postorder): New function. Use it...
12275 (dom_walker::walk): ...here to optimize common cases.
12277 2017-07-25 Martin Liska <mliska@suse.cz>
12280 * ipa-visibility.c (function_and_variable_visibility): Make the
12281 redirection just on target that supports aliasing.
12282 Fix GNU coding style.
12284 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
12287 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
12289 * config/arm/rtems.h: Likewise.
12290 * config/bfin/rtems.h: Likewise.
12291 * config/i386/rtemself.h: Likewise.
12292 * config/lm32/rtems.h: Likewise.
12293 * config/m32c/rtems.h: Likewise.
12294 * config/m68k/rtemself.h: Likewise.
12295 * config/microblaze/rtems.h: Likewise.
12296 * config/mips/rtems.h: Likewise.
12297 * config/moxie/rtems.h: Likewise.
12298 * config/nios2/rtems.h: Likewise.
12299 * config/powerpcspe/rtems.h: Likewise.
12300 * config/rs6000/rtems.h: Likewise.
12301 * config/rtems.h: Likewise.
12302 * config/sh/rtems.h: Likewise.
12303 * config/sh/rtemself.h: Likewise.
12304 * config/sparc/rtemself.h: Likewise.
12306 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
12309 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
12310 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
12311 * tree-ssa-structalias.c (alias_get_name): Same.
12313 2017-07-25 Bin Cheng <bin.cheng@arm.com>
12316 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
12317 instructions if no du chain is found.
12319 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
12321 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
12323 2017-07-25 Richard Biener <rguenther@suse.de>
12325 PR middle-end/81505
12326 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
12329 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
12331 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
12332 upper-regs options.
12333 (ISA_2_7_MASKS_SERVER): Likewise.
12334 (ISA_3_0_MASKS_IEEE): Likewise.
12335 (OTHER_P8_VECTOR_MASKS): Likewise.
12336 (OTHER_VSX_VECTOR_MASKS): Likewise.
12337 (POWERPC_MASKS): Likewise.
12338 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
12339 duplicate list of options.
12340 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
12341 explicit -mupper-regs options.
12342 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
12343 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
12344 alias for -mupper-regs-df.
12345 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12346 (rs6000_init_hard_regno_mode_ok): Likewise.
12347 (rs6000_option_override_internal): Likewise.
12348 (rs6000_opt_masks): Likewise.
12349 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
12350 options in terms of whether -mvsx or -mpower8-vector was used.
12351 (TARGET_UPPER_REGS_DI): Likewise.
12352 (TARGET_UPPER_REGS_SF): Likewise.
12353 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
12354 -mupper-regs-* options.
12356 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
12358 * passes.c (emergency_dump_function): Print some empty lines and a
12359 header before the RTL dump.
12361 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
12363 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
12365 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
12368 * config/aarch64/aarch64.c (aarch64_classify_symbol):
12369 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
12371 2017-07-24 Carl Love <cel@us.ibm.com>
12373 * config/rs6000/rs6000-c.c: Add support for built-in functions
12374 vector float vec_extract_fp32_from_shorth (vector unsigned short);
12375 vector float vec_extract_fp32_from_shortl (vector unsigned short);
12376 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
12377 vec_extract_fp_from_shortl): Add defines for the two builtins.
12378 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
12379 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
12381 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
12382 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
12383 * doc/extend.texi: Update the built-in documentation file for the
12384 new built-in function.
12386 2017-07-24 Jakub Jelinek <jakub@redhat.com>
12389 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
12391 * doc/generic.texi: Likewise.
12392 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
12393 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
12395 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
12397 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
12398 (aarch64_mls_elt_merge<mode>): Likewise.
12400 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
12402 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
12403 having __cxa_atexit.
12405 2017-07-23 Michael Collison <michael.collison@arm.com>
12407 * config/arm/arm.c (arm_option_override): Deprecate
12408 use of -mstructure-size-boundary.
12409 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
12410 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
12412 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12415 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
12416 Reduce cost estimate for direct moves.
12418 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
12421 * config/i386/i386.c (ix86_option_override_internal): Disable
12422 BMI, BMI2 and TBM instructions for -m16.
12424 2017-07-21 Carl Love <cel@us.ibm.com>
12426 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12427 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12428 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12429 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12430 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12431 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12432 VMULOSW): New enum "unspec" values.
12433 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12434 altivec_vmulosw): New patterns.
12435 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12436 VMULOSW): Add definitions.
12438 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
12440 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
12441 (qdf24xx): Likewise.
12442 * config/aarch64/aarch64-options-extensions.def (rdma); New.
12443 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
12444 (AARCH64_FL_V8_1): Renumber.
12445 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
12446 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
12447 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
12448 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
12449 rdma to feature modifiers list.
12451 2017-07-21 Yury Gribov <tetra2005@gmail.com>
12453 PR middle-end/56727
12454 * ipa-visibility (function_and_variable_visibility): Convert
12455 recursive PLT call to direct call if appropriate.
12457 2017-07-21 Andrew Pinski <apinski@cavium.com>
12459 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
12460 operand 1 to see if the types precision matches.
12461 * fold-const.c (operand_equal_p): Likewise.
12463 2017-07-21 Richard Biener <rguenther@suse.de>
12465 PR tree-optimization/81303
12466 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
12467 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
12468 (vect_peeling_hash_get_lowest_cost): Adjust.
12469 (vect_enhance_data_refs_alignment): Likewise. Use
12470 vect_get_peeling_costs_all_drs to compute the penalty for no
12471 peeling to match up costs.
12473 2017-07-21 Richard Biener <rguenther@suse.de>
12475 PR tree-optimization/81500
12476 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
12477 we didn't identify a reduction path.
12479 2017-07-21 Tom de Vries <tom@codesourcery.com>
12480 Cesar Philippidis <cesar@codesourcery.com>
12482 PR gcov-profile/81442
12483 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
12486 2017-07-21 Tom de Vries <tom@codesourcery.com>
12489 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
12491 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
12492 nvptx_override_options_after_change.
12494 2017-07-21 Ulrich Drepper <drepper@redhat.com>
12496 * dwarf2out.c (output_file_names): Avoid double testing for
12497 dwarf_version >= 5.
12499 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
12501 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
12503 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
12505 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
12507 (try_crossjump_to_edge): Do not punt on partitioned functions.
12509 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
12511 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
12512 Put all BBs reachable only via paths crossing cold region to cold
12514 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
12516 2016-07-21 Richard Biener <rguenther@suse.de>
12518 PR tree-optimization/81303
12519 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
12520 into account prologue and epilogue iterations when raising
12521 min_profitable_iters to sth at least covering one vector iteration.
12523 2017-07-21 Tamar Christina <tamar.christina@arm.com>
12525 * config/arm/arm.c (arm_test_cpu_arch_dat):
12528 2017-07-20 Nathan Sidwell <nathan@acm.org>
12530 Remove TYPE_METHODS.
12531 * tree.h (TYPE_METHODS): Delete.
12532 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
12533 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
12534 (dbxout_type_methods): Scan TYPE_FIELDS.
12535 (dbxout_type): Don't check TYPE_METHODS here.
12536 * function.c (use_register_for_decl): Always ignore register for
12537 class types when not optimizing.
12538 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
12539 * tree.c (free_lang_data_in_type): Stitch out member functions and
12540 templates from TYPE_FIELDS.
12541 (build_distinct_type_copy, verify_type_variant,
12542 verify_type): Member fns are on TYPE_FIELDS.
12543 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
12544 * tree-pretty-print.c (dump_generic_node): Likewise.
12546 2017-07-20 Jakub Jelinek <jakub@redhat.com>
12549 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
12550 V2TImode and V4TImode.
12551 (ix86_expand_vector_extract): Likewise.
12552 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
12553 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
12554 (ssescalarmode): Handle V4TImode and V2TImode.
12555 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
12556 (*vec_extractv2ti, *vec_extractv4ti): New insns.
12557 (VEXTRACTI128_MODE): New mode iterator.
12558 (splitter for *vec_extractv?ti first element): New.
12559 (VEC_INIT_MODE): New mode iterator.
12560 (vec_init<mode>): Consolidate 3 expanders into one using
12561 VEC_INIT_MODE mode iterator.
12563 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
12565 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
12566 non_spilled_static_chain_regno_p.
12568 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
12570 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
12572 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
12574 * bb-reorder.c (connect_traces): Allow copying of blocks within
12577 2017-07-20 Richard Biener <rguenther@suse.de>
12579 * gimple.h (gimple_phi_result): Add gphi * overload.
12580 (gimple_phi_result_ptr): Likewise.
12581 (gimple_phi_arg): Likewise. Adjust index assert to only
12582 allow actual argument accesses rather than all slots available
12584 (gimple_phi_arg_def): Add gphi * overload.
12585 * tree-phinodes.c (make_phi_node): Initialize only actual
12587 (resize_phi_node): Clear memory not covered by old node,
12588 do not initialize excess argument slots.
12589 (reserve_phi_args_for_new_edge): Initialize new argument slot
12592 2017-07-20 Bin Cheng <bin.cheng@arm.com>
12594 PR tree-optimization/81388
12596 2016-07-21 Bin Cheng <bin.cheng@arm.com>
12598 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
12599 by removing computation of may_be_zero.
12601 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12602 Tom de Vries <tom@codesourcery.com>
12604 PR middle-end/81030
12605 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
12606 when gimple level profile disagrees with what RTL expander did.
12608 2017-07-20 Richard Biener <rguenther@suse.de>
12610 PR tree-optimization/61171
12611 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
12612 (vect_analyze_stmt): Add slp instance parameter.
12613 (vectorizable_reduction): Likewise.
12614 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
12615 (vect_is_simple_reduction): Deal with chains not detected
12616 as SLP reduction chain, specifically not properly associated
12617 chains containing a mix of plus/minus.
12618 (get_reduction_op): Remove.
12619 (get_initial_defs_for_reduction): Simplify, pass in whether
12620 this is a reduction chain, pass in the SLP node for the PHIs.
12621 (vect_create_epilog_for_reduction): Get the SLP instance as
12623 (vectorizable_reduction): Get the SLP instance as arg.
12624 During analysis remember the SLP node with the PHIs in the
12625 instance. Simplify getting at the vectorized reduction PHIs.
12626 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
12627 through SLP instance.
12628 (vect_slp_analyze_operations): Likewise.
12629 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
12630 (vect_transform_stmt): Likewise.
12632 2017-07-20 Tom de Vries <tom@codesourcery.com>
12634 PR tree-optimization/81489
12635 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
12636 read of phi arg location to before loop that modifies phi.
12638 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12640 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
12643 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12645 PR middle-end/81331
12646 * except.c (execute): Fix ordering issue.
12648 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
12650 PR rtl-optimization/81423
12651 * combine.c (make_compound_operation_int): Don't try to optimize
12652 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
12654 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
12656 PR rtl-optimization/81423
12657 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
12658 with a constant that is -1 in the truncated to mode.
12660 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12662 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
12663 (determine_unlikely_bbs): ... here.
12664 * predict.h (propagate_unlikely_bbs_forward): Declare.
12665 * cfgexpand.c (pass_expand::execute): Use it.
12666 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
12668 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
12669 propagate_unlikely_bbs_forward.
12671 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12673 PR middle-end/81331
12674 * except.c (maybe_add_nop_after_section_switch): New function.
12677 2017-07-19 Tom de Vries <tom@codesourcery.com>
12679 * gimple.h (gimple_phi_set_arg): Make assert more strict.
12681 2017-07-19 Tom de Vries <tom@codesourcery.com>
12683 * gimple.h (gimple_phi_arg): Make assert more strict.
12685 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
12687 * config.gcc (powerpc*-*-*): Add mmintrin.h.
12688 * config/rs6000/mmintrin.h: New file.
12689 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
12691 2017-07-19 Jakub Jelinek <jakub@redhat.com>
12693 PR tree-optimization/81346
12694 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
12696 2017-07-19 Tom de Vries <tom@codesourcery.com>
12698 * config/nvptx/nvptx.md (VECIM): Add V2DI.
12700 2017-07-19 Tom de Vries <tom@codesourcery.com>
12702 * config/nvptx/nvptx-modes.def: Add V2DImode.
12703 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
12704 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
12705 (nvptx_output_mov_insn): Handle lack of mov.b128.
12706 (nvptx_print_operand): Handle 'H' and 'L' codes.
12707 (nvptx_vector_mode_supported): Allow V2DImode.
12708 (nvptx_preferred_simd_mode): New function.
12709 (nvptx_data_alignment): New function.
12710 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
12711 nvptx_preferred_simd_mode.
12712 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
12714 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
12716 2017-07-19 Tom de Vries <tom@codesourcery.com>
12718 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
12719 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
12720 (nvptx_vector_mode_supported): New function. Allow V2SImode.
12721 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
12722 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
12723 (mov<VECIM>_insn): New define_insn.
12724 (define_expand "mov<VECIM>): New define_expand.
12726 2017-07-19 Tom de Vries <tom@codesourcery.com>
12728 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
12730 2017-07-19 Jakub Jelinek <jakub@redhat.com>
12732 PR tree-optimization/81346
12733 * fold-const.h (fold_div_compare, range_check_type): Declare.
12734 * fold-const.c (range_check_type): New function.
12735 (build_range_check): Use range_check_type.
12736 (fold_div_compare): No longer static, rewritten into
12737 a match.pd helper function.
12738 (fold_comparison): Don't call fold_div_compare here.
12739 * match.pd (X / C1 op C2): New optimization using fold_div_compare
12740 as helper function.
12742 2017-07-19 Nathan Sidwell <nathan@acm.org>
12744 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
12745 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
12746 * tree.c (find_decls_types_r, verify_type): Use
12747 TYPE_{MIN,MAX}_VALUE_RAW.
12748 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
12749 (hash_tree): Likewise.
12750 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
12752 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
12755 2017-07-18 Tom de Vries <tom@codesourcery.com>
12757 PR middle-end/81464
12758 * omp-expand.c (expand_omp_for_static_chunk): Handle
12759 equal-argument loop exit phi.
12761 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
12764 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
12765 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
12766 operand 2 predicate.
12767 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
12768 operand 2 predicate.
12769 (ror,rol -> rorx splitters): Use const_int_operand as
12770 operand 2 predicate.
12772 2017-06-18 Richard Biener <rguenther@suse.de>
12774 PR tree-optimization/81410
12775 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
12776 the gap in the ! slp_perm SLP case after each group.
12778 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12780 PR middle-end/81463
12781 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
12784 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12786 PR middle-end/81462
12787 * predict.c (set_even_probabilities): Cleanup; do not affect
12788 probabilities that are already known.
12789 (combine_predictions_for_bb): Call even when count is set.
12791 2017-07-18 Nathan Sidwell <nathan@acm.org>
12793 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
12796 2017-07-18 Bin Cheng <bin.cheng@arm.com>
12799 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
12800 optimization for loop niter analysis.
12802 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
12805 * config/avr/avr.c (avr_optimize_casesi): Don't use
12806 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
12808 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
12810 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
12811 body_cost_vec from _vect_peel_extended_info.
12812 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
12813 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
12816 2017-07-18 Bin Cheng <bin.cheng@arm.com>
12818 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
12820 2017-07-18 Richard Biener <rguenther@suse.de>
12822 PR tree-optimization/80620
12823 PR tree-optimization/81403
12824 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
12825 info when re-using a VN table entry.
12827 2017-07-18 Richard Biener <rguenther@suse.de>
12829 PR tree-optimization/81418
12830 * tree-vect-loop.c (vectorizable_reduction): Properly compute
12831 vectype_in. Verify that with lane-reducing reduction operations
12832 we have a single def-use cycle.
12834 2017-07-17 Carl Love <cel@us.ibm.com>
12836 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
12838 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12839 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12840 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12841 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12842 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12843 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12844 VMULOSW): New enum "unspec" values.
12845 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
12846 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
12847 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12848 altivec_vmulosw): New patterns.
12849 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12850 VMULOSW): Add definitions.
12851 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
12853 * config/alpha/alpha.c: Include predict.h.
12855 2017-07-17 Yury Gribov <tetra2005@gmail.com>
12857 * tree-vrp.c (compare_assert_loc): Fix comparison function
12858 to return predictable results.
12860 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12862 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
12864 (subdi3): Likewise.
12865 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
12866 * doc/invoke.texi (mexpand-adddi): Update text.
12868 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12870 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
12871 that also clobbers the CC register. The old expand code is moved
12873 (*arc_clzsi2): ... here.
12874 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
12875 the CC register. The old expand code is moved to ...
12876 (arc_ctzsi2): ... here.
12878 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12880 * config/arc/arc.opt (mindexed-loads): Use initial value
12881 TARGET_INDEXED_LOADS_DEFAULT.
12882 (mauto-modify-reg): Use initial value
12883 TARGET_AUTO_MODIFY_REG_DEFAULT.
12884 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
12885 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
12886 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
12887 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
12889 2017-07-17 Martin Liska <mliska@suse.cz>
12892 * opts.c (finish_options): Do not allow -fgnu-tm
12893 w/ -fsanitize={kernel-,}address. Say sorry.
12895 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12898 * tree-loop-distribution.c (classify_partition): Only assert on
12899 numer of iterations.
12900 (merge_dep_scc_partitions): Delete prameter. Update function call.
12901 (distribute_loop): Remove code handling loop with unknown niters.
12902 (pass_loop_distribution::execute): Skip loop with unknown niters.
12904 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12907 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
12908 function sort_partitions_by_post_order.
12910 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12912 PR tree-optimization/81374
12913 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
12914 the max index of basic blocks, rather than number of basic blocks.
12916 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12918 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
12920 (arc_legitimate_pic_operand_p): Likewise.
12921 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
12923 (arc_needs_pcl_p): Likewise.
12924 (arc_legitimate_pc_offset_p): Likewise.
12925 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
12926 function is also used in constrains.md.
12927 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
12928 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
12929 PLUS. Only return true/false in known cases, otherwise assert.
12930 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
12931 is already called in arc_legitimate_constant_p.
12932 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
12934 (LEGITIMATE_PIC_OPERAND_P): Use
12935 arc_raw_symbolic_reference_mentioned_p function.
12936 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
12941 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12942 Andrew Burgess <andrew.burgess@embecosm.com>
12944 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
12945 (arc_return_address_register): New function.
12946 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
12947 (arc_handle_fndecl_attribute): Add naked attribute.
12948 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
12949 (TARGET_WARN_FUNC_RETURN): Likewise.
12950 (arc_allocate_stack_slots_for_args): New function.
12951 (arc_warn_func_return): Likewise.
12952 (machine_function): Change type fn_type.
12953 (arc_compute_function_type): Consider new naked function type,
12954 change function return type.
12955 (arc_must_save_register): Adapt to handle new
12956 arc_compute_function_type's return type.
12957 (arc_expand_prologue): Likewise.
12958 (arc_expand_epilogue): Likewise.
12959 (arc_return_address_regs): Delete.
12960 (arc_return_address_register): New function.
12961 (arc_epilogue_uses): Use above function.
12962 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
12963 (arc_function_type): Change encoding, add naked type.
12964 (ARC_INTERRUPT_P): Change to handle the new encoding.
12965 (ARC_FAST_INTERRUPT_P): Likewise.
12966 (ARC_NORMAL_P): Define.
12967 (ARC_NAKED_P): Likewise.
12968 (arc_compute_function_type): Delete prototype.
12969 * config/arc/arc.md (in_ret_delay_slot): Use
12970 arc_return_address_register function.
12971 (simple_return): Likewise.
12972 (p_return_i): Likewise.
12974 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12976 PR tree-optimization/81428
12977 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
12978 can't be built for those types.
12980 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
12982 Remove stuff dead since r239246.
12984 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
12985 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
12986 (avr_inform_devices): Remove dead stuff.
12988 2017-07-17 Tamar Christina <tamar.christina@arm.com>
12990 * config/arm/arm_neon.h: Fix softp typo.
12992 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12994 PR tree-optimization/81365
12995 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
12996 aggregate moves onto bb predecessor edges, make sure there are no
12997 loads that could alias the lhs in between the start of bb and the
13000 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
13003 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
13004 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
13005 [LSHIFTRT, outer_code = TRUNCATE]: Same.
13007 2017-07-17 Jakub Jelinek <jakub@redhat.com>
13009 PR tree-optimization/81396
13010 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
13011 (init_symbolic_number): Initialize it to 1.
13012 (perform_symbolic_merge): Add n_ops from both operands into the new
13014 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
13015 without base_addr as useless if they need more than one operation.
13016 (bswap_replace): Handle !bswap case for NULL base_addr.
13018 2017-07-17 Tom de Vries <tom@codesourcery.com>
13021 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
13024 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
13026 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
13027 conditional builtin define __FIX_LEON3FT_B2BST.
13029 2017-07-17 Daniel Cederman <cederman@gaisler.com>
13031 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
13032 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
13035 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
13037 PR rtl-optimization/81424
13038 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
13039 to remove potential trapping from operands if -fnon-call-exceptions.
13041 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13043 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
13044 profile_proability for scalling.
13045 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
13047 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13049 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
13051 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13053 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
13054 fixpoint arithmetics.
13056 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13058 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
13059 fixpoint arithmetics.
13061 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13063 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
13064 fixpoint arithmetics.
13066 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13068 * profile-count.h (profile_probability::from_reg_br_prob_note,
13069 profile_probability::to_reg_br_prob_note): New functions.
13070 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
13071 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
13072 * predict.c (probability_reliable_p): Update.
13073 (edge_probability_reliable_p): Update.
13074 (br_prob_note_reliable_p): Update.
13075 (invert_br_probabilities): Update.
13076 (add_reg_br_prob_note): New function.
13077 (combine_predictions_for_insn): Update.
13078 * asan.c (asan_clear_shadow): Update.
13079 * cfgbuild.c (compute_outgoing_frequencies): Update.
13080 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
13081 (update_br_prob_note): Update.
13082 (rtl_verify_edges): Update.
13083 (purge_dead_edges): Update.
13084 (fixup_reorder_chain): Update.
13085 * emit-rtl.c (try_split): Update.
13086 * ifcvt.c (cond_exec_process_insns): Update.
13087 (cond_exec_process_if_block): Update.
13088 (dead_or_predicable): Update.
13089 * internal-fn.c (expand_addsub_overflow): Update.
13090 (expand_neg_overflow): Update.
13091 (expand_mul_overflow): Update.
13092 * loop-doloop.c (doloop_modify): Update.
13093 * loop-unroll.c (compare_and_jump_seq): Update.
13094 * optabs.c (emit_cmp_and_jump_insn_1): Update.
13095 * predict.h: Update.
13096 * reorg.c (mostly_true_jump): Update.
13098 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
13099 * config/alpha/alpha.c (emit_unlikely_jump): Update.
13100 * config/arc/arc.c: (emit_unlikely_jump): Update.
13101 * config/arm/arm.c: (emit_unlikely_jump): Update.
13102 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
13103 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
13104 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
13105 (ix86_print_operand): Update.
13106 (ix86_split_fp_branch): Update.
13107 (predict_jump): Update.
13108 * config/ia64/ia64.c (ia64_print_operand): Update.
13109 * config/mmix/mmix.c (mmix_print_operand): Update.
13110 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
13111 (rs6000_expand_split_stack_prologue): Update.
13112 * config/rs6000/rs6000.c: Update.
13113 * config/s390/s390.c (s390_expand_vec_strlen): Update.
13114 (s390_expand_vec_movstr): Update.
13115 (s390_expand_cs_tdsi): Update.
13116 (s390_expand_split_stack_prologue): Update.
13117 * config/sh/sh.c (sh_print_operand): Update.
13118 (expand_cbranchsi4): Update.
13119 (expand_cbranchdi4): Update.
13120 * config/sparc/sparc.c (output_v9branch): Update.
13121 * config/spu/spu.c (get_branch_target): Update.
13122 (ea_load_store_inline): Update.
13123 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
13124 * config/tilepro/tilepro.c: Update.
13126 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
13128 * gimplify.c (mostly_copy_tree_r): Revert latest change.
13129 (gimplify_save_expr): Likewise.
13131 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13133 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
13135 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13137 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
13139 * timevar.def (TV_IPA_FNSUMMARY): Define.
13141 2017-07-16 Daniel Cederman <cederman@gaisler.com>
13143 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
13144 to back store errata sensitive sequence from being generated.
13145 (sqrtdf2_fix): Likewise.
13147 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13149 * tree-ssa-threadupdate.c (compute_path_counts,
13150 update_joiner_offpath_counts): Use profile_probability.
13152 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
13155 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13157 * config/arm/arm-c.c (arm_cpu_builtins): Define
13158 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
13160 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
13162 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13163 array entries to represent __ieee128 versions of the
13164 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
13165 scalar_extract_sig, and scalar_insert_exp built-in functions.
13166 (altivec_resolve_overloaded_builtin): Add special case handling
13167 for the __builtin_scalar_insert_exp function, as represented by
13168 the P9V_BUILTIN_VEC_VSIEDP constant.
13169 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
13170 exponent support for __ieee128 argument.
13171 (VSESQP): Add scalar extract signature support for __ieee128
13173 (VSTDCNQP): Add scalar test negative support for __ieee128
13175 (VSIEQP): Add scalar insert exponent support for __int128 argument
13176 with __ieee128 result.
13177 (VSIEQPF): Add scalar insert exponent support for __ieee128
13178 argument with __ieee128 result.
13179 (VSTDCQP): Add scalar test data class support for __ieee128
13181 (VSTDCNQP): Add overload support for scalar test negative with
13182 __ieee128 argument.
13183 (VSTDCQP): Add overload support for scalar test data class
13184 __ieee128 argument.
13185 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
13186 UNSPEC_VSX_SXSIGDP.
13187 (UNSPEC_VSX_SIEXPQP): New constant.
13188 (xsxexpqp): New insn for VSX scalar extract exponent quad
13190 (xsxsigqp): New insn for VSX scalar extract significand quad
13192 (xsiexpqpf): New insn for VSX scalar insert exponent quad
13193 precision with floating point argument.
13194 (xststdcqp): New expand for VSX scalar test data class quad
13196 (xststdcnegqp): New expand for VSX scalar test negative quad
13198 (xststdcqp): New insn to match expansions for VSX scalar test data
13199 class quad precision and VSX scalar test negative quad precision.
13200 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
13201 special case operand checking to enforce that second operand of
13202 VSX scalar test data class with quad precision argument is a 7-bit
13204 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
13205 prototypes and descriptions of __ieee128 versions of
13206 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
13207 scalar_test_data_class, and scalar_test_neg built-in functions.
13209 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13211 PR tree-optimization/81162
13212 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
13213 replace a negate with an add.
13215 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
13217 * doc/invoke.texi (arm/-mcpu): Document +crypto.
13219 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13221 * config/arm/arm-c.c (arm_cpu_builtins): Define
13222 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
13224 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13226 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
13227 (armv8-r): Set ARM Cortex-R52 as default CPU.
13228 * config/arm/arm-tables.opt: Regenerate.
13229 * config/arm/arm-tune.md: Regenerate.
13230 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
13232 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
13233 extension for -mcpu=cortex-r52.
13235 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13237 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
13238 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
13239 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
13240 (fp-armv8): Define it as FP_ARMv8 only.
13241 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
13242 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
13244 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
13245 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
13246 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
13247 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
13248 than TARGET_FPU_ARMV8.
13249 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
13250 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
13251 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
13253 * config/arm/neon.md (neon_vrint): Likewise.
13254 (neon_vcvt): Likewise.
13255 (neon_<fmaxmin_op><mode>): Likewise.
13256 (<fmaxmin><mode>3): Likewise.
13257 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
13258 * config/arm/predicates.md (arm_cond_move_operator): Check against
13259 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
13261 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
13263 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
13264 to top of function.
13266 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13268 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
13269 loop in comment with memset.
13271 2017-07-14 Martin Liska <mliska@suse.cz>
13273 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
13274 * dwarf2out.c (is_java): Remove the function.
13275 (output_pubname): Remove usage of the function.
13276 (lower_bound_default): Remove usage of DW_LANG_Java.
13277 (gen_compile_unit_die): Likewise.
13278 * gcc.c: Remove compiler defaults for .java and .zip files.
13279 * gimple-expr.c (remove_suffix): Change as there's no longer
13280 extension than 4-letter one.
13281 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
13282 (gimplify_save_expr): Likewise.
13283 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
13284 as it's possible even for other languages than Java.
13285 * langhooks.h (struct lang_hooks): Remove Java from a comment.
13286 * lto-opts.c (lto_write_options): Remove reference to Java.
13287 * opts.c (strip_off_ending): Update file extension handling.
13288 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
13289 * tree-eh.c (lower_resx): Likewise.
13290 * tree.c (free_lang_data_in_type): Remove dead code.
13291 (find_decls_types_r): Likewise.
13292 (build_common_builtin_nodes): Remove Java from a comment.
13293 (verify_type): Remove dead code.
13294 * varasm.c (assemble_external): Remove Java from a comment.
13296 2017-07-14 Martin Liska <mliska@suse.cz>
13298 * opts.c (finish_options): Add quotes.
13299 (common_handle_option): Likewise.
13301 2017-07-14 Martin Liska <mliska@suse.cz>
13303 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
13304 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
13305 Remove N_SO_PASCAL.
13306 * dwarf2out.c (lower_bound_default): Do not handle
13308 (gen_compile_unit_die): Likewise.
13309 * gcc.c: Remove default extension binding for GNU Pascal.
13310 * stmt.c: Remove Pascal language from a comment.
13311 * xcoffout.c: Likewise.
13313 2017-07-13 David Malcolm <dmalcolm@redhat.com>
13316 * diagnostic-show-locus.c (fixit_cmp): New function.
13317 (layout::layout): Sort m_fixit_hints.
13318 (column_range::column_range): Assert that the values are valid.
13319 (struct char_span): New struct.
13320 (correction::overwrite): New method.
13321 (struct source_line): New struct.
13322 (line_corrections::add_hint): Add assertions. Reimplement memcpy
13323 calls in terms of classes source_line and char_span, and
13324 correction::overwrite.
13325 (selftest::test_overlapped_fixit_printing_2): New function.
13326 (selftest::diagnostic_show_locus_c_tests): Call it.
13328 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
13330 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
13331 early if there is no lhs.
13333 2017-07-13 Martin Liska <mliska@suse.cz>
13335 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
13336 (gen_reference_type_die): Likewise.
13337 * stor-layout.c: Remove Pascal-related comment.
13339 2017-07-13 Martin Liska <mliska@suse.cz>
13341 * opts.c (finish_options): Add quotes to error messages.
13342 (parse_sanitizer_options): Likewise.
13344 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
13346 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
13348 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
13350 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
13352 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
13354 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
13356 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
13358 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
13361 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
13362 provides the hardware capability bits, define the macro
13363 __BUILTIN_CPU_SUPPORTS__.
13364 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
13365 if GLIBC does not provide the hardware capability bits. Add a
13366 gcc_unreachable call if the built-in cpu function is neither
13367 __builtin_cpu_is nor __builtin_cpu_supports.
13368 (rs6000_get_function_versions_dispatcher): Change the warning
13369 that an old GLIBC is used which does not export the capability
13370 bits to be an error.
13371 * doc/extend.texi (target_clones attribute): Document the
13372 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
13373 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
13374 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
13375 the macros defined by GCC if the newer GLIBC is available.
13377 2017-07-12 Jeff Law <law@redhat.com>
13379 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
13380 remaining includes slightly.
13381 * config/riscv/riscv-builtins.c: Include profile-count.h.
13383 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
13386 * config/avr/avr.c (avr_set_current_function): In diagnostic
13387 messages: Quote keywords and (parts of) identifiers.
13388 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
13391 2017-07-12 Carl Love <cel@us.ibm.com>
13393 * config/rs6000/rs6000-c.c: Add support for built-in functions
13394 vector bool char vec_revb (vector bool char);
13395 vector bool short vec_revb (vector short char);
13396 vector bool int vec_revb (vector bool int);
13397 vector bool long long vec_revb (vector bool long long);
13398 * doc/extend.texi: Update the built-in documentation file for the
13399 new built-in functions.
13401 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13403 * config/s390/s390.md: Remove movcc splitter.
13405 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13407 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
13408 load/store on condition.
13410 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
13413 * config/avr/avr.c (avr_encode_section_info)
13414 [progmem && !TREE_READONLY]: Error if progmem object needs
13417 2017-07-11 Michael Collison <michael.collison@arm.com>
13419 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
13422 2017-07-11 Carl Love <cel@us.ibm.com>
13424 * config/rs6000/rs6000-c.c: Add support for builtins
13425 vector unsigned int vec_parity_lsbb (vector signed int);
13426 vector unsigned int vec_parity_lsbb (vector unsigned int);
13427 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
13428 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
13429 vector unsigned long long vec_parity_lsbb (vector signed long long);
13430 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
13431 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
13432 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
13433 * doc/extend.texi: Update the built-in documentation file for the
13434 new built-in functions.
13436 2017-07-11 David Malcolm <dmalcolm@redhat.com>
13438 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
13439 (layout::m_primary_loc): New field.
13440 (layout::layout): Initialize new field. Move location filtering
13441 logic from here to...
13442 (layout::maybe_add_location_range): ...this new method. Add
13443 support for filtering to just the lines already specified by other
13445 (layout::will_show_line_p): New method.
13446 (gcc_rich_location::add_location_if_nearby): New method.
13447 (selftest::test_add_location_if_nearby): New test function.
13448 (selftest::diagnostic_show_locus_c_tests): Call it.
13449 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
13452 2017-07-11 Tom de Vries <tom@codesourcery.com>
13454 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
13455 (bb_first_real_insn): New function.
13456 (nvptx_single): Add extra initialization of broadcasted condition
13459 2017-07-11 Nathan Sidwell <nathan@acm.org>
13461 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
13463 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
13465 * doc/extend.texi (AVR Function Attributes): Remove weblink to
13466 Binutils doc as TEXI will mess them up.
13467 * doc/invoke.texi (AVR Options): Same here.
13469 2017-07-11 Daniel Cederman <cederman@gaisler.com>
13471 * config/sparc/sparc.opt (mfix-ut700): New option.
13472 (mfix-gr712rc): Likewise.
13473 (sparc_fix_b2bst): New variable.
13474 * doc/invoke.texi (SPARC options): Document them.
13475 (ARM options): Fix warnings.
13476 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
13477 instructions to prevent sequences that can trigger the store-store
13478 errata for certain LEON3FT processors.
13479 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
13480 (sparc_option_override): Set sparc_fix_b2bst appropriately.
13481 * config/sparc/sparc.md (fix_b2bst): New attribute.
13482 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
13484 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
13487 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
13489 (*rsqrtsf2_sse): Ditto.
13491 (div<mode>3): Macroize insn from divdf3 and divsf3
13492 using MODEF mode iterator.
13494 2017-07-10 Martin Sebor <msebor@redhat.com>
13496 PR tree-optimization/80397
13497 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
13498 instead of testing for equality to INTEGER_TYPE.
13500 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
13502 * config.gcc: Remove uclibc from arc target spec.
13504 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
13506 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
13508 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13511 * lto-wrapper.c (remove_option): New function.
13512 (merge_and_complain): Merge PIC/PIE options more realistically.
13514 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
13516 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
13520 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
13521 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
13522 * config.in: Regenerate.
13523 * configure: Regenerate.
13524 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
13525 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
13526 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
13527 (TARGET_GASISR_PROLOGUES): ...target mask.
13528 * common/config/avr/avr-common.c
13529 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
13530 Set -mgas-isr-prologues.
13531 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
13532 INSERT_PASS_BEFORE for it.
13533 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
13534 * config/avr/avr.c (avr_option_override)
13535 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
13536 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
13537 (avr_attribute_table) <no_gccisr>: Add new function attribute.
13538 (avr_set_current_function) <is_no_gccisr>: Init machine field.
13539 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
13541 (make_avr_pass_pre_proep): New function.
13542 (emit_push_sfr) <treg>: Add argument to function and use it
13543 instead of TMP_REG.
13544 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
13545 and set machine->gasisr.yes.
13546 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
13547 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
13548 __gcc_isr.n_pushed to .L__stack_usage.
13549 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
13550 (avr_asm_final_postscan_insn): ...this new static function.
13551 * config/avr/avr.h (machine_function)
13552 <is_no_gccisr, use_L__stack_usage>: New fields.
13553 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
13554 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
13555 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
13556 (gasisr, *gasisr): New expander and insn.
13557 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
13558 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
13559 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
13561 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
13563 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
13566 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
13568 Move jump-tables out of .text again.
13571 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
13572 (ASM_OUTPUT_ADDR_VEC): New function.
13573 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
13574 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
13575 INSN_ADDRESSes as asm comment.
13576 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
13577 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
13578 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
13579 * config/avr/avr.md (*tablejump): Adjust comment.
13580 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
13581 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
13583 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
13584 (avr_output_addr_vec): New proto.
13585 (avr_log_t) <insn_addresses>: New field.
13587 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
13590 * config/i386/i386.c (ix86_function_arg_advance): Set
13591 outgoing_args_on_stack to true if there are outgoing arguments
13593 (ix86_function_arg): Likewise.
13594 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
13595 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
13596 * config/i386/i386.h (machine_function): Add
13597 outgoing_args_on_stack.
13599 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
13601 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
13602 supporting pthreds.
13603 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
13605 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
13607 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
13608 (REAL_H): Remove $(MACHMODE_H).
13609 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
13611 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
13612 $(MACHMODE_H) and double-int.h.
13613 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
13615 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
13618 2017-07-07 Andrew Pinski <apinski@cavium.com>
13620 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
13621 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
13623 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13625 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
13626 Add warning if GCC was not configured to link against a GLIBC that
13627 exports the hardware capability bits.
13628 (make_resolver_func): Make resolver function private and not a
13629 COMDAT function. Create the name with clone_function_name instead
13630 of make_unique_name.
13633 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
13634 correct operand in doing the split.
13636 2017-07-07 Carl Love <cel@us.ibm.com>
13638 * config/rs6000/rs6000-c: Add support for built-in function
13639 vector unsigned short vec_pack_to_short_fp32 (vector float,
13641 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
13642 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
13643 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
13644 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
13645 (convert_4f32_8i16): Add define_expand.
13646 * doc/extend.texi: Update the built-in documentation file for the
13647 new built-in function.
13649 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13651 * config/sparc/m8.md: New file.
13652 * config/sparc/sparc.md: Include m8.md.
13654 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13656 * config/sparc/sparc.opt: New option -mvis4b.
13657 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
13658 (sparc_option_override): Handle VIS4B.
13659 (enum sparc_builtins): Define
13660 SPARC_BUILTIN_DICTUNPACK{8,16,32},
13661 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
13662 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
13663 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
13664 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
13665 (check_constant_argument): New function.
13666 (sparc_vis_init_builtins): Define builtins
13667 __builtin_vis_dictunpack{8,16,32},
13668 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
13669 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
13670 __builtin_vis_fpcmpde{8,16,32}shl and
13671 __builtin_vis_fpcmpur{8,16,32}shl.
13672 (sparc_expand_builtin): Check that the constant operands to
13673 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
13674 constant and in range.
13675 * config/sparc/sparc-c.c (sparc_target_macros): Handle
13677 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
13678 (SPARC_IMM5_P): Likewise.
13679 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
13680 (enabled): Handle vis4b.
13681 (UNSPEC_DICTUNPACK): New unspec.
13682 (UNSPEC_FPCMPSHL): Likewise.
13683 (UNSPEC_FPUCMPSHL): Likewise.
13684 (UNSPEC_FPCMPDESHL): Likewise.
13685 (UNSPEC_FPCMPURSHL): Likewise.
13686 (cpu_feature): New CPU feature `vis4b'.
13687 (dictunpack{8,16,32}): New insns.
13688 (FPCSMODE): New mode iterator.
13689 (fpcscond): New code iterator.
13690 (fpcsucond): Likewise.
13691 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
13692 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
13693 (fpcmpde{8,16,32}{si,di}shl): Likewise.
13694 (fpcmpur{8,16,32}{si,di}shl): Likewise.
13695 * config/sparc/constraints.md: Define constraints `q' for unsigned
13696 2-bit integer constants and `t' for unsigned 5-bit integer
13698 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
13700 (imm5_operand_dictunpack16): Likewise.
13701 (imm5_operand_dictunpack32): Likewise.
13702 (imm2_operand): Likewise.
13703 * doc/invoke.texi (SPARC Options): Document -mvis4b.
13704 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
13705 ditunpack* and fpcmp*shl builtins.
13707 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13709 * config.gcc: Handle m8 in --with-{cpu,tune} options.
13710 * config.in: Add HAVE_AS_SPARC6 define.
13711 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
13713 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
13715 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
13716 (CPP_CPU_SPEC): Handle m8.
13717 (ASM_CPU_SPEC): Likewise.
13718 * config/sparc/sparc-opts.h (enum processor_type): Add
13720 * config/sparc/sparc.c (m8_costs): New struct.
13721 (sparc_option_override): Handle TARGET_CPU_m8.
13722 (sparc32_initialize_trampoline): Likewise.
13723 (sparc64_initialize_trampoline): Likewise.
13724 (sparc_issue_rate): Likewise.
13725 (sparc_register_move_cost): Likewise.
13726 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
13727 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
13728 (ASM_CPU64_DEFAULT_SPEC): Likewise.
13729 (CPP_CPU_SPEC): Handle M8.
13730 (ASM_CPU_SPEC): Likewise.
13731 (AS_M8_FLAG): Define.
13732 * config/sparc/sparc.md: Add m8 to the cpu attribute.
13733 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
13734 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
13736 * configure: Regenerate.
13737 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
13740 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13742 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
13744 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
13745 ("*movdi_insn_sp32"): Do not set v3pipe.
13746 ("*movsi_insn"): Likewise.
13747 ("*movdi_insn_sp64"): Likewise.
13748 ("*movsf_insn"): Likewise.
13749 ("*movdf_insn_sp32"): Likewise.
13750 ("*movdf_insn_sp64"): Likewise.
13751 ("*zero_extendsidi2_insn_sp64"): Likewise.
13752 ("*sign_extendsidi2_insn"): Likewise.
13753 ("*mov<VM32:mode>_insn"): Likewise.
13754 ("*mov<VM64:mode>_insn_sp64"): Likewise.
13755 ("*mov<VM64:mode>_insn_sp32"): Likewise.
13756 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
13757 ("<vlop:code><VL:mode>3"): Likewise.
13758 ("*not_<vlop:code><VL:mode>3"): Likewise.
13759 ("*nand<VL:mode>_vis"): Likewise.
13760 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
13761 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
13762 ("one_cmpl<VL:mode>2"): Likewise.
13763 ("faligndata<VM64:mode>_vis"): Likewise.
13764 ("alignaddrsi_vis"): Likewise.
13765 ("alignaddrdi_vis"): Likweise.
13766 ("alignaddrlsi_vis"): Likewise.
13767 ("alignaddrldi_vis"): Likewise.
13768 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
13769 ("bmaskdi_vis"): Likewise.
13770 ("bmasksi_vis"): Likewise.
13771 ("bshuffle<VM64:mode>_vis"): Likewise.
13772 ("cmask8<P:mode>_vis"): Likewise.
13773 ("cmask16<P:mode>_vis"): Likewise.
13774 ("cmask32<P:mode>_vis"): Likewise.
13775 ("pdistn<P:mode>_vis"): Likewise.
13776 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
13778 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13780 * config/sparc/sparc.md ("subtype"): New insn attribute.
13781 ("*wrgsr_sp64"): Set insn subtype.
13782 ("*rdgsr_sp64"): Likewise.
13783 ("alignaddrsi_vis"): Likewise.
13784 ("alignaddrdi_vis"): Likewise.
13785 ("alignaddrlsi_vis"): Likewise.
13786 ("alignaddrldi_vis"): Likewise.
13787 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
13788 ("fexpand_vis"): Likewise.
13789 ("fpmerge_vis"): Likewise.
13790 ("faligndata<VM64:mode>_vis"): Likewise.
13791 ("bshuffle<VM64:mode>_vis"): Likewise.
13792 ("cmask8<P:mode>_vis"): Likewise.
13793 ("cmask16<P:mode>_vis"): Likewise.
13794 ("cmask32<P:mode>_vis"): Likewise.
13795 ("fchksm16_vis"): Likewise.
13796 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
13797 ("fmean16_vis"): Likewise.
13798 ("fp<plusminus_insn>64_vis"): Likewise.
13799 ("<plusminus_insn>v8qi3"): Likewise.
13800 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
13801 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
13802 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
13803 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
13804 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
13805 ("*movqi_insn"): Likewise.
13806 ("*movhi_insn"): Likewise.
13807 ("*movsi_insn"): Likewise.
13808 ("movsi_pic_gotdata_op"): Likewise.
13809 ("*movdi_insn_sp32"): Likewise.
13810 ("*movdi_insn_sp64"): Likewise.
13811 ("movdi_pic_gotdata_op"): Likewise.
13812 ("*movsf_insn"): Likewise.
13813 ("*movdf_insn_sp32"): Likewise.
13814 ("*movdf_insn_sp64"): Likewise.
13815 ("*zero_extendhisi2_insn"): Likewise.
13816 ("*zero_extendqihi2_insn"): Likewise.
13817 ("*zero_extendqisi2_insn"): Likewise.
13818 ("*zero_extendqidi2_insn"): Likewise.
13819 ("*zero_extendhidi2_insn"): Likewise.
13820 ("*zero_extendsidi2_insn_sp64"): Likewise.
13821 ("ldfsr"): Likewise.
13822 ("prefetch_64"): Likewise.
13823 ("prefetch_32"): Likewise.
13824 ("tie_ld32"): Likewise.
13825 ("tie_ld64"): Likewise.
13826 ("*tldo_ldub_sp32"): Likewise.
13827 ("*tldo_ldub1_sp32"): Likewise.
13828 ("*tldo_ldub2_sp32"): Likewise.
13829 ("*tldo_ldub_sp64"): Likewise.
13830 ("*tldo_ldub1_sp64"): Likewise.
13831 ("*tldo_ldub2_sp64"): Likewise.
13832 ("*tldo_ldub3_sp64"): Likewise.
13833 ("*tldo_lduh_sp32"): Likewise.
13834 ("*tldo_lduh1_sp32"): Likewise.
13835 ("*tldo_lduh_sp64"): Likewise.
13836 ("*tldo_lduh1_sp64"): Likewise.
13837 ("*tldo_lduh2_sp64"): Likewise.
13838 ("*tldo_lduw_sp32"): Likewise.
13839 ("*tldo_lduw_sp64"): Likewise.
13840 ("*tldo_lduw1_sp64"): Likewise.
13841 ("*tldo_ldx_sp64"): Likewise.
13842 ("*mov<VM32:mode>_insn"): Likewise.
13843 ("*mov<VM64:mode>_insn_sp64"): Likewise.
13844 ("*mov<VM64:mode>_insn_sp32"): Likewise.
13846 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13848 * config/sparc/sparc.md ("type"): New insn type viscmp.
13849 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
13851 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
13852 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
13853 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
13854 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
13856 ("n7_vis_logical_11cycle"): Likewise.
13857 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
13858 * config/sparc/niagara2.md ("niag3_vis": Likewise.
13859 * config/sparc/niagara.md ("niag_vis"): Likewise.
13860 * config/sparc/ultra3.md ("us3_fga"): Likewise.
13861 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
13863 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13865 * config/sparc/sparc.md: New instruction type `bmask'.
13866 (bmaskdi_vis): Use the `bmask' type.
13867 (bmasksi_vis): Likewise.
13868 * config/sparc/ultra3.md (us3_array): Likewise.
13869 * config/sparc/niagara7.md (n7_array): Likewise.
13870 * config/sparc/niagara4.md (n4_array): Likewise.
13871 * config/sparc/niagara2.md (niag2_vis): Likewise.
13872 (niag3_vis): Likewise.
13873 * config/sparc/niagara.md (niag_vis): Likewise.
13875 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13877 * ipa-comdats.c: Remove optimize check from gate.
13878 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
13879 for functions not optimized.
13880 (ipa_fn_summary_read): Skip optimize check.
13881 (ipa_fn_summary_write): Likewise.
13882 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
13884 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
13886 (can_inline_edge_p): Check flag_pcc_struct_return for match.
13887 (check_callers): Give up on caller which is not optimized.
13888 (inline_small_functions): Likewise.
13889 (ipa_inline): Do not give up when not optimizing.
13890 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
13891 away unoptimizes cdtors.
13892 (whole_program_function_and_variable_visibility): Do
13893 ipa_discover_readonly_nonaddressable_vars in LTO mode.
13894 * ipa.c (process_references): Do not check optimize.
13895 (symbol_table::remove_unreachable_nodes): Update optimize check.
13896 (set_writeonly_bit): Update optimize check.
13897 (pass_ipa_cdtor_merge::gate): Do not check optimize.
13898 (pass_ipa_single_use::gate): Remove.
13900 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13902 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
13903 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
13904 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
13905 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
13906 permute_load, permute_store, adjust_extract, adjust_splat,
13907 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
13908 replace_swap_with_copy, dump_swap_insn_table,
13909 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
13910 recombine_lvx_pattern, recombine_stvx_pattern,
13911 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
13912 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
13913 to file rs6000-p8swap.c.
13914 * config/rs6000/rs6000-p8swap.c: New file.
13915 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
13916 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
13917 and rs6000*-*-* targets.
13919 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13921 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
13923 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13925 * lto-wrapper.c (merge_and_complain): Do not merge
13926 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
13927 fsigned_zeros, ftrapping_math, fwrapv.
13928 (append_compiler_options): Do not track these options.
13929 (append_linker_options): Likewie
13931 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13933 * cgraphunit.c (cgraph_node::finalize_function): When
13934 !flag_toplevel_reorde set no_reorder flag.
13935 (varpool_node::finalize_decl): Likewise.
13936 (symbol_table::compile): Drop no toplevel reorder path.
13938 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13940 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
13941 edges; zero probability is not better than uninitialized.
13943 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
13945 * asan.h (asan_sanitize_allocas_p): Declare.
13946 * asan.c (asan_sanitize_allocas_p): New function.
13947 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
13948 (handle_builtin_alloca): Likewise.
13949 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
13950 if !asan_sanitize_allocas_p.
13951 * params.def (asan-instrument-allocas): Add new option.
13952 * params.h (ASAN_PROTECT_ALLOCAS): Define.
13953 * opts.c (common_handle_option): Disable allocas sanitization for
13956 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
13958 * asan.c: Include gimple-fold.h.
13959 (get_last_alloca_addr): New function.
13960 (handle_builtin_stackrestore): Likewise.
13961 (handle_builtin_alloca): Likewise.
13962 (asan_emit_allocas_unpoison): Likewise.
13963 (get_mem_refs_of_builtin_call): Add new parameter, remove const
13964 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
13965 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
13966 (instrument_builtin_call): Pass gimple iterator to
13967 get_mem_refs_of_builtin_call.
13968 (last_alloca_addr): New global.
13969 * asan.h (asan_emit_allocas_unpoison): Declare.
13970 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
13971 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
13972 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
13973 if function calls alloca.
13974 * gimple-fold.c (replace_call_with_value): Remove static keyword.
13975 * gimple-fold.h (replace_call_with_value): Declare.
13976 * internal-fn.c: Include asan.h.
13977 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
13978 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
13980 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13982 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
13983 (C_SELFTEST_FLAGS): New.
13984 (CPP_SELFTEST_FLAGS): New.
13985 (SELFTEST_DEPS): New, from deps of s-selftest.
13986 (C_SELFTEST_DEPS): New, from deps of s-selftest.
13987 (CPP_SELFTEST_DEPS): New.
13988 (selftest): Add dependency on s-selftest-c++.
13989 (s-selftest): Rename to...
13990 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
13991 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
13992 than SELFTEST_FLAGS.
13993 (selftest-gdb): Rename to...
13994 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
13996 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
13997 (selftest-valgrind): Rename to...
13998 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
14000 (selftest-valgrind): Reintroduce as an alias for
14001 selftest-c-valgrind.
14002 (s-selftest-c++): New.
14003 (selftest-c++-gdb): New.
14004 (selftest-c++-valgrind): New.
14006 2017-07-06 Olivier Hainque <hainque@adacore.com>
14008 * gcc.c (process_command): When deciding if undefined variables
14009 should be ignored when processing specs, accept "gcc -v" as well.
14011 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14013 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
14014 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
14016 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
14018 * config/arm/arm-cpus.in (armv8-r): Add new entry.
14019 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
14020 * config/arm/arm-tables.opt: Regenerate.
14021 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
14023 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
14025 2017-07-06 Carl Love <cel@us.ibm.com>
14027 * ChangeLog: Clean up from mid air collision
14029 2017-07-06 Carl Love <cel@us.ibm.com>
14031 * config/rs6000/rs6000-c.c: Add support for built-in functions
14032 vector signed int vec_subc (vector signed int, vector signed int);
14033 vector signed __int128 vec_subc (vector signed __int128,
14034 vector signed __int128);
14035 vector unsigned __int128 vec_subc (vector unsigned __int128,
14036 vector unsigned __int128);
14037 vector signed int vec_sube (vector signed int, vector signed int,
14038 vector signed int);
14039 vector unsigned int vec_sube (vector unsigned int,
14040 vector unsigned int,
14041 vector unsigned int);
14042 vector signed __int128 vec_sube (vector signed __int128,
14043 vector signed __int128,
14044 vector signed__int128);
14045 vector unsigned __int128 vec_sube (vector unsigned __int128,
14046 vector unsigned __int128,
14047 vector unsigned __int128);
14048 vector signed int vec_subec (vector signed int, vector signed int,
14049 vector signed int);
14050 vector unsigned int vec_subec (vector unsigned int,
14051 vector unsigned int,
14052 vector unsigned int);
14053 vector signed __int128 vec_subec (vector signed __int128,
14054 vector signed __int128,
14055 vector signed__int128);
14056 vector unsigned __int128 vec_subec (vector unsigned __int128,
14057 vector unsigned __int128,
14058 vector unsigned __int128);
14059 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
14060 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
14061 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
14062 BU_ALTIVEC_OVERLOAD_X definitions.
14063 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
14064 * doc/extend.texi: Update the built-in documentation file for the new
14065 built-in functions.
14067 2017-07-06 David Malcolm <dmalcolm@redhat.com>
14070 * diagnostic-show-locus.c (layout::layout): Use start and finish
14071 spelling location for the start and finish of each range.
14072 * genmatch.c (linemap_client_expand_location_to_spelling_point):
14073 Add unused aspect param.
14074 * input.c (expand_location_1): Add "aspect" param, and use it
14075 to access the correct part of the location.
14076 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
14078 (expand_location_to_spelling_point): Likewise.
14079 (linemap_client_expand_location_to_spelling_point): Add "aspect"
14080 param, and pass it to expand_location_1.
14082 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
14084 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
14085 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
14086 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
14087 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
14088 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
14089 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
14090 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
14091 _mm_maskz_getmant_ss): New intrinsics.
14092 (__builtin_ia32_getexpss128_mask): Changed to ...
14093 __builtin_ia32_getexpss128_round ... this.
14094 (__builtin_ia32_getexpsd128_mask): Changed to ...
14095 __builtin_ia32_getexpsd128_round ... this.
14096 * config/i386/i386-builtin-types.def
14097 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
14098 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
14099 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
14100 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
14101 __builtin_ia32_getmantss_mask_round): New builtins.
14102 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
14103 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
14104 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
14105 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
14106 * config/i386/sse.md
14107 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
14108 avx512f_sgetexp<mode><mask_scalar_name>
14109 <round_saeonly_scalar_name> ... this.
14110 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
14111 %0, %1, %2<round_saeonly_op3>}): Changed to ...
14112 vgetexp<ssescalarmodesuffix>
14113 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14114 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
14115 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
14116 avx512f_vgetmant<mode><mask_scalar_name>
14117 <round_saeonly_scalar_name> ... this.
14118 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
14119 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
14120 vgetmant<ssescalarmodesuffix>
14121 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
14122 %0<mask_scalar_operand4>, %1, %2
14123 <round_saeonly_scalar_mask_op4>, %3} ... this.
14124 * config/i386/subst.md (mask_scalar_operand4,
14125 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
14126 round_saeonly_scalar_nimm_predicate): New subst attributes.
14128 2017-07-06 Julia Koval <julia.koval@intel.com>
14130 * config/i386/i386.c (ix86_erase_embedded_rounding):
14131 Remove code for old rounding pattern.
14133 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
14135 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
14137 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
14139 * doc/sourcebuild.texi (Test Directives, Variants of
14140 dg-require-support): Add documentation for dg-require-stack-check.
14142 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
14144 * config/i386/subst.md (mask_scalar, round_scalar,
14145 round_saeonly_scalar): New meta-templates.
14146 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
14147 round_scalar_mask_operand3, round_scalar_mask_op3,
14148 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
14149 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
14150 round_saeonly_scalar_constraint,
14151 round_saeonly_scalar_prefix): New subst attribute.
14152 * config/i386/sse.md
14153 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
14154 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
14155 <round_scalar_name> ... this.
14156 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
14157 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
14158 <round_scalar_name> ... this.
14159 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
14160 <sse>_vm<code><mode>3<mask_scalar_name>
14161 <round_saeonly_scalar_name> ... this.
14162 (v<plusminus_mnemonic><ssescalarmodesuffix>
14163 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
14164 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
14165 v<plusminus_mnemonic><ssescalarmodesuffix>
14166 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14167 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
14168 (v<multdiv_mnemonic><ssescalarmodesuffix>
14169 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
14170 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
14171 v<multdiv_mnemonic><ssescalarmodesuffix>
14172 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14173 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
14174 (v<maxmin_float><ssescalarmodesuffix>
14175 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
14176 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
14177 v<maxmin_float><ssescalarmodesuffix>
14178 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14179 %0<mask_scalar_operand3>, %1, %<iptr>2
14180 <round_saeonly_scalar_mask_op3>} ... this.
14182 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
14184 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
14185 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
14187 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
14188 Alan Hayward <alan.hayward@arm.com>
14189 David Sherwood <david.sherwood@arm.com>
14191 * combine.c (simplify_if_then_else): Remove "enum" before
14193 * compare-elim.c (can_eliminate_compare): Likewise.
14194 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
14196 (aarch64_lookup_simd_builtin_type): Likewise.
14197 (aarch64_simd_builtin_type): Likewise.
14198 (aarch64_init_simd_builtin_types): Likewise.
14199 (aarch64_simd_expand_args): Likewise.
14200 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
14202 (aarch64_reverse_mask): Likewise.
14203 (aarch64_simd_emit_reg_reg_move): Likewise.
14204 (aarch64_gen_adjusted_ldpstp): Likewise.
14205 (aarch64_ccmp_mode_to_code): Likewise.
14206 (aarch64_operands_ok_for_ldpstp): Likewise.
14207 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14208 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
14210 (aarch64_min_divisions_for_recip_mul): Likewise.
14211 (aarch64_reassociation_width): Likewise.
14212 (aarch64_get_condition_code_1): Likewise.
14213 (aarch64_simd_emit_reg_reg_move): Likewise.
14214 (aarch64_simd_attr_length_rglist): Likewise.
14215 (aarch64_reverse_mask): Likewise.
14216 (aarch64_operands_ok_for_ldpstp): Likewise.
14217 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14218 (aarch64_gen_adjusted_ldpstp): Likewise.
14219 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
14221 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
14222 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
14223 (arm_lookup_simd_builtin_type): Likewise.
14224 (arm_simd_builtin_type): Likewise.
14225 (arm_init_simd_builtin_types): Likewise.
14226 (arm_expand_builtin_args): Likewise.
14227 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
14228 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
14229 (ft32_setup_incoming_varargs): Likewise.
14230 (ft32_function_arg): Likewise.
14231 (ft32_function_arg_advance): Likewise.
14232 (ft32_pass_by_reference): Likewise.
14233 (ft32_arg_partial_bytes): Likewise.
14234 (ft32_valid_pointer_mode): Likewise.
14235 (ft32_addr_space_pointer_mode): Likewise.
14236 (ft32_addr_space_legitimate_address_p): Likewise.
14237 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
14239 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
14240 (ix86_emit_outlined_ms2sysv_restore): Likewise.
14241 (iamcu_alignment): Likewise.
14242 (canonicalize_vector_int_perm): Likewise.
14243 (ix86_noce_conversion_profitable_p): Likewise.
14244 (ix86_mpx_bound_mode): Likewise.
14245 (ix86_operands_ok_for_move_multiple): Likewise.
14246 * config/microblaze/microblaze-protos.h
14247 (microblaze_expand_conditional_branch_reg): Likewise.
14248 * config/microblaze/microblaze.c
14249 (microblaze_expand_conditional_branch_reg): Likewise.
14250 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
14252 (rs6000_reassociation_width): Likewise.
14253 (rs6000_invalid_binary_op): Likewise.
14254 (fusion_p9_p): Likewise.
14255 (emit_fusion_p9_load): Likewise.
14256 (emit_fusion_p9_store): Likewise.
14257 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
14259 (riscv_hard_regno_mode_ok_p): Likewise.
14260 (riscv_address_insns): Likewise.
14261 (riscv_split_symbol): Likewise.
14262 (riscv_legitimize_move): Likewise.
14263 (riscv_function_value): Likewise.
14264 (riscv_hard_regno_nregs): Likewise.
14265 (riscv_expand_builtin): Likewise.
14266 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
14267 (riscv_build_integer): Likewise.
14268 (riscv_split_integer): Likewise.
14269 (riscv_legitimate_constant_p): Likewise.
14270 (riscv_cannot_force_const_mem): Likewise.
14271 (riscv_regno_mode_ok_for_base_p): Likewise.
14272 (riscv_valid_base_register_p): Likewise.
14273 (riscv_valid_offset_p): Likewise.
14274 (riscv_valid_lo_sum_p): Likewise.
14275 (riscv_classify_address): Likewise.
14276 (riscv_legitimate_address_p): Likewise.
14277 (riscv_address_insns): Likewise.
14278 (riscv_load_store_insns): Likewise.
14279 (riscv_force_binary): Likewise.
14280 (riscv_split_symbol): Likewise.
14281 (riscv_force_address): Likewise.
14282 (riscv_legitimize_address): Likewise.
14283 (riscv_move_integer): Likewise.
14284 (riscv_legitimize_const_move): Likewise.
14285 (riscv_legitimize_move): Likewise.
14286 (riscv_address_cost): Likewise.
14287 (riscv_subword): Likewise.
14288 (riscv_output_move): Likewise.
14289 (riscv_canonicalize_int_order_test): Likewise.
14290 (riscv_emit_int_order_test): Likewise.
14291 (riscv_function_arg_boundary): Likewise.
14292 (riscv_pass_mode_in_fpr_p): Likewise.
14293 (riscv_pass_fpr_single): Likewise.
14294 (riscv_pass_fpr_pair): Likewise.
14295 (riscv_get_arg_info): Likewise.
14296 (riscv_function_arg): Likewise.
14297 (riscv_function_arg_advance): Likewise.
14298 (riscv_arg_partial_bytes): Likewise.
14299 (riscv_function_value): Likewise.
14300 (riscv_pass_by_reference): Likewise.
14301 (riscv_setup_incoming_varargs): Likewise.
14302 (riscv_print_operand): Likewise.
14303 (riscv_elf_select_rtx_section): Likewise.
14304 (riscv_save_restore_reg): Likewise.
14305 (riscv_for_each_saved_reg): Likewise.
14306 (riscv_register_move_cost): Likewise.
14307 (riscv_hard_regno_mode_ok_p): Likewise.
14308 (riscv_hard_regno_nregs): Likewise.
14309 (riscv_class_max_nregs): Likewise.
14310 (riscv_memory_move_cost): Likewise.
14311 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
14312 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
14313 (rl78_addr_space_address_mode): Likewise.
14314 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14316 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14317 (rs6000_reassociation_width): Likewise.
14318 (rs6000_invalid_binary_op): Likewise.
14319 (fusion_p9_p): Likewise.
14320 (emit_fusion_p9_load): Likewise.
14321 (emit_fusion_p9_store): Likewise.
14322 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
14323 (ok_for_simple_move_operands): Likewise.
14324 (ok_for_simple_move_strict_operands): Likewise.
14325 (ok_for_simple_arith_logic_operands): Likewise.
14326 (visium_legitimize_reload_address): Likewise.
14327 (visium_select_cc_mode): Likewise.
14328 (output_cbranch): Likewise.
14329 (visium_split_double_move): Likewise.
14330 (visium_expand_copysign): Likewise.
14331 (visium_expand_int_cstore): Likewise.
14332 (visium_expand_fp_cstore): Likewise.
14333 * config/visium/visium.c (visium_pass_by_reference): Likewise.
14334 (visium_function_arg): Likewise.
14335 (visium_function_arg_advance): Likewise.
14336 (visium_libcall_value): Likewise.
14337 (visium_setup_incoming_varargs): Likewise.
14338 (visium_legitimate_constant_p): Likewise.
14339 (visium_legitimate_address_p): Likewise.
14340 (visium_legitimize_address): Likewise.
14341 (visium_secondary_reload): Likewise.
14342 (visium_register_move_cost): Likewise.
14343 (visium_memory_move_cost): Likewise.
14344 (prepare_move_operands): Likewise.
14345 (ok_for_simple_move_operands): Likewise.
14346 (ok_for_simple_move_strict_operands): Likewise.
14347 (ok_for_simple_arith_logic_operands): Likewise.
14348 (visium_function_value_1): Likewise.
14349 (rtx_ok_for_offset_p): Likewise.
14350 (visium_legitimize_reload_address): Likewise.
14351 (visium_split_double_move): Likewise.
14352 (visium_expand_copysign): Likewise.
14353 (visium_expand_int_cstore): Likewise.
14354 (visium_expand_fp_cstore): Likewise.
14355 (visium_split_cstore): Likewise.
14356 (visium_select_cc_mode): Likewise.
14357 (visium_split_cbranch): Likewise.
14358 (output_cbranch): Likewise.
14359 (visium_print_operand_address): Likewise.
14360 * expmed.c (flip_storage_order): Likewise.
14361 * expmed.h (emit_cstore): Likewise.
14362 (flip_storage_order): Likewise.
14363 * genrecog.c (validate_pattern): Likewise.
14364 * hsa-gen.c (gen_hsa_addr): Likewise.
14365 * internal-fn.c (expand_arith_overflow): Likewise.
14366 * ira-color.c (allocno_copy_cost_saving): Likewise.
14367 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14368 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
14369 (process_invariant_for_inheritance): Likewise.
14370 * lra-eliminations.c (move_plus_up): Likewise.
14371 * omp-low.c (lower_oacc_reductions): Likewise.
14372 * simplify-rtx.c (simplify_subreg): Likewise.
14373 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
14374 (TARGET_CHKP_BOUND_MODE): Likewise..
14375 * targhooks.c (default_chkp_bound_mode): Likewise.
14376 (default_setup_incoming_vararg_bounds): Likewise.
14377 * targhooks.h (default_chkp_bound_mode): Likewise.
14378 (default_setup_incoming_vararg_bounds): Likewise.
14379 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
14380 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
14381 (have_whole_vector_shift): Likewise.
14382 * tree-vect-stmts.c (vectorizable_load): Likewise.
14383 * doc/tm.texi: Regenerate.
14385 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
14387 Graceful degrade if Binutils PR21472 is not available.
14390 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
14391 .rodata in flash test fails.
14392 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
14393 * confgure: Regenerate.
14394 * config.in: Regenerate.
14395 * config/avr/avr.c (avr_asm_named_section)
14396 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
14397 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
14398 (avr_asm_init_sections): Same.
14400 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14402 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
14403 (fma<VH:mode>4_intrinsic): Likewise.
14404 (*fmsub<VCVTF:mode>4): Likewise.
14405 (*fmsub<VH:mode>4_intrinsic): Likewise.
14407 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
14410 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
14411 Don't depend on "optimize > 0".
14412 (out_movhi_r_mr, out_movqi_mr_r): Same.
14413 (out_movhi_mr_r, out_movqi_r_mr): Same.
14414 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
14415 io_address_operand on "optimize > 0".
14417 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14419 * tree-loop-distribution.c: Add general explanantion on the pass.
14420 (generate_loops_for_partition): Mark distributed loop.
14421 (pg_add_dependence_edges): New parameter. Handle alias data
14422 dependence specially and record it in the parameter if asked.
14423 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
14424 (init_partition_graph_vertices, add_partition_graph_edge): New.
14425 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
14426 (free_partition_graph_vdata, build_partition_graph): New.
14427 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
14428 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
14429 (data_ref_segment_size, latch_dominated_by_data_ref): New.
14430 (compute_alias_check_pairs, version_loop_by_alias_check): New.
14431 (version_for_distribution_p, finalize_partitions): New.
14432 (distribute_loop): Handle alias data dependence specially. Factor
14433 out loop fusion code as functions and call these functions.
14435 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14437 * tree-loop-distribution.c (classify_partition): New parameter and
14438 better handle reduction statement.
14439 (rdg_build_partitions): Revise comment.
14440 (distribute_loop): Compute statements in all partitions and pass it
14441 to classify_partition.
14443 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14445 * tree-loop-distribution.c (enum partition_type): New.
14446 (struct partition): New field type.
14447 (partition_merge_into): Add parameter. Update partition type.
14448 (data_dep_in_cycle_p, update_type_for_merge): New functions.
14449 (build_rdg_partition_for_vertex): Compute partition type.
14450 (rdg_build_partitions): Dump partition type.
14451 (distribute_loop): Update calls to partition_merge_into.
14453 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14455 * tree-loop-distribution.c (struct ddr_hasher): New.
14456 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
14458 (classify_partition): Call get_data_dependence.
14459 (pg_add_dependence_edges): Ditto.
14460 (distribute_loop): Release data dependence hash table.
14462 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14464 * tree-loop-distribution.c (ref_base_address): Delete.
14465 (similar_memory_accesses): Rename ...
14466 (share_memory_accesses): ... to this. Check if partitions access
14467 the same memory reference.
14468 (distribute_loop): Call share_memory_accesses.
14470 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14472 * tree-loop-distribution.c (struct partition): New field recording
14473 its data reference.
14474 (partition_alloc, partition_free): Init and release data refs.
14475 (partition_merge_into): Merge data refs.
14476 (build_rdg_partition_for_vertex): Collect data refs for partition.
14477 (pg_add_dependence_edges): Change parameters from vector to bitmap.
14479 (distribute_loop): Remve data refs from vertice data of partition
14482 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14484 * tree-loop-distribution.c (params.h): Include header file.
14485 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
14486 (datarefs_vec): New global var.
14487 (create_rdg_vertices): Use datarefs_vec directly.
14488 (free_rdg): Don't free data references.
14489 (build_rdg): Update use. Don't free data references.
14490 (distribute_loop): Compute global variable for data references.
14491 Bail out if there are too many data references.
14493 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14495 * tree-loop-distribution.c (loop_nest): New global var.
14496 (build_rdg): Use loop directly, rather than loop nest.
14497 (pg_add_dependence_edges): Remove loop nest parameter. Use global
14499 (distribute_loop): Compute global variable loop nest. Update use.
14501 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14503 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
14504 (partition_merge_into): New parameter. Dump reason for fusion.
14505 (distribute_loop): Update use of partition_merge_into.
14507 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14509 * tree-loop-distribution.c (bb_top_order_index): New.
14510 (bb_top_order_index_size, bb_top_order_cmp): New.
14511 (stmts_from_loop): Use topological order.
14512 (pass_loop_distribution::execute): Compute and release topological
14513 order for basic blocks.
14515 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14517 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
14520 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14522 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
14523 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
14524 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
14525 * internal-fn.def (LOOP_DIST_ALIAS): New.
14526 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
14527 (fold_loop_internal_call): ... this.
14528 (vect_loop_dist_alias_call): New function.
14529 (set_uid_loop_bbs): Call fold_loop_internal_call.
14530 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
14533 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
14536 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
14537 Require dead FLAGS_REG at the beginning of a peephole.
14539 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
14542 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
14543 arguments in the call to __builtin_ia32_sbb_u32.
14544 (_subborrow_u64): Swap _X and _Y arguments in the call to
14545 __builtin_ia32_sbb_u64.
14547 2017-07-04 Jakub Jelinek <jakub@redhat.com>
14550 * tree-vrp.c (compare_assert_loc): Turn into a function template
14551 with stable template parameter. Only test if a->e is NULL,
14552 !a->e == !b->e has been verified already. Use e == NULL or
14553 e != NULL instead of e or ! e tests. If stable is true, don't use
14554 iterative_hash_expr, on the other side allow a or b or both NULL
14555 and sort the NULLs last.
14556 (process_assert_insertions): Sort using compare_assert_loc<false>
14557 instead of compare_assert_loc, later sort using
14558 compare_assert_loc<true> before calling process_assert_insertions_for
14559 in a loop. Use break instead of continue once seen NULL pointer.
14561 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
14563 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
14564 Cortex-R7 and Cortex-R8 processors.
14566 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14568 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
14569 uninitialized while src is not.
14571 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
14573 * common/config/arm/arm-common.c: Adjust include path for
14575 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
14576 (arm-cpu.h): Create in build directory. Adjust dependency rules.
14577 (arm-cpu-data.h): Likewise.
14578 (arm-cpu-cdata.h): Likewise.
14579 * config/arm/arm-cpu.h: Delete.
14580 * config/arm/arm-cpu-cdata.h: Delete.
14581 * config/arm/arm-cpu-data.h: Delete.
14583 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
14585 * config/arm/arm-cpus.in (cortex-a55): New.
14586 (cortex-a75): Likewise.
14587 (cortex-a75.cortex-a55): Likewise.
14588 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
14590 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
14591 * config/arm/arm-cpu-cdata.h: Regenerate.
14592 * config/arm/arm-cpu-data.h: Regenerate.
14593 * config/arm/arm-cpu.h: Regenerate.
14594 * config/arm/arm-tables.opt: Regenerate.
14595 * config/arm/arm-tune.md: Regenerate.
14597 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14599 * haifa-sched.c (sched_create_recovery_edges): Update profile.
14601 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14603 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
14606 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
14608 PR tree-optimization/81292
14609 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
14610 full_string_p, also call adjust_related_strinfos if the adjustment
14611 is simple, otherwise invalidate related strinfos.
14613 2017-07-04 Martin Liska <mliska@suse.cz>
14616 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
14617 newly created variable as DECL_IGNORED_P.
14619 2017-07-04 Martin Liska <mliska@suse.cz>
14622 * ipa-inline.c (inline_small_functions):
14623 Use xstrdup_for_dump.
14625 2017-07-04 Tom de Vries <tom@codesourcery.com>
14627 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
14629 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
14632 * config/darwin.c (darwin_function_switched_text_sections):
14635 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
14637 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
14639 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
14641 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
14643 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14645 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
14646 min_profitable_iters, and th as inclusive lower bounds.
14647 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
14648 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
14649 for min_profitable_iters and min_profitable_estimate.
14650 (vect_transform_loop): Treat th as an inclusive lower bound.
14651 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
14653 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
14656 * config/darwin.c (darwin_function_switched_text_sections):
14657 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
14658 in two pieces, and suppress the use of buf.
14660 2017-07-03 Nathan Sidwell <nathan@acm.org>
14662 * hash-table.h (hash_table_mod1): Fix indentation.
14664 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14666 PR middle-end/81290
14667 * predict.c (force_edge_cold): Be more careful about propagation
14669 * profile-count.h (profile_probability::guessed,
14670 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
14672 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
14674 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
14676 * doc/invoke.texi (rcpc architecture extension): Document it.
14678 2017-07-03 Richard Biener <rguenther@suse.de>
14680 PR tree-optimization/60510
14681 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
14682 the scalar reduction PHI and use it.
14683 (vectorizable_reduction): Properly guard the single_defuse_cycle
14684 path for non-SLP reduction chains where we cannot use it.
14685 Rework reduc_def/index and vector type deduction. Rework
14686 vector operand gathering during reduction op code-gen.
14687 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
14688 chains dissolve the chain and leave it to non-SLP reduction
14691 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14693 * tree-data-ref.h (dr_alignment): Declare.
14694 * tree-data-ref.c (dr_alignment): New function.
14695 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
14696 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
14698 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
14700 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14702 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
14703 and base_misalignment fields.
14704 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
14705 * tree-data-ref.c: Include builtins.h.
14706 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
14707 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
14708 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
14709 * tree-vect-data-refs.c: Include tree-cfg.h.
14710 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
14711 fields instead of calculating an alignment here.
14712 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
14713 innermost_loop_behavior fields.
14715 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14717 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
14719 (DR_STEP_ALIGNMENT): New macro.
14720 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
14721 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
14722 (create_data_ref): Print it.
14723 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
14724 to tell whether the step preserves vector (mis)alignment.
14725 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
14726 Move the check for an integer step and generalise to all INTEGER_CST.
14727 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
14728 Print the outer step alignment.
14730 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14732 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
14733 with offset_alignment.
14734 (DR_ALIGNED_TO): Delete.
14735 (DR_OFFSET_ALIGNMENT): New macro.
14736 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
14737 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
14738 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
14739 (create_data_ref): Likewise.
14740 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
14741 (vect_analyze_data_refs): Likewise.
14742 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
14743 creating dummy innermost behavior.
14745 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14747 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
14748 with a "innermost_loop_behavior *" and refeence tree.
14749 * tree-data-ref.c (dr_analyze_innermost): Likewise.
14750 (create_data_ref): Update call accordingly.
14751 * tree-predcom.c (find_looparound_phi): Likewise.
14753 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14755 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
14756 fields with dr_wrt_vec_loop.
14757 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
14758 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
14759 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
14760 (vect_dr_behavior): New function.
14761 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
14762 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
14763 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
14764 track whether the step preserves the misalignment.
14765 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
14766 Use vect_dr_behavior.
14767 (vect_setup_realignment): Update call accordingly.
14768 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
14769 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
14770 call to vect_create_addr_base_for_vector_ref.
14771 (vect_create_cond_for_align_checks): Likewise.
14772 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
14773 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
14774 (vect_recog_mask_conversion_pattern): Likewise.
14775 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
14776 (new_stmt_vec_info): Remove redundant zeroing.
14778 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
14780 * common/config/arm/arm-common.c (arm_be8_option): New function.
14781 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
14782 (ISA_ARMv6): Add isa_bit_be8.
14783 * config/arm/arm.h (arm_be8_option): Add prototype.
14784 (BE8_SPEC_FUNCTION): New define.
14785 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
14786 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
14787 (mlittle-endian): Similarly.
14788 (mbe8, mbe32): New options.
14789 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
14790 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
14792 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14794 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
14796 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14798 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
14799 (cleanup_tree_cfg_bb): Use it.
14800 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
14802 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
14804 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14807 * loop-doloop.c (add_test): Update profile.
14809 2017-07-03 Martin Liska <mliska@suse.cz>
14812 * sanopt.c (rewrite_usage_of_param): New function.
14813 (sanitize_rewrite_addressable_params): Likewise.
14814 (pass_sanopt::execute): Call rewrite_usage_of_param.
14816 2017-07-03 Richard Biener <rguenther@suse.de>
14818 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
14819 back to using VIEW_CONVERT_EXPR.
14821 2017-07-03 Martin Liska <mliska@suse.cz>
14824 * doc/extend.texi: Document when a resolver function is
14825 generated for target_clones.
14827 2017-07-03 Martin Liska <mliska@suse.cz>
14829 * asan.c (asan_emit_stack_protection): Unpoison just red zones
14830 and shadow memory of auto variables which are subject of
14831 use-after-scope sanitization.
14832 (asan_expand_mark_ifn): Add do set only when is_poison.
14834 2016-07-03 Richard Biener <rguenther@suse.de>
14836 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
14838 (vect_force_simple_reduction): Record reduction def -> phi mapping.
14839 (vectorizable_reduction): Perform reduction PHI creation when
14840 visiting a reduction PHI and adjust and simplify code generation
14841 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
14842 (vect_transform_loop): Visit reduction PHIs.
14843 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
14844 defs into the SLP tree.
14845 (vect_build_slp_tree): Reduction defs terminate the recursion.
14846 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
14848 (vect_get_vec_defs_for_stmt_copy): Export.
14849 (vect_get_vec_defs): Likewise.
14850 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
14852 (vect_get_vec_defs_for_stmt_copy): Declare.
14853 (vect_get_vec_defs): Likewise.
14855 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14857 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
14858 parameter with a "loop" parameter and use it instead of the
14859 loop containing DR_STMT. Don't check simple_iv when doing
14860 BB analysis. Describe the two analysis modes in the comment.
14862 2017-07-03 Tom de Vries <tom@codesourcery.com>
14864 PR tree-optimization/69468
14865 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
14866 (find_same_succ_bb): Handle ignore_edge_flags.
14868 2017-07-03 Tom de Vries <tom@codesourcery.com>
14870 PR tree-optimization/81192
14871 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
14873 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
14875 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
14877 2017-07-03 Tom de Vries <tom@codesourcery.com>
14879 PR tree-optimization/81192
14880 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
14881 BB_SAME_SUCC (bb) == NULL.
14883 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14885 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
14888 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14890 * dumpfile.c: Include profile-count.h
14891 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
14893 (insert_cond_bb): Update profile.
14894 * tree-cfg.h (insert_cond_bb): Update prototype.
14895 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
14896 * tree-dump.c: Do not include tree-cfg.
14898 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14900 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
14902 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14904 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
14907 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14909 * tree-complex.c (expand_complex_div_wide): update profile.
14911 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14912 Alan Hayward <alan.hayward@arm.com>
14913 David Sherwood <david.sherwood@arm.com>
14915 * Makefile.in (MACHMODE_H): Remove insn-modes.h
14916 (CORETYPES_H): New define.
14917 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
14918 (insn-modes-inline.h, s-modes-inline-h): New rules.
14919 (generated_files): Add insn-modes-inline.h.
14920 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
14921 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
14922 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
14923 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
14924 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
14925 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
14926 (build/gencodes.o, build/genconditions.o): Likewise.
14927 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
14928 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
14929 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
14930 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
14931 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
14932 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
14933 * coretypes.h: Include everything up to real.h for generators.
14934 Include insn-modes.h first. Include wide-int-print.h after
14935 wide-int.h. Include insn-modes-inline.h and then machmode.h.
14936 * machmode.h: Don't include insn-modes.h here.
14937 * function-tests.c: Remove includes of signop.h, machmode.h,
14938 double-int.h and wide-int.h.
14940 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
14942 * optc-save-gen.awk: Likewise.
14943 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
14944 * godump.c: Remove include of wide-int-print.h.
14945 * pretty-print.h: Likewise.
14946 * wide-int-print.cc: Likewise.
14947 * wide-int.cc: Likewise.
14948 * hash-map-tests.c: Remove include of signop.h.
14949 * hash-set-tests.c: Likewise.
14950 * rtl-tests.c: Likewise.
14951 * mkconfig.sh: Remove include of machmode.h.
14952 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
14954 (emit_insn_modes_inline_h): ...this new function. Emit the code
14955 into an insn-modes-inline.h header file, adding appropriate
14956 include guards and end comments.
14957 (emit_insn_modes_c_header): Remove include of machmode.h.
14958 (emit_min_insn_modes_c_header): Include coretypes.h rather than
14960 (main): Handle -i flag and call emit_insn_modes_inline_h when
14963 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14965 * tree-ssa-strlen.c (strinfo): Rename the length field to
14966 nonzero_chars. Add a full_string_p field.
14967 (compare_nonzero_chars, zero_length_string_p): New functions.
14968 (get_addr_stridx): Add an offset_out parameter.
14969 Use compare_nonzero_chars.
14970 (get_stridx): Update accordingly. Use compare_nonzero_chars.
14971 (new_strinfo): Update after above changes to strinfo.
14972 (set_endptr_and_length): Set full_string_p.
14973 (get_string_length): Update after above changes to strinfo.
14974 (unshare_strinfo): Update call to new_strinfo.
14975 (maybe_invalidate): Likewise.
14976 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
14977 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
14978 as a uhwi instead of an shwi. Update after above changes to
14979 strinfo and new_strinfo.
14980 (zero_length_string): Assert that chainsi contains full strings.
14981 Use zero_length_string_p. Update call to new_strinfo.
14982 (adjust_related_strinfos): Update after above changes to strinfo.
14983 Copy full_string_p from origsi.
14984 (adjust_last_stmt): Use zero_length_string_p.
14985 (handle_builtin_strlen): Update after above changes to strinfo and
14986 new_strinfo. Install the lhs as the string length if the previous
14987 entry didn't describe a full string.
14988 (handle_builtin_strchr): Update after above changes to strinfo
14990 (handle_builtin_strcpy): Likewise.
14991 (handle_builtin_strcat): Likewise.
14992 (handle_builtin_malloc): Likewise.
14993 (handle_pointer_plus): Likewise.
14994 (handle_builtin_memcpy): Likewise. Track nonzero characters
14995 that aren't necessarily followed by a nul terminator.
14996 (handle_char_store): Likewise.
14998 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15000 PR tree-optimization/80769
15001 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
15002 for malloc and calloc. Document the new invariant that all related
15003 strinfos have delayed lengths or none do.
15004 (verify_related_strinfos): Move earlier in file.
15005 (set_endptr_and_length): New function, split out from...
15006 (get_string_length): ...here. Also set the lengths of related
15008 (zero_length_string): Assert that chainsi has known (rather than
15010 (adjust_related_strinfos): Likewise.
15012 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15014 PR tree-optimization/81136
15015 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
15016 assert that two references with the same misalignment have the same
15017 compile-time misalignment if those compile-time misalignments
15020 2017-07-01 Andi Kleen <ak@linux.intel.com>
15022 * print-tree.c (print_node): Print all attributes.
15024 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15026 * cfg.c (scale_bbs_frequencies): New function.
15027 * cfg.h (scale_bbs_frequencies): Declare it.
15028 * cfgloopanal.c (single_likely_exit): Cleanup.
15029 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
15031 (scale_loop_profile): Likewise.
15032 (loop_version): Likewise.
15033 (create_empty_loop_on_edge): Update.
15034 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
15035 scale_loop_frequencies, scale_loop_profile, loopify,
15036 loop_version): Update prototypes.
15037 * modulo-sched.c (sms_schedule): Update.
15038 * predict.c (unlikely_executed_edge_p): Also check probability.
15039 (probably_never_executed_edge_p): Fix typo.
15040 * tree-if-conv.c (version_loop_for_if_conversion): Update.
15041 * tree-parloops.c (gen_parallel_loop): Update.
15042 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
15043 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
15044 * tree-ssa-loop-split.c (split_loop): Update.
15045 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
15046 * tree-vect-loop-manip.c (vect_do_peeling): Update.
15047 (vect_loop_versioning): Update.
15048 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
15050 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15052 * trans-mem.c (split_bb_make_tm_edge): Update profile.
15054 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15056 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
15057 to keep profile consistent.
15059 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15061 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
15062 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
15063 * profile-count.h (max_safe_multiplier): Make unsigned.
15064 (profile_count::guessed_zero): New.
15066 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15068 * bb-reorder.c (fix_up_crossing_landing_pad,
15069 fix_crossing_conditional_branches): Use make_single_succ_edge
15070 to keep profile consistent.
15072 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15074 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
15077 2017-07-01 Jakub Jelinek <jakub@redhat.com>
15080 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
15081 the right scopes, make sure cond_jump isn't preserved between multiple
15082 iterations. Search for fallthru edge whenever there are 3+ edges and
15083 use find_fallthru_edge for it.
15085 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15087 Patch by Alexander Monakov <amonakov@ispras.ru>
15088 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
15089 probabilities consistently.
15091 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15093 * pa.c (pa_expand_compare_and_swap_loop): Update call of
15094 emit_cmp_and_jump_insns.
15096 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15099 * tree-inline.c (expand_call_inline): Combine profile statuses.
15101 2017-06-30 Andrew Pinski <apinski@cavium.com>
15103 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
15104 fold_stmt returned true.
15106 2017-06-30 Nathan Sidwell <nathan@acm.org>
15108 * ggc.h (empty_string): Delete.
15109 * cfgexpand.c (expand_asm_stmt): Use plain "".
15110 * optabs.c (expand_asm_memory_barrier): Likewise.
15111 * stringpool.c (empty_string): Delete.
15112 (digit_vector, digit_string): Delete.
15113 (ggc_alloc_string): Use plain "", don't optimize single digit
15114 strings. Use ggc_alloc_atomic.
15116 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
15118 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
15119 comparison set and one other set, use the cost of the non-comparison
15122 2017-06-30 Nathan Sidwell <nathan@acm.org>
15124 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
15127 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
15129 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
15130 loops. Remove now unneeded calls to gimple_switch_set_label() that
15131 just set removed labels to NULL_TREE.
15133 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
15135 * tree-ssanames.c (set_range_info_raw): Abstract from ...
15136 (set_range_info): ...here. Only call set_range_info_raw if domain
15138 (set_nonzero_bits): Call set_range_info_raw.
15139 * tree-ssanames.h (set_range_info_raw): New.
15141 2017-06-30 Jakub Jelinek <jakub@redhat.com>
15144 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
15145 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
15146 of nonimmediate_operand and <store_mask_constraint> instead of m
15147 for the input operand. For V8FI iterator, always split if input
15148 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
15149 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
15150 <store_mask_predicate> instead of register_operand and
15151 <store_mask_constraint> instead of v for the input operand. Make
15152 sure both operands aren't MEMs for if not <mask_applied>.
15154 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
15156 * lto-wrapper.c (copy_file) Close both file descriptors before
15159 2017-06-30 Martin Liska <mliska@suse.cz>
15162 * multiple_target.c (create_dispatcher_calls): Make ifunc
15163 also for function that don't have calls or are not referenced.
15165 2017-06-30 Richard Biener <rguenther@suse.de>
15167 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
15168 analyze the first scalar stmt. Move vector type computation
15169 for the BB case here from ...
15170 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
15171 live operation processing in the SLP case properly.
15173 2017-06-30 Richard Biener <rguenther@suse.de>
15175 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
15177 2017-06-30 Martin Liska <mliska@suse.cz>
15180 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
15181 before BUILT_IN_UNWIND_RESUME when ASAN is used.
15183 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
15185 * doc/invoke.texi (AArch64): Add missing options and remove redundant
15188 2017-06-30 Richard Biener <rguenther@suse.de>
15190 PR tree-optimization/81249
15191 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
15192 condition reduction result to original scalar type.
15194 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15196 * profile-count.h (enum profile_quality): Fix typos and whitespace
15199 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15201 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
15202 type for branch probabilities.
15204 2017-06-29 Julian Brown <julian@codesourcery.com>
15205 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15207 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
15208 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
15209 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
15210 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
15212 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15214 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
15215 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
15216 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
15217 CC usage from generic code to here.
15218 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
15219 CC usage into the target macros.
15221 2017-06-29 Maya Rashish <coypu@sdf.org>
15223 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
15226 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15228 * arm/arm-builtins.c: Include profile-count.h
15229 * except.c (sjlj_emit_function_enter): Use
15230 profile_probability::unlikely.
15232 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15234 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
15235 and tocrel_offset be pointer args rather than implicitly using
15237 (legitimate_constant_pool_address_p, rs6000_emit_move,
15238 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
15239 tocrel_offset and use in toc_relative_expr_p call.
15240 (print_operand, print_operand_address): Use static tocrel_base_oac
15241 and tocrel_offset_oac.
15242 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
15245 2017-06-29 Maya Rashish <coypu@sdf.org>
15247 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
15249 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
15251 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
15252 objects, take into account only the alignment of 'op0' and 'mode1' if
15255 2017-06-29 Steve Ellcey <sellcey@cavium.com>
15257 * ccmp.c (ccmp_tree_comparison_p): New function.
15258 (ccmp_candidate_p): Update to use above function.
15259 (get_compare_parts): New function.
15260 (expand_ccmp_next): Update to use new functions.
15261 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
15263 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
15264 take mode as argument.
15265 * ccmp.h (expand_ccmp_expr): Add mode as argument.
15266 * expr.c (expand_expr_real_1): Pass mode as argument.
15268 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
15270 * combine.c (combine_instructions): Print insns to dump_file, together
15273 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15275 * asan.c (asan_emit_stack_protection): Update.
15276 (create_cond_insert_point): Update.
15277 * auto-profile.c (afdo_propagate_circuit): Update.
15278 * basic-block.h (struct edge_def): Turn probability to
15279 profile_probability.
15280 (EDGE_FREQUENCY): Update.
15281 * bb-reorder.c (find_traces_1_round): Update.
15282 (better_edge_p): Update.
15283 (sanitize_hot_paths): Update.
15284 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
15285 (make_single_succ_edge): Update.
15286 (check_bb_profile): Update.
15287 (dump_edge_info): Update.
15288 (update_bb_profile_for_threading): Update.
15289 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
15290 probabilitycount to 0.
15291 * cfgbuild.c (compute_outgoing_frequencies): Update.
15292 * cfgcleanup.c (try_forward_edges): Update.
15293 (outgoing_edges_match): Update.
15294 (try_crossjump_to_edge): Update.
15295 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
15296 (expand_gimple_tailcall): Update.
15297 (construct_init_block): Use make_single_succ_edge.
15298 (construct_exit_block): Use make_single_succ_edge.
15299 * cfghooks.c (verify_flow_info): Update.
15300 (redirect_edge_succ_nodup): Update.
15301 (split_edge): Update.
15302 (account_profile_record): Update.
15303 * cfgloopanal.c (single_likely_exit): Update.
15304 * cfgloopmanip.c (scale_loop_profile): Update.
15305 (set_zero_probability): Remove.
15306 (duplicate_loop_to_header_edge): Update.
15307 * cfgloopmanip.h (loop_version): Update prototype.
15308 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
15309 (force_nonfallthru_and_redirect): Update.
15310 (update_br_prob_note): Update.
15311 (rtl_verify_edges): Update.
15312 (purge_dead_edges): Update.
15313 (rtl_lv_add_condition_to_bb): Update.
15314 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
15315 * cgraphunit.c (init_lowered_empty_function): Update.
15316 (cgraph_node::expand_thunk): Update.
15317 * cilk-common.c: Include profile-count.h
15318 * dojump.c (inv): Remove.
15319 (jumpifnot): Update.
15320 (jumpifnot_1): Update.
15321 (do_jump_1): Update.
15323 (do_jump_by_parts_greater_rtx): Update.
15324 (do_compare_rtx_and_jump): Update.
15325 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
15326 do_jump_1. do_compare_rtx_and_jump): Update prototype.
15327 * dwarf2cfi.c: Include profile-count.h
15328 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
15329 (sjlj_emit_dispatch_table): Likewise.
15330 * explow.c: Include profile-count.h
15331 * expmed.c (emit_store_flag_force): Update.
15332 (do_cmp_and_jump): Update.
15333 * expr.c (compare_by_pieces_d::generate): Update.
15334 (compare_by_pieces_d::finish_mode): Update.
15335 (emit_block_move_via_loop): Update.
15336 (store_expr_with_bounds): Update.
15337 (store_constructor): Update.
15338 (expand_expr_real_2): Update.
15339 (expand_expr_real_1): Update.
15340 * expr.h (try_casesi, try_tablejump): Update prototypes.
15341 * gimple-pretty-print.c (dump_probability): Update.
15342 (dump_profile): New.
15343 (dump_gimple_label): Update.
15344 (dump_gimple_bb_header): Update.
15345 * graph.c (draw_cfg_node_succ_edges): Update.
15346 * hsa-gen.c (convert_switch_statements): Update.
15347 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
15348 (find_if_case_1): Update.
15349 (find_if_case_2): Update.
15350 * internal-fn.c (expand_arith_overflow_result_store): Update.
15351 (expand_addsub_overflow): Update.
15352 (expand_neg_overflow): Update.
15353 (expand_mul_overflow): Update.
15354 (expand_vector_ubsan_overflow): Update.
15355 * ipa-cp.c (good_cloning_opportunity_p): Update.
15356 * ipa-split.c (split_function): Use make_single_succ_edge.
15357 * ipa-utils.c (ipa_merge_profiles): Update.
15358 * loop-doloop.c (add_test): Update.
15359 (doloop_modify): Update.
15360 * loop-unroll.c (compare_and_jump_seq): Update.
15361 (unroll_loop_runtime_iterations): Update.
15362 * lra-constraints.c (lra_inheritance): Update.
15363 * lto-streamer-in.c (input_cfg): Update.
15364 * lto-streamer-out.c (output_cfg): Update.
15365 * mcf.c (adjust_cfg_counts): Update.
15366 * modulo-sched.c (sms_schedule): Update.
15367 * omp-expand.c (expand_omp_for_init_counts): Update.
15368 (extract_omp_for_update_vars): Update.
15369 (expand_omp_ordered_sink): Update.
15370 (expand_omp_for_ordered_loops): Update.
15371 (expand_omp_for_generic): Update.
15372 (expand_omp_for_static_nochunk): Update.
15373 (expand_omp_for_static_chunk): Update.
15374 (expand_cilk_for): Update.
15375 (expand_omp_simd): Update.
15376 (expand_omp_taskloop_for_outer): Update.
15377 (expand_omp_taskloop_for_inner): Update.
15378 * omp-simd-clone.c (simd_clone_adjust): Update.
15379 * optabs.c (expand_doubleword_shift): Update.
15380 (expand_abs): Update.
15381 (emit_cmp_and_jump_insn_1): Update.
15382 (expand_compare_and_swap_loop): Update.
15383 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
15384 * predict.c (predictable_edge_p): Update.
15385 (edge_probability_reliable_p): Update.
15386 (set_even_probabilities): Update.
15387 (combine_predictions_for_insn): Update.
15388 (combine_predictions_for_bb): Update.
15389 (propagate_freq): Update.
15390 (estimate_bb_frequencies): Update.
15391 (force_edge_cold): Update.
15392 * profile-count.c (profile_count::dump): Add missing space into dump.
15393 (profile_count::debug): Add newline.
15394 (profile_count::differs_from_p): Explicitly convert to unsigned.
15395 (profile_count::stream_in): Update.
15396 (profile_probability::dump): New member function.
15397 (profile_probability::debug): New member function.
15398 (profile_probability::differs_from_p): New member function.
15399 (profile_probability::differs_lot_from_p): New member function.
15400 (profile_probability::stream_in): New member function.
15401 (profile_probability::stream_out): New member function.
15402 * profile-count.h (profile_count_quality): Rename to ...
15403 (profile_quality): ... this one.
15404 (profile_probability): New.
15405 (profile_count): Update.
15406 * profile.c (compute_branch_probabilities): Update.
15407 * recog.c (peep2_attempt): Update.
15408 * sched-ebb.c (schedule_ebbs): Update.
15409 * sched-rgn.c (find_single_block_region): Update.
15410 (compute_dom_prob_ps): Update.
15411 (schedule_region): Update.
15412 * sel-sched-ir.c (compute_succs_info): Update.
15413 * stmt.c (struct case_node): Update.
15414 (do_jump_if_equal): Update.
15415 (get_outgoing_edge_probs): Update.
15416 (conditional_probability): Update.
15417 (emit_case_dispatch_table): Update.
15418 (expand_case): Update.
15419 (expand_sjlj_dispatch_table): Update.
15420 (emit_case_nodes): Update.
15421 * targhooks.c: Update.
15422 * tracer.c (better_p): Update.
15423 (find_best_successor): Update.
15424 * trans-mem.c (expand_transaction): Update.
15425 * tree-call-cdce.c: Update.
15426 * tree-cfg.c (gimple_split_edge): Upate.
15427 (move_sese_region_to_fn): Upate.
15428 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
15429 * tree-eh.c (lower_resx): Upate.
15430 (cleanup_empty_eh_move_lp): Upate.
15431 * tree-if-conv.c (version_loop_for_if_conversion): Update.
15432 * tree-inline.c (copy_edges_for_bb): Update.
15433 (copy_cfg_body): Update.
15434 * tree-parloops.c (gen_parallel_loop): Update.
15435 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
15436 (gimple_gen_time_profiler): Update.
15437 * tree-ssa-dce.c (remove_dead_stmt): Update.
15438 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
15439 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
15440 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
15441 (unloop_loops): Update.
15442 (try_peel_loop): Update.
15443 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
15444 * tree-ssa-loop-split.c (connect_loops): Update.
15445 (split_loop): Update.
15446 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
15447 (hoist_guard): Update.
15448 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
15449 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
15450 (value_replacement): Update.
15451 * tree-ssa-reassoc.c (branch_fixup): Update.
15452 * tree-ssa-tail-merge.c (replace_block_by): Update.
15453 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
15454 (create_edge_and_update_destination_phis): Update.
15455 (compute_path_counts): Update.
15456 (recompute_probabilities): Update.
15457 (update_joiner_offpath_counts): Update.
15458 (freqs_to_counts_path): Update.
15459 (duplicate_thread_path): Update.
15460 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
15461 (struct switch_conv_info): Update.
15462 (gen_inbound_check): Update.
15463 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
15464 (vect_do_peeling): Update.
15465 (vect_loop_versioning): Update.
15466 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
15467 (optimize_mask_stores): Update.
15468 * ubsan.c (ubsan_expand_null_ifn): Update.
15469 * value-prof.c (gimple_divmod_fixed_value): Update.
15470 (gimple_divmod_fixed_value_transform): Update.
15471 (gimple_mod_pow2): Update.
15472 (gimple_mod_pow2_value_transform): Update.
15473 (gimple_mod_subtract): Update.
15474 (gimple_mod_subtract_transform): Update.
15475 (gimple_ic): Update.
15476 (gimple_stringop_fixed_value): Update.
15477 (gimple_stringops_transform): Update.
15478 * value-prof.h: Update.
15480 2017-06-29 Carl Love <cel@us.ibm.com>
15482 * config/rs6000/rs6000-c.c: Add support for built-in functions
15483 vector signed int vec_signed (vector float);
15484 vector signed long long vec_signed (vector double);
15485 vector signed int vec_signed2 (vector double, vector double);
15486 vector signed int vec_signede (vector double);
15487 vector signed int vec_signedo (vector double);
15488 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
15489 instruction generator.
15490 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
15491 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
15492 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
15494 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
15495 vunsignede_v2df): Add define_expands.
15496 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
15497 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
15498 VEC_UNSIGNEDO): Add definitions.
15499 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
15500 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
15501 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
15502 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
15503 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
15504 * config/rs6000/altivec.h (vec_signed, vec_signed2,
15505 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
15506 vec_unsignede, vec_unsignedo): Add builtin defines.
15507 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
15509 * doc/extend.texi: Update the built-in documentation file for the
15510 new built-in functions.
15512 2017-06-29 Richard Biener <rguenther@suse.de>
15514 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
15515 reduction chains to LOOP_VINFO_REDUCTIONS.
15516 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
15517 SLP reductions after processing reduction chains.
15519 2017-06-29 Nathan Sidwell <nathan@acm.org>
15521 * builtins.c (fold_builtin_FUNCTION): Use
15522 lang_hooks.decl_printable_name.
15524 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
15526 PR middle-end/81194
15527 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
15528 with only one label.
15529 * stmt.c (expand_case): Assert NCASES is greater than one.
15531 2017-06-29 Richard Biener <rguenther@suse.de>
15533 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
15535 (group_case_labels): Likewise.
15536 (find_taken_edge): Push sanity checking on val to workers...
15537 (find_taken_edge_cond_expr): ... here
15538 (find_taken_edge_switch_expr): ... and here, handle cases
15539 with just a default label.
15540 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
15541 (group_case_labels): Likewise.
15542 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
15543 group_case_labels does anything cleanup the CFG again.
15545 2017-06-29 Bin Cheng <bin.cheng@arm.com>
15547 PR tree-optimization/81196
15548 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
15549 exit condition comparing two IVs.
15551 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
15553 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
15554 profile to the dummy entry at the end of the list of architectures.
15555 * config/arm/arm-cpu-cdata.h: Regenerated.
15557 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15558 Michael Collison <michael.collison@arm.com>
15561 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
15563 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
15564 (*aarch64_reg_<mode>3_minus_mask): New pattern.
15565 (*aarch64_<optab>_reg_di3_mask2): New pattern.
15566 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
15567 of shift when the shift amount is masked with constant equal to
15568 the size of the mode.
15569 * config/aarch64/predicates.md (subreg_lowpart_operator): New
15572 2017-06-29 Martin Liska <mliska@suse.cz>
15574 * config/i386/i386.opt: Change range from [1,5] to [0,5].
15576 2017-06-29 Yury Gribov <tetra2005@gmail.com>
15579 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
15582 (edge_growth_cache_entry::edge_growth_cache_entry): New
15584 (reset_edge_growth_cache): Update to use constructor.
15586 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15588 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
15589 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
15590 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
15592 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
15594 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
15595 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
15597 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
15599 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
15600 (*-linux-uclibc*): Add t-uclibc tmake_file.
15601 * config/t-musl: New.
15602 * config/t-uclibc: New.
15604 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
15606 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
15608 (gen_comm_data): Emit architectural setting of arch_prof.
15609 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
15611 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
15612 (armv8-m.base, armv8-m.main): Likewise.
15613 * arm-protos.h (arm_build_target): Add profile field.
15614 (arch_option): Likewise.
15615 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
15617 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
15618 arm_active_target.profile.
15620 2017-06-28 Richard Biener <rguenther@suse.de>
15622 PR middle-end/81227
15623 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
15624 TYPE_OVERFLOW_WRAPS.
15625 * match.pd (negate_expr_p): Likewise.
15626 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
15627 fold_build2, not fold_binary.
15629 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
15631 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
15632 Convert memory address to Pmode.
15633 (aarch64_print_operand): Assert MEM operands are always Pmode.
15635 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
15638 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
15639 Remove redundant if.
15640 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
15641 * config/arm/aarch-common-protos.h
15642 (aarch_forward_to_shift_is_not_shifted_re): Remove.
15643 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
15645 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
15648 * multiple_target.c (create_dispatcher_calls): Set the default
15649 clone to be static, not public.
15651 2017-06-28 Richard Biener <rguenther@suse.de>
15653 * tree-vect-loop.c (vectorizable_reduction): Move special
15654 cond reduction IV var creation ...
15655 (vect_create_epilog_for_reduction): ... here. Remove induction_index
15656 parameter. Use STMT_VINFO_VECTYPE.
15657 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
15660 2017-06-28 Martin Liska <mliska@suse.cz>
15663 * ipa-visibility.c (non_local_p): Handle visibility.
15665 2017-06-28 Martin Liska <mliska@suse.cz>
15668 * common.opt: Add IntegerRange to various options.
15669 * opt-functions.awk (integer_range_info): New function.
15670 * optc-gen.awk: Add integer_range_info to cl_options struct.
15671 * opts-common.c (decode_cmdline_option): Handle
15672 CL_ERR_INT_RANGE_ARG.
15673 (cmdline_handle_error): Likewise.
15674 * opts.c (print_filtered_help): Show valid interval in
15675 when --help is provided.
15676 * opts.h (struct cl_option): Add range_min and range_max fields.
15677 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
15679 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
15681 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
15682 (x * C EQ/NE y * C): New transformation.
15684 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
15686 * genmultilib (combination_space): Accept '+' in option names.
15688 2017-06-28 Martin Liska <mliska@suse.cz>
15691 * asan.c (instrument_derefs): Bail out inner references
15692 that are hard register variables.
15694 2017-06-28 Jakub Jelinek <jakub@redhat.com>
15697 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
15698 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
15700 2017-06-28 Richard Biener <rguenther@suse.de>
15702 * tree-vectorizer.h (vect_get_vec_defs): Remove.
15703 (vect_get_slp_defs): Adjust.
15704 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
15706 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
15708 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
15709 get_initial_defs_for_reduction instead of vect_get_vec_defs.
15710 (vectorizable_reduction): Adjust.
15711 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
15713 (vect_get_slp_defs): Likewise.
15714 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
15715 (vectorizable_bswap): Adjust.
15716 (vectorizable_call): Likewise.
15717 (vectorizable_conversion): Likewise.
15718 (vectorizable_assignment): Likewise.
15719 (vectorizable_shift): Likewise.
15720 (vectorizable_operation): Likewise.
15721 (vectorizable_store): Likewise.
15722 (vectorizable_condition): Likewise.
15723 (vectorizable_comparison): Likewise.
15725 2017-06-28 Michael Collison <michael.collison@arm.com>
15728 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
15730 (arm_gen_movmemqi): Removed unused variable 'i'.
15731 Convert 'for' loop into 'while' loop.
15732 (arm_expand_prologue): Remove last unnecessary set of insn.
15733 (thumb_pop): Remove unused variable 'pushed_words'.
15734 (thumb_exit): Remove last unnecessary set of regs_to_pop.
15736 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15738 * config/s390/predicates.md: Use s390_rel_address_ok_p.
15739 * config/s390/s390-protos.h: Add prototype of
15740 s390_rel_address_ok_p.
15741 * config/s390/s390.c (s390_got_symbol): New function.
15742 (s390_rel_address_ok_p): New function.
15743 (legitimize_pic_address): Use s390_rel_address_ok_p.
15744 (s390_load_got): Use s390_got_symbol.
15745 (s390_option_override): Issue error if
15746 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
15747 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
15749 * config/s390/s390.opt: New option mpic-data-is-text-relative.
15751 2017-06-27 Andrew Pinski <apinski@cavium.com>
15753 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
15754 (X * copysign (1.0, X)): New pattern.
15755 (X * copysign (1.0, -X)): New pattern.
15756 (copysign (-1.0, CST)): New pattern.
15758 2017-06-27 Joseph Myers <joseph@codesourcery.com>
15760 * genmultilib (combination_space): Remove variable.
15761 Validate reuse rules against regular expression for any sequence
15762 of multilib options in any order.
15764 2017-06-27 Michael Collison <michael.collison@arm.com>
15766 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
15767 call aarch64_split_simd_combine.
15768 * (aarch64_combine_internal<mode>): Delete pattern.
15769 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
15770 Allow register and subreg operands.
15772 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15774 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
15775 specific need, just fallback on defaults.
15776 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
15778 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15779 Olivier Hainque <hainque@adacore.com>
15781 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
15783 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
15784 targets. Pick a default if no particular attempt applied.
15785 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
15788 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15790 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
15791 (x86_64-wrs-vxworks7): Likewise.
15793 2017-06-27 Marek Polacek <polacek@redhat.com>
15796 * ubsan.c (instrument_null): Check get_base_address's result for null.
15798 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
15800 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
15802 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
15804 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
15805 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
15806 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
15807 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
15808 New function types.
15809 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
15810 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
15811 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
15812 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
15813 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
15814 BUILT_IN_FEUPDATEENV): New builtins.
15815 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
15816 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
15817 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
15818 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
15820 (builtin_structptr_types): Adjust size.
15821 * tree.c (builtin_structptr_types): Add four entries.
15823 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15824 Olivier Hainque <hainque@adacore.com>
15826 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
15827 (TLS_SYM): New local macro, forcing reference to __tls__ on
15828 link command lines for VxWorks 7 RTPs, triggering initialization
15830 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
15831 OS features TLS support, true for RTPs on VxWorks 7.
15832 * config/vxworks.c (vxworks_override_options): Setup emutls
15835 2017-06-27 Jakub Jelinek <jakub@redhat.com>
15837 * predict.c (test_prediction_value_range): Use -1U instead of -1
15838 to avoid narrowing conversion warning.
15839 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
15840 to avoid narrowing conversion warning.
15841 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
15843 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
15845 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
15847 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
15848 64bit configurations.
15849 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
15850 (SIZE_TYPE): Likewise.
15851 * config/vxworks.c (vxworks_emutls_var_fields): Use
15852 long_unsigned_type_node instead of unsigned_type_node as the offset
15853 field type, which is "pointer" mode in emutls.c.
15855 2017-06-27 Jakub Jelinek <jakub@redhat.com>
15858 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
15860 PR middle-end/81207
15861 * gimple-fold.c (replace_call_with_call_and_fold): Handle
15862 gimple_vuse copying separately from gimple_vdef copying.
15864 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15866 * value-prof.c (free_hist): Remove call to memset and the enclosing if
15869 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
15870 Olivier Hainque <hainque@adacore.com>
15872 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
15873 for all vxworks7 targets.
15874 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
15875 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
15876 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
15877 variations for VX6/VX7 and 32/64bits later on in ...
15878 (VXWORKS_LIB_SPEC): Leverage new macros.
15879 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
15880 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
15882 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
15884 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
15885 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
15887 2017-06-26 Carl Love <cel@us.ibm.com>
15889 * config/rs6000/rs6000-c.c: Add support for built-in functions
15890 vector bool char vec_reve (vector bool char);
15891 vector signed char vec_reve (vector signed char);
15892 vector unsigned char vec_reve (vector unsigned char);
15893 vector bool int vec_reve (vector bool int);
15894 vector signed int vec_reve (vector signed int);
15895 vector unsigned int vec_reve (vector unsigned int);
15896 vector bool long long vec_reve (vector bool long long);
15897 vector signed long long vec_reve (vector signed long long);
15898 vector unsigned long long vec_reve (vector unsigned long long);
15899 vector bool short vec_reve (vector bool short);
15900 vector signed short vec_reve (vector signed short);
15901 vector double vec_reve (vector double);
15902 vector float vec_reve (vector float);
15903 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
15904 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
15905 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
15906 (altivec_vreve): New pattern.
15907 * config/rs6000/altivec.h (vec_reve): New define.
15908 * doc/extend.texi (vec_rev): Update the built-in documentation file
15909 for the new built-in functions.
15911 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15913 PR tree-optimization/71815
15914 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
15916 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
15918 (slsr_process_phi): Likewise.
15919 (replace_uncond_cands_and_profitable_phis): Don't replace a
15920 multiply candidate with a stride of 1 (copy or cast).
15921 (phi_incr_cost): Call uses_consumed_by_stmt rather than
15923 (lowest_cost_path): Likewise.
15924 (total_savings): Likewise.
15926 2017-06-26 Richard Biener <rguenther@suse.de>
15929 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
15930 Use def_builtin_pure for all gather builtins.
15932 2017-06-26 Richard Biener <rguenther@suse.de>
15934 PR tree-optimization/81203
15935 * tree-tailcall.c (find_tail_calls): Do not move stmts into
15936 non-dominating BBs.
15938 2017-06-26 Marek Polacek <polacek@redhat.com>
15941 * doc/invoke.texi: Document -Wmultistatement-macros.
15943 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
15945 * doc/sourcebuild.texi (ARM-specific attributes): Document new
15946 arm_neon_ok_no_float_abi effective target.
15948 2017-06-26 Richard Biener <rguenther@suse.de>
15950 PR tree-optimization/80928
15951 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
15952 (copy_bbs): Set BB_DUPLICATED flag early.
15953 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
15955 (execute_on_shrinking_pred): Likewise.
15956 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
15957 BB_DUPLICATED blocks.
15958 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
15959 iterate over all PHIs considering removal of *gsi.
15961 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
15963 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
15966 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15968 * config/rs6000/rs6000-string.c: (expand_block_clear,
15969 do_load_for_compare, select_block_compare_mode,
15970 compute_current_alignment, expand_block_compare,
15971 expand_strncmp_align_check, expand_strn_compare,
15972 expand_block_move, rs6000_output_load_multiple)
15973 Move functions related to string/block move/compare
15974 to a separate file.
15975 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
15976 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
15977 for this function which is now used in two files.
15978 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
15979 * config.gcc: Add rs6000-string.o to extra_objs for
15980 targets powerpc*-*-* and rs6000*-*-*.
15982 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
15985 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
15986 32-bit, since indexed is not valid for DImode.
15987 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
15988 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
15989 (define_peephole2 for Altivec d-form load): Add 32-bit support.
15990 (define_peephole2 for Altivec d-form store): Likewise.
15993 * multiple_target.c (create_dispatcher_calls): Only create the
15994 dispatcher call if the function is the default clone of a
15995 versioned function.
15997 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
15999 PR middle-end/80902
16000 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
16001 a call, force the call to not be a tail call.
16003 2017-06-23 Jeff Law <law@redhat.com>
16005 * doc/contrib.texi: Add entry for Steven Pemberton's work on
16008 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
16010 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
16011 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
16012 handling for early expansion of vector shifts (sl,sr,sra,rl).
16013 (builtin_function_type): Add vector shift right instructions
16014 to the unsigned argument list.
16016 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
16018 rtl-optimizatoin/79286
16019 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
16020 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
16021 trap. PIC register plus a const unspec without offset can never trap.
16023 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
16025 * tree.h (builtin_structptr_type): New type.
16026 (builtin_structptr_types): Declare new array.
16027 * tree.c (builtin_structptr_types): New array.
16028 (free_lang_data, build_common_tree_nodes): Use it.
16030 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
16033 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
16036 2017-06-22 Matt Turner <mattst88@gmail.com>
16038 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
16039 Lake models to skylake case. Assume skylake for unknown
16040 models with clflushopt.
16042 2017-06-22 Jeff Law <law@redhat.com>
16044 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
16045 frame sizes that do not satisfy aarch64_uimm12_shift.
16047 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
16049 * profile-count.h (apply_probability,
16050 apply_scale, probability_in): Fix checks for zero.
16052 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16054 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
16055 * doc/cppdiropts.texi (-I @var{dir}): Document it.
16057 2016-06-22 Richard Biener <rguenther@suse.de>
16059 * tree-vect-loop.c (vect_model_reduction_cost): Handle
16060 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
16061 REDUC_MAX_EXPR support.
16062 (vectorizable_reduction): Likewise.
16063 (vect_create_epilog_for_reduction): Likewise.
16065 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
16067 * match.pd (A / (1 << B) -> A >> B): New.
16068 * generic-match-head.c: Include optabs-tree.h.
16069 * gimple-match-head.c: Likewise.
16070 * optabs-tree.h (target_supports_op_p): New.
16071 * optabs-tree.c (target_supports_op_p): New.
16073 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16075 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
16076 $gcc_cv_ld --help output.
16077 (gcc_cv_ld_demangle): Likewise.
16078 (gcc_cv_ld_eh_frame_hdr): Likewise.
16079 (gcc_cv_ld_pie): Likewise.
16080 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
16081 (gcc_cv_ld_buildid): Likewise.
16082 (gcc_cv_ld_sysroot): Likewise.
16083 (ld_bndplt_support): Likewise.
16084 (ld_pushpopstate_support): Likewise.
16085 * configure: Regenerate.
16086 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
16088 2017-06-21 Jakub Jelinek <jakub@redhat.com>
16091 * config/i386/sse.md (round<mode>2): Renumber match_dup and
16092 operands indexes to avoid gap between operands and match_dups.
16094 2017-06-21 Andrew Pinski <apinski@cavium.com>
16096 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
16097 Increment Arith_shift and Arith_shift_reg by 1.
16098 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
16100 * config/aarch64/aarch64.c (thunderx_tunings): Enable
16101 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
16102 (aarch64_strip_extend): Add new argument and test for it.
16103 (aarch64_cheap_mult_shift_p): New function.
16104 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
16105 add a cost if it is true.
16106 Update calls to aarch64_strip_extend.
16107 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
16109 2017-06-21 Andrew Pinski <apinski@cavium.com>
16111 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
16113 (thunderxt88): Likewise.
16114 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
16115 (thunderx_prefetch_tune): New variable.
16116 (thunderx2t99_prefetch_tune): Update for the correct values.
16117 (thunderxt88_tunings): New variable.
16118 (thunderx_tunings): Use thunderx_prefetch_tune instead of
16119 generic_prefetch_tune.
16120 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
16122 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16124 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
16125 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
16126 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
16127 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
16128 (aarch64_atomic_cas<mode>, GPI): Likewise.
16130 2017-06-21 Martin Liska <mliska@suse.cz>
16132 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
16133 statements on cold and hot labels.
16134 * predict.c (tree_estimate_probability_bb): Remove the
16135 prediction from this place.
16137 2017-06-21 Martin Liska <mliska@suse.cz>
16139 PR tree-optimization/79489
16140 * gimplify.c (maybe_add_early_return_predict_stmt): New
16142 (gimplify_return_expr): Call the function.
16143 * predict.c (tree_estimate_probability_bb): Remove handling
16145 * predict.def: Update comment about early return predictor.
16146 * gimple-predict.h (is_gimple_predict): New function.
16147 * predict.def: Change default value of early return to 66.
16148 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
16150 * passes.def: Put pass_strip_predict_hints to the beginning of
16153 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
16155 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
16156 FUNCTION_DECL declarations.
16157 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
16159 (dwaf2out_decl): Likewise.
16160 * godump.c (go_early_global_decl): Skip call to the real debug hook
16161 for FUNCTION_DECL declarations.
16162 * passes.c (rest_of_decl_compilation): Skip call to the
16163 early_global_decl debug hook for FUNCTION_DECL declarations, unless
16164 -fdump-go-spec is passed.
16166 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
16168 * config/i386/i386.c (struct builtin_isa): New field pure_p.
16169 Reorder for compactness.
16170 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
16171 (def_builtin_pure, def_builtin_pure2): New functions.
16172 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
16174 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
16176 * match.pd (nop_convert): New predicate.
16177 ((A +- CST1) +- CST2): Allow some NOP conversions.
16179 2017-06-21 Jakub Jelinek <jakub@redhat.com>
16182 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
16183 with ctors/dtors if GOVD_SHARED is set.
16185 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16187 * config/aarch64/aarch64.md (movti_aarch64):
16188 Emit mov rather than orr.
16189 (movtf_aarch64): Likewise.
16190 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
16191 Emit mov rather than orr.
16193 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16195 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
16196 Swap alternatives, make integer dup more expensive.
16198 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16200 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
16201 Return true for non-tls symbols.
16203 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
16205 * config/aarch64/aarch64-cores.def (cortex-a55): New.
16206 (cortex-a75): Likewise.
16207 (cortex-a75.cortex-a55): Likewise.
16208 * config/aarch64/aarch64-tune.md: Regenerate.
16209 * doc/invoke.texi (-mtune): Document new values for -mtune.
16211 2017-06-21 Tom de Vries <tom@codesourcery.com>
16213 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
16214 stack_size feature.
16215 (Effective-Target Keywords, Other attributes): Suggest using
16216 dg-add-options stack_size feature to get stack limit in stack_size
16217 effective target documentation.
16219 2017-06-21 Julian Brown <julian@codesourcery.com>
16220 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16222 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
16223 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
16224 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
16226 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
16227 attribute type list for neon_multiply.
16228 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
16229 attribute type list for neon_multiply.
16230 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
16231 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
16232 attribute type list for neon_multiply.
16233 * config/arm/types.md (crypto_pmull): Add.
16234 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
16235 attribute type list.
16237 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
16239 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
16242 2017-06-20 Jakub Jelinek <jakub@redhat.com>
16244 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
16245 to make sure not to dereference a NULL cost_classes_ptr pointer.
16247 2017-06-20 Carl Love <cel@us.ibm.com>
16249 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16250 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16251 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16252 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16253 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16254 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16255 VMULOSW): New enum "unspec" values.
16256 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16257 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16258 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16259 altivec_vmulosw): New patterns.
16260 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16261 VMULOSW): Add definitions.
16263 2017-06-20 Julia Koval <julia.koval@intel.com>
16265 * config/i386/i386.c: Fix rounding expand for new pattern.
16266 * config/i386/subst.md: Fix pattern (parallel -> unspec).
16268 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16270 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
16271 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
16273 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16275 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
16278 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16280 * config/aarch64/aarch64-cores.def: Rearrange to sort by
16281 architecture, then by implementer ID.
16282 * config/aarch64/aarch64-tune.md: Regenerate.
16284 2017-06-20 Richard Biener <rguenther@suse.de>
16286 PR middle-end/81097
16287 * fold-const.c (split_tree): Fold to type before negating.
16289 2017-06-20 David Malcolm <dmalcolm@redhat.com>
16291 * diagnostic-show-locus.c
16292 (selftest::test_fixit_deletion_affecting_newline): New function.
16293 (selftest::diagnostic_show_locus_c_tests): Call it.
16295 2017-06-20 Andreas Schwab <schwab@suse.de>
16298 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
16301 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16303 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
16304 __ARM_FEATURE_COPROC according to support.
16306 2017-06-20 Jakub Jelinek <jakub@redhat.com>
16308 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
16309 Rewritten to avoid overflow for > 32-bit pointers.
16312 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
16313 by removing enum keyword.
16314 (ubsan_type_descriptor): Likewise. Formatting fix.
16317 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
16318 splitter): Require TARGET_SSE2 in the condition.
16320 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
16323 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
16324 for doing vector set of SFmode on ISA 3.0.
16325 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
16326 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
16328 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
16329 SFmode value into a V4SF variable that was extracted from another
16330 V4SF variable without converting the element to double precision
16331 and back to single precision vector format.
16332 (vsx_insert_extract_v4sf_p9_2): Likewise.
16334 2017-06-19 Jakub Jelinek <jakub@redhat.com>
16336 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
16337 in UWHI to avoid undefined overflow.
16340 * ubsan.h (enum ubsan_encode_value_phase): New.
16341 (ubsan_encode_value): Change second argument to
16342 enum ubsan_encode_value_phase with default value of
16343 UBSAN_ENCODE_VALUE_GENERIC.
16344 * ubsan.c (ubsan_encode_value): Change second argument to
16345 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
16346 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
16347 create_tmp_var_raw instead of create_tmp_var and use a
16349 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
16350 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
16351 ubsan_encode_value callers.
16354 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
16355 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
16356 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
16358 2017-06-19 Richard Biener <rguenther@suse.de>
16360 PR middle-end/81118
16361 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
16362 estimates if we changed anything.
16364 2017-06-19 Richard Biener <rguenther@suse.de>
16366 PR tree-optimization/80887
16367 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
16368 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
16369 simplified lookups, then reset mprts_hook.
16370 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
16372 (try_to_simplify): Likewise.
16374 2017-06-19 Martin Liska <mliska@suse.cz>
16377 * gimplify.c (gimplify_switch_expr):
16378 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
16380 2017-06-19 Martin Liska <mliska@suse.cz>
16382 * doc/install.texi: Document that PGO runs in 4 stages.
16384 2017-06-19 Martin Liska <mliska@suse.cz>
16387 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
16388 to dispatcher function name.
16389 * multiple_target.c (replace_function_decl): New function.
16390 (create_dispatcher_calls): Redirect both edges and references.
16392 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
16394 * profile-count.c (profile_count::dump): Dump quality.
16395 (profile_count::differs_from_p): Update for unsigned val.
16396 * profile-count.h (profile_count_quality): New enum.
16397 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
16399 2017-06-19 Richard Biener <rguenther@suse.de>
16401 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
16402 struct function as arg.
16403 (estimate_numbers_of_iterations): Export overload with loop arg.
16404 (free_numbers_of_iterations_estimates_loop): Use an overload of
16405 free_numbers_of_iterations_estimates instead.
16406 * tree-cfg.c (remove_bb): Adjust.
16407 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
16408 * tree-parloops.c (gen_parallel_loop): Likewise.
16409 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
16411 (tree_unroll_loops_completely): Likewise.
16412 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
16413 Use an overload instead and export.
16414 (estimated_loop_iterations): Adjust.
16415 (max_loop_iterations): Likewise.
16416 (likely_max_loop_iterations): Likewise.
16417 (estimate_numbers_of_iterations): Take struct function as arg
16419 (loop_exits_before_overflow): Adjust.
16420 (free_numbers_of_iterations_estimates_loop): Use an overload.
16421 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
16422 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
16424 2017-06-19 Richard Biener <rguenther@suse.de>
16427 * ipa-prop.c (find_constructor_constant_at_offset): Handle
16428 RANGE_EXPR conservatively.
16430 2017-06-16 Carl Love <cel@us.ibm.com>
16432 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16433 definitions for vec_float, vec_float2, vec_floato,
16434 vec_floate built-ins.
16435 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
16436 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
16438 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
16439 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
16440 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
16441 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
16442 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
16443 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
16444 vec_floato): Add builtin defines.
16445 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
16446 Update the built-in documentation file for the new built-in
16449 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16451 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
16452 (mthumb): Mark as the negative of -marm.
16454 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16456 * doc/invoke.texi (ARM Options, -mcpu): Document supported
16458 (ARM Options, -mtune): Document that this accepts the same
16459 extension options as -mcpu.
16460 (ARM Options, -mfpu): Document addition of -mfpu=auto.
16462 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16464 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
16465 permitted extensions.
16467 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16469 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
16470 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
16471 (armv8-m.main): Add option +nodsp.
16472 * config/arm/arm-cpu-cdata.h: Regenerated.
16474 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16476 * config/arm/t-fuchsia: New file.
16477 * config.gcc (arm*-*-fuchsia*): Use it.
16479 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16481 * config/arm/t-symbian: Rewrite for new option infrastructure.
16483 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16485 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
16486 (MULTILIB_REQUIRED): Likewise.
16488 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16490 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
16491 (MULTILIB_RESUE): Likewise.
16492 (MULTILIB_MATCHES): Likewise.
16493 (MULTLIB_REQUIRED): Likewise.
16495 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16497 * config/arm/t-rtems: Rewrite for new option framework.
16499 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16501 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
16502 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
16503 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
16504 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
16505 * config/arm/t-multilib: ... here.
16506 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
16507 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
16508 armv7-a and armv8*-a when A-profile libraries have not been built.
16509 * config/arm/t-rmprofile: Rewrite.
16511 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16513 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
16514 with a backslash. Remove the backslash after substituting unescaped
16516 * doc/fragments.texi (MULTILIB_REUSE): Document it.
16518 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16520 * config.gcc: (arm*-*-*): When building a-profile libraries, force
16521 the driver to pass through the default setting of -mfloat-abi.
16522 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
16524 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
16525 (all_feat_combs): New rule.
16526 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
16528 * config/arm/t-aprofile: Rewrite.
16530 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16532 * config/arm/arm.h (FPUTYPE_AUTO): Define.
16533 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
16534 fpu is not specified by the user/command-line.
16535 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
16536 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
16537 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
16538 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
16539 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
16540 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
16542 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16544 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
16545 * config/arm/t-arm-elf: Rewritten.
16547 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16549 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
16550 have some floating-point instructions.
16551 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
16552 (TARGET_MAYBE_HARD_FLOAT): New macro.
16553 * config/arm/arm-builtins.c (arm_init_builtins): Use
16554 TARGET_MAYBE_HARD_FLOAT.
16555 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
16557 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16559 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
16560 (configargs.h): Include it.
16561 (arm_print_hint_for_fpu_option): New function.
16562 (arm_parse_fpu_option): New function.
16563 (candidate_extension): New class.
16564 (arm_canon_for_multilib): New function.
16565 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
16566 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
16567 (ARCH_CANONICAL_SPECS): New macro.
16568 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
16570 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16572 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
16573 are set after handling multilib fragments. Set target_cpu_default2
16576 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16578 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
16580 (arm*-*-*): Set target_cpu_default2 to a quoted string.
16581 * config/arm/parsecpu.awk (check_cpu): Validate any extension
16583 (check_arch): Likewise.
16584 * config/arm/arm.c (arm_configure_build_target): Handle
16585 TARGET_CPU_DEFAULT being a string constant. Scan any feature
16586 options in the default.
16588 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16590 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
16591 when an option is an alias of another.
16592 * config/arm/parsecpu.awk (optalias): New parser token.
16593 (gen_comm_data): Mark non-alias options as such. Emit entries
16594 for extension aliases.
16595 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
16596 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
16597 (armv6kz, armv6zk, armv6t2): Likewise.
16598 (armv7): Make vfpv3-d16 an alias.
16599 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
16601 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
16602 Sort in canonical order.
16603 (armv8-a): Sort in canonical order.
16604 (armv8.1-a, armv8.2-a): Likewise.
16605 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
16607 (cortex-a9): Sort in canonical order.
16608 * config/arm/arm.c (selftests.h): Include it.
16609 (arm_test_cpu_arch_data): New function.
16610 (arm_run_self_tests): New function.
16611 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
16612 (targetm): Move declaration to the end of the file.
16613 * arm-cpu-cdata.h: Regenerated.
16615 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16617 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
16618 call to target_mode_check describing the type of option passed.
16619 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
16620 (arm_target_thumb_only): Use arm_parse_arch_option_name or
16621 arm_parse_cpu_option_name to match parameters against list of
16623 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
16624 arm_arch_core_flags data structure.
16625 * config/arm/arm-cpu_cdata.h: Regenerated.
16627 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16629 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
16631 (arm_print_hint_for_cpu_option): Likewise.
16632 (arm_print_hint_for_arch_option): Likewise.
16633 (arm_parse_cpu_option_name): Likewise.
16634 (arm_parse_arch_option_name): Likewise.
16635 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
16636 of entries in the all_fpus list.
16637 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
16638 (arm_parse_cpu_option_name): Declare.
16639 (arm_parse_arch_option_name): Declare.
16640 (arm_parse_option_features): Declare.
16641 (arm_intialize_isa): Declare.
16642 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
16644 (gen_comm_data): ... here. Make definitions non-static.
16645 * config/arm/arm-cpu-data.h: Regenerated.
16646 * config/arm/arm-cpu-cdata.h: Regenerated.
16648 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16650 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
16651 (cpu_arch_extension): New structure.
16652 (cpu_arch_option, arch_option, cpu_option): New structures.
16653 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
16654 architecture types.
16655 (gen_data): Generate new format data tables.
16656 * config/arm/arm.c (cpu_tune): New structure.
16657 (cpu_option, processors): Delete.
16658 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
16659 (arm_print_hint_for_cpu_option): ... this and ...
16660 (arm_print_hint_for_arch_option): ... this.
16661 (arm_parse_arch_cpu_name): Delete. Replace with ...
16662 (arm_parse_cpu_option_name): ... this and ...
16663 (arm_parse_arch_option_name): ... this.
16664 (arm_unrecognized_feature): Change type of target parameter to
16666 (arm_parse_arch_cpu_features): Delete. Replace with ...
16667 (arm_parse_option_features): ... this.
16668 (arm_configure_build_target): Rework to use new configuration data
16670 (arm_print_tune_info): Rework for new configuration data tables.
16671 * config/arm/arm-cpu-data.h: Regenerated.
16672 * config/arm/arm-cpu.h: Regenerated.
16674 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16676 * Makefile.in (OBJS): Move sbitmap.o from here ...
16677 (OBJS-libcommon): ... to here.
16679 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16681 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
16682 (ISA_ALL_CRYPTO): New macro.
16683 (ISA_ALL_SIMD): New macro
16684 (ISA_ALL_FP): New macro.
16685 * config/arm/arm.c (fpu_bitlist): Update initializer.
16686 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
16688 (arm9e): Add fpu. Add option for nofp
16689 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
16690 (arm926ej-s, arm1026ej-s): Likewise.
16691 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
16692 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
16693 neon-fp16, neon-vfpv4, nofp and nosimd.
16694 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
16695 (cortex-a8): Add fpu. Add option for nofp.
16696 (cortex-a9): Add fpu. Add options for nosimd and nofp.
16697 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
16698 (cortex-r4f): Add fpu.
16699 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
16700 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
16702 (cortex-r8): Likewise.
16703 (cortex-m4): Add fpu. Add option for nofp.
16704 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
16705 (cortex-a17.cortex-a7): Likewise.
16706 (cortex-a32): Add fpu. Add options for crypto and nofp.
16707 (cortex-a35, cortex-a53): Likewise.
16708 (cortex-a57): Add fpu. Add option for crypto.
16709 (cortex-a72, cortex-a73): Likewise.
16710 (exynos-m1): Likewise.
16711 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
16712 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
16713 (cortex-m33): Add fpu. Add option for nofp.
16714 * config/arm/arm-cpu-cdata.h: Regenerated
16715 * config/arm/arm-cpu-data.h: Regenerated.
16717 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16719 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
16720 (armv5te, armv5tej): Likewise.
16721 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
16722 (armv7): Add options fp and vfpv3-d16.
16723 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
16724 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
16726 (armv7ve): Likewise.
16727 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
16728 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
16729 (armv8-a): Add nocrypto option.
16730 (armv8.1-a, armv8.2-a): Likewise.
16731 (armv8-m.main): add options fp, fp.dp and nofp.
16733 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16735 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
16737 (armv8-a+crc): Delete.
16738 (armv8.1-a): Add options simd, crypto and nofp.
16739 (armv8.2-a): Add options fp16, simd, crypto and nofp.
16740 (armv8.2-a+fp16): Delete.
16741 (armv8-m.main): Add option dsp.
16742 (armv8-m.main+dsp): Delete.
16743 (cortex-a8): Add fpu. Add nofp option.
16744 (cortex-a9): Add fpu. Add nofp and nosimd options.
16745 * config/arm/parsecpu.awk (gen_data): Generate option tables and
16746 link to main cpu and architecture data structures.
16747 (gen_comm_data): Only put isa attributes from the main architecture
16749 (option): New statement for architecture and CPU entries.
16750 * arm.c (struct cpu_option): New structure.
16751 (struct processors): Add entry for options.
16752 (arm_unrecognized_feature): New function.
16753 (arm_parse_arch_cpu_name): Ignore any characters after the first
16755 (arm_parse_arch_cpu_feature): New function.
16756 (arm_configure_build_target): Separate out any CPU and architecture
16757 features and parse separately. Don't error out if -mfpu=auto is
16758 used with only an architecture string.
16759 (arm_print_asm_arch_directives): New function.
16760 (arm_file_start): Call it.
16761 * config/arm/arm-cpu-cdata.h: Regenerated.
16762 * config/arm/arm-cpu-data.h: Likewise.
16763 * config/arm/arm-tables.opt: Likewise.
16765 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16767 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
16768 assembler when it is not -mfpu=auto.
16770 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16772 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
16773 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
16774 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
16775 (ASM_CPU_SPEC): Rewrite.
16776 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
16777 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
16778 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
16779 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
16780 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
16781 copied string is NUL-terminated. Also strip any characters prefixed
16783 (arm_rewrite_selected_arch): New function.
16784 (arm_rewrite_march): New function.
16786 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16788 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
16789 (x_arm_cpu_string, x_arm_tune_string): Likewise.
16790 (march, mcpu, mtune): Convert to string-based options.
16791 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
16792 (arm_parse_arch_cpu_name): New function.
16793 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
16794 identify selected architecture or CPU.
16795 (arm_option_save): New function.
16796 (TARGET_OPTION_SAVE): Redefine.
16797 (arm_option_restore): Restore string options.
16798 (arm_option_print): Print string options.
16800 2017-06-16 Martin Sebor <msebor@redhat.com>
16802 PR tree-optimization/80933
16803 PR tree-optimization/80934
16804 * builtins.c (fold_builtin_3): Do not handle bcmp here.
16805 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
16806 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
16807 (gimple_fold_builtin): Call them.
16809 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16811 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
16812 as unlikely; update profile.
16814 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16816 * predict.c (force_edge_cold): Handle declaring edges impossible
16819 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16821 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
16823 (try_unroll_loop_completely): Fix reporting.
16825 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16827 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
16829 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
16832 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
16833 if given a non-constant argument for an intrinsic which requires a
16836 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16838 * profile.c (compare_freqs): New function.
16839 (branch_prob): Sort edge list.
16840 (find_spanning_tree): Assume that the list is priority sorted.
16842 2017-06-16 Richard Biener <rguenther@suse.de>
16844 PR tree-optimization/81090
16845 * passes.def (pass_record_bounds): Remove.
16846 * tree-pass.h (make_pass_record_bounds): Likewise.
16847 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
16848 make_pass_record_bounds): Likewise.
16849 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
16850 not free niter estimates at the beginning but at the end.
16851 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
16853 2017-06-16 Richard Biener <rguenther@suse.de>
16855 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
16856 initializer to workaround ICE in host GCC 4.8.
16858 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
16860 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
16862 (clone_inlined_nodes): Update.
16864 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16866 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
16867 prefetch settings, and enable prefetching by default at -O3.
16869 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16871 * config/aarch64/aarch64.c (aarch64_override_options_internal):
16872 Set flag_prefetch_loop_arrays according to tuning data.
16874 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16876 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
16877 New tune structure.
16878 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
16879 [Unrelated to main purpose of the patch] Place the pointer field last
16880 to enable type checking errors when tune structure are wrongly merged.
16881 * config/aarch64/aarch64.c (generic_prefetch_tune,)
16882 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
16883 (thunderx2t99_prefetch_tune): New tune constants.
16884 (tune_params *_tunings): Update all tunings (no functional change).
16885 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
16886 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
16887 from tunings structures.
16889 2017-06-16 Jakub Jelinek <jakub@redhat.com>
16892 * ubsan.c (instrument_null): Add T argument, use it instead
16893 of computing it based on IS_LHS.
16894 (instrument_object_size): Likewise.
16895 (pass_ubsan::execute): Adjust instrument_null and
16896 instrument_object_size callers to pass gimple_get_lhs or
16897 gimple_assign_rhs1 result to it. Use instrument_null instead of
16898 calling get_base_address and instrument_mem_ref. Handle
16899 aggregate call arguments for object-size sanitization.
16901 2017-06-16 Yury Gribov <tetra2005@gmail.com>
16903 PR tree-optimization/81089
16904 * tree-vrp.c (is_masked_range_test): Validate operands of
16907 2017-06-15 Martin Sebor <msebor@redhat.com>
16910 * dumpfile.c (dump_register): Avoid calling memset to initialize
16911 a class with a default ctor.
16912 * gcc.c (struct compiler): Remove const qualification.
16913 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
16914 * hash-table.h: Ditto.
16915 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
16917 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
16918 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
16920 * params.h (struct param_info): Make struct members non-const.
16921 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
16922 with default initialization.
16923 * vec.h (vec_copy_construct, vec_default_construct): New helper
16925 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
16926 with vec_copy_construct.
16927 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
16928 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
16929 * doc/invoke.texi (-Wclass-memaccess): Document.
16931 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16933 * emit-rtl.h (is_leaf): Update comment about local
16934 register allocator.
16936 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16939 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
16940 for a variable to have a section before checking if the section has a
16942 Set section to.persistent if persistent attribute is set.
16943 Warn if .persistent attribute is used on an automatic variable.
16945 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
16947 PR rtl-optimization/80474
16948 * reorg.c (update_block): Do not ignore instructions in a delay slot.
16950 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
16952 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
16955 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
16957 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
16958 (casesi): Emit bounds checking as RTL.
16959 (casesi_internal_mips16_<mode>): Remove bounds checking.
16961 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
16963 * config/xtensa/xtensa.c (xtensa_option_override): Append
16964 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
16965 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
16966 xtensa_doloop_hooks): Define unconditionally.
16967 (xtensa_reorg_loops): Only call reorg_loops in the presence of
16969 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
16970 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
16971 for it in xtensa_option_override.
16972 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
16973 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
16975 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
16977 * doc/cppopts.texi: Document '-' special value to -MF.
16979 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
16981 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
16982 (cortex_a53_fconst): Likewise.
16983 (cortex_a53_fpmul): Likewise.
16984 (cortex_a53_f_load_64): Likewise.
16985 (cortex_a53_f_load_many): Likewise.
16986 (cortex_a53_advsimd_alu): Likewise.
16987 (cortex_a53_advsimd_alu_q): Likewise.
16988 (cortex_a53_advsimd_mul): Likewise.
16989 (cortex_a53_advsimd_mul_q): Likewise.
16990 (fpmac bypass): Add new bypass for fpmac-fpmac case.
16991 Add missing fmul, r2f_cvt and fconst cases.
16993 2017-06-14 Richard Biener <rguenther@suse.de>
16995 PR middle-end/81088
16996 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
16998 (fold_binary_loc): When associating do not treat pre-existing
16999 TREE_OVERFLOW on literal constants as a reason to allow
17000 TREE_OVERFLOW on associated literal constants.
17002 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
17004 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
17005 (MASK_FEATURES): New macro.
17006 * config/sparc/sparc.c (sparc_option_override): Remove the special
17007 handling of -mfpu and generalize it to all MASK_FEATURES switches.
17009 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
17011 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
17012 a division of 0 if non-call exceptions are enabled.
17014 2017-06-14 Andrew Pinski <apinski@cavium.com>
17015 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17018 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
17019 Improve vector initialization code gen for only variable case.
17021 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
17023 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
17025 2017-06-14 Richard Biener <rguenther@suse.de>
17027 PR tree-optimization/81083
17028 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
17031 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17033 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
17034 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
17035 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
17036 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
17037 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
17038 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
17040 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17042 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
17043 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
17045 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17047 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
17049 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17051 * config/rs6000/t-rtems: Don't handle SPE.
17053 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17055 * config/rs6000/t-linux: Don't handle SPE.
17057 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17059 * config/rs6000/eabispe.h: Delete file.
17061 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17063 * config/rs6000/t-spe: Delete file.
17065 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17067 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
17068 (rs6000_legitimate_offset_address_p): Return false for anything in
17069 V2SImode or V2SFmode.
17071 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17073 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
17074 except V2SF and V2SI. Rearrange the vector modes, and add comments.
17075 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
17077 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
17078 (rs6000_legitimate_offset_address_p): Ditto.
17079 (rs6000_emit_move): Ditto.
17080 (rs6000_init_builtins): Remove V4HI_type_node.
17082 2017-06-13 Martin Liska <mliska@suse.cz>
17085 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
17086 (gate_asan): Likewise.
17087 * asan.h (asan_no_sanitize_address_p): Remove the function.
17088 (sanitize_flags_p): New function.
17089 * builtins.def: Fix coding style.
17090 * common.opt: Use renamed enum value.
17091 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
17092 * doc/extend.texi: Document no_sanitize attribute.
17093 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
17094 to SANITIZE_UNDEFINED_NONDEFAULT.
17095 * gcc.c (sanitize_spec_function): Use the renamed enum value.
17096 * gimple-fold.c (optimize_atomic_compare_exchange_p):
17097 Use sanitize_flags_p.
17098 * gimplify.c (gimplify_function_tree): Likewise.
17099 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
17100 * opts.c (parse_no_sanitize_attribute): New function.
17101 (common_handle_option): Use renamed enum value.
17102 * opts.h (parse_no_sanitize_attribute): Declare.
17103 * tree.c (sanitize_flags_p): New function.
17104 * tree.h: Declared here.
17105 * tsan.c: Use sanitize_flags_p.
17106 * ubsan.c (ubsan_expand_null_ifn): Likewise.
17107 (instrument_mem_ref): Likewise.
17108 (instrument_bool_enum_load): Likewise.
17109 (do_ubsan_in_current_function): Remove the function.
17110 (pass_ubsan::execute): Use sanitize_flags_p.
17111 * ubsan.h: Remove do_ubsan_in_current_function
17112 * tree-cfg.c (print_no_sanitize_attr_value): New function.
17113 (dump_function_to_file): Use it here.
17115 2017-06-13 Martin Jambor <mjambor@suse.cz>
17117 PR tree-optimization/80803
17118 PR tree-optimization/81063
17119 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
17120 (propagate_subaccesses_across_link): Enqueue subtree whenever
17121 necessary instead of relying on the caller.
17123 2017-06-13 Martin Jambor <mjambor@suse.cz>
17125 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
17126 that have a first_link.
17127 (sort_and_splice_var_accesses): Do not check first_link before
17129 (subtree_mark_written_and_enqueue): Likewise.
17130 (propagate_all_subaccesses): Likewise and do not stop at first
17131 parent with a first_link.
17133 2017-06-13 Martin Jambor <mjambor@suse.cz>
17135 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
17138 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17140 * match.pd: New pattern.
17142 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17144 * tree-vrp.c (is_masked_range_test): New function.
17145 (register_edge_assert_for): Determine ranges for
17148 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17150 PR tree-optimization/67328
17151 * fold-const.c (maskable_range_p): New function.
17152 (build_range_check): Generate bittests if possible.
17154 2017-06-13 Martin Liska <mliska@suse.cz>
17156 * gimple-pretty-print.c (dump_probability): Add new argument.
17157 (dump_edge_probability): Dump both probability and count.
17158 (dump_gimple_label): Likewise.
17159 (dump_gimple_bb_header): Likewise.
17161 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
17164 * config/avr/avr-devices.c: Fix indentation.
17165 * config/avr/gen-avr-mmcu-specs.c: Dito.
17167 2017-06-13 Richard Biener <rguenther@suse.de>
17169 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
17170 instead get vector type from stmt_info.
17171 (vectorizable_reduction): Adjust. Remove dead code.
17173 2017-06-13 Richard Biener <rguenther@suse.de>
17175 PR middle-end/81065
17176 * fold-const.c (extract_muldiv_1): Remove bogus distribution
17177 case of C * (x * C2 + C3).
17178 (fold_addr_of_array_ref_difference): Properly fold index difference.
17180 2017-06-12 David S. Miller <davem@davemloft.net>
17183 * config/sparc/sparc.md (return expander): Emit frame blockage if
17184 function uses alloca.
17186 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
17188 * combine.c (make_field_assignment): Check len rather than the mode
17189 precision when calling force_to_mode.
17191 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
17193 Support multilibs and devices that see flash in RAM address range.
17196 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
17197 (avr_mcu_t) <flash_pm_offset>: New field.
17198 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
17199 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
17200 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
17201 (AVR_TINY_PM_OFFSET): Remove macro.
17202 * config/avr/avr.opt (-mshort-calls): New option.
17203 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17204 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
17205 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
17206 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
17207 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
17208 instead of avr_arch->have_jmp_call.
17209 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
17210 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
17211 avr_arch->flash_pm_offset to define.
17212 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
17213 new field flash_pm_offset. Add entry for avrxmega3.
17214 (avr_texinfo): Add entry for avrxmega3.
17215 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
17216 attiny212, attiny214,
17217 attiny412, attiny414, attiny416, attiny417,
17218 attiny814, attiny816, attiny817,
17219 attiny1614, attiny1616, attiny1617,
17220 attiny3214, attiny3216, attiny3217.
17221 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
17222 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
17223 (avr_print_operand_address) [AVR_TINY]: Same.
17224 (avr_asm_init_sections) <readonly_data_section>: Only patch
17225 callback if avr_arch->flash_pm_offset = 0.
17226 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
17227 for rodata if avr_arch->flash_pm_offset != 0.
17228 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
17229 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
17230 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
17231 (m_options): Append opt_rcall.
17232 (m_dirnames): Append dir_rcall.
17233 * config/avr/t-multilib: Regenerate.
17235 * configure.ac [target=avr]: Check whether avrxmega3 default
17236 linker description file works as needed.
17237 * configure: Regenerate.
17238 * doc/avr-mmcu.texi: Regenerate.
17239 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
17240 <__AVR_ARCH__>: Document avrxmega3 and 103.
17241 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
17242 <__AVR_SHORT_CALLS__>: Document it.
17243 <__AVR_PM_BASE_ADDRESS__>: Document it.
17244 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
17245 (AVR Variable Attributes) <progmem>: Document this is
17246 not needed for avrxmega3.
17247 (AVR Named Address Spaces) <__flash>: Dito.
17249 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
17251 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
17253 2017-06-12 Doug Rupp <rupp@adacore.com>
17255 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
17256 Append vxworks-stdint.h to the tm_file list.
17257 * config/vxworks-stdint.h: New file.
17259 2017-06-12 Martin Liska <mliska@suse.cz>
17261 PR tree-optimization/81041
17262 * tree-profile.c (gimple_gen_ic_func_profiler):
17263 Create an extra BB in profile-generate
17264 (gimple_gen_time_profiler): Likewise.
17266 2017-06-12 Jakub Jelinek <jakub@redhat.com>
17268 PR tree-optimization/81003
17269 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
17270 (update_range_test): Use it instead of force_gimple_operand_gsi.
17272 2017-06-12 Richard Biener <rguenther@suse.de>
17274 PR tree-optimization/81053
17275 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
17276 with backedge value not defined in loop. Simplify def stmt
17279 2017-06-11 Tom de Vries <tom@codesourcery.com>
17282 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
17284 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
17285 nvptx_cannot_force_const_mem.
17287 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17289 * opts.c (finish_options): Move test for flag_split_stack after
17290 it has been initialized.
17292 2017-06-11 Jason Merrill <jason@redhat.com>
17294 * tree.h (id_equal): New.
17295 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
17296 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
17297 instead of strcmp of IDENTIFIER_POINTER.
17299 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17301 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
17302 (mark_all_inlined_calls_cdtor): Fix formating.
17303 (inline_transform): Rescale profile before inlining.
17305 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17307 * cgraph.h (cgraph_edge::clone): Update prototype.
17308 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
17309 (cgraph_node::create_clone): Update.
17310 (cgraph_node::create_version_clone): Update.
17311 * tree-inline.c (copy_bb): Update.
17312 (expand_call_inline): Update.
17314 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
17316 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
17317 factored out from ...
17318 (rs6000_emit_prologue): ... here.
17320 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
17322 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
17323 factored out from ...
17324 (rs6000_emit_prologue): ... here.
17326 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17328 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
17330 (handle_missing_profiles): Fix computation of tp_first_run.
17331 (counts_to_freqs): Do not touch freqs when count is 0.
17333 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17335 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
17338 2017-06-10 Tom de Vries <tom@codesourcery.com>
17340 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
17341 attributes): Document signal effective target.
17343 2017-06-10 Tom de Vries <tom@codesourcery.com>
17345 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
17346 Document effective target stack_size.
17348 2017-06-09 David Malcolm <dmalcolm@redhat.com>
17350 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
17351 to the edit_context if they can be auto-applied.
17353 2017-06-9 Ian Lance Taylor <iant@golang.org>
17355 * opts.c (finish_options): If -fsplit-stack, disable implicit
17356 -forder-blocks-and-partition.
17357 * doc/invoke.texi (Optimize Options): Document that when using
17358 -fsplit-stack -forder-blocks-and-partition is not implicitly
17361 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17363 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
17364 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
17365 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
17366 * builtins.def (abort, trap, unreachable): Declare cold.
17367 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
17368 * tree-core.h (ECF_COLD): New.
17369 * tree.c (set_call_expr_flags): Handle ECF_COLD.
17370 (build_common_builtin_nodes): Mark unreachable and abort as cold.
17372 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17374 * predict.c (unlikely_executed_stmt_p): Cleanup.
17376 2017-06-09 Richard Biener <rguenther@suse.de>
17378 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
17379 model if the ref is always written to.
17381 2017-06-09 Tamar Christina <tamar.christina@arm.com>
17383 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
17385 2017-06-09 Tamar Christina <tamar.christina@arm.com>
17387 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
17390 2017-06-09 Tom de Vries <tom@codesourcery.com>
17393 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
17394 "target cannot support label values" when encountering LABEL_REF.
17396 2017-06-09 Martin Liska <mliska@suse.cz>
17398 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
17399 (gimple_gen_ic_func_profiler): Emit direct comparison
17400 of __gcov_indirect_call_callee with NULL.
17401 (gimple_gen_time_profiler): Change probability from
17402 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
17404 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17406 * profile.c (edge_gcov_counts): Turn to pointer.
17407 (compute_branch_probabilities, compute_branch_probabilities): Update.
17408 (branch_prob): Do not clear edge_gcov_count.
17409 * profile.h (edge_gcov_counts): Turn to pointer.
17410 (edge_gcov_count): Update.
17412 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17414 * gimple.h (gimple_check_failed): Mark cold.
17416 2017-06-09 Richard Biener <rguenther@suse.de>
17418 PR tree-optimization/66623
17419 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
17420 refactor check_reduction into two parts, properly computing
17421 whether we have to check reduction validity for outer loop
17424 2017-06-09 Richard Biener <rguenther@suse.de>
17426 PR tree-optimization/79483
17427 * graphite-scop-detection.c (order): New global.
17428 (get_order): Compute bb to order mapping that satisfies code
17429 generation constraints.
17430 (cmp_pbbs): New helper.
17431 (build_scops): Start domwalk at entry block, sort generated
17434 2017-06-09 Richard Biener <rguenther@suse.de>
17436 PR middle-end/81007
17437 * ipa-polymorphic-call.c
17438 (ipa_polymorphic_call_context::restrict_to_inner_class):
17439 Skip FIELD_DECLs with error_mark_node type.
17440 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
17443 2017-06-09 Martin Liska <mliska@suse.cz>
17445 * predict.c (struct branch_predictor): New struct.
17446 (test_prediction_value_range): New test.
17447 (predict_c_tests): New function.
17448 * selftest-run-tests.c (selftest::run_tests): Run the function.
17449 * selftest.h: Declare new tests.
17451 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
17454 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
17455 gen_add3_insn did not fail.
17456 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
17457 r0, construct that number in a temporary reg and add that reg to r0.
17458 If asked to put the result in r0 as well, fail.
17460 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
17462 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17463 for early expansion of vec_eqv.
17465 2017-06-08 Jakub Jelinek <jakub@redhat.com>
17467 PR middle-end/81005
17468 * ubsan.c (instrument_null): Avoid pointless code temporary.
17469 (pass_ubsan::execute): Instrument aggregate arguments of calls.
17471 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
17475 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
17478 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
17481 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17483 * predict.c (unlikely_executed_edge_p): Move ahead.
17484 (probably_never_executed_edge_p): Use it.
17486 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
17488 PR middle-end/79988
17489 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
17490 gimple_call_builtin_p call.
17492 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17494 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
17495 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
17496 rtl_check_failed_type2, rtl_check_failed_code1,
17497 rtl_check_failed_code2, rtl_check_failed_code_mode,
17498 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
17499 rtvec_check_failed_bounds, rtl_check_failed_flag,
17500 _fatal_insn_not_found, _fatal_insn): Likewise.
17501 * tree.h (tree_contains_struct_check_failed,
17502 tree_check_failed, tree_not_check_failed,
17503 tree_class_check_failed, tree_range_check_failed,
17504 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
17505 tree_vec_elt_check_failed, phi_node_elt_check_failed,
17506 tree_operand_check_failed, omp_clause_check_failed,
17507 omp_clause_operand_check_failed, omp_clause_range_check_failed):
17510 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17512 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
17513 flag_branch_probabilities.
17514 * ipa-inline.c (edge_badness): Likewise.
17515 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
17516 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
17517 * predict.c (maybe_hot_frequency_p): Likewise.
17518 (probably_never_executed): Likewise.
17519 * sched-ebb.c (schedule_ebbs): Likewise.
17520 * sched-rgn.c (find_single_block_region): Likewise.
17521 * tracer.c (tail_duplicate): Likewise.
17523 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17525 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
17526 longer requires x_flag_profile_use.
17528 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17530 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
17531 instead of flag_reorder_blocks_and_partition.
17532 * dbxout.c (dbxout_function_end): Likewise.
17533 * dwarf2out.c (gen_subprogram_die): Likewise.
17534 * haifa-sched.c (sched_create_recovery_edges): Likewise.
17535 * hw-doloop.c (reorg_loops): Likewise.
17536 * varasm.c (assemble_start_function,
17537 assemble_end_function): Likewise.
17538 (decide_function_section): Do not check for
17539 flag_reorder_blocks_and_partition.
17541 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
17543 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
17545 (chkp_get_hard_register_fake_addr_expr): Ditto.
17546 (chkp_build_addr_expr): Add check for hard reg case.
17547 (chkp_parse_array_and_component_ref): Ditto.
17548 (chkp_find_bounds_1): Ditto.
17549 (chkp_process_stmt): Don't generate bounds store for
17552 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17554 * predict.c (maybe_hot_bb_p): Do not check profile status.
17555 (maybe_hot_edge_p): Likewise.
17556 (probably_never_executed): Check for zero counts even if profile
17558 (unlikely_executed_edge_p): New function.
17559 (unlikely_executed_stmt_p): New function.
17560 (unlikely_executed_bb_p): New function.
17561 (set_even_probabilities): Use unlikely predicates.
17562 (combine_predictions_for_bb): Likewise.
17563 (predict_paths_for_bb): Likewise.
17564 (predict_paths_leading_to_edge): Likewise.
17565 (determine_unlikely_bbs): New function.
17566 (estimate_bb_frequencies): Use it.
17567 (compute_function_frequency): Use zero counts even if profile is
17569 * profile-count.h: Fix typo.
17571 2017-08-08 Julia Koval <julia.koval@intel.com>
17573 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
17574 _mm512_mask_cvtsepi16_storeu_epi8,
17575 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
17576 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
17577 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
17578 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
17579 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
17580 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
17581 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
17582 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
17583 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
17584 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
17585 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
17586 __builtin_ia32_pmovuswb256mem_mask,
17587 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
17588 __builtin_ia32_pmovwb512mem_mask): New builtins.
17590 2017-08-08 Julia Koval <julia.koval@intel.com>
17592 PR target/73350,80862
17593 * config/i386/subst.md (round): Fix round pattern.
17594 * config/i386/i386.c (ix86_erase_embedded_rounding):
17595 Fix erasing rounding for the fixed pattern.
17597 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17599 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
17601 2017-06-08 Martin Liska <mliska@suse.cz>
17603 PR gcov-profile/80911
17604 * gcov.c (block_info::block_info): New constructor.
17606 2017-06-07 Carl Love <cel@us.ibm.com>
17608 * config/rs6000/rs6000-c: The return type of the following
17609 built-in functions was implemented as int not long long. Fix sign
17610 of return value for the unsigned version of vec_mulo and vec_mule.
17611 vector unsigned long long vec_bperm (vector unsigned long long,
17612 vector unsigned char)
17613 vector signed long long vec_mule (vector signed int,
17615 vector unsigned long long vec_mule (vector unsigned int,
17616 vector unsigned int)
17617 vector signed long long vec_mulo (vector signed int,
17619 vector unsigned long long vec_mulo (vector unsigned int,
17620 vector unsigned int)
17621 * doc/extend.texi: Fix the documentation for the built-in
17624 2017-06-07 Carl Love <cel@us.ibm.com>
17627 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
17630 2017-06-07 Carl Love <cel@us.ibm.com>
17632 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
17633 support, Generate doublehv for signed int/float for BE case only.
17635 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
17637 * doc/invoke.texi (mcx16): Rewrite.
17639 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17641 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
17642 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
17643 *mov<mode>_softfloat, and an anonymous splitter): Use
17644 nonimmediate_operand instead of rs6000_nonimmediate_operand.
17646 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17648 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
17650 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
17651 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
17652 (rs6000_debug_reg_global): Adjust.
17653 (rs6000_init_hard_regno_mode_ok): Adjust.
17654 (rs6000_dbx_register_number): Adjust.
17655 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
17656 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
17657 Remove SPE_ACC and SPEFSCR.
17658 (REG_ALLOC_ORDER): Ditto.
17659 (FRAME_POINTER_REGNUM): Change to 111.
17660 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
17661 (REG_CLASS_NAMES): Ditto.
17662 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
17663 (REGISTER_NAMES): Ditto.
17664 (ADDITIONAL_REG_NAMES): Ditto.
17665 (rs6000_reg_names): Ditto.
17666 * config/rs6000/rs6000.md: Renumber some register number
17669 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17671 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
17673 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
17674 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
17676 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
17677 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
17678 Delete the SPE high registers.
17679 (REG_ALLOC_ORDER): Ditto.
17680 (enum reg_class): Remove SPE_HIGH_REGS.
17681 (REG_CLASS_NAMES): Ditto.
17682 (REG_CLASS_CONTENTS): Delete the SPE high registers.
17683 (REGISTER_NAMES): Ditto.
17684 (rs6000_reg_names): Ditto.
17685 * doc/tm.texi.in: Remove SPE as example.
17686 * doc/tm.texi: Regenerate.
17688 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17690 * config/rs6000/8540.md (ppc8540_brinc): Delete.
17691 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
17692 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
17693 * config/rs6000/rs6000.md (type): Remove "brinc".
17695 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17697 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
17698 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
17699 * config/rs6000/linuxspe.h: Delete file.
17700 * config/rs6000/rs6000.md: Don't include spe.md.
17701 * config/rs6000/spe.h: Delete file.
17702 * config/rs6000/spe.md: Delete file.
17703 * config/rs6000/t-rs6000: Remove spe.md.
17705 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17707 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
17708 (reg_or_none500mem_operand): Delete.
17709 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
17710 instead of reg_or_none500mem_operand.
17712 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17714 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
17715 handling of SPE flags.
17716 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
17718 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17720 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
17722 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
17723 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
17724 paired_divv2sf3): Similar.
17725 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
17726 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
17727 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
17729 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
17730 Rename the paired_* instruction patterns.
17731 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
17733 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
17734 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
17735 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
17736 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
17737 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
17738 PAIRED_VECTOR_MODE.
17739 (struct machine_function): Delete field spe_insn_chain_scanned_p.
17740 (spe_func_has_64bit_regs_p): Delete.
17741 (spe_expand_predicate_builtin): Delete.
17742 (spe_expand_evsel_builtin): Delete.
17743 (TARGET_DWARF_REGISTER_SPAN): Do not define.
17744 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
17745 (invalid_e500_subreg): Delete.
17746 (rs6000_legitimize_address): Always force_reg op2 as well, for
17747 paired single memory accesses.
17748 (rs6000_member_type_forces_blk): Delete.
17749 (rs6000_spe_function_arg): Delete.
17750 (rs6000_expand_unop_builtin): Delete SPE handling.
17751 (rs6000_expand_binop_builtin): Ditto.
17752 (spe_expand_stv_builtin): Delete.
17753 (bdesc_2arg_spe): Delete.
17754 (spe_expand_builtin): Delete.
17755 (spe_expand_predicate_builtin): Delete.
17756 (spe_expand_evsel_builtin): Delete.
17757 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
17758 (spe_init_builtins): Delete.
17759 (spe_func_has_64bit_regs_p): Delete.
17760 (savres_routine_name): Delete "info" parameter. Adjust callers.
17761 (rs6000_emit_stack_reset): Ditto.
17762 (rs6000_dwarf_register_span): Delete.
17763 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
17764 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
17765 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
17767 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
17769 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
17770 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
17771 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
17772 mulv2sf3, divv2sf3): Delete expanders.
17774 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17776 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
17778 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17780 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
17781 * config/rs6000/rs6000.c: Ditto.
17783 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17785 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
17786 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
17787 comparison_operator.
17789 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17791 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
17792 * config/rs6000/rs6000.opt: Ditto.
17793 * config/rs6000/t-rtems: Ditto.
17795 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17797 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
17798 TARGET_E500_SINGLE by 0, simplify.
17799 * config/rs6000/rs6000.c: Ditto.
17800 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
17801 (spe_build_register_parallel): Delete.
17802 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
17803 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
17804 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
17805 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
17806 (E500_CONVERT): Delete.
17807 * config/rs6000/spe.md: Remove many patterns and all define_constants.
17809 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17811 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
17812 * config/rs6000/dfp.md: Ditto.
17813 (negdd2, *negdd2_fpr): Merge.
17814 (absdd2, *absdd2_fpr): Merge.
17815 (negtd2, *negtd2_fpr): Merge.
17816 (abstd2, *abstd2_fpr): Merge.
17817 * config/rs6000/e500.h: Delete file.
17818 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
17819 TARGET_FPRS by 1 and simplify.
17820 * config/rs6000/rs6000-c.c: Ditto.
17821 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
17822 TARGET_DF_SPE by 0.
17823 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
17825 * config/rs6000/rs6000.md: Ditto.
17826 (floatdidf2, *floatdidf2_fpr): Merge.
17827 (move_from_CR_gt_bit): Delete.
17828 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
17829 (E500_CR_IOR_COMPARE): Delete.
17830 (All patterns that require !TARGET_FPRS): Delete.
17831 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
17833 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17835 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
17837 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17839 * graphds.c (add_edge): Intitialize edge's attached data.
17840 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
17841 pointer parameter. Call pointed function on each edge during
17842 graph traversing. Skip traversing the edge when the function
17844 (graphds_dfs, graphds_scc): Ditto.
17845 (for_each_edge): New parameter. Pass the new parameter to callback
17847 * graphds.h (skip_edge_callback): New function pointer type.
17848 (graphds_dfs, graphds_scc): New function pointer parameter.
17849 (graphds_edge_callback, for_each_edge): New parameter.
17851 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17853 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
17854 out code checking if runtime alias check is possible to below ...
17855 Call the new function.
17856 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
17857 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
17859 2017-06-07 Marek Polacek <polacek@redhat.com>
17862 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
17863 TYPE_OVERFLOW_WRAPS check.
17865 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17867 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
17868 if versioning is required.
17869 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
17870 peeling with the check for versioning.
17872 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17874 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
17875 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
17876 Set true to new parameter if new ssa variable is defined.
17877 (vect_gen_vector_loop_niters): Refactor. Set range information
17878 for the new vector loop bound variable.
17879 (vect_do_peeling): Ditto.
17881 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17883 * tree-affine.c (ssa.h): Include header file.
17884 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
17885 has wrapping overflow behavior.
17887 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17889 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
17891 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17893 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
17894 (tree_to_aff_combination): ... here.
17896 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17898 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
17899 reg_pressure model function.
17900 (ivopts_global_cost_for_size): Delete.
17901 (determine_set_costs, iv_ca_recount_cost): Call new model function
17902 ivopts_estimate_reg_pressure.
17904 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17906 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
17907 expensive than udiv. Remove floating point cases from mod.
17909 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17911 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
17912 Increase idiv cost.
17914 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17916 * config/aarch64/aarch64.md
17917 (copysignsf3): Fix mask generation.
17919 2017-06-07 Jakub Jelinek <jakub@redhat.com>
17921 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
17923 (class dump_manager): Add register_dumps method.
17924 * dumpfile.c: Include langhooks.h.
17925 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
17926 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
17927 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
17928 (dump_manager::dump_register): Start with 512 entries instead of 32.
17929 (dump_manager::register_dumps): New method.
17930 * toplev.c (general_init): Instead of invoking register_dumps
17931 langhook, invoke register_dumps method on the dump manager.
17932 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
17935 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
17937 * doc/md.texi: Clarify the restrictions on a define_insn condition.
17938 Say that # requires an associated define_split to exist, and that
17939 the define_split must be suitable for use after register allocation.
17941 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17943 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
17944 (compute_outgoing_frequencies): Also initialize zero counts.
17945 (find_many_sub_basic_blocks): Do not produce uninitialized profile
17946 around loops; preserve more of profile when nothing changes.
17948 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
17950 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
17952 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
17953 * config/arm/arm-cpu-cdata.h: Regenerate.
17954 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
17955 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
17956 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
17957 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
17958 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
17960 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
17962 * config/arm/t-rmprofile: Likewise.
17963 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
17965 2017-06-06 David S. Miller <davem@davemloft.net>
17968 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
17969 blockage if function uses alloca.
17971 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17973 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
17974 New "uid" fields to hold pretty-print IDs of group and ref.
17975 Memory references are now identified as <group_id>:<ref_id>
17976 instead of using [random] addresses.
17977 (dump_mem_details): Simplify, no functional change.
17978 (dump_mem_ref): Simplify and make output more concise.
17979 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
17980 (find_or_create_group): Initialize group uid.
17981 (record_ref): Initialize ref uid. Improve debug output.
17982 (prune_group_by_reuse, should_issue_prefetch_p,)
17983 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
17984 (mark_nontemporal_store, determine_loop_nest_reuse):
17985 Improve debug output.
17987 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17989 * dbgcnt.def (prefetch): New debug counter.
17990 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
17991 (schedule_prefetches): Stop issueing prefetches if debug counter
17994 2017-06-06 Tom de Vries <tom@codesourcery.com>
17996 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
17997 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
17999 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18001 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
18002 Use aarch64_reg_or_zero predicate for operand 4.
18003 (aarch64_compare_and_swap<mode> define_insn_and_split):
18004 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
18005 (aarch64_store_exclusive<mode>): Likewise for operand 2.
18007 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18009 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
18010 (arm_compute_save_core_reg_mask): This.
18011 (thumb1_compute_save_reg_mask): Rename into ...
18012 (thumb1_compute_save_core_reg_mask): This.
18013 (arm_compute_save_reg0_reg12_mask): Adapt comment.
18014 (arm_compute_frame_layout): Likewise.
18016 2017-06-06 Richard Biener <rguenther@suse.de>
18018 PR tree-optimization/80974
18019 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
18020 keep or clear leaders SSA info.
18022 2017-06-06 Tom de Vries <tom@codesourcery.com>
18024 * config/nvptx/nvptx.c (split_mode_p): New function.
18025 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
18027 2017-06-06 Tom de Vries <tom@codesourcery.com>
18029 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
18031 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18034 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
18037 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18039 * shrink-wrap.c (handle_simple_exit): Update profile.
18040 (try_shrink_wrapping): Upate profile.
18042 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18044 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
18045 (tree_guess_outgoing_edge_probabilities): New.
18046 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
18047 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
18049 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18051 * ipa-split.c (split_function): Initialize return bb profile.
18053 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18055 * profile.c (compute_branch_probabilities): Also initialize
18056 EXIT_BLOCK profile.
18058 2017-06-06 Richard Biener <rguenther@suse.de>
18060 PR tree-optimization/80928
18061 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
18062 (vect_analyze_loop_operations): Properly guard analysis for
18064 (vect_transform_loop): Likewise.
18065 (vect_analyze_loop_2): Also reset SLP type on PHIs.
18066 (vect_model_induction_cost): Do not cost for pure SLP.
18067 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
18068 of induction in inner loop vectorization.
18069 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
18070 (vect_get_and_check_slp_defs): Handle vect_induction_def.
18071 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
18073 (vect_analyze_slp_cost_1): Cost induction.
18074 (vect_detect_hybrid_slp_stmts): Handle PHIs.
18075 (vect_get_slp_vect_defs): Likewise.
18076 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
18077 (vect_transform_stmt): Handle SLP reductions.
18078 * tree-vectorizer.h (vectorizable_induction): Adjust.
18080 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18082 * config/rs6000/rs6000.c (make_resolver_func): Update
18083 init_lowered_empty_function call.
18085 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
18087 * doc/invoke.texi: Document the -fprofile-abs-path option.
18088 * common.opt (fprofile-abs-path): New option.
18089 * gcov-io.h (gcov_write_filename): Declare.
18090 * gcov-io.c (gcov_write_filename): New function.
18091 * coverage.c (coverage_begin_function): Use gcov_write_filename.
18092 * profile.c (output_location): Likewise.
18094 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18096 * shring-wrap.c: Revert accidental commit.
18098 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
18100 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
18102 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18104 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
18106 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
18107 profile in callgraph edge.
18108 * profile-count.h (apply_probability): If THIS is 0, then result is 0
18109 (apply_scale): Likewise.
18110 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
18111 Also scale profile when inlining function with zero profile.
18112 (initialize_cfun): Update exit block profile even when it is zero.
18113 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
18114 when profile is read.
18116 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18118 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
18119 (CLONE_*): New constants to define the processors we can generate
18120 code for with the target_clone attribute.
18121 (rs6000_clone_map): New array to identify which clone processors
18122 the current program is running on.
18123 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
18124 target_clone attribute.
18125 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
18126 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
18127 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
18128 (cpu_expand_builtin): Add support for target_clone attribute.
18129 (rs6000_valid_attribute_p): Allow "default" attribute.
18130 (get_decl_name): New debug function to simplify printing the
18131 current function name in debugging statements.
18132 (rs6000_clone_priority): New functions to support the target_clone
18133 attribute, and be able to generate code to switch between ISA 2.05
18134 through ISA 3.0 (power6 through power9).
18135 (rs6000_compare_version_priority): Likewise.
18136 (rs6000_get_function_versions_dispatcher): Likewise.
18137 (make_resolver_func): Likewise.
18138 (add_condition_to_bb): Likewise.
18139 (dispatch_function_versions): Likewise.
18140 (rs6000_generate_version_dispatcher_body): Likewise.
18141 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
18142 (fusion_gpr_load_p): Fix a spacing issue.
18143 * doc/extend.texi (Common Function Attributes): Document that the
18144 PowerPC supports the target_clone attribute.
18146 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
18148 * config/arm/arm.h: explain F symbol found in description of ARM
18149 register allocation in its legend.
18151 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18153 * config/mips/frame-header-opt.c: Include profile-count.h.
18154 * config/riscv/riscv.c: Include profile-count.h
18156 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18158 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
18160 (sm_set_flag_if_changed): Add bbs field.
18161 (execute_sm_if_changed_flag_set): Pass BBS.
18162 (execute_sm): Update.
18164 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18166 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
18169 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18171 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
18172 (peephole2): New peephole2 to emit the above.
18173 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
18175 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18177 * config/aarch64/aarch64.c (define_peephole2 above
18178 *sub_<shift>_<mode>): New peephole.
18180 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18182 * config/i386/i386.c (make_resolver_func): Update.
18183 * Makefile.in: Add profile-count.h and profile-count.o
18184 * auto-profile.c (afdo_indirect_call): Update to new API.
18185 (afdo_set_bb_count): Update.
18186 (afdo_propagate_edge): Update.
18187 (afdo_propagate_circuit): Update.
18188 (afdo_calculate_branch_prob): Update.
18189 (afdo_annotate_cfg): Update.
18190 * basic-block.h: Include profile-count.h
18191 (struct edge_def): Turn count to profile_count.
18192 (struct basic_block_def): Likewie.
18193 (REG_BR_PROB_BASE): Move to profile-count.h
18194 (RDIV): Move to profile-count.h
18195 * bb-reorder.c (max_entry_count): Turn to profile_count.
18196 (find_traces): Update.
18197 (rotate_loop):Update.
18198 (connect_traces):Update.
18199 (sanitize_hot_paths):Update.
18200 * bt-load.c (migrate_btr_defs): Update.
18201 * cfg.c (RDIV): Remove.
18202 (init_flow): Use alloc_block.
18203 (alloc_block): Uninitialize count.
18204 (unchecked_make_edge): Uninitialize count.
18205 (check_bb_profile): Update.
18206 (dump_edge_info): Update.
18207 (dump_bb_info): Update.
18208 (update_bb_profile_for_threading): Update.
18209 (scale_bbs_frequencies_int): Update.
18210 (scale_bbs_frequencies_gcov_type): Update.
18211 (scale_bbs_frequencies_profile_count): New.
18212 * cfg.h (update_bb_profile_for_threading): Update.
18213 (scale_bbs_frequencies_profile_count): Declare.
18214 * cfgbuild.c (compute_outgoing_frequencies): Update.
18215 (find_many_sub_basic_blocks): Update.
18216 * cfgcleanup.c (try_forward_edges): Update.
18217 (try_crossjump_to_edge): Update.
18218 * cfgexpand.c (expand_gimple_tailcall): Update.
18219 (construct_exit_block): Update.
18220 * cfghooks.c (verify_flow_info): Update.
18221 (dump_bb_for_graph): Update.
18222 (split_edge): Update.
18223 (make_forwarder_block): Update.
18224 (duplicate_block): Update.
18225 (account_profile_record): Update.
18226 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
18227 (get_estimated_loop_iterations): Update.
18228 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
18229 (single_likely_exit): Update.
18230 * cfgloopmanip.c (scale_loop_profile): Update.
18232 (set_zero_probability): Update.
18233 (lv_adjust_loop_entry_edge): Update.
18234 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
18235 (purge_dead_edges): Update.
18236 (rtl_account_profile_record): Update.
18237 * cgraph.c (cgraph_node::create): Uninitialize count.
18238 (symbol_table::create_edge): Uninitialize count.
18239 (cgraph_update_edges_for_call_stmt_node): Update.
18240 (cgraph_edge::dump_edge_flags): Update.
18241 (cgraph_node::dump): Update.
18242 (cgraph_edge::maybe_hot_p): Update.
18243 * cgraph.h: Include profile-count.h
18244 (create_clone), create_edge, create_indirect_edge): Update.
18245 (cgraph_node): Turn count to profile_count.
18246 (cgraph_edge0: Likewise.
18247 (make_speculative, clone): Update.
18248 (create_edge): Update.
18249 (init_lowered_empty_function): Update.
18250 * cgraphclones.c (cgraph_edge::clone): Update.
18251 (duplicate_thunk_for_node): Update.
18252 (cgraph_node::create_clone): Update.
18253 * cgraphunit.c (cgraph_node::analyze): Update.
18254 (cgraph_node::expand_thunk): Update.
18255 * final.c (dump_basic_block_info): Update.
18256 * gimple-streamer-in.c (input_bb): Update.
18257 * gimple-streamer-out.c (output_bb): Update.
18258 * graphite.c (print_global_statistics): Update.
18259 (print_graphite_scop_statistics): Update.
18260 * hsa-brig.c: Include basic-block.h.
18261 * hsa-dump.c: Include basic-block.h.
18262 * hsa-gen.c (T sum_slice): Update.
18263 (convert_switch_statements):Update.
18264 * hsa-regalloc.c: Include basic-block.h.
18265 * ipa-chkp.c (chkp_produce_thunks): Update.
18266 * ipa-cp.c (struct caller_statistics): Update.
18267 (init_caller_stats): Update.
18268 (gather_caller_stats): Update.
18269 (ipcp_cloning_candidate_p): Update.
18270 (good_cloning_opportunity_p): Update.
18271 (get_info_about_necessary_edges): Update.
18272 (dump_profile_updates): Update.
18273 (update_profiling_info): Update.
18274 (update_specialized_profile): Update.
18275 (perhaps_add_new_callers): Update.
18276 (decide_about_value): Update.
18277 (ipa_cp_c_finalize): Update.
18278 * ipa-devirt.c (struct odr_type_warn_count): Update.
18279 (struct decl_warn_count): Update.
18280 (struct final_warning_record): Update.
18281 (possible_polymorphic_call_targets): Update.
18282 (ipa_devirt): Update.
18283 * ipa-fnsummary.c (redirect_to_unreachable): Update.
18284 * ipa-icf.c (sem_function::merge): Update.
18285 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
18286 * ipa-inline.c (compute_uninlined_call_time): Update.
18287 (compute_inlined_call_time): Update.
18288 (want_inline_small_function_p): Update.
18289 (want_inline_self_recursive_call_p): Update.
18290 (edge_badness): Update.
18291 (lookup_recursive_calls): Update.
18292 (recursive_inlining): Update.
18293 (inline_small_functions): Update.
18294 (dump_overall_stats): Update.
18295 (dump_inline_stats): Update.
18296 * ipa-profile.c (ipa_profile_generate_summary): Update.
18297 (ipa_propagate_frequency): Update.
18298 (ipa_profile): Update.
18299 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18300 * ipa-utils.c (ipa_merge_profiles): Update.
18301 * loop-doloop.c (doloop_modify): Update.
18302 * loop-unroll.c (report_unroll): Update.
18303 (unroll_loop_runtime_iterations): Update.
18304 * lto-cgraph.c (lto_output_edge): Update.
18305 (lto_output_node): Update.
18306 (input_node): Update.
18307 (input_edge): Update.
18308 (merge_profile_summaries): Update.
18309 * lto-streamer-in.c (input_cfg): Update.
18310 * lto-streamer-out.c (output_cfg): Update.
18311 * mcf.c (create_fixup_graph): Update.
18312 (adjust_cfg_counts): Update.
18313 (sum_edge_counts): Update.
18314 * modulo-sched.c (sms_schedule): Update.
18315 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
18316 * predict.c (maybe_hot_count_p): Update.
18317 (probably_never_executed): Update.
18318 (dump_prediction): Update.
18319 (combine_predictions_for_bb): Update.
18320 (propagate_freq): Update.
18321 (handle_missing_profiles): Update.
18322 (counts_to_freqs): Update.
18323 (rebuild_frequencies): Update.
18324 (force_edge_cold): Update.
18325 * predict.h: Include profile-count.h
18326 (maybe_hot_count_p, counts_to_freqs): UPdate.
18327 * print-rtl-function.c: Do not include cfg.h
18328 * print-rtl.c: Include basic-block.h
18329 * profile-count.c: New file.
18330 * profile-count.h: New file.
18331 * profile.c (is_edge_inconsistent): Update.
18332 (correct_negative_edge_counts): Update.
18333 (is_inconsistent): Update.
18334 (set_bb_counts): Update.
18335 (read_profile_edge_counts): Update.
18336 (compute_frequency_overlap): Update.
18337 (compute_branch_probabilities): Update; Initialize and deinitialize
18339 (branch_prob): Update.
18340 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
18341 (edge_gcov_count): New.
18342 (bb_gcov_count): New.
18343 * shrink-wrap.c (try_shrink_wrapping): Update.
18344 * tracer.c (better_p): Update.
18345 * trans-mem.c (expand_transaction): Update.
18346 (ipa_tm_insert_irr_call): Update.
18347 (ipa_tm_insert_gettmclone_call): Update.
18348 * tree-call-cdce.c: Update.
18349 * tree-cfg.c (gimple_duplicate_sese_region): Update.
18350 (gimple_duplicate_sese_tail): Update.
18351 (gimple_account_profile_record): Update.
18352 (execute_fixup_cfg): Update.
18353 * tree-inline.c (copy_bb): Update.
18354 (copy_edges_for_bb): Update.
18355 (initialize_cfun): Update.
18356 (freqs_to_counts): Update.
18357 (copy_cfg_body): Update.
18358 (expand_call_inline): Update.
18359 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18360 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
18361 (try_unroll_loop_completely): Update.
18362 (try_peel_loop): Update.
18363 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18364 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
18365 * tree-ssa-loop-split.c (connect_loops): Update.
18366 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
18367 * tree-ssa-reassoc.c (branch_fixup): Update.
18368 * tree-ssa-tail-merge.c (replace_block_by): Update.
18369 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
18370 (compute_path_counts): Update.
18371 (update_profile): Update.
18372 (recompute_probabilities): Update.
18373 (update_joiner_offpath_counts): Update.
18374 (estimated_freqs_path): Update.
18375 (freqs_to_counts_path): Update.
18376 (clear_counts_path): Update.
18377 (ssa_fix_duplicate_block_edges): Update.
18378 (duplicate_thread_path): Update.
18379 * tree-switch-conversion.c (case_bit_test_cmp): Update.
18380 (struct switch_conv_info): Update.
18381 * tree-tailcall.c (decrease_profile): Update.
18382 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18383 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18384 * value-prof.c (check_counter): Update.
18385 (gimple_divmod_fixed_value): Update.
18386 (gimple_mod_pow2): Update.
18387 (gimple_mod_subtract): Update.
18388 (gimple_ic_transform): Update.
18389 (gimple_stringop_fixed_value): Update.
18390 * value-prof.h (gimple_ic): Update.
18392 2017-06-02 Carl Love <cel@us.ibm.com>
18394 * config/rs6000/rs6000-c: Add support for built-in functions
18395 vector double vec_doublee (vector signed int);
18396 vector double vec_doublee (vector unsigned int);
18397 vector double vec_doublee (vector float);
18398 vector double vec_doubleh (vector signed int);
18399 vector double vec_doubleh (vector unsigned int);
18400 vector double vec_doubleh (vector float);
18401 vector double vec_doublel (vector signed int);
18402 vector double vec_doublel (vector unsigned int);
18403 vector double vec_doublel (vector float);
18404 vector double vec_doubleo (vector signed int);
18405 vector double vec_doubleo (vector unsigned int);
18406 vector double vec_doubleo (vector float);.
18407 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
18408 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
18410 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
18411 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
18412 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
18414 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
18415 vec_doublel, vec_doubleh.
18416 * doc/extend.texi: Update the built-in documentation file for the
18417 new built-in functions.
18419 2017-06-02 David Malcolm <dmalcolm@redhat.com>
18422 * ipa-inline-analysis.c (free_growth_caches): Set
18423 edge_removal_hook_holder to NULL after removing it.
18425 2017-06-02 Sudakshina Das <sudi.das@arm.com>
18427 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
18428 comparision with zero.
18430 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
18431 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18432 for early expansion of vec_min and vec_max builtins.
18433 (builtin_function_type): Add min/max unsigned variants to those
18434 identified as having unsigned arguments.
18436 2017-06-02 Olivier Hainque <hainque@adacore.com>
18438 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
18440 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18442 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
18443 Use VALL_F16 iterator rather than VALL.
18445 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18447 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
18448 Emit CBNZ inside loop when doing a strong exchange and comparing
18449 against zero. Generate the CC flags after the loop.
18451 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
18453 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
18454 (dl_section_ref): New.
18455 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
18456 On AIX, append an expression to subtract the size of the
18457 section length to dl_section_ref.
18459 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
18461 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18462 for early expansion of vector absolute builtins.
18464 2017-06-02 Richard Biener <rguenther@suse.de>
18466 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
18467 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
18469 2017-06-02 Richard Biener <rguenther@suse.de>
18471 PR tree-optimization/80948
18472 * tree-tailcall.c (find_tail_calls): Track stmts to move in
18473 stmt order as well.
18475 2017-06-02 Richard Biener <rguenther@suse.de>
18477 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
18479 * tree-vect-stmts.c (process_use): Do not mark backedge defs
18480 for inductions as relevant.
18482 2017-06-02 Richard Biener <rguenther@suse.de>
18484 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
18485 (vectorizable_induction): ... this. Remove dead code.
18487 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
18489 * builtins. (expand_builtin_alloca): Remove second parameter and
18490 infer its value from the first parameter instead.
18491 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
18493 2017-06-02 Jakub Jelinek <jakub@redhat.com>
18495 PR rtl-optimization/80903
18496 * loop-doloop.c (add_test): Unshare sequence.
18498 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
18500 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
18502 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
18504 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
18506 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
18507 xlogue_layout::get_instance, logue_layout::xlogue_layout,
18508 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
18509 (xlogue_layout::get_stub_rtx): Make static.
18510 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
18511 (xlogue_layout::compute_stub_managed_regs): Rename to...
18512 (xlogue_layout::count_stub_managed_regs): ...this.
18513 (xlogue_layout::is_stub_managed_reg): New function.
18514 (xlogue_layout::m_stub_names): Rename to...
18515 (xlogue_layout::s_stub_names): ...this, make static.
18516 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
18517 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
18518 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
18519 xlogue_layout::s_stub_names): Instantiate statics.
18520 (stub_managed_regs): Remove.
18521 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
18522 (disable_call_ms2sysv_xlogues): Rename to...
18523 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
18524 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
18526 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
18527 change after reload_completed.
18528 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
18530 (ix86_expand_prologue): Likewise.
18531 (ix86_expand_epilogue): Likewise.
18532 (ix86_expand_split_stack_prologue): Likewise.
18533 (ix86_compute_frame_layout): Remove frame parameter ...
18534 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
18535 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
18537 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
18538 (ix86_frame): Move from here ...
18539 * config/i386/i386.h (ix86_frame): ... to here.
18540 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
18541 complete ix86_frame data structure instead. Remove some_ld_name.
18543 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
18545 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
18546 symbols that hold a DECL_VALUE_EXPR.
18548 2017-06-01 Martin Jambor <mjambor@suse.cz>
18550 PR tree-optimization/80898
18551 * tree-sra.c (process_subtree_disqualification): Removed.
18552 (disqualify_candidate): Do not acll
18553 process_subtree_disqualification.
18554 (subtree_mark_written_and_enqueue): New function.
18555 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
18556 RHS has been disqualified and re-queue LHS if necessary. Apart
18557 from that, ignore disqualified RHS.
18559 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18561 * config/s390/s390.c (s390_emit_epilogue): Disable early return
18562 address fetch for z10 or later.
18564 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18566 * config/arc/arc.md (tst_movb): Add guard when splitting.
18568 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18570 * config/arc/arc.c (arc_can_eliminate): Test against
18571 arc_frame_pointer_needed.
18573 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18575 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
18576 to prevent store reordering.
18577 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
18578 (type): Add block type.
18579 (stack_tie): Define special instruction to be used in
18582 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18584 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
18585 constraint. It is not valid for the pattern.
18586 (noncommutative_binary_comparison): Likewise.
18588 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18590 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
18593 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18595 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
18596 be used by the reg-alloc.
18598 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18600 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
18601 reg-alloc when having mul64 or mul32x16 instructions.
18602 (mulsidi3): Likewise.
18603 (umulsidi3): Likewise.
18604 (mulsi32x16): New pattern.
18605 (mulsi64): Likewise.
18606 (mulsidi64): Likewise.
18607 (umulsidi64): Likewise.
18608 (MUL32x16_REG): Define.
18609 (mul64_600): Use MUL32x16_REG.
18610 (mac64_600): Likewise.
18611 (umul64_600): Likewise.
18612 (umac64_600): Likewise.
18614 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18616 * config/arc/arc.md (mulsi3_700): Make it commutative.
18618 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
18620 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
18622 (*sign_extendsidi2_insn): Likewise for movstosw.
18624 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
18626 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
18627 the type of the input discriminant value. Convert the
18628 discriminant value of signedness vary.
18630 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
18632 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
18633 Add to -Wall section.
18635 2017-06-01 Richard Biener <rguenther@suse.de>
18637 PR middle-end/66313
18638 * fold-const.c (fold_plusminus_mult_expr): If the factored
18639 factor may be zero use a wrapping type for the inner operation.
18640 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
18641 and handle moved defs.
18642 (process_assignment): Properly guard the unary op case. Return a
18643 tri-state indicating that moving the stmt before the call may allow
18644 to continue. Pass through to_move.
18645 (find_tail_calls): Handle moving unrelated defs before
18648 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
18651 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
18652 splitter result in the canonical way.
18654 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
18656 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
18657 also for 32bit target. Update insn attributes.
18658 (zero-extendsidi2 splitter): Allow all registers for operand 1.
18660 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
18662 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
18663 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
18664 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
18665 (_mm_maskz_min_ss): New intrinsics.
18667 2017-05-31 Martin Liska <mliska@suse.cz>
18669 * tree-vect-loop.c (vect_create_epilog_for_reduction):
18670 Change comment style to one we normally use.
18671 (vectorizable_reduction): Likewise.
18672 (vectorizable_induction): Likewise.
18673 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
18674 (vectorizable_call): Likewise.
18675 (vectorizable_simd_clone_call): Likewise.
18676 (vectorizable_conversion): Likewise.
18677 (vectorizable_assignment): Likewise.
18678 (vectorizable_shift): Likewise.
18679 (vectorizable_operation): Likewise.
18680 (vectorizable_store): Likewise.
18681 (vectorizable_load): Likewise.
18682 * tree-vectorizer.h: Likewise.
18684 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
18686 * passes.c (emergency_dump_function): New.
18687 * tree-pass.h (emergency_dump_function): Declare.
18688 * plugin.c (plugins_internal_error_function): Remove.
18689 * plugin.h (plugins_internal_error_function): Remove declaration.
18690 * toplev.c (internal_error_function): New static function. Use it...
18691 (general_init): ...here.
18693 2017-05-31 Graham Markall <graham.markall@embecosm.com>
18695 * config/arc/arc.c (arc_print_operand): Handle constant operands.
18696 (arc_rtx_costs): Add costs for new patterns.
18697 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
18698 * config/arc/predicates.md: Add _1_2_3_operand predicate.
18700 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
18702 * tree-ssa-strlen.c (get_next_strinfo): New function.
18703 (get_stridx_plus_constant): Use it.
18704 (zero_length_string): Likewise.
18705 (adjust_related_strinfos): Likewise.
18706 (adjust_last_stmt): Likewise.
18708 2017-05-31 Richard Biener <rguenther@suse.de>
18711 * config/i386/i386.c (ix86_expand_builtin): Remove assert
18712 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
18714 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
18716 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
18717 loop_vinfo argument and use of dependence distance vectors.
18718 Check instead whether the two references differ only in their
18719 initial value and assume that they have the same alignment if the
18720 difference is a multiple of the vector alignment.
18721 (vect_analyze_data_refs_alignment): Update call accordingly.
18723 2017-05-31 Martin Liska <mliska@suse.cz>
18726 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
18728 2017-05-31 Bin Cheng <bin.cheng@arm.com>
18730 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
18731 (create_intersect_range_checks): Move from ...
18732 * tree-data-ref.c (create_intersect_range_checks_index)
18733 (create_intersect_range_checks): ... to here.
18734 (create_runtime_alias_checks): New function factored from ...
18735 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
18736 here. Call above function.
18737 * tree-data-ref.h (create_runtime_alias_checks): New function.
18739 2017-05-31 Bin Cheng <bin.cheng@arm.com>
18741 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
18742 segment length for dr_b and compute it in wide_int.
18744 2017-05-31 Richard Biener <rguenther@suse.de>
18746 PR tree-optimization/80906
18747 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
18748 and pass through iv_map.
18749 (copy_bb_and_scalar_dependences): Adjust.
18750 (translate_pending_phi_nodes): Likewise.
18751 (copy_loop_close_phi_args): Handle code-generating IVs instead
18754 2017-05-30 David Malcolm <dmalcolm@redhat.com>
18756 * diagnostic-color.c (color_dict): Add "type-diff".
18757 (parse_gcc_colors): Update comment.
18758 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
18759 -fdiagnostics-show-template-tree and -fno-elide-type.
18760 (GCC_COLORS): Add type-diff to example.
18762 (-fdiagnostics-show-template-tree): New.
18763 (-fno-elide-type): New.
18764 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
18765 the pp_format_decoder callback. Call any m_format_postprocessor's
18767 (pretty_printer::pretty_printer): Initialize
18768 m_format_postprocessor.
18769 (pretty_printer::~pretty_printer): Delete any
18770 m_format_postprocessor.
18771 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
18772 (class format_postprocessor): New class.
18773 (struct pretty_printer::format_decoder): Document the new parameters.
18774 (struct pretty_printer::m_format_postprocessor): New field.
18775 * tree-diagnostic.c (default_tree_printer): Update for new
18776 bool and const char ** params.
18777 * tree-diagnostic.h (default_tree_printer): Likewise.
18779 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
18781 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
18782 (lwa_operand): Delete rs6000_gen_cell_microcode test.
18783 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
18784 rs6000_gen_cell_microcode code.
18785 (rs6000_final_prescan_insn): Delete.
18786 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
18787 "warn-cell-microcode" entries.
18788 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
18789 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
18790 throughout. Change cc_reg_not_micro_cr0_operand to
18791 cc_reg_not_cr0_operand throughout.
18792 (*extendhi<mode>2_noload): Delete.
18793 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
18794 (mwarn-cell-microcode): Delete.
18795 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
18796 -mgen-cell-microcode and -mwarn-cell-microcode.
18798 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
18801 * config/i386/constraints.md (Yd): New constraint.
18803 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
18804 and (?Yd, r) alternatives. Update insn attributes.
18805 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
18806 and (?*Yd, r) alternatives. Update insn attributes.
18807 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
18809 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
18811 * gimplify.c (gimplify_modify_expr): Don't create a
18812 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
18815 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
18817 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
18819 2017-05-30 Richard Biener <rguenther@suse.de>
18821 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
18822 and reduc_def fields.
18823 (STMT_VINFO_REDUC_TYPE): New define.
18824 (STMT_VINFO_REDUC_DEF): Likewise.
18825 (vect_force_simple_reduction): Adjust prototype.
18826 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
18827 (vect_is_simple_reduction): Remove check_reduction argument.
18828 (vect_force_simple_reduction): Adjust and set
18829 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
18830 (vectorizable_reduction): Do not re-do reduction analysis
18831 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
18832 * tree-parloops.c (gather_scalar_reductions): Adjust.
18834 2017-05-30 Richard Biener <rguenther@suse.de>
18836 PR middle-end/80901
18837 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
18840 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
18842 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
18843 Introduce unknown_misalignment parameter and remove vf.
18844 (vect_peeling_hash_get_lowest_cost):
18845 Pass unknown_misalignment parameter.
18846 (vect_enhance_data_refs_alignment):
18847 Fix unsupportable data ref treatment.
18849 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18851 * tree-vect-data-refs.c (vect_get_data_access_cost):
18852 Workaround for SLP handling.
18853 (vect_enhance_data_refs_alignment):
18854 Compute costs for doing no peeling at all, compare to the best
18855 peeling costs so far and avoid peeling if cheaper.
18857 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18859 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
18860 Return peeling info and set costs to zero for unlimited cost
18862 (vect_enhance_data_refs_alignment): Also inspect all datarefs
18863 with unknown misalignment. Compute and costs for unknown
18864 misalignment, compare them to the costs for known misalignment
18865 and choose the cheapest for peeling.
18867 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18869 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
18870 (vect_get_peeling_costs_all_drs): Create function.
18871 (vect_peeling_hash_get_lowest_cost):
18872 Use vect_get_peeling_costs_all_drs.
18873 (vect_peeling_supportable): Create function.
18874 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
18876 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18878 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
18879 DR_HAS_NEGATIVE_STEP.
18880 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
18881 (vect_enhance_data_refs_alignment): Use.
18882 (vect_duplicate_ssa_name_ptr_info): Use.
18883 * tree-vectorizer.h (dr_misalignment): Use.
18884 (known_alignment_for_access_p): Use.
18886 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18889 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
18891 (has_section_name): New function.
18893 2017-05-30 Martin Liska <mliska@suse.cz>
18896 * auto-profile.c (get_function_decl_from_block): Fix
18899 2017-05-30 Richard Biener <rguenther@suse.de>
18901 PR middle-end/80876
18902 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
18904 2017-05-30 Martin Liska <mliska@suse.cz>
18906 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
18907 * dumpfile.h (struct dump_file_info): Remove ctors.
18909 2017-05-30 Martin Liska <mliska@suse.cz>
18911 * predict.def: Fix GNU coding style.
18913 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
18915 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
18916 Mark 'to' argument with ATTRIBUTE_UNUSED.
18918 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
18920 * config/xtensa/xtensa.c (xtensa_emit_call): Use
18921 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
18922 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
18925 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
18927 * doc/install.texi (Options specification): Restore entry of
18928 --enable-sjlj-exceptions.
18930 2017-05-27 Michael Eager <eager@eagercon.com>
18933 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
18935 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
18937 * config/microblaze/microblaze.h
18938 (FIXED_REGISTERS): Update in macro.
18939 (CALL_USED_REGISTERS): Update in macro.
18941 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
18943 * doc/install.texi: Add links to macOS binary distributions.
18945 2017-05-27 Jakub Jelinek <jakub@redhat.com>
18949 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18951 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
18953 2017-05-26 Martin Liska <mliska@suse.cz>
18955 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
18957 2017-05-26 Martin Liska <mliska@suse.cz>
18959 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
18960 always leading ';; '.
18961 (dump_bb_info): Likewise.
18962 (brief_dump_cfg): Likewise.
18963 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
18964 * dumpfile.c: Remove usage of TDF_VERBOSE.
18965 * dumpfile.h (enum dump_kind): Likewise.
18966 (dump_gimple_bb_header): Do not use TDF_COMMENT.
18967 * print-tree.c (debug_verbose): Remove.
18968 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
18969 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
18970 * tree-diagnostic.c (default_tree_printer): Replace
18971 TDF_DIAGNOSTIC with TDF_SLIM.
18973 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18975 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
18976 in parameter loop, rather than loop_vinfo.
18977 (create_intersect_range_checks): Ditto.
18978 (vect_create_cond_for_alias_checks): Update call to above functions.
18980 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18982 PR tree-optimization/80815
18983 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
18984 for merging runtime alias checks. Handle negative DR_STEPs.
18986 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18988 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
18990 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
18991 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
18992 out code pruning runtime alias checks.
18993 * tree-data-ref.c (prune_runtime_alias_test_list): New function
18994 factored out from above.
18995 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
18997 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
18999 (prune_runtime_alias_test_list): New decalaration.
19001 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19003 * tree-vect-data-refs.c (compare_tree): Rename and move ...
19004 * tree-data-ref.c (data_ref_compare_tree): ... to here.
19005 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
19006 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
19007 (operator==, comp_dr_with_seg_len_pair): Ditto.
19008 (vect_prune_runtime_alias_test_list): Ditto.
19010 2017-05-26 Martin Liska <mliska@suse.cz>
19013 * params.def: Bound partial-inlining-entry-probability param.
19015 2017-05-26 Marek Polacek <polacek@redhat.com>
19018 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
19021 2017-05-26 Richard Biener <rguenther@suse.de>
19023 PR tree-optimization/80842
19024 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
19027 2017-05-26 Richard Biener <rguenther@suse.de>
19029 PR tree-optimization/80844
19030 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
19032 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
19034 * doc/md.texi (Machine Constraints): Update x86 family
19035 machine constraints section to match 'config/i386/constraints.md'.
19037 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
19039 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
19041 2017-05-25 Nathan Sidwell <nathan@acm.org>
19043 * doc/invoke.texi (--enable-languages): Update documentation.
19045 2017-05-25 Martin Liska <mliska@suse.cz>
19047 * dumpfile.c: Add TDF_FOLDING.
19048 * dumpfile.h (enum dump_kind): Likewise.
19049 * genmatch.c (dt_simplify::gen_1): Use it.
19051 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19053 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
19055 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19057 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
19058 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
19060 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19062 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
19063 * match.pd (X == C): Rewrite it here.
19064 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
19065 with_certain_nonzero_bits2): New predicates.
19066 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
19068 2017-05-24 Nathan Sidwell <nathan@acm.org>
19070 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
19073 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
19076 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
19078 * config/powerpcspe: New port. Files are copied from the rs6000
19079 port, with "rs6000" in filenames replaced by "powerpcspe".
19081 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
19083 PR rtl-optimization/80754
19084 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
19086 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
19088 * config/sparc/sparc.md (length): Return the correct value for -mflat
19089 sibcalls to match output_sibcall.
19091 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
19095 * config/rs6000/rs6000.c (struct machine_function): Add new field
19097 (rs6000_get_separate_components): Init that field, use it.
19098 (rs6000_components_for_bb): Use the field.
19100 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
19102 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
19104 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
19106 PR middle-end/80823
19107 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
19109 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19112 * config/s390/s390.c (s390_check_qrst_address): Check incoming
19113 address against address_operand predicate.
19114 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
19116 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
19118 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
19119 parameters passed indirectly.
19121 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
19123 * config/i386/i386.md (*movdi_internal): Remove SSE4
19124 alternative 18 (?r, *v). Update insn attributes.
19125 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
19126 Update insn attributes.
19127 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
19128 Update insn attributes.
19129 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
19130 alternative 1 (r, v). Remove isa attribute.
19131 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
19132 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
19133 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
19135 2017-05-23 Tom de Vries <tom@codesourcery.com>
19137 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
19140 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19142 * cgraphunit.c (symbol_table::process_new_functions): Update.
19143 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
19144 (inline_generate_summary): Rename to ...
19145 (ipa_fn_summary_generate): ... this one.
19146 (inline_read_summary): Rename to ...
19147 (ipa_fn_summary_read): ... this one.
19148 (inline_write_summary): Rename to ...
19149 (ipa_fn_summary_write): ... this one.
19150 (inline_free_summary): Rename to ...
19151 (ipa_free_fn_summary): ... this one.
19152 (pass_data_local_fn_summary, pass_local_fn_summary,
19153 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
19154 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
19155 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
19156 make_pass_ipa_fn_summary): New.
19157 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
19158 inline_write_summary, inline_free_summary): Remove.
19159 (ipa_free_fn_summary) : New.
19160 * ipa-inline.c (ipa_inline): Update.
19161 (pass_ipa_inline): Do not generate summaries.
19162 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
19164 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
19165 and add pass_ipa_fn_summary.
19166 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
19168 (make_pass_inline_parameters): Remove.
19170 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
19172 * omp-low.c (struct omp_context): Remove "default_kind" member.
19175 * omp-offload.c (execute_oacc_device_lower): Remove the
19176 parallelism dimensions function attributes for unparallelized
19177 OpenACC kernels constructs.
19179 2017-05-23 Martin Liska <mliska@suse.cz>
19181 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
19183 (cgraph_edge::make_speculative): Likewise.
19184 (cgraph_edge::resolve_speculation): Likewise.
19185 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
19186 (cgraph_node::dump): Likewise.
19187 * cgraph.h: Likewise.
19188 * cgraphunit.c (analyze_functions): Likewise.
19189 (symbol_table::compile): Likewise.
19190 * ipa-cp.c (print_all_lattices): Likewise.
19191 (determine_versionability): Likewise.
19192 (initialize_node_lattices): Likewise.
19193 (ipcp_verify_propagated_values): Likewise.
19194 (estimate_local_effects): Likewise.
19195 (update_profiling_info): Likewise.
19196 (create_specialized_node): Likewise.
19197 (perhaps_add_new_callers): Likewise.
19198 (decide_about_value): Likewise.
19199 (decide_whether_version_node): Likewise.
19200 (identify_dead_nodes): Likewise.
19201 (ipcp_store_bits_results): Likewise.
19202 * ipa-devirt.c (dump_targets): Likewise.
19203 (ipa_devirt): Likewise.
19204 * ipa-icf.c (sem_item::dump): Likewise.
19205 (sem_function::equals): Likewise.
19206 (sem_variable::equals): Likewise.
19207 (sem_item_optimizer::read_section): Likewise.
19208 (sem_item_optimizer::execute): Likewise.
19209 (congruence_class::dump): Likewise.
19210 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
19211 (dump_inline_summary): Likewise.
19212 (estimate_node_size_and_time): Likewise.
19213 (inline_analyze_function): Likewise.
19214 * ipa-inline-transform.c (inline_call): Likewise.
19215 * ipa-inline.c (report_inline_failed_reason): Likewise.
19216 (want_early_inline_function_p): Likewise.
19217 (edge_badness): Likewise.
19218 (update_edge_key): Likewise.
19219 (inline_small_functions): Likewise.
19220 * ipa-profile.c (ipa_profile): Likewise.
19221 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
19222 (ipa_make_edge_direct_to_target): Likewise.
19223 (remove_described_reference): Likewise.
19224 (ipa_impossible_devirt_target): Likewise.
19225 (propagate_controlled_uses): Likewise.
19226 (ipa_print_node_params): Likewise.
19227 (ipcp_transform_function): Likewise.
19228 * ipa-pure-const.c (pure_const_read_summary): Likewise.
19229 (propagate_pure_const): Likewise.
19230 * ipa-reference.c (generate_summary): Likewise.
19231 (read_write_all_from_decl): Likewise.
19232 (propagate): Likewise.
19233 (ipa_reference_read_optimization_summary): Likewise.
19234 * ipa-utils.c (ipa_merge_profiles): Likewise.
19235 * ipa.c (walk_polymorphic_call_targets): Likewise.
19236 (symbol_table::remove_unreachable_nodes): Likewise.
19237 (ipa_single_use): Likewise.
19238 * passes.c (execute_todo): Likewise.
19239 * predict.c (drop_profile): Likewise.
19240 * symtab.c (symtab_node::get_dump_name): New function.
19241 (symtab_node::dump_name): Likewise.
19242 (symtab_node::dump_asm_name): Likewise.
19243 (symtab_node::dump_references): Likewise.
19244 (symtab_node::dump_referring): Likewise.
19245 (symtab_node::dump_base): Likewise.
19246 (symtab_node::debug_symtab): Likewise.
19247 * tree-sra.c (convert_callers_for_node): Likewise.
19248 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
19249 * value-prof.c (init_node_map): Likewise.
19251 2017-05-23 Martin Liska <mliska@suse.cz>
19253 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
19254 and symtab_node::debug_symtab to symbol_table::debug.
19255 * cgraphunit.c (analyze_functions): Use the renamed function.
19256 (symbol_table::compile): Likewise.
19257 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
19258 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
19259 * passes.c (execute_todo): Likewise.
19260 * symtab.c (symbol_table::dump): New function.
19261 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
19263 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19265 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
19266 that nonconst implies exec.
19268 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19270 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
19271 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
19272 (inline_edge_summary_vec): Turn into ...
19273 (ipa_call_summaries): ... this one.
19274 (redirect_to_unreachable, edge_set_predicate,
19275 evaluate_properties_for_edge, inline_summary_alloc,
19276 reset_ipa_call_summary, reset_inline_summary,
19277 inline_summary_t::duplicate): Update.
19278 (inline_edge_duplication_hook): Turn to ...
19279 (ipa_call_summary_t::duplicate): ... this one.
19280 (inline_edge_removal_hook): Turn to ...
19281 (ipa_call_summary_t::remove): ... this one.
19282 (dump_inline_edge_summary): Turn to ...
19283 (dump_ipa_call_summary): ... this one.
19284 (estimate_function_body_sizes): Update.
19285 (inline_update_callee_summaries): Update.
19286 (remap_edge_change_prob): Update.
19287 (remap_edge_summaries): Update.
19288 (inline_merge_summary): Update.
19289 (do_estimate_edge_time): Update.
19290 (inline_generate_summary): Update.
19291 (inline_read_section): Update.
19292 (inline_read_summary): Update.
19293 (inline_free_summary): Update.
19294 * ipa-inline.c (can_inline_edge_p): Update.
19295 (compute_inlined_call_time): Update.
19296 (want_inline_small_function_p): Update.
19297 (edge_badness): Update.
19298 (early_inliner): Update.
19299 * ipa-inline.h (inline_edge_summary): Turn to ...
19300 (ipa_call_summary): ... this one.
19301 (ipa_call_summary_t): New class.
19302 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
19303 (ipa_call_summaries): New.
19304 (inline_edge_summary): Remove.
19305 (estimate_edge_growth): Update.
19306 * ipa-profile.c (ipa_propagate_frequency_1): Update.
19307 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19308 * ipa-split.c (execute_split_functions): Update.
19309 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
19311 2017-05-23 Tom de Vries <tom@codesourcery.com>
19313 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
19314 attributes): Document rdrand effective target.
19316 2017-05-23 Tom de Vries <tom@codesourcery.com>
19318 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
19319 attributes): Sort alphabetically.
19321 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
19323 * config/avr/genmultilib.awk: Use gsub instead of gensub.
19325 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
19328 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
19329 V2DF/V2DI splat into two separate patterns, one that handles
19330 registers, and the other that only handles memory. Drop support
19331 for splatting from a GPR on ISA 2.07 and then splitting the
19332 splat into direct move and splat.
19333 (vsx_splat_<mode>_reg): Likewise.
19334 (vsx_splat_<mode>_mem): Likewise.
19336 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
19338 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
19340 2017-05-22 Jakub Jelinek <jakub@redhat.com>
19342 PR middle-end/80809
19343 * omp-low.c (finish_taskreg_remap): New function.
19344 (finish_taskreg_scan): If unit size of ctx->record_type
19345 is non-constant, unshare the size expression and replace
19346 decls in it with possible outer var refs.
19348 PR middle-end/80809
19349 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
19350 GOVD_SHARED rather than GOVD_PRIVATE with it.
19351 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
19352 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
19354 PR middle-end/80853
19355 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
19356 as last argument to build_outer_var_ref for pointer bases of array
19357 section reductions.
19359 2017-05-19 Martin Sebor <msebor@redhat.com>
19361 * print-tree.c (print_node): Print DECL_READ_P flag.
19363 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19365 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
19366 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
19367 * cgraph.c: Likewise.
19368 * cgraphunit.c: Likewise.
19369 * gengtype.c: Likewise.
19370 * ipa-cp.c: Likewise.
19371 * ipa-devirt.c: Likewise.
19372 * ipa-icf.c: Likewise.
19373 * ipa-predicate.c: Likewise.
19374 * ipa-profile.c: Likewise.
19375 * ipa-prop.c: Likewise.
19376 * ipa-split.c: Likewise.
19378 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
19379 edge_predicate_pool, dump_inline_hints,
19380 inline_summary::account_size_time, redirect_to_unreachable,
19381 edge_set_predicate, set_hint_predicate,
19382 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
19383 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
19384 inline_summary_t::remove, remap_hint_predicate_after_duplication,
19385 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
19386 ipa_call_summary_t::remove, initialize_growth_caches,
19387 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
19388 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
19389 mark_modified, unmodified_parm_1, unmodified_parm,
19390 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
19391 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
19392 compute_bb_predicates, will_be_nonconstant_expr_predicate,
19393 will_be_nonconstant_predicate, record_modified_bb_info,
19394 get_minimal_bb, record_modified, param_change_prob,
19395 phi_result_unknown_predicate, predicate_for_phi_result,
19396 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
19397 estimate_function_body_sizes, compute_inline_parameters,
19398 compute_inline_parameters_for_curren, pass_data_inline_parameters,
19399 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
19400 inline_update_callee_summaries, remap_edge_change_prob,
19401 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
19402 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
19403 inline_analyze_function, inline_summary_t::insert,
19404 inline_generate_summary, read_ipa_call_summary, inline_read_section,
19405 inline_read_summary, write_ipa_call_summary, inline_write_summary,
19406 inline_free_summary): Move to ipa-fnsummary.h
19407 (predicate_t): Remove.
19408 * ipa-fnsummary.c: New file.
19409 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
19410 (enum inline_hints_vals, inline_hints, agg_position_info,
19411 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
19412 inline_summaries, ipa_call_summary, ipa_call_summary_t,
19413 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
19414 dump_inline_summary, dump_inline_hints, inline_generate_summary,
19415 inline_read_summary, inline_write_summary, inline_free_summary,
19416 inline_analyze_function, initialize_inline_failed,
19417 inline_merge_summary, inline_update_overall_summary,
19418 compute_inline_parameters): Move to ipa-fnsummary.h
19419 * ipa-fnsummary.h: New file.
19420 * ipa-inline-transform.h: Include ipa-inline.h.
19421 * ipa-inline.c: LIkewise.
19423 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19425 * ipa-inline.c (edge_badness): Use inlined_time instead of
19426 inline_summaries->get.
19428 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19430 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
19432 2017-05-22 Nathan Sidwell <nathan@acm.org>
19434 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
19435 (fdump-lang): Document 'raw' option.
19436 * dumpfile.h (TDI_tu): Delete.
19437 * dumpfile.c (dump_files): Remove translation-unit.
19438 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
19440 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
19442 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
19443 command option from $(AWK) call.
19444 * config/avr/genmultilib.awk: Simplify and rewrite so that it
19445 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
19446 [FORMAT]: Remove handling of variable.
19447 * config/avr/t-multilib: Regenerate.
19449 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19451 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
19453 (dump_inline_summary): Do not print self_time.
19454 (estimate_function_body_sizes): Do not set self_time.
19455 (compute_inline_parameters): Likewise.
19456 (inline_read_section, inline_write_summary): Do not stream self_time.
19457 * ipa-inline.h (inline_summary): Drop self_time.
19459 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19461 * ipa-inline-analysis.c (account_size_time): Rename to ...
19462 (inline_summary::account_size_time): ... this one.
19463 (reset_ipa_call_summary): Turn to ...
19464 (ipa_call_summary::reset): ... this one.
19465 (reset_inline_summary): Turn to ...
19466 (inline_summary::reset): ... this one.
19467 (inline_summary_t::remove): Update.
19468 (inline_summary_t::duplicate): Update.
19469 (ipa_call_summary_t::remove): Update.
19470 (dump_inline_summary): Update.
19471 (estimate_function_body_sizes): Update.
19472 (compute_inline_parameters): Update.
19473 (estimate_node_size_and_time): Update.
19474 (inline_merge_summary): Update.
19475 (inline_update_overall_summary): Update.
19476 (inline_read_section): Update.
19477 (inline_write_summary): Update.
19478 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
19479 add account_size_time and reset member functions.
19480 (ipa_call_summary): Add reset function.
19481 * ipa-predicate.h (predicate::operator &): Constify.
19483 2017-05-22 Richard Biener <rguenther@suse.de>
19485 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
19487 2017-05-19 Jason Merrill <jason@redhat.com>
19489 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
19491 2017-05-19 Marek Polacek <polacek@redhat.com>
19494 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
19495 TYPE_OVERFLOW_WRAPS checks.
19497 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
19499 * tree-core.h (enum omp_clause_default_kind): Add
19500 "OMP_CLAUSE_DEFAULT_PRESENT".
19501 * tree-pretty-print.c (dump_omp_clause): Handle it.
19502 * gimplify.c (enum gimplify_omp_var_data): Add
19503 "GOVD_MAP_FORCE_PRESENT".
19504 (gimplify_adjust_omp_clauses_1): Map it to
19505 "GOMP_MAP_FORCE_PRESENT".
19506 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
19508 * gimplify.c (oacc_default_clause): Clarify.
19510 2017-05-19 Nathan Sidwell <nathan@acm.org>
19512 LANG_HOOK_REGISTER_DUMPS
19513 * toplev.c (general_init): Call register dump lang hook.
19514 * doc/invoke.texi: Document -fdump-lang option family.
19515 * dumpfile.c (dump_files): Remove class dump here.
19516 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
19517 * dumpfile.h (tree_dump_index): Remove TDI_class.
19518 * langhooks-def.h (lhd_register_dumps): Declare.
19519 (LANG_HOOKS_REGISTER_DUMPS): Define.
19520 (LANG_HOOKS_INITIALIZER): Add it.
19521 * langhooks.c (lhd_register_dumps): Define.
19522 * langhooks.h (struct lang_hooks): Add register_dumps.
19524 2017-05-19 Nathan Sidwell <nathan@acm.org>
19526 * context.h (context::set_passes): New.
19527 * context.c (context::context): Do not create pass manager.
19528 * toplev.c (general_init): Create pass manager here.
19530 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
19532 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
19533 use this splitter if two add or or instructions would also work for
19534 the constant we want to generate.
19536 2017-05-19 Richard Biener <rguenther@suse.de>
19539 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
19540 predicate evaluation.
19542 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
19544 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
19546 * ipa-inline.c (want_inline_small_function_p): Do not cast to
19549 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
19551 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
19552 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
19553 (inline_edge_summary_vec): Turn into ...
19554 (ipa_call_summaries): ... this one.
19555 (redirect_to_unreachable, edge_set_predicate,
19556 evaluate_properties_for_edge, inline_summary_alloc,
19557 reset_ipa_call_summary, reset_inline_summary,
19558 inline_summary_t::duplicate): Update.
19559 (inline_edge_duplication_hook): Turn to ...
19560 (ipa_call_summary_t::duplicate): ... this one.
19561 (inline_edge_removal_hook): Turn to ...
19562 (ipa_call_summary_t::remove): ... this one.
19563 (dump_inline_edge_summary): Turn to ...
19564 (dump_ipa_call_summary): ... this one.
19565 (estimate_function_body_sizes): Update.
19566 (inline_update_callee_summaries): Update.
19567 (remap_edge_change_prob): Update.
19568 (remap_edge_summaries): Update.
19569 (inline_merge_summary): Update.
19570 (do_estimate_edge_time): Update.
19571 (inline_generate_summary): Update.
19572 (inline_read_section): Update.
19573 (inline_read_summary): Update.
19574 (inline_free_summary): Update.
19575 * ipa-inline.c (can_inline_edge_p): Update.
19576 (compute_inlined_call_time): Update.
19577 (want_inline_small_function_p): Update.
19578 (edge_badness): Update.
19579 (early_inliner): Update.
19580 * ipa-inline.h (inline_edge_summary): Turn to ...
19581 (ipa_call_summary): ... this one.
19582 (ipa_call_summary_t): New class.
19583 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
19584 (ipa_call_summaries): New.
19585 (inline_edge_summary): Remove.
19586 (estimate_edge_growth): Update.
19587 * ipa-profile.c (ipa_propagate_frequency_1): Update.
19588 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19589 * ipa-split.c (execute_split_functions): Update.
19590 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
19592 2017-05-19 Richard Biener <rguenther@suse.de>
19594 PR middle-end/80764
19595 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
19597 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
19599 * config/rs6000/rs6000.c (struct machine_function): Add field
19600 fpr_is_wrapped_separately.
19601 (rs6000_get_separate_components): Use 64 components. Handle the
19602 new FPR components.
19603 (rs6000_components_for_bb): Handle the FPR components.
19604 (rs6000_emit_prologue_components): Handle the FPR components.
19605 (rs6000_emit_epilogue_components): Handle the FPR components.
19606 (rs6000_set_handled_components): Handle the FPR components.
19607 (rs6000_emit_prologue): Don't output prologue code for those FPRs
19608 that are already separately shrink-wrapped.
19609 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
19610 that are already separately shrink-wrapped.
19612 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
19615 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
19618 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
19619 (define_peephole2 for Altivec d-form load): Add peepholes to catch
19620 cases where the register allocator uses a move and an offsettable
19621 memory operation to/from a FPR register on ISA 2.06/2.07.
19622 (define_peephole2 for Altivec d-form store): Likewise.
19624 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
19627 * config/i386/mmx.md (*mov<mode>_internal): Enable
19628 alternatives 11, 12, 13 and 14 also for 32bit targets.
19629 Remove alternatives 15, 16, 17 and 18.
19630 * config/i386/sse.md (vec_concatv2di): Change
19631 alternative (!x, *y) to (x, ?!*Yn).
19633 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
19635 * dumpfile.h (enum dump_kind): Remove stray comma.
19637 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
19639 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
19640 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
19641 predicate::num_conditions
19642 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
19643 (CHANGED): turn into predicate::changed.
19644 (agg_position_info): Move to ipa-predicate.h
19645 (add_condition, predicate::add_clause, predicate::operator &=,
19646 predicate::or_with, predicate::evaluate, predicate::probability,
19647 dump_condition, dump_clause, predicate::dump,
19648 predicate::remap_after_duplication, predicate::remap_after_inlining,
19649 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
19650 (evaluate_conditions_for_known_args): Update.
19651 (set_cond_stmt_execution_predicate): Update.
19652 * ipa-inline.h: Include ipa-predicate.h
19653 (condition, inline_param_summary, conditions, agg_position_info,
19654 predicate): Move to ipa-predicate.h
19655 * ipa-predicate.c: New file.
19656 * ipa-predicate.h: New file.
19658 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
19660 * final.c (leaf_function_p): Check we are not in a sequence.
19662 2017-05-18 Martin Liska <mliska@suse.cz>
19664 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
19665 * dumpfile.c (dump_register): Use new enum dump_kind.
19666 (get_dump_file_name): Likewise.
19667 (dump_enable_all): Likewise.
19668 (dump_switch_p_1): Likewise.
19669 (enable_rtl_dump_file): Remove usage of TDF_RTL.
19670 * dumpfile.h (enum dump_kind): New enum type.
19671 (struct dump_file_info): Create constructor and
19672 format fields and comments.
19673 * passes.c (pass_manager::register_one_dump_file):
19675 * statistics.c (statistics_early_init): Likewise.
19676 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
19677 TDF_TREE with TDF_SLIM.
19678 (gather_memory_references_ref): Likewise.
19680 2017-05-18 Martin Liska <mliska@suse.cz>
19682 * vec.h (struct vnull): Use it.
19684 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
19686 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
19687 (true_predicate, false_predicate, true_predicate_p,
19688 false_predicate_p): Remove.
19689 (single_cond_predicate, not_inlined_predicate): Turn to member function
19691 (add_condition): Update.
19692 (add_clause): Turn to...
19693 (predicate::add_clause): ... this one; update; allow passing NULL
19695 (and_predicates): Turn to ...
19696 (predicate::operator &=): ... this one.
19697 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
19698 (or_predicates): Turn to ...
19699 (predicate::or_with): ... this one.
19700 (evaluate_predicate): Turn to ...
19701 (predicate::evaluate): ... this one.
19702 (predicate_probability): Turn to ...
19703 (predicate::probability): ... this one.
19704 (dump_condition): Update.
19705 (dump_predicate): Turn to ...
19706 (predicate::dump): ... this one.
19707 (account_size_time): Update.
19708 (edge_set_predicate): Update.
19709 (set_hint_predicate): UPdate.
19710 (evaluate_conditions_for_known_args): Update.
19711 (evaluate_properties_for_edge): Update.
19712 (remap_predicate_after_duplication): Turn to...
19713 (predicate::remap_after_duplication): ... this one.
19714 (remap_hint_predicate_after_duplication): Update.
19715 (inline_summary_t::duplicate): UPdate.
19716 (dump_inline_edge_summary): Update.
19717 (dump_inline_summary): Update.
19718 (set_cond_stmt_execution_predicate): Update.
19719 (set_switch_stmt_execution_predicate): Update.
19720 (compute_bb_predicates): Update.
19721 (will_be_nonconstant_expr_predicate): Update.
19722 (will_be_nonconstant_predicate): Update.
19723 (phi_result_unknown_predicate): Update.
19724 (predicate_for_phi_result): Update.
19725 (array_index_predicate): Update.
19726 (estimate_function_body_sizes): Update.
19727 (estimate_node_size_and_time): Update.
19728 (estimate_ipcp_clone_size_and_time): Update.
19729 (remap_predicate): Rename to ...
19730 (predicate::remap_after_inlining): ... this one.
19731 (remap_hint_predicate): Update.
19732 (inline_merge_summary): Update.
19733 (inline_update_overall_summary): Update.
19734 (estimate_size_after_inlining): Update.
19735 (read_predicate): Rename to ...
19736 (predicate::stream_in): ... this one.
19737 (read_inline_edge_summary): Update.
19738 (write_predicate): Rename to ...
19739 (predicate::stream_out): ... this one.
19740 (write_inline_edge_summary): Update.
19741 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
19742 (clause_t): Turn to uint32_t
19743 (predicate): Turn to class; implement constructor and operators
19745 (size_time_entry): Update.
19746 (inline_summary): Update.
19747 (inline_edge_summary): Update.
19749 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
19751 * fold-const.c (fold_binary_loc): Move transformation...
19752 * match.pd (C - X CMP X): ... here.
19754 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
19756 * config/sparc/sparc.c (sparc_option_override): Set function
19757 alignment for -mcpu=niagara7 to 64 to match the I$ line.
19758 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
19760 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
19762 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
19764 2017-05-18 Marek Polacek <polacek@redhat.com>
19767 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
19768 (pass_ubsan::execute): Call gimple_assign_single_p instead of
19769 gimple_assign_load_p.
19771 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
19773 PR middle-end/80692
19774 * real.c (do_compare): Give decimal_do_compare preference over
19775 comparing just the signs.
19777 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
19779 * doc/md.texi (Canonicalization of Instructions): Describe the
19780 canonical form of instructions that inherently set a condition
19783 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
19785 PR middle-end/80775
19786 * tree-cfg.c: Move deletion of unreachable case statements to after
19787 the merging of consecutive case labels.
19789 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
19791 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
19792 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
19793 restoring of callee-saved registers.
19795 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
19797 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
19798 * config/visium/visium.c (single_set_and_flags): Likewise.
19799 * config/visium/visium.md (Substitutions): Likewise.
19801 2017-05-17 Martin Liska <mliska@suse.cz>
19803 * cfg.c: Introduce dump_flags_t type and
19804 use it instead of int type.
19806 * cfghooks.c: Likewise.
19807 * cfghooks.h (struct cfg_hooks): Likewise.
19808 * cfgrtl.c: Likewise.
19809 * cfgrtl.h: Likewise.
19810 * cgraph.c (cgraph_node::get_body): Likewise.
19811 * coretypes.h: Likewise.
19812 * domwalk.c: Likewise.
19813 * domwalk.h: Likewise.
19814 * dumpfile.c (struct dump_option_value_info): Likewise.
19815 (dump_enable_all): Likewise.
19816 (dump_switch_p_1): Likewise.
19817 (opt_info_switch_p): Likewise.
19818 * dumpfile.h (enum tree_dump_index): Likewise.
19819 (struct dump_file_info): Likewise.
19820 * genemit.c: Likewise.
19821 * generic-match-head.c: Likewise.
19822 * gengtype.c (open_base_files): Likewise.
19823 * gimple-pretty-print.c: Likewise.
19824 * gimple-pretty-print.h: Likewise.
19825 * graph.c (print_graph_cfg): Likewise.
19826 * graphite-scop-detection.c (dot_all_sese): Likewise.
19827 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
19828 * loop-unroll.c (report_unroll): Likewise.
19829 * passes.c (pass_manager::register_one_dump_file): Likewise.
19830 * print-tree.c: Likewise.
19831 * statistics.c: Likewise.
19832 * tree-cfg.c: Likewise.
19833 * tree-cfg.h: Likewise.
19834 * tree-dfa.c: Likewise.
19835 * tree-dfa.h: Likewise.
19836 * tree-dump.c (dump_function): Likewise.
19837 * tree-dump.h (struct dump_info): Likewise.
19838 * tree-pretty-print.c: Likewise.
19839 * tree-pretty-print.h: Likewise.
19840 * tree-ssa-live.c: Likewise.
19841 * tree-ssa-live.h: Likewise.
19842 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
19843 * tree-vect-loop.c: Likewise.
19844 * tree-vect-slp.c: Likewise.
19846 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
19847 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19849 PR tree-optimization/80457
19850 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
19851 of all arguments to a statement as scalar_to_vec operations.
19852 (vectorizable_call): Adjust call to vect_model_simple_cost for
19854 (vectorizable_conversion): Likewise.
19855 (vectorizable_assignment): Likewise.
19856 (vectorizable_shift): Likewise.
19857 (vectorizable_operation): Likewise.
19858 (vectorizable_comparison): Likewise.
19859 (vect_is_simple_cond): Record the def types for operands.
19860 (vectorizable_condition): Likewise, call vect_model_simple_cost.
19861 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
19862 for statement argument count.
19864 2017-05-16 Carl Love <cel@us.ibm.com>
19866 * config/rs6000/rs6000-c: Add support for built-in functions
19867 vector unsigned long long vec_bperm (vector unsigned long long,
19868 vector unsigned char)
19869 vector signed long long vec_mule (vector signed int,
19871 vector unsigned long long vec_mule (vector unsigned int,
19872 vector unsigned int)
19873 vector signed long long vec_mulo (vector signed int,
19875 vector unsigned long long vec_mulo (vector unsigned int,
19876 vector unsigned int)
19877 vector signed char vec_sldw (vector signed char,
19878 vector signed char,
19880 vector unsigned char vec_sldw (vector unsigned char,
19881 vector unsigned char,
19883 vector signed short vec_sldw (vector signed short,
19884 vector signed short,
19886 vector unsigned short vec_sldw (vector unsigned short,
19887 vector unsigned short,
19889 vector signed int vec_sldw (vector signed int,
19892 vector unsigned int vec_sldw (vector unsigned int,
19893 vector unsigned int,
19895 vector signed long long vec_sldw (vector signed long long,
19896 vector signed long long,
19898 vector unsigned long long vec_sldw (vector unsigned long long,
19899 vector unsigned long long,
19901 * config/rs6000/rs6000-c: Add support for built-in functions
19902 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
19903 * config/rs6000/altivec.h: Add defintion for vec_sldw.
19904 * doc/extend.texi: Update the built-in documentation for the
19905 new built-in functions.
19907 2017-05-16 Marek Polacek <polacek@redhat.com>
19911 * tree.c (save_expr): Don't fold the expression.
19913 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
19915 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
19916 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
19917 and (?*y,m). Update insn attributes.
19919 2017-05-16 Martin Liska <mliska@suse.cz>
19921 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
19922 flags argument of print_gimple_stmt, print_gimple_expr,
19923 print_generic_stmt and print_generic_expr.
19924 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
19925 * coretypes.h: Likewise.
19926 * except.c (dump_eh_tree): Likewise.
19927 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
19928 * gimple-pretty-print.h: Likewise.
19929 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
19930 (backprop::push_to_worklist): Likewise.
19931 (backprop::pop_from_worklist): Likewise.
19932 (backprop::process_use): Likewise.
19933 (backprop::intersect_uses): Likewise.
19934 (note_replacement): Likewise.
19935 * gimple-ssa-store-merging.c
19936 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
19937 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
19938 (pass_store_merging::execute): Likewise.
19939 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
19940 (ssa_base_cand_dump_callback): Likewise.
19941 (dump_incr_vec): Likewise.
19942 (replace_refs): Likewise.
19943 (replace_mult_candidate): Likewise.
19944 (create_add_on_incoming_edge): Likewise.
19945 (create_phi_basis): Likewise.
19946 (insert_initializers): Likewise.
19947 (all_phi_incrs_profitable): Likewise.
19948 (introduce_cast_before_cand): Likewise.
19949 (replace_one_candidate): Likewise.
19950 * gimplify.c (gimplify_expr): Likewise.
19951 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
19952 (set_rename): Likewise.
19953 (rename_uses): Likewise.
19954 (copy_loop_phi_nodes): Likewise.
19955 (add_close_phis_to_merge_points): Likewise.
19956 (copy_loop_close_phi_args): Likewise.
19957 (copy_cond_phi_args): Likewise.
19958 (graphite_copy_stmts_from_block): Likewise.
19959 (translate_pending_phi_nodes): Likewise.
19960 * graphite-poly.c (print_pdr): Likewise.
19961 (dump_gbb_cases): Likewise.
19962 (dump_gbb_conditions): Likewise.
19963 (print_scop_params): Likewise.
19964 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
19965 (build_cross_bb_scalars_use): Likewise.
19966 (gather_bbs::before_dom_children): Likewise.
19967 * hsa-dump.c (dump_hsa_immed): Likewise.
19968 * ipa-cp.c (print_ipcp_constant_value): Likewise.
19969 (get_replacement_map): Likewise.
19970 * ipa-inline-analysis.c (dump_condition): Likewise.
19971 (estimate_function_body_sizes): Likewise.
19972 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
19973 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
19974 * ipa-prop.c (ipa_dump_param): Likewise.
19975 (ipa_print_node_jump_functions_for_edge): Likewise.
19976 (ipa_modify_call_arguments): Likewise.
19977 (ipa_modify_expr): Likewise.
19978 (ipa_dump_param_adjustments): Likewise.
19979 (ipa_dump_agg_replacement_values): Likewise.
19980 (ipcp_modif_dom_walker::before_dom_children): Likewise.
19981 * ipa-pure-const.c (check_stmt): Likewise.
19982 (pass_nothrow::execute): Likewise.
19983 * ipa-split.c (execute_split_functions): Likewise.
19984 * omp-offload.c (dump_oacc_loop_part): Likewise.
19985 (dump_oacc_loop): Likewise.
19986 * trans-mem.c (tm_log_emit): Likewise.
19987 (tm_memopt_accumulate_memops): Likewise.
19988 (dump_tm_memopt_set): Likewise.
19989 (dump_tm_memopt_transform): Likewise.
19990 * tree-cfg.c (gimple_verify_flow_info): Likewise.
19991 (print_loop): Likewise.
19992 * tree-chkp-opt.c (chkp_print_addr): Likewise.
19993 (chkp_gather_checks_info): Likewise.
19994 (chkp_get_check_result): Likewise.
19995 (chkp_remove_check_if_pass): Likewise.
19996 (chkp_use_outer_bounds_if_possible): Likewise.
19997 (chkp_reduce_bounds_lifetime): Likewise.
19998 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
19999 (chkp_mark_completed_bounds): Likewise.
20000 (chkp_register_incomplete_bounds): Likewise.
20001 (chkp_mark_invalid_bounds): Likewise.
20002 (chkp_maybe_copy_and_register_bounds): Likewise.
20003 (chkp_build_returned_bound): Likewise.
20004 (chkp_get_bound_for_parm): Likewise.
20005 (chkp_build_bndldx): Likewise.
20006 (chkp_get_bounds_by_definition): Likewise.
20007 (chkp_generate_extern_var_bounds): Likewise.
20008 (chkp_get_bounds_for_decl_addr): Likewise.
20009 * tree-chrec.c (chrec_apply): Likewise.
20010 * tree-data-ref.c (dump_data_reference): Likewise.
20011 (dump_subscript): Likewise.
20012 (dump_data_dependence_relation): Likewise.
20013 (analyze_overlapping_iterations): Likewise.
20014 * tree-inline.c (expand_call_inline): Likewise.
20015 (tree_function_versioning): Likewise.
20016 * tree-into-ssa.c (dump_defs_stack): Likewise.
20017 (dump_currdefs): Likewise.
20018 (dump_names_replaced_by): Likewise.
20019 (dump_update_ssa): Likewise.
20020 (update_ssa): Likewise.
20021 * tree-object-size.c (pass_object_sizes::execute): Likewise.
20022 * tree-parloops.c (build_new_reduction): Likewise.
20023 (try_create_reduction_list): Likewise.
20024 (ref_conflicts_with_region): Likewise.
20025 (oacc_entry_exit_ok_1): Likewise.
20026 (oacc_entry_exit_single_gang): Likewise.
20027 * tree-pretty-print.h: Likewise.
20028 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
20029 (get_scalar_evolution): Likewise.
20030 (add_to_evolution): Likewise.
20031 (get_loop_exit_condition): Likewise.
20032 (analyze_evolution_in_loop): Likewise.
20033 (analyze_initial_condition): Likewise.
20034 (analyze_scalar_evolution): Likewise.
20035 (instantiate_scev): Likewise.
20036 (number_of_latch_executions): Likewise.
20037 (gather_chrec_stats): Likewise.
20038 (final_value_replacement_loop): Likewise.
20039 (scev_const_prop): Likewise.
20040 * tree-sra.c (dump_access): Likewise.
20041 (disqualify_candidate): Likewise.
20042 (create_access): Likewise.
20043 (reject): Likewise.
20044 (maybe_add_sra_candidate): Likewise.
20045 (create_access_replacement): Likewise.
20046 (analyze_access_subtree): Likewise.
20047 (analyze_all_variable_accesses): Likewise.
20048 (sra_modify_assign): Likewise.
20049 (initialize_constant_pool_replacements): Likewise.
20050 (find_param_candidates): Likewise.
20051 (decide_one_param_reduction): Likewise.
20052 (replace_removed_params_ssa_names): Likewise.
20053 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
20054 * tree-ssa-copy.c (dump_copy_of): Likewise.
20055 (copy_prop_visit_cond_stmt): Likewise.
20056 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
20057 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
20058 (record_equivalences_from_stmt): Likewise.
20059 * tree-ssa-dse.c (compute_trims): Likewise.
20060 (delete_dead_call): Likewise.
20061 (delete_dead_assignment): Likewise.
20062 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
20063 (forward_propagate_into_cond): Likewise.
20064 (pass_forwprop::execute): Likewise.
20065 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
20066 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
20068 (move_computations_worker): Likewise.
20069 (execute_sm): Likewise.
20070 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
20071 (remove_exits_and_undefined_stmts): Likewise.
20072 (remove_redundant_iv_tests): Likewise.
20073 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
20074 (adjust_iv_update_pos): Likewise.
20075 * tree-ssa-math-opts.c (bswap_replace): Likewise.
20076 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
20077 (value_replacement): Likewise.
20078 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
20079 * tree-ssa-pre.c (print_pre_expr): Likewise.
20080 (get_representative_for): Likewise.
20081 (create_expression_by_pieces): Likewise.
20082 (insert_into_preds_of_block): Likewise.
20083 (eliminate_insert): Likewise.
20084 (eliminate_dom_walker::before_dom_children): Likewise.
20085 (eliminate): Likewise.
20086 (remove_dead_inserted_code): Likewise.
20087 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
20088 * tree-ssa-reassoc.c (get_rank): Likewise.
20089 (eliminate_duplicate_pair): Likewise.
20090 (eliminate_plus_minus_pair): Likewise.
20091 (eliminate_not_pairs): Likewise.
20092 (undistribute_ops_list): Likewise.
20093 (eliminate_redundant_comparison): Likewise.
20094 (update_range_test): Likewise.
20095 (optimize_range_tests_var_bound): Likewise.
20096 (optimize_vec_cond_expr): Likewise.
20097 (rewrite_expr_tree): Likewise.
20098 (rewrite_expr_tree_parallel): Likewise.
20099 (linearize_expr): Likewise.
20100 (break_up_subtract): Likewise.
20101 (linearize_expr_tree): Likewise.
20102 (attempt_builtin_powi): Likewise.
20103 (attempt_builtin_copysign): Likewise.
20104 (transform_stmt_to_copy): Likewise.
20105 (transform_stmt_to_multiply): Likewise.
20106 (dump_ops_vector): Likewise.
20107 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
20108 (print_scc): Likewise.
20109 (set_ssa_val_to): Likewise.
20110 (visit_reference_op_store): Likewise.
20111 (visit_use): Likewise.
20112 (sccvn_dom_walker::before_dom_children): Likewise.
20113 (run_scc_vn): Likewise.
20114 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
20116 (expr_hash_elt::print): Likewise.
20117 (const_and_copies::pop_to_marker): Likewise.
20118 (const_and_copies::record_const_or_copy_raw): Likewise.
20119 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
20120 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
20121 (dump_predicates): Likewise.
20122 (find_uninit_use): Likewise.
20123 (warn_uninitialized_phi): Likewise.
20124 (pass_late_warn_uninitialized::execute): Likewise.
20125 * tree-ssa.c (verify_vssa): Likewise.
20126 (verify_ssa): Likewise.
20127 (maybe_optimize_var): Likewise.
20128 * tree-vrp.c (dump_value_range): Likewise.
20129 (dump_all_value_ranges): Likewise.
20130 (dump_asserts_for): Likewise.
20131 (register_edge_assert_for_2): Likewise.
20132 (vrp_visit_cond_stmt): Likewise.
20133 (vrp_visit_switch_stmt): Likewise.
20134 (vrp_visit_stmt): Likewise.
20135 (vrp_visit_phi_node): Likewise.
20136 (simplify_cond_using_ranges_1): Likewise.
20137 (fold_predicate_in): Likewise.
20138 (evrp_dom_walker::before_dom_children): Likewise.
20139 (evrp_dom_walker::push_value_range): Likewise.
20140 (evrp_dom_walker::pop_value_range): Likewise.
20141 (execute_early_vrp): Likewise.
20143 2017-05-16 Richard Biener <rguenther@suse.de>
20145 * dwarf2out.c (loc_list_from_tree_1): Do not create
20146 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
20148 2017-05-16 Richard Biener <rguenther@suse.de>
20150 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
20152 (note_variable_value_in_expr): If we resolved the decl ref
20153 do not push to the stack.
20155 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
20157 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
20158 operations in fast-math mode.
20159 (vaddq_f16): Likewise.
20160 (vmul_f16): Likewise.
20161 (vmulq_f16): Likewise.
20162 (vsub_f16): Likewise.
20163 (vsubq_f16): Likewise.
20164 * config/arm/neon.md (add<mode>3): New.
20166 (fma:<VH:mode>3): New. Also remove outdated comment.
20169 2017-05-16 Martin Liska <mliska@suse.cz>
20173 * ipa-devirt.c (warn_types_mismatch): Fix typo.
20174 (odr_types_equivalent_p): Likewise.
20176 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
20178 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
20180 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
20183 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
20184 non-interunit SSE move alternatives with '?'.
20185 (zero-extendsidi peephole2): New peephole to skip intermediate
20186 general register in SSE zero-extend sequence.
20188 2017-05-15 Jeff Law <law@redhat.com>
20190 * reorg.c (relax_delay_slots): Create a new variable to hold
20191 the temporary target rather than clobbering TARGET_LABEL.
20193 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
20194 missing argument to extract_bit_field call.
20195 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
20197 2017-05-15 Martin Liska <mliska@suse.cz>
20200 * gcc.c (process_command): Do not allow empty argument of -o option.
20202 2017-05-15 Renlin Li <renlin.li@arm.com>
20204 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
20205 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
20206 * config/aarch64/constraints.md (Usf): Add long call check.
20207 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
20208 (call_value): Likewise.
20209 (sibcall): Likewise.
20210 (sibcall_value): Likewise.
20212 (call_value_insn): New.
20213 (sibcall_insn): Update rtx pattern.
20214 (sibcall_value_insn): Likewise.
20215 (call_internal): Remove.
20216 (call_value_internal): Likewise.
20217 (sibcall_internal): Likewise.
20218 (sibcall_value_internal): Likewise.
20219 (call_reg): Likewise.
20220 (call_symbol): Likewise.
20221 (call_value_reg): Likewise.
20222 (call_value_symbol): Likewise.
20224 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
20227 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
20229 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
20231 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
20232 compatible with CCGOCmode and with CCZmode.
20234 2017-05-14 Martin Sebor <msebor@redhat.com>
20236 PR middle-end/77671
20237 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
20238 (gimple_fold_builtin_snprintf): Same.
20239 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
20240 (gimple_fold_builtin_snprintf): Same.
20241 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
20242 of character types.
20243 (is_call_safe): New function.
20244 (try_substitute_return_value): Call it.
20245 (try_simplify_call): New function.
20246 (pass_sprintf_length::handle_gimple_call): Call it.
20248 2017-05-14 Martin Sebor <msebor@redhat.com>
20250 PR middle-end/80669
20251 * builtins.c (expand_builtin_stpncpy): Simplify.
20253 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
20255 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
20256 * config/i386/i386.h
20257 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
20258 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
20259 (struct machine_function): Add new members call_ms2sysv,
20260 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
20261 (struct machine_frame_state): New fields sp_realigned and
20262 sp_realigned_offset.
20263 * config/i386/i386.c
20264 (enum xlogue_stub): New enum.
20265 (enum xlogue_stub_sets): New enum.
20266 (class xlogue_layout): New class.
20267 (struct ix86_frame): New fields stack_realign_allocate_offset,
20268 stack_realign_offset and outlined_save_offset. Modify comments to
20269 detail stack layout when using out-of-line stubs.
20270 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
20271 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
20272 -mcall-ms2sysv-xlogues.
20273 (stub_managed_regs): New static variable.
20274 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
20275 registers managed by out-of-line stub.
20276 (disable_call_ms2sysv_xlogues): New function.
20277 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
20278 m->call_ms2sysv when appropriate and compute frame layout for
20280 (sp_valid_at, fp_valid_at): New inline functions.
20281 (choose_basereg): New function.
20282 (choose_baseaddr): Add align parameter, use choose_basereg and modify
20284 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
20285 Use align parameter of choose_baseaddr to generated aligned SSE movs
20287 (pro_epilogue_adjust_stack): Modify to track
20288 machine_frame_state::sp_realigned.
20289 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
20290 (ix86_nsaved_sseregs): Likewise.
20291 (ix86_emit_save_regs): Likewise.
20292 (ix86_emit_save_regs_using_mov): Likewise.
20293 (ix86_emit_save_sse_regs_using_mov): Likewise.
20294 (get_scratch_register_on_entry): Likewise.
20295 (gen_frame_set): New function.
20296 (gen_frame_load): Likewise.
20297 (gen_frame_store): Likewise.
20298 (emit_outlined_ms2sysv_save): Likewise.
20299 (emit_outlined_ms2sysv_restore): Likewise.
20300 (ix86_expand_prologue): Modify stack re-alignment code and call
20301 emit_outlined_ms2sysv_save when appropriate.
20302 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
20303 parameter rtx_insn *insn, which allows the function to be used to only
20304 generate the notes.
20305 (ix86_expand_epilogue): Modify validity checks of frame and stack
20306 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
20307 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
20308 * config/i386/predicates.md
20309 (save_multiple): New predicate.
20310 (restore_multiple): Likewise.
20311 * config/i386/sse.md
20312 (save_multiple<mode>): New pattern.
20313 (save_multiple_realign<mode>): Likewise.
20314 (restore_multiple<mode>): Likewise.
20315 (restore_multiple_and_return<mode>): Likewise.
20316 (restore_multiple_leave_return<mode>): Likewise.
20317 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
20319 2017-05-14 Julia Koval <julia.koval@intel.com>
20321 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
20322 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
20323 (__builtin_ia32_xsetbv): New builtins.
20324 * config/i386/i386.c (ix86_expand_special_args_builtin):
20326 (ix86_expand_builtin): Special expand for new intrinsics.
20327 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
20328 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
20329 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
20331 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20333 * cfganal.c (inverted_post_order_compute): Change argument type
20335 * cfganal.h (inverted_post_order_compute): Adjust prototype.
20336 * df-core.c (rest_of_handle_df_initialize): Adjust.
20337 (rest_of_handle_df_finish): Likewise.
20338 (df_analyze_1): Likewise.
20339 (df_analyze): Likewise.
20340 (loop_inverted_post_order_compute): Change argument to be a vec *.
20341 (df_analyze_loop): Adjust.
20342 (df_get_n_blocks): Likewise.
20343 (df_get_postorder): Likewise.
20344 * df.h (struct df_d): Change field to be a vec.
20345 * lcm.c (compute_laterin): Adjust.
20346 (compute_available): Likewise.
20347 * lra-lives.c (lra_create_live_ranges_1): Likewise.
20348 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
20349 * tree-ssa-pre.c (compute_antic): Likewise.
20351 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20353 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
20354 (depth_first_search::depth_first_search): Change structure init
20355 function to this constructor.
20356 (depth_first_search::add_bb): Rename function to this member.
20357 (depth_first_search::execute): Likewise.
20358 (flow_dfs_compute_reverse_finish): Adjust.
20360 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20362 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
20363 (longest_simple_path): Likewise.
20364 * shrink-wrap.c (spread_components): Likewise.
20365 (disqualify_problematic_components): Likewise.
20366 (emit_common_heads_for_components): Likewise.
20367 (emit_common_tails_for_components): Likewise.
20368 (insert_prologue_epilogue_for_components): Likewise.
20370 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20372 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
20375 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20377 * df-core.c (df_set_blocks): Start using auto_bitmap.
20378 (df_compact_blocks): Likewise.
20379 * df-problems.c (df_rd_confluence_n): Likewise.
20380 * df-scan.c (df_insn_rescan_all): Likewise.
20381 (df_process_deferred_rescans): Likewise.
20382 (df_update_entry_block_defs): Likewise.
20383 (df_update_exit_block_uses): Likewise.
20384 (df_entry_block_bitmap_verify): Likewise.
20385 (df_exit_block_bitmap_verify): Likewise.
20386 (df_scan_verify): Likewise.
20387 * lra-constraints.c (lra_constraints): Likewise.
20388 (undo_optional_reloads): Likewise.
20389 (lra_undo_inheritance): Likewise.
20390 * lra-remat.c (calculate_gen_cands): Likewise.
20391 (do_remat): Likewise.
20392 * lra-spills.c (assign_spill_hard_regs): Likewise.
20393 (spill_pseudos): Likewise.
20394 * tree-ssa-pre.c (bitmap_set_and): Likewise.
20395 (bitmap_set_subtract_values): Likewise.
20397 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20399 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
20400 management with auto_bitmap.
20401 (fix_inter_tick): Likewise.
20402 (fix_recovery_deps): Likewise.
20403 * ira.c (add_store_equivs): Likewise.
20404 (find_moveable_pseudos): Likewise.
20405 (split_live_ranges_for_shrink_wrap): Likewise.
20406 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
20407 (rtx_reuse_manager::seen_def_p): Likewise.
20408 (rtx_reuse_manager::set_seen_def): Likewise.
20409 * print-rtl.h (class rtx_reuse_manager): Likewise.
20411 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20413 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
20415 (migrate_btr_def): Likewise.
20416 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
20417 * df-core.c (loop_post_order_compute): Likewise.
20418 (loop_inverted_post_order_compute): Likewise.
20419 * hsa-common.h: Likewise.
20420 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
20421 * init-regs.c (initialize_uninitialized_regs): Likewise.
20422 * ipa-inline.c (resolve_noninline_speculation): Likewise.
20423 (inline_small_functions): Likewise.
20424 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
20425 * ira.c (combine_and_move_insns): Likewise.
20426 (build_insn_chain): Likewise.
20427 * loop-invariant.c (find_invariants): Likewise.
20428 * lower-subreg.c (propagate_pseudo_copies): Likewise.
20429 * predict.c (tree_predict_by_opcode): Likewise.
20430 (predict_paths_leading_to): Likewise.
20431 (predict_paths_leading_to_edge): Likewise.
20432 (estimate_loops_at_level): Likewise.
20433 (estimate_loops): Likewise.
20434 * shrink-wrap.c (try_shrink_wrapping): Likewise.
20435 (spread_components): Likewise.
20436 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
20437 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
20438 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
20439 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
20440 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
20441 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
20442 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
20443 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
20444 (mark_threaded_blocks): Likewise.
20445 (thread_through_all_blocks): Likewise.
20446 * tree-ssa.c (verify_ssa): Likewise.
20447 (execute_update_addresses_taken): Likewise.
20448 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
20450 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20452 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
20454 (post_order_compute): Likewise.
20455 (inverted_post_order_compute): Likewise.
20456 (pre_and_rev_post_order_compute_fn): Likewise.
20458 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20460 * genrecog.c (int_set::int_set): Explicitly construct our
20461 auto_vec base class.
20462 * vec.h (auto_vec::auto_vec): New constructor.
20464 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20466 * bitmap.h (class auto_bitmap): New constructor taking
20467 bitmap_obstack * argument.
20469 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20471 * bitmap.h (class auto_bitmap): Change type of m_bits to
20472 bitmap_head, and adjust ctor / dtor and member operators.
20474 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
20476 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
20477 when returned register mode doesn't match original mode.
20479 2017-05-12 Jeff Law <law@redhat.com>
20480 Jakub Jelinek <jakub@redhat.com>
20482 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
20483 we look for cc setter after the compare-elim changes.
20484 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
20485 within the vector to match what compare-elim now expects.
20486 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
20487 (xorsi3_flags, one_cmplsi2_flags): Likewise.
20489 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
20490 after the compare-elim changes.
20491 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
20492 the vector to match what compare-elim now expects.
20493 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
20494 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
20495 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
20496 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
20497 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
20499 * config/visium/visium.c (single_set_and_flags): Fix where
20500 we look for cc setter after the compare-elim changes.
20501 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
20502 with the vector to match what compare-elim now expects.
20503 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
20504 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
20505 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
20506 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
20507 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
20508 (neg<mode>2_insn_set_overflow): Likewise.
20510 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
20512 PR middle-end/79794
20513 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
20514 maybe_expand_insn call, set ops[0].target. If still set after call,
20515 set alt_rtl. Add extra arg to recursive calls.
20516 (extract_bit_field): Add alt_rtl argument. Pass to
20518 * expmed.h (extract_bit_field): Fix prototype.
20519 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
20520 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
20521 to extract_bit_field_calls.
20522 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
20523 Pass alt_rtl to extract_bit_field calls.
20524 * calls.c (store_unaligned_arguments_into_psuedos)
20525 load_register_parameters): Pass extra NULL to extract_bit_field calls.
20526 * optabs.c (maybe_legitimize_operand): Clear op->target when call
20528 * optabs.h (struct expand_operand): Add target bitfield.
20530 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
20532 * compare-elim.c (try_eliminate_compare): Canonicalize
20533 operation with embedded compare to
20534 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
20535 (set (reg) (operation)].
20537 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
20539 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
20542 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
20543 cost of adding a carry flag for ADC instruction.
20544 [case MINUS]: Ignore the cost of subtracting a carry flag
20545 for SBB instruction.
20547 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
20549 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
20551 * config/rs6000/bmiintrin.h: New file.
20552 * config/rs6000/bmi2intrin.h: New file.
20553 * config/rs6000/x86intrin.h: New file.
20555 2017-05-12 Jeff Law <law@redhat.com>
20557 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
20560 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
20562 PR middle-end/80707
20563 * tree-cfg.c: Remove cfg edges of unreachable case statements.
20565 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20567 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
20568 early expansion of vector divide builtins.
20569 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
20570 builtins identified as having unsigned arguments.
20572 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20574 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
20575 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
20576 expansion of vector logical operations (and, andc, or, xor,
20579 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20581 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
20582 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
20584 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20586 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
20587 early GIMPLE expansion of vector multiplies.
20589 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20591 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
20592 TARGET_HAVE_MOVT conditional.
20593 (movt splitter): Likewise.
20595 2017-05-12 Richard Biener <rguenther@suse.de>
20597 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
20598 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
20599 Fold all stmts not inplace.
20601 2017-05-12 Richard Biener <rguenther@suse.de>
20603 PR tree-optimization/80713
20604 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
20605 inserted_exprs bit for not removed stmts.
20607 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
20609 PR middle-end/69921
20610 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
20611 parallelized" attribute for parallelized OpenACC kernels.
20612 * omp-offload.c (execute_oacc_device_lower): Use it.
20614 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
20615 Set "oacc kernels" attribute.
20616 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
20617 parameter. Adjust all users.
20618 (oacc_fn_attrib_kernels_p): Remove function.
20619 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
20620 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
20621 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
20622 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
20623 assert "oacc kernels" attribute is set.
20625 2017-05-11 Carl Love <cel@us.ibm.com>
20627 * config/rs6000/rs6000-c: Add support for built-in functions
20628 vector unsigned char vec_popcnt (vector signed char)
20629 vector unsigned char vec_popcnt (vector unsigned char)
20630 vector unsigned short vec_popcnt (vector signed short)
20631 vector unsigned short vec_popcnt (vector unsigned short)
20632 vector unsigned int vec_popcnt (vector signed int)
20633 vector unsigned int vec_popcnt (vector unsigned int)
20634 vector unsigned long long vec_popcnt (vector signed long long)
20635 vector unsigned long long vec_popcnt (vector unsigned long long)
20636 vector signed long long vec_slo (vector signed long long,
20637 vector signed char)
20638 vector signed long long vec_slo (vector signed long long,
20639 vector unsigned char)
20640 vector unsigned long long vec_slo (vector unsigned long long,
20641 vector signed char)
20642 vector unsigned long long vec_slo (vector unsigned long long,
20643 vector unsigned char)
20644 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
20645 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
20646 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
20647 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
20648 * doc/extend.texi: Update the built-in documentation file for the
20649 new built-in functions.
20651 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
20653 * attribs.h (sorted_attr_string): Move machine independent
20654 functions for target clone support from the i386 port to common
20655 code. Rename ix86_function_versions to common_function_versions.
20656 Rename make_name to make_unique_name.
20657 (common_function_versions): Likewise.
20658 (make_unique_name): Likewise.
20659 (make_dispatcher_decl): Likewise.
20660 (is_function_default_version): Likewise.
20661 * attribs.c (attr_strcmp): Likewise.
20662 (sorted_attr_string): Likewise.
20663 (common_function_versions): Likewise.
20664 (make_unique_name): Likewise.
20665 (make_dispatcher_decl): Likewise.
20666 (is_function_default_version): Likewise.
20667 * config/i386/i386.c (attr_strcmp): Likewise.
20668 (sorted_attr_string): Likewise.
20669 (ix86_function_versions): Likewise.
20670 (make_name): Likewise.
20671 (make_dispatcher_decl): Likewise.
20672 (is_function_default_version): Likewise.
20673 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
20675 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20678 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20679 Account for direct move costs for vec_construct of integer
20682 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
20685 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
20686 (UNSPEC_STX_ATOMIC): Ditto.
20687 (loaddi_via_sse): New insn.
20688 (storedi_via_sse): Ditto.
20689 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
20690 Update corresponding peephole2 patterns.
20691 (atomic_storedi_fpu): Ditto.
20693 2017-05-11 Julia Koval <julia.koval@intel.com>
20695 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
20696 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
20698 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
20699 (__builtin_ia32_rsqrt14ss_mask): New builtins.
20700 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
20702 2017-05-11 Nathan Sidwell <nathan@acm.org>
20704 * graphite-poly.c: Include dumpfile.h.
20706 * dumpfle.h (dump_function): Declare here ...
20707 * tree-dump.h (dump_function): ... not here.
20708 * dumpfile.c: #include tree-cfg.h.
20709 (dump_function): Move here from ...
20710 * tree-dump.c (dump_function): ... here.
20711 * gimplify.c: #include splay-tree.h, not tree-dump.h.
20712 * graphite-poly.c: Don't include tree-dump.h.
20713 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
20714 * print-tree.c: Likewise.
20715 * stor-layout.c: Likewise.
20716 * tree-nested.c: Likewise.
20718 * dumpfile.c (dump_start): Use TDF_FLAGS.
20719 (dump_enable_all): Fix TDF_KIND check thinko.
20721 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
20723 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20724 array entries to represent two legal parameterizations of the
20725 overloaded __builtin_cmpb function, as represented by the
20726 P6_OV_BUILTIN_CMPB constant.
20727 (altivec_resolve_overloaded_builtin): Add special case handling
20728 for the __builtin_cmpb function, as represented by the
20729 P6_OV_BUILTIN_CMPB constant.
20730 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
20731 (BU_P6_64BIT_2): New macro.
20732 (BU_P6_OVERLOAD_2): New macro
20733 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
20734 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
20735 (CMPB): Add overload support to represent both 32-bit and 64-bit
20736 compare-bytes function.
20737 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
20738 support for TARGET_CMPB.
20739 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
20740 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20741 documentation of the __builtin_cmpb overloaded built-in function.
20743 2017-05-11 Richard Biener <rguenther@suse.de>
20745 PR tree-optimization/80705
20746 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
20747 bases are not vectorizable.
20749 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20751 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
20752 when counting register pressure.
20754 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20756 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
20757 (iv_ca_more_deps): Renamed to ...
20758 (iv_ca_compare_deps): ... this.
20759 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
20761 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20763 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
20765 (determine_group_iv_costs): ... here.
20766 (find_inv_vars_cb): Record inv var if it's not recorded before.
20768 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20770 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
20771 (get_shiftadd_cost): Ditto.
20773 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20775 * tree-ssa-address.c: Include header file.
20776 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
20778 (add_to_parts): Refactor.
20779 (addr_to_parts): New parameter. Update use of move_hint_to_base.
20780 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
20783 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20785 PR tree-optimization/53090
20786 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
20788 (extract_cond_operands): Detect condition with IV on both sides
20789 and return COMP_IV_EXPR_2.
20790 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
20791 (rewrite_use_compare): Simplify by removing call to function
20792 extract_cond_operands.
20794 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20796 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
20797 (extract_cond_operands): Detect condition comparing against non-
20798 invariant bound and return appropriate enum value.
20799 (find_interesting_uses_cond): Update use of extract_cond_operands.
20800 Handle its return value accordingly.
20801 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
20803 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20805 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
20806 nonlinear iv_use computation in loop invariant sensitive way.
20808 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20810 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
20811 (find_iv_candidates): Call relate_compare_use_with_all_cands.
20813 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20815 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
20816 (dump_cand): Support iv_cand.inv_exprs.
20817 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
20819 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
20822 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20824 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
20826 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
20827 as local function. Include necessary header files.
20828 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
20830 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20832 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
20834 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20836 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
20837 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
20838 RSHIFT_EXPR and BIT_NOT_EXPR.
20840 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20842 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
20843 (adjust_setup_cost): New parameter supporting round up adjustment.
20844 (struct address_cost_data): Delete.
20845 (force_expr_to_var_cost): Don't bound cost with spill_cost.
20846 (split_address_cost, ptr_difference_cost): Delete.
20847 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
20848 (struct ainc_cost_data): New struct.
20849 (get_address_cost_ainc): New function.
20850 (get_address_cost, get_computation_cost): Reimplement.
20851 (determine_group_iv_cost_address): Record inv_expr for all uses of
20853 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
20854 (iv_ca_has_deps): Reimplemented to ...
20855 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
20857 (iv_ca_extend): Call iv_ca_more_deps.
20859 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20861 * tree-ssa-address.c (struct mem_address): Move to header file.
20862 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
20863 * tree-ssa-address.h (struct mem_address): Move from C file.
20864 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
20866 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20868 * tree-affine.h (aff_combination_type): New interface.
20869 (aff_combination_zero_p): Remove static.
20870 (aff_combination_const_p): New interface.
20871 (aff_combination_singleton_var_p): New interfaces.
20873 2017-05-11 Richard Biener <rguenther@suse.de>
20875 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
20876 Skip unreachable blocks and destinations.
20877 (eliminate): Move stmt removal and fixup ...
20878 (fini_eliminate): ... here. Skip inserted exprs.
20879 (pass_pre::execute): Move fini_pre after fini_eliminate.
20880 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
20881 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
20882 PRE to get rid of dead code that has invalid SSA form and
20883 split critical edges again.
20885 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20887 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20889 2017-05-11 Richard Biener <rguenther@suse.de>
20891 * passes.c (execute_function_todo): Verify loops if they are
20892 said to be up-to-date.
20893 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
20894 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
20896 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
20899 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
20900 handle calling assemble_external ourself.
20903 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
20904 modes with zero size. Enhance comment.
20906 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20908 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
20909 built-ins for vec_xl and vec_xst with short and char pointer
20912 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
20914 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
20915 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
20916 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
20917 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
20918 (_mm_maskz_min_round_ss): New intrinsics.
20919 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
20920 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
20921 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
20922 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
20923 (__builtin_ia32_minss_mask_round): New builtins.
20924 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20925 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20926 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
20928 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
20929 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
20931 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
20934 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
20936 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
20937 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
20938 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
20939 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
20940 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
20941 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
20942 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
20943 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20944 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
20945 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
20946 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
20947 (__builtin_ia32_mulss_mask_round): New builtins.
20948 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20949 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20950 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
20952 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
20953 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
20955 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
20958 2017-05-10 Julia Koval <julia.koval@intel.com>
20960 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
20961 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
20962 (_mm256_setr_m128i): New intrinsics.
20964 2017-05-10 Julia Koval <julia.koval@intel.com>
20966 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
20967 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
20968 (_mm_maskz_rcp14_ss): New intrinsics.
20969 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
20970 (__builtin_ia32_rcp14ss_mask): New builtins.
20971 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
20973 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
20975 PR tree-optimization/51513
20976 * tree-cfg.c (gimple_seq_unreachable_p): New function.
20977 (assert_unreachable_fallthru_edge_p): Use it.
20978 (group_case_labels_stmt): Likewise.
20979 * tree-cfg.h: Prototype it.
20980 * stmt.c: Include cfghooks.h and tree-cfg.h.
20981 (emit_case_dispatch_table) <gap_label>: New local variable.
20982 Use it to fill dispatch table gaps.
20983 Test for default_label before updating probabilities.
20984 (expand_case) <default_label>: Remove unneeded initialization.
20985 Test for unreachable default case statement and remove its edge.
20986 Set default_label accordingly.
20987 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
20989 2017-05-10 Carl Love <cel@us.ibm.com>
20991 * config/rs6000/rs6000-c: Add support for built-in functions
20992 vector signed char vec_neg (vector signed char)
20993 vector signed short int vec_neg (vector short int)
20994 vector signed int vec_neg (vector signed int)
20995 vector signed long long vec_neg (vector signed long long)
20996 vector float vec_neg (vector float)
20997 vector double vec_neg (vector double)
20998 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
21000 * config/rs6000/altivec.h: Add define for vec_neg
21001 * doc/extend.texi: Update the built-in documentation for the
21002 new built-in functions.
21004 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21006 PR tree-optimization/77644
21007 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
21009 2017-05-10 Nathan Sidwell <nathan@acm.org>
21011 * dumpfile.h (TDI_lang_all): New.
21012 (TDF_KIND): New. Renumber others
21013 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
21015 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
21017 (get_dump_file_name): Adjust suffix generation.
21018 (dump_enable_all): Use TDF_KIND.
21019 * doc/invoke.texi (-fdump-lang-all): Document.
21021 * dumpfile.h: Tabify.
21023 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
21026 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
21027 Move member access before delete.
21029 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
21031 * tree-inline.c (expand_call_inline): Split block at stmt
21034 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
21037 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
21038 are now unused after splitting mov{sf,sd}_hardfloat.
21039 (f32_lr2): Likewise.
21040 (f32_lm): Likewise.
21041 (f32_lm2): Likewise.
21042 (f32_li): Likewise.
21043 (f32_li2): Likewise.
21044 (f32_lv): Likewise.
21045 (f32_sr): Likewise.
21046 (f32_sr2): Likewise.
21047 (f32_sm): Likewise.
21048 (f32_sm2): Likewise.
21049 (f32_si): Likewise.
21050 (f32_si2): Likewise.
21051 (f32_sv): Likewise.
21052 (f32_dm): Likewise.
21053 (f32_vsx): Likewise.
21054 (f32_av): Likewise.
21055 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
21056 For movsf, order stores so the VSX stores occur before the GPR
21057 store which encourages the register allocator to use a traditional
21058 FPR instead of a GPR. For movsd, order the stores so that the GPR
21059 store comes before the VSX stores to allow the power6 to work.
21060 This is due to the power6 not having a 32-bit integer store
21061 instruction from a FPR.
21062 (movsf_hardfloat): Likewise.
21063 (movsd_hardfloat): Likewise.
21065 2017-05-09 Martin Sebor <msebor@redhat.com>
21067 PR translation/80280
21068 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
21071 PR translation/80280
21072 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
21073 data member added in r247778.
21074 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
21076 2017-05-09 Nathan Sidwell <nathan@acm.org>
21078 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
21080 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
21083 2017-05-09 Marek Polacek <polacek@redhat.com>
21085 * doc/invoke.texi: Fix typo.
21087 2017-05-09 Richard Biener <rguenther@suse.de>
21089 * tree-vrp.c (vrp_val_is_max): Adjust comment.
21090 (vrp_val_is_min): Likewise.
21091 (set_value_range_to_value): Likewise.
21092 (set_value_range_to_nonnegative): Likewise.
21093 (gimple_assign_nonzero_p): Likewise.
21094 (gimple_stmt_nonzero_p): Likewise.
21095 (vrp_int_const_binop): Likewise. Remove unreachable case.
21096 (adjust_range_with_scev): Adjust comments.
21097 (compare_range_with_value): Likewise.
21098 (extract_range_from_phi_node): Likewise.
21099 (test_for_singularity): Likewise.
21101 2017-05-09 Richard Biener <rguenther@suse.de>
21103 * tree-vrp.c (get_single_symbol): Add assert that we don't
21104 get overflowed constants as invariant part.
21105 (compare_values_warnv): Add comment before the TREE_NO_WARNING
21106 checks. Use wi::cmp instead of recursing for integer constants.
21107 (compare_values): Just ignore whether we assumed undefined
21108 overflow instead of failing the compare.
21109 (extract_range_for_var_from_comparison_expr): Add comment before the
21110 TREE_NO_WARNING sets.
21111 (test_for_singularity): Likewise.
21112 (extract_range_from_comparison): Do not disable optimization
21113 when we assumed undefined overflow.
21114 (extract_range_basic): Remove init of unused var.
21116 2017-05-09 Richard Biener <rguenther@suse.de>
21118 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
21119 (extract_range_from_multiplicative_op_1): Adjust.
21120 (extract_range_from_binary_expr_1): Use int_const_binop.
21122 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
21125 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
21126 rs6000_store_data_bypass_p in seven define_bypass directives and
21127 in several comments.
21128 * config/rs6000/rs6000-protos.h: Add prototype for
21129 rs6000_store_data_bypass_p function.
21130 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
21131 function implements slightly different (rs6000-specific) semantics
21132 than store_data_bypass_p, returning false rather than aborting
21133 with assertion error when arguments do not satisfy the
21134 requirements of store data bypass.
21135 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
21136 rs6000_store_data_bypass_p.
21138 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
21140 * config/xtensa/xtensa-protos.h
21141 (xtensa_initial_elimination_offset): New declaration.
21142 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21143 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
21144 macro definition, add case for FRAME_POINTER_REGNUM when
21145 FRAME_GROWS_DOWNWARD.
21146 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
21147 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
21148 xtensa_initial_elimination_offset.
21150 2017-05-08 Nathan Sidwell <nathan@acm.org>
21152 * doc/invoke.texi: Alphabetize -fdump options.
21154 2017-05-08 Martin Sebor <msebor@redhat.com>
21156 PR translation/80280
21157 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
21159 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
21161 * target.def (compute_frame_layout): New optional target hook.
21162 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
21163 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
21164 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
21166 * reload1.c (verify_initial_elim_offsets): Likewise.
21167 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
21168 (use_simple_return_p): Call arm_compute_frame_layout if needed.
21169 (arm_get_frame_offsets): Split up into this ...
21170 (arm_compute_frame_layout): ... and this function.
21172 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
21174 * config/aarch64/constraints.md (Usa): New constraint.
21175 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
21177 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
21179 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
21180 with_multilib_list after it has been checked.
21182 2017-05-08 Richard Biener <rguenther@suse.de>
21184 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
21185 (bitmap_set_subtract_values): Likewise.
21187 2017-05-08 Richard Biener <rguenther@suse.de>
21189 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
21190 (gimple_assign_nonzero): ... this and remove strict_overflow_p
21192 (gimple_stmt_nonzero_warnv_p): Rename to ...
21193 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
21195 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
21196 (extract_range_basic): Adjust, do not disable propagation on
21197 strict overflow sensitive simplification.
21198 (vrp_visit_cond_stmt): Likewise.
21200 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
21202 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
21203 body size unconditionally.
21205 2017-05-07 Jeff Law <law@redhat.com>
21208 2017-05-06 Jeff Law <law@redhat.com>
21209 PR tree-optimization/78496
21210 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
21213 PR tree-optimization/78496
21214 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
21215 (simplify_stmt_using_ranges): Call it.
21216 (vrp_dom_walker::before_dom_children): Extract equivalences
21217 from an ASSERT_EXPR with an equality comparison against a
21220 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 Richard Sandiford <richard.sandiford@linaro.org>
21235 * lra-constraints.c (lra_copy_reg_equiv): New function.
21236 (split_reg): Use it to copy equivalence information from the
21237 original register to the spill register.
21239 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
21241 PR rtl-optimization/75964
21242 * simplify-rtx.c (simplify_const_relational_operation): Remove
21243 invalid handling of comparisons of integer ABS.
21245 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
21247 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
21248 initialize to zero.
21249 (init_regs): Remove declaration.
21250 (function_arg_advance_32): Initialize error_p as boolean variable.
21252 2017-05-05 Nathan Sidwell <nathan@acm.org>
21254 * store-motion.c (remove_reachable_equiv_notes): Reformat long
21255 lines. Use for (;;).
21257 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21259 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
21260 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
21261 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
21262 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
21263 VF=2 that require versioning.
21265 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21267 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
21270 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21272 * diagnostic.h (diagnostic_override_option_index): Convert from
21273 macro to inline function.
21275 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21277 * diagnostic.c (last_module_changed_p): New function.
21278 (set_last_module): New function.
21279 (diagnostic_report_current_module): Convert macro usage to
21280 the above functions.
21281 * diagnostic.h (diagnostic_context::last_module): Strengthen
21282 from const line_map * to const line_map_ordinary *.
21283 (diagnostic_last_module_changed): Delete macro.
21284 (diagnostic_set_last_module): Delete macro.
21286 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21288 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
21289 with diagnostic_report_diagnostic.
21290 (diagnostic_n_impl_richloc): Likewise.
21291 * diagnostic.h (report_diagnostic): Delete macro.
21292 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
21293 with diagnostic_report_diagnostic.
21294 * substring-locations.c (format_warning_va): Likewise.
21296 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21298 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
21299 save/restor of format_spec. Move option-printing code to...
21300 (print_option_information): ...this new function, and
21301 reimplement by simply printing to the pretty_printer,
21302 rather than appending to the format string.
21304 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21306 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
21307 handling logic into...
21308 (update_effective_level_from_pragmas): ...this new function.
21310 2017-05-04 Andrew Waterman <andrew@sifive.com>
21312 * config/riscv/riscv.opt (mstrict-align): New option.
21313 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
21314 (SLOW_UNALIGNED_ACCESS): Define.
21315 (riscv_slow_unaligned_access): Declare.
21316 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
21318 (riscv_slow_unaligned_access): New variable.
21319 (rocket_tune_info): Set slow_unaligned_access to true.
21320 (optimize_size_tune_info): Set slow_unaligned_access to false.
21321 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
21322 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
21323 (riscv_option_override): Set riscv_slow_unaligned_access.
21324 * doc/invoke.texi: Add -mstrict-align to RISC-V.
21326 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
21328 * config/riscv/riscv.md: Unify indentation.
21330 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21335 * config/rs6000/rs6000.md (u code attribute): Add FIX and
21337 (extendsi<mode>2): Add support for doing sign extension via
21338 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
21339 don't have ISA 3.0 instructions.
21340 (extendsi<mode>2 splitter): Likewise.
21341 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
21342 generate the normal insns since SImode can now go in vector
21343 registers. Disallow the special UNSPECs needed for previous
21344 machines to hide SImode being used. Add new insns
21345 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
21346 (fix_trunc<mode>si2_stfiwx): Likewise.
21347 (fix_trunc<mode>si2_internal): Likewise.
21348 (fixuns_trunc<mode>si2): Likewise.
21349 (fixuns_trunc<mode>si2_stfiwx): Likewise.
21350 (fctiw<u>z_<mode>_smallint): Likewise.
21351 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
21352 of floating point to 32-bit integer from doing a direct move to
21353 the GPR registers to do a store.
21354 (fctiwz_<mode>): Break long line.
21356 2017-05-05 Bin Cheng <bin.cheng@arm.com>
21358 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
21359 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
21360 (addr_list, addr_offset_valid_p): New.
21361 (split_address_groups): Check offset validity with above function.
21362 (gt-tree-ssa-loop-ivopts.h): Include header file.
21364 2017-05-05 Nathan Sidwell <nathan@acm.org>
21366 * config.gcc (arm*-*-*): Add missing 'fi'.
21368 2017-05-05 Steve Ellcey <sellcey@cavium.com>
21370 * doc/invoke.texi (-fopt-info): Explicitly say order of options
21371 included in -fopt-info does not matter.
21372 * doc/optinfo.texi (-fopt-info): Fix description of default
21373 behavour. Explicitly say order of options included in -fopt-info
21376 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21378 * config.gcc: Allow combinations of aprofile and rmprofile values for
21379 --with-multilib-list.
21380 * config/arm/t-multilib: New file.
21381 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
21382 variables. Remove setting of ISA and floating-point ABI in
21383 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
21384 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
21385 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
21386 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
21388 * config/arm/t-rmprofile: Likewise except for the matches changes.
21389 * doc/install.texi (--with-multilib-list): Document the combination of
21390 aprofile and rmprofile values and warn about pitfalls in doing that.
21392 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
21394 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
21395 (movdi_aarch64): Likewise.
21397 2017-05-05 Jakub Jelinek <jakub@redhat.com>
21399 PR tree-optimization/80632
21400 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
21402 (build_arrays): Initialize it for virtual phis.
21403 (fix_phi_nodes): Use it for virtual phis.
21405 PR tree-optimization/80558
21406 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
21407 [x, y] op z into [x op, y op z] for op & or | if conditions
21410 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
21411 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21414 * config/arm/arm.md (use_literal_pool): Remove.
21415 (64-bit immediate split): No longer takes cost into consideration
21416 if arm_disable_literal_pool is enabled.
21417 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
21418 used when arm_disable_literal_pool is enabled.
21419 (arm_max_const_double_inline_cost): Remove use of
21420 arm_disable_literal_pool.
21421 (push_minipool_fix): Add assert.
21422 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
21423 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
21424 (no_literal_pool_sf_immediate): New.
21426 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21428 PR tree-optimization/80613
21429 * tree-ssa-dce.c (propagate_necessity): Remove cases for
21430 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
21432 2017-05-05 Richard Biener <rguenther@suse.de>
21434 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
21436 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
21438 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
21439 of this flag from insn conditions due to removal from r247495.
21441 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
21443 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
21445 (arm_early_store_addr_dep_ptr): Likewise.
21446 * config/arm/aarch-common-protos.h
21447 (arm_early_load_addr_dep_ptr): Add prototype.
21448 (arm_early_store_addr_dep_ptr): Likewise.
21449 * config/arm/cortex-a53.md: Add new bypasses.
21451 2017-05-05 Jakub Jelinek <jakub@redhat.com>
21453 * tree.c (next_type_uid): Change type to unsigned.
21454 (type_hash_canon): Decrement back next_type_uid if
21455 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
21456 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
21459 2017-05-04 Martin Sebor <msebor@redhat.com>
21461 * builtins.c: Fix a trivial typo in a comment.
21463 PR middle-end/79234
21464 * builtins.c (check_sizes): Adjust to handle reading past the end.
21465 Avoid printing excessive upper bound of ranges. Use %E to print
21466 tree nodes instead of converting them to %wu.
21467 (expand_builtin_memchr): New function.
21468 (compute_dest_size): Rename...
21469 (compute_objsize): ...to this.
21470 (expand_builtin_memcpy): Adjust.
21471 (expand_builtin_mempcpy): Adjust.
21472 (expand_builtin_strcat): Adjust.
21473 (expand_builtin_strcpy): Adjust.
21474 (check_strncat_sizes): Adjust.
21475 (expand_builtin_strncat): Adjust.
21476 (expand_builtin_strncpy): Adjust and simplify.
21477 (expand_builtin_memset): Adjust.
21478 (expand_builtin_bzero): Adjust.
21479 (expand_builtin_memcmp): Adjust.
21480 (expand_builtin): Handle memcmp.
21481 (maybe_emit_chk_warning): Check strncat just once.
21483 2017-05-04 Martin Sebor <msebor@redhat.com>
21485 PR preprocessor/79214
21486 PR middle-end/79222
21487 PR middle-end/79223
21488 * builtins.c (check_sizes): Add inlining context and issue
21489 warnings even when -Wno-system-headers is set.
21490 (check_strncat_sizes): Same.
21491 (expand_builtin_strncat): Same.
21492 (expand_builtin_memmove): New function.
21493 (expand_builtin_stpncpy): Same.
21494 (expand_builtin): Handle memmove and stpncpy.
21496 2017-05-04 Bin Cheng <bin.cheng@arm.com>
21498 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
21499 which is not used any more.
21501 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21503 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
21505 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21507 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
21508 (cortexa53_tunings): Likewise.
21509 (cortexa57_tunings): Likewise.
21510 (cortexa72_tunings): Likewise.
21511 (cortexa73_tunings): Likewise.
21513 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21515 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
21516 Set loop alignment to 8.
21518 2017-05-04 Martin Sebor <msebor@redhat.com>
21520 PR translation/80280
21521 * builtins.c (expand_builtin_object_size): Add missing quoting to
21522 %D and like directives.
21523 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
21524 (hsa_type_for_tree_type): Same.
21525 (verify_function_arguments): Same.
21526 * symtab.c (symbol_table::change_decl_assembler_name): Same.
21527 * varasm.c (get_section): Same.
21530 2017-05-04 Martin Sebor <msebor@redhat.com>
21532 PR translation/80280
21533 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
21535 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21537 * config/aarch64/aarch64.c (generic_addrcost_table):
21538 Change HI/TI mode setting.
21540 2017-05-04 Martin Jambor <mjambor@suse.cz>
21542 PR tree-optimization/80622
21543 * tree-sra.c (comes_initialized_p): New function.
21544 (build_accesses_from_assign): Only set write lazily when
21545 comes_initialized_p is false.
21546 (analyze_access_subtree): Use comes_initialized_p.
21547 (propagate_subaccesses_across_link): Assert !comes_initialized_p
21548 instead of testing for PARM_DECL.
21550 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21552 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
21553 constraint on operand 0 to allow more general addressing modes.
21554 Adjust output template.
21555 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
21557 * config/aarch64/aarch64-protos.h
21558 (aarch64_address_valid_for_prefetch_p): Declare prototype.
21559 * config/aarch64/constraints.md (Dp): New address constraint.
21560 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
21563 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
21565 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
21566 update use of estimate_ipcp_clone_size_and_time.
21567 (estimate_local_effects): Update use of
21568 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
21569 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
21570 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
21571 Return nonspecialized time.
21573 2017-05-04 Richard Biener <rguenther@suse.de>
21575 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
21576 for the last VUSE which def dominates the PHI. Directly call
21578 (get_continuation_for_phi_1): Remove.
21580 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
21582 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
21583 to explain the use of truncating division. Cap the number of
21584 iterations to the maximum given by nb_iterations_upper_bound,
21587 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21589 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
21590 * configure: Regenerate.
21591 * config.in: Regenerate.
21592 * config/i386/driver-mingw32.c: new file.
21593 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
21594 * config.host: Link driver-mingw32.o on MinGW host.
21595 * doc/install.texi: Document new --enable-mingw-wildcard configure
21598 2017-05-04 Marek Polacek <polacek@redhat.com>
21600 PR tree-optimization/80612
21601 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
21603 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21604 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
21606 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
21607 (movt splitter): Likewise.
21608 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
21609 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
21610 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
21611 block for Thumb-1 with MOVT.
21612 (thumb2_legitimate_address_p): Move code block ...
21613 (can_avoid_literal_pool_for_label_p): ... into this new function.
21614 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
21616 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
21617 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
21618 "M-profile targets with the MOVT instruction".
21620 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21622 * config/arm/arm-builtins.c (arm_init_builtins): Rename
21623 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
21624 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
21626 2017-05-04 Martin Liska <mliska@suse.cz>
21628 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
21629 variable cond_code.
21631 2017-05-04 Richard Biener <rguenther@suse.de>
21633 * tree.c (array_at_struct_end_p): Handle arrays at struct
21634 end with flexarrays more conservatively. Refactor and treat
21635 arrays of arrays or aggregates more strict. Fix
21636 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
21637 * tree.c (array_at_struct_end_p): Adjust prototype.
21638 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
21639 * gimple-fold.c (get_range_strlen): Likewise.
21640 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
21642 2017-05-04 Richard Biener <rguenther@suse.de>
21644 PR tree-optimization/31130
21645 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
21647 (supports_overflow_infinity): Likewise.
21648 (is_negative_overflow_infinity): Likewise.
21649 (is_positive_overflow_infinity): Likewise.
21650 (is_overflow_infinity): Likewise.
21651 (stmt_overflow_infinity): Likewise.
21652 (overflow_infinity_range_p): Likewise.
21653 (usable_range_p): Remove as always returning true.
21654 (make_overflow_infinity): Remove.
21655 (negative_overflow_infinity): Likewise.
21656 (positive_overflow_infinity): Likewise.
21657 (avoid_overflow_infinity): Likewise.
21658 (set_value_range): Adjust accordingly.
21659 (set_value_range_to_nonnegative): Likewise, remove now unused
21660 overflow_infinity arg.
21661 (vrp_operand_equal_p): Adjust.
21662 (update_value_range): Likewise.
21663 (range_int_cst_singleton_p): Likewise.
21664 (operand_less_p): Likewise.
21665 (compare_values_warnv): Likewise.
21666 (extract_range_for_var_from_comparison_expr): Likewise.
21667 (vrp_int_const_binop): Likewise.
21668 (zero_nonzero_bits_from_vr): Likewise.
21669 (extract_range_from_multiplicative_op_1): Likewise.
21670 (extract_range_from_binary_expr_1): Likewise.
21671 (extract_range_from_unary_expr): Likewise.
21672 (extract_range_from_comparison): Likewise.
21673 (extract_range_basic): Likewise.
21674 (adjust_range_with_scev): Likewise.
21675 (compare_ranges): Likewise.
21676 (compare_range_with_value): Likewise.
21677 (dump_value_range): Likewise.
21678 (test_for_singularity): Likewise, remove strict_overflow_p parameter
21680 (simplify_cond_using_ranges): Adjust.
21682 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
21684 * brig-builtins.def: Added a builtin for class_f64.
21685 * builtin-types.def: Added a builtin type needed by class_f64.
21687 2017-05-03 Jason Merrill <jason@redhat.com>
21689 * timevar.def: Add TV_CONSTEXPR.
21691 2017-05-03 David Malcolm <dmalcolm@redhat.com>
21693 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
21695 2017-05-03 Martin Jambor <mjambor@suse.cz>
21697 * ipa-prop.c (ipa_update_after_lto_read): Removed.
21698 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
21699 * ipa-cp.c (ipcp_propagate_stage): Do not call
21700 ipa_update_after_lto_read.
21701 * ipa-inline.c (ipa_inline): Likewise.
21703 2017-05-03 Martin Jambor <mjambor@suse.cz>
21705 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
21706 tag. Added a default constructor and a destructor.
21707 (ipa_edge_args_sum_t): New class;
21708 (ipa_edge_args_sum): Declare.
21709 (ipa_edge_args_vector): Remove declaration.
21710 (IPA_EDGE_REF): Use ipa_edge_args_sum.
21711 (ipa_free_edge_args_substructures): Remove declaration.
21712 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
21713 (ipa_edge_args_info_available_for_edge_p): Likewise.
21714 * ipa-prop.c (ipa_edge_args_vector): Removed.
21715 (edge_removal_hook_holder): Likewise.
21716 (edge_duplication_hook_holder): Likewise.
21717 (ipa_edge_args_sum): New variable.
21718 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
21719 ipa_edge_args_vector.
21720 (ipa_free_edge_args_substructures): Likewise.
21721 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
21722 ipa_edge_args_vector.
21723 (ipa_edge_removal_hook): Turned into method
21724 ipa_edge_args_sum_t::remove.
21725 (ipa_edge_duplication_hook): Turned into method
21726 ipa_edge_args_sum_t::duplicate.
21727 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
21728 registering edge hooks.
21729 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
21730 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
21731 ipa_edge_args_sum instead of ipa_edge_args_vector.
21732 * ipa-profile.c (ipa_profile): Likewise.
21734 2017-05-03 Martin Jambor <mjambor@suse.cz>
21736 * symbol-summary.h (function_summary): New method exists.
21737 (function_summary::symtab_removal): Deallocate through release.
21738 (call_summary): New class.
21739 (gt_ggc_mx): New overload.
21740 (gt_pch_nx): Likewise.
21741 (gt_pch_nx): Likewise.
21743 2017-05-03 Jeff Law <law@redhat.com>
21745 PR tree-optimization/78496
21746 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
21747 from simplify_cond_using_ranges. Split off code to walk
21748 backwards through casts into ...
21749 (simplify_cond_using_ranges_2): New function.
21750 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
21751 (execute_vrp): After identifying jump threads, call
21752 simplify_cond_using_ranges_2.
21754 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
21757 * ipa-inline.h (inline_summary): Add ctor.
21758 (create_ggc): Do not use ggc_cleared_alloc.
21760 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
21761 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21763 * gcc.c (handle_braces): Support escaping in switch matching
21765 * doc/invoke.texi (Spec Files): Document it.
21766 Remove superfluous @code markup in items.
21768 2017-05-03 David Malcolm <dmalcolm@redhat.com>
21770 * diagnostic-show-locus.c (struct column_range): New struct.
21771 (get_affected_columns): New function.
21772 (get_printed_columns): New function.
21773 (struct correction): New struct.
21774 (correction::ensure_capacity): New function.
21775 (correction::ensure_terminated): New function.
21776 (struct line_corrections): New struct.
21777 (line_corrections::~line_corrections): New dtor.
21778 (line_corrections::add_hint): New function.
21779 (layout::print_trailing_fixits): Reimplement in terms of the new
21781 (selftest::test_overlapped_fixit_printing): New function.
21782 (selftest::diagnostic_show_locus_c_tests): Call it.
21784 2017-05-03 Nathan Sidwell <nathan@acm.org>
21786 Canonicalize canonical type hashing
21787 * tree.h (type_hash_canon_hash): Declare.
21788 * tree.c (type_hash_list, attribute_hash_list): Move into
21789 type_hash_canon_hash.
21790 (build_type_attribute_qual_variant): Break out hash code calc into
21791 type_hash_canon_hash.
21792 (type_hash_canon_hash): New. Generic type hash computation.
21793 (build_range_type_1, build_array_type_1, build_function_type,
21794 build_method_type_directly, build_offset_type, build_complex_type,
21795 make_vector_type): Call it.
21797 2017-05-03 Richard Biener <rguenther@suse.de>
21799 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
21800 When all DRs have unknown misaligned do not always peel
21801 when there is a store but apply the same costing model as if
21802 there were only loads.
21804 2017-05-03 Richard Biener <rguenther@suse.de>
21807 PR tree-optimization/80492
21808 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
21809 compare_base_decls returning dont-know properly.
21811 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
21813 * config/arm/iterators.md (CCSI): New mode iterator.
21814 (arch): New mode attribute.
21815 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
21816 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
21817 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
21818 code iterator for success result mode.
21819 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
21820 the corresponding new insn generators.
21822 2017-05-03 Bin Cheng <bin.cheng@arm.com>
21825 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21826 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21828 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
21830 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
21831 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
21832 (DDR_A): Wrap DDR argument in brackets.
21833 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
21834 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
21835 (DDR_REVERSED_P): Likewise.
21837 2017-05-03 Jakub Jelinek <jakub@redhat.com>
21839 PR tree-optimization/79472
21840 * tree-switch-conversion.c (struct switch_conv_info): Add
21841 contiguous_range and default_case_nonstandard fields.
21842 (collect_switch_conv_info): Compute contiguous_range and
21843 default_case_nonstandard fields, don't clear final_bb if
21844 contiguous_range and only the default case doesn't have the required
21846 (check_all_empty_except_final): Set default_case_nonstandard instead
21847 of failing if contiguous_range and the default case doesn't have empty
21849 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
21850 and only the default case doesn't have the required constants. Skip
21852 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
21853 if default_case_nonstandard.
21854 (build_constructors): Build constant 1 just once. Assert that default
21855 values aren't inserted in between cases if contiguous_range. Skip
21857 (build_arrays): Skip virtual phis.
21858 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
21859 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
21860 Handle virtual phis.
21861 (gen_inbound_check): Handle default_case_nonstandard case.
21862 (process_switch): Adjust check_final_bb caller. Call
21863 gather_default_values with the first non-default case instead of
21864 default case if default_case_nonstandard.
21866 2017-05-02 Nathan Sidwell <nathan@acm.org>
21868 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
21869 check. Fix formatting.
21871 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
21873 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
21874 errors when comparing specialized and unspecialized times.
21876 2017-05-02 David Malcolm <dmalcolm@redhat.com>
21878 * diagnostic-show-locus.c
21879 (layout::should_print_annotation_line_p): Make private.
21880 (layout::print_annotation_line): Make private.
21881 (layout::annotation_line_showed_range_p): Make private.
21882 (layout::show_ruler): Make private.
21883 (layout::print_source_line): Make private. Pass in line and
21884 line_width, rather than calling location_get_source_line. Drop
21886 (layout::print_leading_fixits): New method.
21887 (layout::print_any_fixits): Rename to...
21888 (layout::print_trailing_fixits): ...this, and make private.
21889 Don't print newline fixits.
21890 (diagnostic_show_locus): Move logic for printing one row into...
21891 (layout::print_line): ...this new function. Move the
21892 location_get_source_line call and error-handling from
21893 print_source_line to here. Call print_leading_fixits, and rename
21894 print_any_fixits to print_trailing_fixits.
21895 (selftest::test_fixit_insert_containing_newline): Update now that
21896 newlines are partially supported.
21897 (selftest::test_fixit_insert_containing_newline_2): New test.
21898 (selftest::test_fixit_replace_containing_newline): Update comments.
21899 (selftest::diagnostic_show_locus_c_tests): Call the new test.
21900 * edit-context.c (class added_line): New class.
21901 (class edited_line): Describe newline handling in comment.
21902 (edited_line::actually_edited_p): New method.
21903 (edited_line::print_content): Delete redundant decl.
21904 (edited_line::m_predecessors): New field.
21905 (edited_file::print_content): Call edited_line::print_content.
21906 (edited_file::print_diff): Update to support newlines.
21907 (edited_file::print_diff_hunk): Likewise.
21908 (edited_file::print_run_of_changed_lines): New function.
21909 (edited_file::print_diff_line): Convert to...
21910 (print_diff_line): ...this.
21911 (edited_file::get_effective_line_count): New function.
21912 (edited_line::edited_line): Initialize new field m_predecessors.
21913 (edited_line::~edited_line): Clean up m_predecessors.
21914 (edited_line::apply_fixit): Handle newlines.
21915 (edited_line::get_effective_line_count): New function.
21916 (edited_line::print_content): New function.
21917 (edited_line::print_diff_lines): New function.
21918 (selftest::test_applying_fixits_insert_containing_newline): New
21920 (selftest::test_applying_fixits_replace_containing_newline): New
21922 (selftest::insert_line): New function.
21923 (selftest::test_applying_fixits_multiple_lines): Add example of
21925 (selftest::edit_context_c_tests): Call the new tests.
21927 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21929 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
21930 parameter cand. Update dump information.
21931 (get_computation_cost): Update uses.
21933 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21935 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
21936 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
21937 (get_computation_at, rewrite_use_address): Update use of
21938 get_computation_aff.
21940 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21942 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
21943 (get_computation): Delete.
21944 (get_computation_cost): Implement like get_computation_cost_at.
21945 Use get_computation_at.
21946 (get_computation_cost_at): Delete.
21947 (rewrite_use_nonlinear_expr): Use get_computation_at.
21948 (rewrite_use_compare, remove_unused_ivs): Ditto.
21950 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21952 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
21954 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21956 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
21957 (ivopts_global_cost_for_size): Rename parameter and update uses.
21958 (iv_ca_recount_cost): Update uses.
21959 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
21960 candidates seperately in n_invs and n_cands.
21961 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
21963 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21965 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
21966 (find_inv_vars_cb): New.
21967 (find_depends): Renamed to ...
21968 (find_inv_vars): ... this.
21969 (add_candidate_1, force_var_cost): Call find_inv_vars.
21970 (split_address_cost, determine_group_iv_cost_cond): Ditto.
21972 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21974 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
21975 inv_vars. Add inv_exprs.
21976 (struct iv_cand): Rename depends_on to inv_vars.
21977 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
21978 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
21979 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
21980 (dump_cand): Dump inv_vars.
21981 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
21982 (record_invariant, find_depends, add_candidate_1): Ditto.
21983 (set_group_iv_cost, force_var_cost): Ditto.
21984 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
21985 (get_computation_cost_at, get_computation_cost): Ditto.
21986 (determine_group_iv_cost_generic): Ditto.
21987 (determine_group_iv_cost_address): Ditto.
21988 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
21989 (determine_group_iv_costs): Ditto.
21990 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
21991 (iv_ca_set_remove_invariants): Renamed to ...
21992 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
21993 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
21994 (iv_ca_set_add_invariants): Renamed to ...
21995 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
21996 (iv_ca_set_cp): Use iv_ca_set_add_invs.
21997 (iv_ca_has_deps): Support inv_vars and inv_exprs.
21998 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
21999 (create_new_ivs): Remove useless dump.
22001 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22003 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
22005 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
22006 (iv_ca_set_no_cp, create_new_iv): Ditto.
22008 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22010 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22012 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
22014 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
22015 function tree_check2.
22017 2017-05-02 Martin Liska <mliska@suse.cz>
22019 * doc/gcov.texi: Add missing preposition.
22020 * gcov.c (function_info::function_info): Properly fill up
22021 all member variables.
22023 2017-05-02 Tamar Christina <tamar.christina@arm.com>
22025 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
22027 2017-05-02 Tamar Christina <tamar.christina@arm.com>
22029 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
22031 2017-05-02 Martin Liska <mliska@suse.cz>
22034 * lto-streamer-in.c (lto_read_tree_1): Remove
22035 LTO_STREAMER_DEBUG.
22036 * lto-streamer.c (struct tree_hash_entry): Likewise.
22037 (struct tree_entry_hasher): Likewise.
22038 (tree_entry_hasher::hash): Likewise.
22039 (tree_entry_hasher::equal): Likewise.
22040 (lto_streamer_init): Likewise.
22041 (lto_orig_address_map): Likewise.
22042 (lto_orig_address_get): Likewise.
22043 (lto_orig_address_remove): Likewise.
22044 * lto-streamer.h: Likewise.
22045 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
22046 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
22048 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
22050 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
22051 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
22052 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
22053 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
22054 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
22055 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
22056 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
22057 (mm_maskz_sub_ss): New intrinsics.
22058 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22059 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
22060 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
22061 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
22062 (__builtin_ia32_subss_mask_round): New builtins.
22063 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22064 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
22065 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
22067 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
22068 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
22070 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
22073 2017-05-02 Martin Jambor <mjambor@suse.cz>
22075 PR tree-optimization/78687
22076 * tree-sra.c (access): New field parent.
22077 (process_subtree_disqualification): New function.
22078 (disqualify_candidate): Call it.
22079 (build_accesses_from_assign): Reset write flag if creating an
22081 (build_access_subtree): Fill in parent field and also prpagate
22082 down grp_write flag.
22083 (create_artificial_child_access): New parameter set_grp_write, set
22084 grp_write to its value.
22085 (propagate_subaccesses_across_link): Also propagate grp_write flag
22087 (propagate_all_subaccesses): Push the closest parent back to work
22088 queue if add_access_to_work_queue returned true.
22090 2017-05-02 Richard Biener <rguenther@suse.de>
22092 * common.opt (fstrict-overflow): Alias negative to fwrapv.
22093 * doc/invoke.texi (fstrict-overflow): Remove all traces of
22094 -fstrict-overflow documentation.
22095 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
22096 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
22097 flag_strict_overflow.
22098 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
22099 * lto-opts.c (lto_write_options): Do not stream it.
22100 * lto-wrapper.c (merge_and_complain): Do not handle it.
22101 * opts.c (default_options_table): Do not set -fstrict-overflow.
22102 (finish_options): Likewise do not clear it when sanitizing.
22103 * simplify-rtx.c (simplify_const_relational_operation): Do not
22104 test flag_strict_overflow.
22106 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
22108 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
22109 using enabled attribute.
22110 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
22111 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
22112 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
22113 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
22114 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
22115 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
22116 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
22117 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
22118 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
22119 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
22121 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
22123 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
22125 2017-05-02 Richard Biener <rguenther@suse.de>
22127 PR tree-optimization/80591
22129 2017-04-10 Richard Biener <rguenther@suse.de>
22131 * tree-ssa-structalias.c (find_func_aliases): Properly handle
22134 2017-05-02 Richard Biener <rguenther@suse.de>
22136 PR tree-optimization/80549
22137 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
22138 (cleanup_tree_cfg_noloop): Create forwarders to known loop
22139 headers if they do not have a preheader.
22141 2017-05-02 Martin Liska <mliska@suse.cz>
22144 * common.opt: Fix typo.
22145 * doc/invoke.texi: Likewise.
22147 2017-05-01 Jan Beulich <jbeulich@suse.com>
22149 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
22150 swapping, add (x,x,m,x,n) alternative.
22152 2017-05-01 Nathan Sidwell <nathan@acm.org>
22154 * calls.c (combine_pending_stack_adjustment_and_call): Remove
22155 unnecessary unadjusted_alignment check.
22157 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
22160 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
22162 * gimplify.c (gimplify_cilk_detach): New function.
22163 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
22164 * tree-core.h: Document EXPR_CILK_SPAWN.
22165 * tree.h (EXPR_CILK_SPAWN): Define.
22167 2017-05-01 David Malcolm <dmalcolm@redhat.com>
22169 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
22170 to use new fixit_hint representation, using the "replace" logic.
22171 (get_line_span_for_fixit_hint): Likewise.
22172 (layout::print_any_fixits): Likewise.
22173 (selftest::test_one_liner_many_fixits): Rename to...
22174 (selftest::test_one_liner_many_fixits_1): ...this, and update
22175 comment and expected output to reflect that the multiple fix-it
22176 hints are now consolidated into one insertion.
22177 (selftest::test_one_liner_many_fixits_2): New test.
22178 (selftest::test_diagnostic_show_locus_one_liner): Update for
22180 (selftest::test_fixit_consolidation): Update for fix-it API
22182 * diagnostic.c (print_parseable_fixits): Likewise.
22183 * edit-context.c (edited_line::m_line_events): Convert from
22184 auto_vec <line_event *> to auto_vec <line_event>.
22185 (class line_event): Convert from abstract base class to a concrete
22186 class, taking over the role of replace_event.
22187 (class insert_event): Delete.
22188 (class replace_event): Rename to class line_event. Convert to
22190 (edit_context::add_fixits): Reimplement.
22191 (edit_context::apply_insert): Delete.
22192 (edit_context::apply_replace): Rename to...
22193 (edit_context::apply_fixit): ...this. Convert to half-open range.
22194 (edited_file::apply_insert): Delete.
22195 (edited_file::apply_replace): Rename to...
22196 (edited_file::apply_fixit): ...this.
22197 (edited_line::~edited_line): Drop deletion of events.
22198 (edited_line::apply_insert): Delete.
22199 (edited_line::apply_replace): Rename to...
22200 (edited_line::apply_fixit): ...this. Convert to half-open range.
22201 Update for change to type of m_line_events.
22202 * edit-context.h (edit_context::apply_insert): Delete.
22203 (edit_context::apply_replace): Rename to...
22204 (edit_context::apply_fixit): ...this.
22206 2017-05-01 Martin Sebor <msebor@redhat.com>
22208 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
22211 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
22214 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
22215 __get_cpuid_max returns 0.
22216 (__get_cpuid_count): Ditto.
22218 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
22220 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
22221 replacement expression is another instance of one of its arguments.
22223 2017-05-01 Jakub Jelinek <jakub@redhat.com>
22226 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
22227 check for stack push/pop autoinc.
22228 * config/i386/i386.c (ix86_agi_dependent): Return false
22229 if the only reason why modified_in_p returned true is that
22230 addr is SP based and set_insn is a push or pop.
22232 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
22234 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
22237 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
22240 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
22241 (account_size_time): Use two predicates - exec_pred and
22243 (evaluate_conditions_for_known_args): Compute both clause and
22245 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
22246 (inline_summary_t::duplicate): Update.
22247 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
22249 (compute_inline_parameters): Likewise.
22250 (estimate_edge_size_and_time): Update caluclation of time.
22251 (estimate_node_size_and_time): Compute both time and nonspecialized
22253 (estimate_ipcp_clone_size_and_time): Update.
22254 (inline_merge_summary): Update.
22255 (do_estimate_edge_time): Update.
22256 (do_estimate_edge_size): Update.
22257 (do_estimate_edge_hints): Update.
22258 (inline_read_section, inline_write_summary): Stream both new predicates.
22259 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
22261 (compute_inlined_call_time): Cleanup.
22262 (big_speedup_p): Update.
22263 (edge_badness): Update.
22264 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
22265 (size_time_entry): Replace predicate by exec_predicate and
22266 nonconst_predicate.
22267 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
22268 (estimate_edge_time): Return also nonspec_time.
22269 (reset_edge_growth_cache): Update.
22271 2017-04-29 Jakub Jelinek <jakub@redhat.com>
22273 PR rtl-optimization/80491
22274 * ifcvt.c (noce_process_if_block): When looking for x setter
22275 with missing else_bb, don't check only the insn right before
22276 cond_earliest, but look for the last insn that x is modified in
22277 within the same bb.
22279 PR rtl-optimization/80491
22280 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
22282 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
22284 PR tree-optimization/80487
22285 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
22287 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22289 PR tree-optimization/79697
22290 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
22291 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
22292 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
22294 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
22295 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
22297 2017-04-28 Martin Sebor <msebor@redhat.com>
22299 PR tree-optimization/80523
22300 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
22301 (init_target_to_host_charmap, target_to_host, target_strtol10): New
22303 (maybe_warn, format_directive, parse_directive): Use new functions.
22304 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
22306 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
22308 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
22310 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
22312 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
22313 target_header_dir): Set correctly.
22314 * configure: Regenerated.
22315 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22316 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22317 instead of SYSTEM_HEADER_DIR.
22319 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
22321 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
22322 (estimate_local_effects): Likewise.
22323 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
22324 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
22325 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
22326 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
22327 do_estimate_edge_time, estimate_edge_time): Likewise.
22328 * ipa-inline-analysis.c (estimate_node_size_and_time,
22329 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
22330 (estimate_time_after_inlining): Remove.
22332 2017-04-28 Martin Liska <mliska@suse.cz>
22334 * doc/gcov.texi: Enhance documentation of gcov.
22336 2017-04-28 Martin Liska <mliska@suse.cz>
22338 * doc/gcov.texi: Sort options in alphabetic order.
22339 * doc/gcov-dump.texi: Likewise.
22340 * doc/gcov-tool.texi: Likewise.
22341 * gcov.c (print_usage): Likewise.
22342 * gcov-dump.c (print_usage): Likewise.
22343 * gcov-tool.c (print_merge_usage_message): Likewise.
22344 (print_rewrite_usage_message): Likewise.
22345 (print_overlap_usage_message): Likewise.
22347 2017-04-28 Martin Liska <mliska@suse.cz>
22349 PR gcov-profile/53915
22350 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
22352 2017-04-28 Martin Liska <mliska@suse.cz>
22354 PR gcov-profile/79891
22355 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
22356 is marked by compiler as living on a line.
22357 (get_cycles_count): Remove usage of the union.
22358 (output_intermediate_file): Likewise.
22359 (find_source): Fix GNU coding style.
22360 (accumulate_line_counts): Remove old non-all block mode.
22361 (output_lines): Remove usage of the union.
22362 * profile.c (output_location): Include all BBs, even if
22363 belonging to a same line (and file) as a previous BB.
22365 2017-04-28 Martin Liska <mliska@suse.cz>
22367 * gcov.c (process_args): Handle new argument 'w'.
22368 (read_graph_file): Assign ID to BBs.
22369 (output_branch_count): Display BB # if verbose flag is set.
22370 (output_lines): Likewise for arcs.
22371 (print_usage): Add '--verbose' option help.
22372 * doc/gcov.texi: Document --verbose (-w) option.
22374 2017-04-28 Martin Liska <mliska@suse.cz>
22376 * gcov.c (struct block_location_info): New struct.
22377 (process_file): Fill up the new structure.
22378 (read_graph_file): Replace usage of encoding by the newly added
22380 (add_line_counts): Likewise.
22381 (accumulate_line_counts): Remove usage of the union.
22382 (function_info::function_info): New function.
22383 (function_info::~function_info): Likewise.
22384 (process_file): Call delete instead of release_function.
22385 (release_function): Release the function.
22386 (release_structures): Call delete instead of release_function.
22387 (solve_flow_graph): Replace usage of num_blocks.
22388 (find_exception_blocks): Likewise.
22389 (output_lines): Fix GNU coding style.
22391 2017-04-28 Martin Liska <mliska@suse.cz>
22394 * coverage.c (coverage_remove_note_file): New function.
22395 * coverage.h: Declare the function.
22396 * toplev.c (finalize): Clean if an error has been seen.
22398 2017-04-28 Martin Liska <mliska@suse.cz>
22400 PR gcov-profile/80031
22401 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
22402 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
22403 * gcov.c (read_graph_file): Read just number of blocks.
22404 * profile.c (branch_prob): Do not stream 0 flags per a basic
22407 2017-04-28 Martin Liska <mliska@suse.cz>
22409 * gcov-dump.c (tag_*): Add new argument to declarations.
22410 (dump_gcov_file): Likewise.
22411 (tag_blocks): Add and use new argument depth.
22412 (tag_arcs): Likewise.
22413 (tag_lines): Likewise.
22414 (tag_counters): Likewise.
22415 (tag_summary): Likewise.
22416 (dump_working_sets): Use depth to do a proper indentation.
22418 2017-04-28 Jakub Jelinek <jakub@redhat.com>
22421 * cgraph.h (symtab_node::debug_symtab): No longer inline.
22422 * symtab.c (symtab_node::debug_symtab): Move definition here.
22424 2017-04-28 Richard Biener <rguenther@suse.de>
22426 * lto-streamer.h (LTO_major_version): Bump to 7.
22428 2017-04-28 Richard Biener <rguenther@suse.de>
22430 * tree-vrp.c (assert_info): New struct.
22431 (add_assert_info): New helper.
22432 (register_edge_assert_for_2): Refactor to add asserts to a vector
22434 (register_edge_assert_for_1): Likewise.
22435 (register_edge_assert_for): Likewise.
22436 (finish_register_edge_assert_for): New helper actually registering
22437 asserts where live on edge.
22438 (find_conditional_asserts): Adjust.
22439 (find_switch_asserts): Likewise.
22440 (evrp_dom_walker::try_find_new_range): Generalize.
22441 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
22443 2017-04-27 Marek Polacek <polacek@redhat.com>
22446 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
22447 arg10 and arg11 to itype.
22449 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
22451 * doc/extend.texi (Object Size Checking): Improve grammar.
22453 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
22456 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
22457 that the logic for permitting reciprocal estimates matches that
22460 2017-04-27 Jakub Jelinek <jakub@redhat.com>
22463 * tree.c (type_cache_hasher::equal): Only compare
22464 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
22465 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
22466 non-aggregate element types.
22467 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
22468 about the flag on ARRAY_TYPEs in the comment, formatting fix.
22470 2017-04-27 Richard Biener <rguenther@suse.de>
22472 PR middle-end/80533
22473 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
22474 stripping ARRAY_REFs from MEM_EXPR make sure we're not
22475 keeping a reference to a trailing array.
22477 2017-04-27 Richard Biener <rguenther@suse.de>
22479 PR middle-end/80539
22480 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
22481 being in loop-closed SSA form conservatively.
22482 (chrec_fold_multiply_poly_poly): Likewise.
22484 2017-04-27 Tamar Christina <tamar.christina@arm.com>
22486 PR middle-end/79665
22487 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
22488 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
22490 2017-04-27 Jakub Jelinek <jakub@redhat.com>
22493 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
22494 (aarch64_function_arg_alignment): Return unsigned int again, but still
22495 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
22496 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
22497 Don't emit -Wpsabi note.
22498 (aarch64_function_arg_boundary): Likewise.
22499 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
22502 2017-04-26 Nathan Sidwell <nathan@acm.org>
22504 * tree.h (crc32_unsigned_n): Declare.
22505 (crc32_unsigned, crc32_unsigned): Make inline.
22506 * tree.c (crc32_unsigned_bits): Replace with ...
22507 (crc32_unsigned_n): ... this.
22508 (crc32_unsigned, crc32_byte): Remove.
22509 (crc32_string): Remove unnecessary braces.
22511 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
22513 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
22514 * ipa-inline-analysis.c (MAX_TIME): Remove.
22515 (account_size_time): Use sreal for time.
22516 (dump_inline_summary): Update.
22517 (estimate_function_body_sizes): Update.
22518 (estimate_edge_size_and_time): Update.
22519 (estimate_calls_size_and_time): Update.
22520 (estimate_node_size_and_time): Update.
22521 (inline_merge_summary): Update.
22522 (inline_update_overall_summary): Update.
22523 (estimate_time_after_inlining): Update.
22524 (inline_read_section): Update.
22525 (inline_write_summary): Update.
22526 * ipa-inline.c (compute_uninlined_call_time): Update.
22527 (compute_inlined_call_time): Update.
22528 (recursive_inlining): Update.
22529 (inline_small_functions): Update.
22530 (dump_overall_stats): Update.
22531 * ipa-inline.h: Include sreal.h.
22532 (size_time_entry): Turn time to sreal.
22533 (inline_summary): Turn self_time nad time to sreal.
22535 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
22537 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
22539 (sreal::stream_out, sreal::stream_in): New.
22540 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
22542 2017-04-25 Jakub Jelinek <jakub@redhat.com>
22544 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
22547 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
22550 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
22552 (dimode_scalar_chain::compute_convert_gain): Ditto.
22553 (dimode_scalar_chain::make_vector_copies): Ditto.
22554 (dimode_scalar_chain::convert_reg): Ditto.
22555 (dimode_scalar_chain::convert_insn): Ditto.
22556 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
22557 (VI248_AVX512BW_1): New mode iterator.
22558 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
22559 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
22562 2017-04-25 Martin Sebor <msebor@redhat.com>
22564 PR tree-optimization/80497
22565 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
22566 constants are representable in HOST_WIDE_INT.
22567 (parse_directive): Ditto.
22569 2017-04-25 Martin Sebor <msebor@redhat.com>
22572 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
22573 (new_zero_array): Adjust signature.
22574 (dom_info::dom_init): Used unsigned rather that size_t.
22575 (dom_info::dom_info): Same.
22577 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22578 Jakub Jelinek <jakub@redhat.com>
22581 * config/arm/arm.c: Include gimple.h.
22582 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
22583 returns negative, increment ncrn only if it returned positive.
22584 (arm_needs_doubleword_align): Return int instead of bool,
22585 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
22586 members, but if there is any such non-FIELD_DECL
22587 > PARM_BOUNDARY aligned decl, return -1 instead of false.
22588 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
22589 returns negative, increment nregs only if it returned positive.
22590 (arm_setup_incoming_varargs): Likewise.
22591 (arm_function_arg_boundary): Emit -Wpsabi note if
22592 arm_needs_doubleword_align returns negative, return
22593 DOUBLEWORD_ALIGNMENT only if it returned positive.
22595 2017-04-25 Marek Polacek <polacek@redhat.com>
22598 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
22599 first argument to type.
22601 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
22604 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
22605 type checks to test for compatibility instead of equality.
22607 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22608 Jakub Jelinek <jakub@redhat.com>
22611 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
22613 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
22614 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
22615 the alignment computation, but return their maximum in warn_alignment.
22616 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
22617 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
22619 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
22620 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
22623 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22625 * config/arc/simdext.md (dmpyh): Fix typo.
22627 2017-04-25 Richard Biener <rguenther@suse.de>
22629 PR tree-optimization/80492
22630 * alias.c (compare_base_decls): Handle registers with asm
22631 specification conservatively.
22632 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
22633 compare_base_decls returning dont-know properly.
22635 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22637 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
22638 (legitimate_offset_address_p): New function.
22639 (arc_legitimate_address_p): Use above function.
22641 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22643 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
22645 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22647 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
22648 ACCH registers whenever they are available.
22650 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22652 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
22653 double regs fix when not used.
22655 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22657 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
22659 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
22660 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
22662 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22664 * config/arc/arc.c (arc_output_addsi): Check for h-register class
22665 when emitting short ADD instructions.
22667 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22669 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
22671 (cmpsi_cc_c_insn): Likewise.
22672 (cbranchsi4_scratch): Compute proper instruction length using
22673 compact_hreg_operand.
22674 * config/arc/predicates.md (compact_hreg_operand): New predicate.
22676 2017-04-25 Richard Biener <rguenther@suse.de>
22678 PR middle-end/80509
22679 * passes.c (pass_manager::pass_manager): Initialize
22680 m_name_to_pass_map.
22682 2017-04-25 Richard Biener <rguenther@suse.de>
22684 PR tree-optimization/79201
22685 * tree-ssa-sink.c (statement_sink_location): Handle calls.
22687 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22690 * config/s390/vector.md: Split MEM->GPR vector moves for
22691 non-s_operand addresses.
22693 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22696 * config/s390/predicates.md (reload_const_wide_int_operand): New
22698 * config/s390/s390.md ("movti"): Remove d/P alternative.
22699 ("movti_bigconst"): New pattern definition.
22701 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22704 * s390-protos.h (s390_expand_cs_hqi): Removed.
22705 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
22706 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
22707 modes as well as CCZ1mode and CCZmode.
22708 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
22709 signature of s390_emit_compare_and_swap.
22710 (s390_expand_cs_hqi): Likewise, make static.
22711 (s390_expand_cs_tdsi): Generate an explicit compare before trying
22712 compare-and-swap, in some cases.
22713 (s390_expand_cs): Wrapper function.
22714 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
22716 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
22717 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
22718 patterns for small and large integers. Forbid symref memory operands.
22719 Move expander to s390.c. Require cc register.
22720 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
22721 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
22722 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
22723 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
22724 symref memory operands. Remove CC mode and call s390_match_ccmode
22726 ("atomic_exchange<mode>"): Allow and implement all integer modes.
22728 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22730 * config/s390/s390.md (define_peephole2): New peephole to help
22731 combining the load-and-test pattern with volatile memory.
22733 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22735 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
22736 with CCZmode for TARGET_Z196.
22738 2017-04-25 Jakub Jelinek <jakub@redhat.com>
22740 PR rtl-optimization/80501
22741 * combine.c (make_compound_operation_int): Set subreg_code to SET
22742 even for AND with mask of the sign bit of mode.
22744 PR rtl-optimization/80500
22745 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
22746 sum's initial value.
22748 2017-04-25 Julian Brown <julian@codesourcery.com>
22749 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22751 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
22753 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
22755 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
22757 2017-04-25 Julian Brown <julian@codesourcery.com>
22758 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22760 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
22761 (thunderx2t99_sha): New Reservation.
22763 2017-04-25 Julian Brown <julian@codesourcery.com>
22764 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22766 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
22767 type for 1-element load.
22769 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
22771 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
22773 2017-04-24 Martin Jambor <mjambor@suse.cz>
22775 PR tree-optimization/80293
22776 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
22777 char arrays not totally scalarizable if it is false.
22778 (analyze_all_variable_accesses): Pass correct value in the new
22779 parameter. Add a statistics counter.
22781 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
22783 PR middle-end/79931
22784 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
22786 2017-04-24 Richard Biener <rguenther@suse.de>
22788 PR tree-optimization/80494
22789 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
22790 out for complex types.
22792 2017-04-24 Richard Biener <rguenther@suse.de>
22794 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
22795 * tree-ssa-sccvn.c (print_scc): Print SCC size.
22796 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
22797 (DFS): Adjust and never fail.
22798 (sccvn_dom_walker::fail): Remove.
22799 (sccvn_dom_walker::before_dom_children): Adjust.
22800 (run_scc_vn): Likewise and never fail.
22801 * tree-ssa-pre.c (pass_pre::execute): Adjust.
22802 (pass_fre::execute): Likewise.
22804 2017-04-24 Richard Biener <rguenther@suse.de>
22806 PR tree-optimization/79725
22807 * tree-ssa-sink.c (statement_sink_location): Return whether
22808 failure reason was zero uses. Move that check later.
22809 (sink_code_in_bb): Deal with zero uses by removing the stmt
22812 2017-04-24 Richard Biener <rguenther@suse.de>
22815 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
22816 pointer-based references.
22818 2017-04-24 Richard Biener <rguenther@suse.de>
22821 * pass_manager.h (pass_manager::operator new): Remove.
22822 (pass_manager::operator delete): Likewise.
22823 * passes.c (pass_manager::operator new): Remove.
22824 (pass_manager::operator delete): Likewise.
22825 (pass_manager::pass_manager): Zero individual pass members.
22827 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
22830 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
22831 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
22832 Check "XEXP (src, 1)" operand here.
22833 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
22834 Check "XEXP (src, 1)" operand here.
22835 (dimode_scalar_chain::make_vector_copies): Detect count register
22836 of a shift instruction. Zero extend count register from QImode
22837 to DImode to satisfy vector shift pattern count operand predicate.
22838 Substitute vector shift count operand with a DImode copy.
22839 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
22842 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
22844 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
22845 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
22846 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
22847 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
22848 (UNSPEC_NOREX_MEM): Remove definition.
22850 2017-04-21 Richard Biener <rguenther@suse.de>
22852 PR tree-optimization/79547
22853 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
22854 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
22855 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
22856 without any constraints.
22858 2017-04-21 Richard Biener <rguenther@suse.de>
22860 PR tree-optimization/78847
22861 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
22863 2017-04-21 Richard Biener <rguenther@suse.de>
22865 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
22866 (build_distinct_type_copy): Likewise.
22867 (build_variant_type_copy): Likewise.
22868 * tree.c (build_qualified_type): Pass down mem-stat info.
22869 (build_distinct_type_copy): Likewise.
22870 (build_variant_type_copy): Likewise.
22872 2017-04-21 Richard Biener <rguenther@suse.de>
22874 PR tree-optimization/80237
22875 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
22877 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
22878 for a simplified result.
22880 2016-04-21 Richard Biener <rguenther@suse.de>
22882 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
22883 sth as strict as a simple_iv but a chrec without symbols and an
22884 operand defined in the loop we are peeling (and not some subloop).
22885 (propagate_constants_for_unrolling): Propagate all constants.
22887 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
22890 * config/i386/i386.c (print_reg): Remove assert for disalowed
22891 regno values, call output_operand_lossage instead.
22893 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
22896 * config/i386/constraints.md (Yc): New register constraint.
22897 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
22898 Use Yc constraint for alternative 2 of operand 0. Remove
22899 preferred_for_speed attribute.
22901 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
22903 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
22904 lastprivate clauses in SIMT case.
22906 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
22908 * doc/invoke.texi (-Wextra-semi): Document new warning option.
22910 2017-04-20 Richard Biener <rguenther@suse.de>
22912 PR tree-optimization/57796
22913 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
22914 as N scalar stores.
22915 (vect_model_load_cost): Cost gathers as N scalar loads.
22917 2017-04-20 Richard Biener <rguenther@suse.de>
22919 * ggc-page.c (ggc_allocated_p): Rename to ...
22920 (safe_lookup_page_table_entry): ... this and return the lookup
22922 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
22924 2017-04-20 Richard Biener <rguenther@suse.de>
22926 PR tree-optimization/80453
22927 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
22928 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
22929 from the conditions.
22930 (vn_phi_eq): Pass them down.
22931 (vn_phi_lookup): Record them.
22932 (vn_phi_insert): Likewise.
22934 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
22936 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
22937 uninitialized variable warning to avoid buffer overrun.
22939 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
22942 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
22943 is suppressed for '{ 0 }' in C.
22945 2017-04-20 Jakub Jelinek <jakub@redhat.com>
22947 * BASE-VER: Set to 8.0.0.
22949 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
22951 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
22952 priority .init_array and .fini_array section with SECTION_NOTYPE
22955 2017-04-20 Jakub Jelinek <jakub@redhat.com>
22957 PR middle-end/80423
22958 * tree.h (build_array_type): Add typeless_storage default argument.
22959 * tree.c (type_cache_hasher::equal): Also compare
22960 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
22961 (build_array_type): Add typeless_storage argument, set
22962 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
22964 (build_nonshared_array_type): Adjust build_array_type_1 caller.
22965 (build_array_type): Likewise. Add typeless_storage argument.
22967 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
22968 Jakub Jelinek <jakub@redhat.com>
22970 PR tree-optimization/80426
22971 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
22972 operation on symbolic operands, also compute the overflow for the
22973 invariant part when the operation degenerates into a negation.
22975 2017-04-19 Jakub Jelinek <jakub@redhat.com>
22978 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
22979 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
22982 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
22984 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
22987 * config/avr/avr.c (tree.h): Include it.
22988 (cgraph.h): Include it.
22989 (avr_encode_section_info): Don't warn for uninitialized progmem
22990 variable if it's just an alias.
22992 2017-04-19 Richard Biener <rguenther@suse.de>
22995 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
22996 when needed by AutoPGO.
22998 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
23001 * doc/lto.texi: Remove an extra 'that'.
23003 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
23005 PR rtl-optimization/80429
23006 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
23007 are only used in debug insns.
23009 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
23010 Vladimir Makarov <vmakarov@redhat.com>
23012 * config/sparc/predicates.md (input_operand): Add comment. Return
23013 true for any memory operand when LRA is in progress.
23014 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
23016 2017-04-18 Jeff Law <law@redhat.com>
23019 * mips.md ({return,simple_return}_internal): Do not overwrite
23022 2017-04-18 Jakub Jelinek <jakub@redhat.com>
23024 PR tree-optimization/80443
23025 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
23026 instead of adding 1, subtract -1 and similarly instead of subtracting
23029 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
23031 PR rtl-optimization/80357
23032 * haifa-sched.c (tmp_bitmap): New variable.
23033 (model_recompute): Handle duplicate use records.
23034 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
23035 (free_global_sched_pressure_data): Free it.
23037 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
23040 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
23041 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23042 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23043 instead of SYSTEM_HEADER_DIR.
23045 2017-04-18 Jeff Law <law@redhat.com>
23047 PR middle-end/80422
23048 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
23049 predecessors after walking up the insn chain.
23051 2017-04-18 Jakub Jelinek <jakub@redhat.com>
23054 * dwarf2out.c (modified_type_die): Try harder not to emit internal
23055 sizetype type into debug info.
23057 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23060 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
23061 unneeded test for TARGET_UPPER_REGS_SF.
23062 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
23064 2017-04-18 Jakub Jelinek <jakub@redhat.com>
23067 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
23068 instead of gsi_after_labels.
23070 2017-04-18 Jeff Law <law@redhat.com>
23072 * regcprop.c (maybe_mode_change): Avoid creating copies of the
23076 2017-04-13 Jeff Law <law@redhat.com>
23077 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
23078 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
23080 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
23083 * config/avr/avr.c (intl.h): Include it.
23084 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
23086 2017-04-18 Martin Liska <mliska@suse.cz>
23088 PR gcov-profile/78783
23089 * gcov-tool.c (gcov_output_files): Validate that destination
23090 file is either removed by the tool or by a user.
23092 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
23093 Guy Benyei <guybe@mellanox.com>
23095 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
23096 block, and do not negate it, the stored id is already negative.
23098 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
23100 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
23102 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
23105 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
23106 masks of options that should be turned off if the VSX vector
23107 options are turned off.
23108 (OTHER_P8_VECTOR_MASKS): Likewise.
23109 (OTHER_VSX_VECTOR_MASKS): Likewise.
23110 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
23111 rs6000_disable_incompatible_switches to validate no type switches
23113 (rs6000_incompatible_switch): New function to disallow turning on
23114 other vector options if -mno-vsx, -mno-power8-vector, or
23115 -mno-power9-vector are specified.
23117 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23119 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
23121 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23123 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
23124 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
23125 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
23126 (ARG_POINTER_CFA_OFFSET): Likewise.
23128 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23130 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
23131 conditions to take advantage of various optimizations.
23133 2017-04-13 Jeff Law <law@redhat.com>
23135 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
23136 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
23137 (zero_extendsidi2_dext): Likewise.
23139 2017-04-13 Jakub Jelinek <jakub@redhat.com>
23142 * fold-const.c (fold_ternary_loc): Revert
23143 use op0 instead of fold_convert_loc (loc, type, arg0) part of
23146 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
23148 PR rtl-optimization/80343
23149 * lra-remat.c (update_scratch_ops): Assign original hard reg to
23150 new scratch pseudo.
23152 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
23155 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
23156 to ubsan_encode_value.
23158 2017-04-13 Jeff Law <law@redhat.com>
23160 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
23161 appearing in DEBUG_INSNs.
23163 2017-04-13 Martin Liska <mliska@suse.cz>
23165 PR gcov-profile/80413
23166 * gcov-io.c (gcov_write_string): Copy to buffer just when
23167 allocated size is greater than zero.
23169 2017-04-13 Jakub Jelinek <jakub@redhat.com>
23172 * dwarf2out.c (decls_for_scope): Ignore declarations of
23173 current_function_decl in BLOCK_NONLOCALIZED_VARS.
23175 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
23178 * ipa-visibility.c (non_local_p): Fix typos.
23179 (localize_node): When localizing symbol in same comdat group,
23180 dissolve the group only when we know external symbols are going
23182 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
23184 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23186 PR tree-optimization/79390
23187 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
23188 order does not result in usable sequence, retry with reversed operand
23194 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
23195 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
23196 op0 instead of fold_convert_loc (loc, type, arg0).
23198 2017-04-12 Jeff Law <law@redhat.com>
23200 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
23201 has a delay slot in the generated code.
23203 * config/cris/cris.md (cris_preferred_reload_class): Return
23204 GENNONACR_REGS rather than GENERAL_REGS.
23206 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23209 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
23210 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
23211 signedness of the result type.
23213 2017-04-12 Richard Biener <rguenther@suse.de>
23214 Jeff Law <law@redhat.com>
23216 PR tree-optimization/80359
23217 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
23218 trim stores to TARGET_MEM_REFs.
23220 2017-04-12 Richard Biener <rguenther@suse.de>
23222 PR tree-optimization/79390
23223 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
23224 threading case even more.
23226 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
23229 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
23230 for quad_address_p for TImode, instead of just not indexed_address.
23232 2017-04-12 Richard Biener <rguenther@suse.de>
23233 Bernd Edlinger <bernd.edlinger@hotmail.de>
23235 PR middle-end/79671
23236 * alias.c (component_uses_parent_alias_set_from): Handle
23237 TYPE_TYPELESS_STORAGE.
23238 (get_alias_set): Likewise.
23239 * tree-core.h (tree_type_common): Add typeless_storage flag.
23240 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
23241 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
23242 for types containing members with TYPE_TYPELESS_STORAGE.
23243 (place_field): Likewise.
23244 (layout_type): Likewise for ARRAY_TYPE.
23245 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
23246 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
23247 TYPE_TYPELESS_STORAGE.
23248 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
23250 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23253 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
23254 first argument to type.
23256 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23260 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
23261 CONST0_RTX (mode) rather than const0_rtx where appropriate.
23262 (rs6000_expand_binop_builtin): Likewise.
23263 (rs6000_expand_ternop_builtin): Likewise; also add missing
23264 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
23265 vshasigma built-ins.
23266 * doc/extend.texi: Document that vec_xxpermdi's third argument
23267 must be a constant.
23269 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
23271 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
23272 Use shift_const cost parameter when calculating gain of STV shifts.
23274 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
23276 PR rtl-optimization/70478
23277 * lra-constraints.c (process_alt_operands): Check memory for
23278 disfavoring memory insn operand.
23280 2017-04-11 Jakub Jelinek <jakub@redhat.com>
23282 PR middle-end/80100
23283 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
23284 left shift in unsigned HOST_WIDE_INT type.
23286 PR rtl-optimization/80385
23287 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
23288 (not (neg X)) into (plus X -1) for complex or non-integral modes.
23291 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
23292 if they have any depend clauses.
23294 2017-04-11 Martin Liska <mliska@suse.cz>
23297 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
23298 * ipa-split.c (split_function): Create a local comdat symbol
23299 if caller is in a comdat group.
23301 2017-04-11 Martin Liska <mliska@suse.cz>
23304 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
23307 2017-04-11 Martin Sebor <msebor@redhat.com>
23309 PR middle-end/80364
23310 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
23311 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
23313 (directive::set_width, directive::set_precision, format_character):
23315 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
23318 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
23321 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
23322 conflict, set target->arch_name instead of target->cpu_name.
23324 2017-04-11 Richard Biener <rguenther@suse.de>
23326 PR tree-optimization/80374
23327 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
23328 build_zero_cst, remove fold_convertible_p check again.
23330 2017-04-11 Martin Liska <mliska@suse.cz>
23333 * ubsan.c (instrument_object_size): Do not instrument register
23336 2017-04-11 Jakub Jelinek <jakub@redhat.com>
23339 * config/i386/i386-builtin-types.def
23340 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
23341 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
23342 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
23343 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
23344 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
23345 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
23346 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
23347 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
23348 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
23349 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
23350 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
23351 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
23352 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
23353 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
23354 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
23355 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
23356 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
23357 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
23358 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
23359 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
23360 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
23361 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
23362 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
23363 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
23364 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
23365 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
23366 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
23367 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
23368 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
23369 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
23370 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
23371 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
23372 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
23373 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
23374 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
23375 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
23376 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
23377 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
23378 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
23379 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
23380 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
23381 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
23382 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
23383 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
23384 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
23385 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
23387 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
23388 flag to second_arg_count, handle 4 argument function type _COUNT
23389 aliases, handle second_arg_count on second argument rather than last.
23391 2017-04-10 Jeff Law <law@redhat.com>
23393 PR tree-optimization/80374
23394 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
23395 record anything if we can not convert integer_zero_node to the
23398 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
23401 * config/rs6000/rs6000.c (rs6000_option_override_internal):
23402 Enhance special handling given to the TARGET_P9_MINMAX option in
23403 relation to certain other options.
23405 2017-04-10 Bin Cheng <bin.cheng@arm.com>
23407 PR tree-optimization/80153
23408 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
23409 remove POINTER_PLUS_EXPR's base part directly, rather than through
23412 2017-04-10 Richard Biener <rguenther@suse.de>
23413 Bin Cheng <bin.cheng@arm.com>
23415 PR tree-optimization/80153
23416 * tree-affine.c (aff_combination_to_tree): Get base pointer from
23417 the first element of pointer type aff_tree. Build result expr in
23419 (add_elt_to_tree): Convert to type unconditionally. Remove other
23420 fold_convert calls.
23421 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
23422 (rewrite_use_nonlinear_expr): Check invariant using iv information.
23424 2017-04-10 Richard Biener <rguenther@suse.de>
23426 * tree-ssa-structalias.c (find_func_aliases): Properly handle
23429 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
23431 PR rtl-optimization/70478
23432 * lra-constraints.c (curr_small_class_check): New.
23433 (update_and_check_small_class_inputs): New.
23434 (process_alt_operands): Update curr_small_class_check. Disfavor
23435 alternative insn memory operands. Check available regs for small
23438 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
23441 * config/mips/mips.opt (-mvirt): Update description.
23442 * doc/invoke.texi (-mvirt): Likewise.
23444 2017-04-10 Richard Biener <rguenther@suse.de>
23446 PR middle-end/80362
23447 * fold-const.c (fold_binary_loc): Look at unstripped ops when
23448 looking for NEGATE_EXPR in -A / -B to A / B folding.
23450 2017-04-10 Martin Liska <mliska@suse.cz>
23452 PR gcov-profile/80224
23453 * gcov.c (print_usage): Fix usage string.
23454 (get_gcov_intermediate_filename): Remove.
23455 (output_gcov_file): Use both for normal and intermediate format.
23456 (generate_results): Do not initialize special file for
23457 intermediate format.
23459 2017-04-10 Richard Biener <rguenther@suse.de>
23461 PR tree-optimization/80304
23462 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
23465 2017-04-10 Nathan Sidwell <nathan@acm.org>
23468 * config/rs6000/rs6000.c (rs6000_vector_type): New.
23469 (rs6000_init_builtins): Use it.
23471 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23473 * config/arm/arm.md (<mrc>): Add mode to SET source.
23474 (<mrrc>): Likewise.
23476 2017-04-10 Richard Biener <rguenther@suse.de>
23478 PR middle-end/80344
23479 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
23481 2017-04-10 Jakub Jelinek <jakub@redhat.com>
23484 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
23485 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
23486 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
23487 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
23488 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
23489 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
23490 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
23491 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
23492 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
23493 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
23494 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
23495 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
23496 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
23497 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
23498 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
23499 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
23500 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
23501 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
23502 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
23503 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
23504 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
23505 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
23506 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
23508 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
23510 PR rtl-optimization/70478
23511 * lra-constraints.c: Reverse the last patch.
23513 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
23515 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
23516 Add comment for WCHAR_T.
23518 2017-04-08 Martin Liska <mliska@suse.cz>
23521 2017-04-07 Martin Liska <mliska@suse.cz>
23524 * ipa-split.c (split_function): Add function part to a same comdat
23527 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23530 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
23532 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
23534 * rs6000/rs6000.c (vec_load_pendulum): Rename...
23535 (vec_pairing): ...to this.
23536 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
23537 (rs6000_sched_init): Adjust for name change.
23538 (struct rs6000_sched_context): Likewise.
23539 (rs6000_init_sched_context): Likewise.
23540 (rs6000_set_sched_context): Likewise.
23542 2017-04-07 Jakub Jelinek <jakub@redhat.com>
23548 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
23550 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
23551 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
23552 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
23554 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
23556 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
23558 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
23560 PR rtl-optimization/70703
23561 * ira-color.c (update_conflict_hard_regno_costs): Use
23562 int64_t instead of HOST_WIDE_INT.
23564 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
23566 PR rtl-optimization/70478
23567 * lra-constraints.c (process_alt_operands): Disfavor alternative
23568 insn memory operands.
23570 2017-04-07 Jeff Law <law@redhat.com>
23572 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
23573 CALL and NOTE_INSN_CALL_ARG_LOCATION.
23575 2017-04-07 Martin Liska <mliska@suse.cz>
23578 * config/aarch64/aarch64.c (aarch64_process_target_attr):
23579 Show error message instead of an ICE.
23581 2017-04-07 Martin Liska <mliska@suse.cz>
23584 * ipa-split.c (split_function): Add function part to a same comdat
23587 2017-04-07 Richard Biener <rguenther@suse.de>
23589 PR middle-end/80341
23590 * tree.c (get_unwidened): Also handle ! for_type case for
23592 * convert.c (do_narrow): Split out from ...
23593 (convert_to_integer_1): ... here. Do not pass final truncation
23594 type to get_unwidened for TRUNC_DIV_EXPR.
23596 2017-04-07 Richard Biener <rguenther@suse.de>
23598 * tree-affine.c (wide_int_ext_for_comb): Take type rather
23600 (aff_combination_const): Adjust.
23601 (aff_combination_scale): Likewise.
23602 (aff_combination_add_elt): Likewise.
23603 (aff_combination_add_cst): Likewise.
23604 (aff_combination_convert): Likewise.
23605 (add_elt_to_tree): Likewise. Remove unused argument.
23606 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
23608 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
23610 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
23612 * config/arm/arm.c (arm_default_short_enums): Use
23613 ARM_DEFAULT_SHORT_ENUMS.
23614 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
23616 2017-04-06 Jakub Jelinek <jakub@redhat.com>
23619 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
23620 members with redundant out-of-class redeclaration.
23622 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
23625 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
23626 * config/i386/i386.md (*zero_extendsidi2):
23627 Add (?*x,*x) and (?*v,*v) alternatives.
23629 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
23632 * config/i386/i386.c (ix86_expand_builtin)
23633 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
23634 mode from insn data. Convert operands to insn operand mode.
23635 Copy operands that don't satisfy insn predicate to a register.
23637 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
23639 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
23642 2017-04-06 Richard Biener <rguenther@suse.de>
23644 PR tree-optimization/80334
23645 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
23646 preserve alignment of accesses.
23648 2017-04-06 Richard Biener <rguenther@suse.de>
23650 PR tree-optimization/80262
23651 * tree-sra.c (build_ref_for_offset): Preserve address-space
23653 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
23654 Drop useless address-space information on MEM_REF offsets.
23656 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
23658 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
23660 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
23662 PR rtl-optimization/70703
23663 * ira-color.c (update_conflict_hard_regno_costs): Use
23664 HOST_WIDE_INT instead of long.
23666 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
23669 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
23670 not defined for x86_64 target. Add -mmmx target option when __SSE2__
23672 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
23673 for x86_64 target. Handle -m3dnowa option.
23675 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
23677 PR rtl-optimization/70703
23678 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
23679 (update_conflict_hard_regno_costs): Use long instead of unsigned
23680 arithmetic for cost calculation.
23682 2017-04-05 Jakub Jelinek <jakub@redhat.com>
23683 Bernd Edlinger <bernd.edlinger@hotmail.de>
23686 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
23689 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
23692 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
23693 ptr_mode with Pmode throughout.
23694 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
23695 into probe_stack_range and use DImode.
23697 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
23700 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
23701 call_eh_return is true.
23703 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23705 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
23706 Initialize last_match_fntype_index.
23708 2017-04-05 Jakub Jelinek <jakub@redhat.com>
23711 * tree-nvr.c: Include internal-fn.h.
23712 (pass_return_slot::execute): Ignore internal calls without
23715 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23716 Richard Biener <rguenther@suse.de>
23719 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
23720 captures used multiple times, except for the last use.
23721 * generic-match-head.c: Include gimplify.h.
23723 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23725 PR tree-optimization/79390
23726 * target.h (struct noce_if_info): Declare.
23727 * targhooks.h (default_noce_conversion_profitable_p): Declare.
23728 * target.def (noce_conversion_profitable_p): New target hook.
23729 * ifcvt.h (struct noce_if_info): New type, moved from ...
23730 * ifcvt.c (struct noce_if_info): ... here.
23731 (noce_conversion_profitable_p): Renamed to ...
23732 (default_noce_conversion_profitable_p): ... this. No longer
23734 (noce_try_store_flag_constants, noce_try_addcc,
23735 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
23736 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
23737 instead of noce_conversion_profitable_p.
23738 * config/i386/i386.c: Include ifcvt.h.
23739 (ix86_option_override_internal): Don't override
23740 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
23741 (ix86_noce_conversion_profitable_p): New function.
23742 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
23743 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
23744 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
23745 * doc/tm.texi: Regenerated.
23747 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23749 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
23752 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23755 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
23756 instructions for small multiply cores.
23758 2017-04-04 Jeff Law <law@redhat.com>
23760 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
23762 (mips_expand_vec_perm_const): Initialize elements in orig_perm
23763 that are not set by the loop over the elements.
23765 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23768 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
23769 int mode, convert_modes it to mode as unsigned, otherwise use
23770 lowpart_subreg to mode rather than SImode.
23771 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
23772 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
23773 Use DImode instead of SImode for the shift count operand.
23774 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
23777 2017-04-04 Richard Biener <rguenther@suse.de>
23779 PR middle-end/80281
23780 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
23781 arithmetic done for the negate or the plus. Simplify.
23782 (A - (-B) -> A + B): Likewise.
23783 * fold-const.c (split_tree): Make sure to not negate pointers.
23785 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
23787 PR rtl-optimization/60818
23788 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
23789 a compare of comparisons with the thing compared if this results
23790 in a different machine mode.
23792 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
23794 * alias.c (base_alias_check): Fix typo in comment.
23795 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
23796 * cgraphunit.c (symbol_table::compile): Likewise.
23797 * collect2.c (maybe_run_lto_and_relink): Likewise.
23798 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
23799 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
23800 * config/avr/avr.c (avr_map_op_t): Likewise.
23801 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
23802 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
23803 * config/epiphany/epiphany.md (movcc): Likewise.
23804 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
23805 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
23807 * config/mips/mips.c (mips_save_restore_reg): Likewise.
23808 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
23809 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
23810 * config/sh/sh.c (sh_rtx_costs): Likewise.
23811 * fold-const.c (fold_truth_andor): Likewise.
23812 * genautomata.c (collapse_flag): Likewise.
23813 * gengtype.h (struct type::u::s): Likewise.
23814 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
23815 * input.c (FORMAT_AMOUNT): Likewise.
23816 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
23817 (known_aggs_to_agg_replacement_list): Likewise.
23818 * ipa-inline-analysis.c: Likewise.
23819 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
23820 * ipa-polymorphic-call.c
23821 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
23822 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
23823 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
23825 * modulo-sched.c (apply_reg_moves): Likewise.
23826 * omp-expand.c (build_omp_regions_1): Likewise.
23827 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
23828 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
23829 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
23830 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
23831 * value-prof.c: Likewise.
23832 * var-tracking.c (val_reset): Likewise.
23834 2017-04-03 Richard Biener <rguenther@suse.de>
23836 PR tree-optimization/80275
23837 * fold-const.c (split_address_to_core_and_offset): Handle
23840 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
23842 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
23843 descriptors is at least equal to that of functions.
23845 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
23847 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
23849 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
23852 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
23853 (mov<IMOD4:mode>): New expander.
23854 (*mov<IMOD4:mode>_internal): New insn and split pattern.
23856 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
23858 PR rtl-optimization/79405
23859 * fwprop.c (propagations_left): New variable.
23860 (forward_propagate_into): Decrement it.
23861 (fwprop_init): Initialize it.
23862 (fw_prop): If the variable has reached zero, stop propagating.
23863 (fwprop_addr): Ditto.
23865 2017-03-31 Jakub Jelinek <jakub@redhat.com>
23868 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
23869 a FUNCTION_DECL, pass it as decl instead of origin to
23872 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
23874 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
23877 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
23880 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
23881 TARGET_VSX_SMALL_INTEGER.
23883 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23885 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23886 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
23888 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
23890 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
23891 extraction from odd-numbered MSA register.
23893 2017-03-31 Jakub Jelinek <jakub@redhat.com>
23895 PR middle-end/80173
23896 * expmed.c (store_bit_field_1): Don't attempt to create
23897 a word subreg out of hard registers wider than word if they
23898 have HARD_REGNO_NREGS of 1 for their mode.
23900 PR middle-end/80163
23901 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
23902 conversions to integer types wider than word and pointer.
23905 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
23906 (rtx_equal_for_cselib_p): Pass 0 to it.
23907 * cselib.c (cselib_hasher::equal): Likewise.
23908 (rtx_equal_for_cselib_1): Add depth argument. If depth
23909 is 128, don't look up VALUE locs and punt. Increment
23910 depth in recursive calls when walking VALUE locs.
23912 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
23914 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
23915 (make_gcov_file_name): Use the canonical path name for generating
23917 (read_line): Fix handling of files with ascii null bytes.
23919 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
23921 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
23922 to initialise a vector register instead
23923 of using a const_int.
23925 2017-03-30 Jakub Jelinek <jakub@redhat.com>
23927 PR translation/80189
23928 * gimplify.c (omp_default_clause): Use %qs instead of %s in
23929 diagnostic messages.
23931 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
23934 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
23935 (dfp_diex_<mode>): Update mode of operand 1.
23936 * doc/extend.texi (dxex, dxexq): Document change to return type.
23937 (diex, diexq): Document change to argument type.
23939 2017-03-30 Martin Jambor <mjambor@suse.cz>
23942 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
23943 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
23944 it reflects the signature changes performed at the callee side.
23945 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
23946 to cgraph_build_function_type_skip_args.
23947 (build_function_decl_skip_args): Adjust call to the above function.
23949 2017-03-30 Jakub Jelinek <jakub@redhat.com>
23952 * config/i386/sse.md
23953 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
23954 register as dest whenever it is a MEM not rtx_equal_p to the
23955 corresponding dup operand, and when forcing into reg move the
23956 reg into the memory afterwards.
23957 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
23958 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
23959 for the force_reg mode.
23960 (avx512vl_vextractf128<mode>): Use register as dest either
23961 always when a MEM, or when it is a MEM not rtx_equal_p to the
23962 corresponding dup operand, or even not when it is a CONST_VECTOR
23963 depending on the mode and lo vs. hi.
23964 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
23966 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
23967 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
23968 Likewise. Require that operands[2] is even.
23969 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
23970 Remove extraneous parens. Require that operands[2] is a multiple
23972 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
23973 operands[0] is a MEM if <mask_applied>, the predicates/constraints
23974 disallow memory then.
23976 2017-03-30 Richard Biener <rguenther@suse.de>
23978 PR tree-optimization/77498
23979 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
23980 to non-constants over backedges.
23982 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
23984 PR rtl-optimization/80233
23985 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
23986 as last_combined_insn. Do not test for BARRIER_P separately.
23988 2017-03-29 Andreas Schwab <schwab@suse.de>
23991 * calls.c (prepare_call_address): Convert funexp to Pmode before
23992 copying to temp reg.
23994 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23996 PR tree-optimization/80158
23997 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
23998 Handle possible future case of more than one alternate
24000 (replace_rhs_if_not_dup): Likewise.
24001 (replace_one_candidate): Likewise.
24003 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
24005 PR rtl-optimization/80193
24006 * ira.c (ira): Do not check allocation for LRA.
24008 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
24010 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
24011 (nvptx_output_simt_exit): Declare.
24012 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
24013 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
24014 (init_softstack_frame): Move initialization of crtl->is_leaf to...
24015 (nvptx_declare_function_name): ...here. Emit declaration of local
24016 memory space buffer for omp_simt_enter insn.
24017 (nvptx_output_unisimt_switch): New.
24018 (nvptx_output_softstack_switch): New.
24019 (nvptx_output_simt_enter): New.
24020 (nvptx_output_simt_exit): New.
24021 * config/nvptx/nvptx.h (struct machine_function): New fields
24022 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
24023 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
24024 (UNSPECV_SIMT_EXIT): Ditto.
24025 (omp_simt_enter_insn): New insn.
24026 (omp_simt_enter): New expansion.
24027 (omp_simt_exit): New insn.
24028 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
24030 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
24031 (expand_GOMP_SIMT_ENTER_ALLOC): New.
24032 (expand_GOMP_SIMT_EXIT): New.
24033 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
24034 (GOMP_SIMT_ENTER_ALLOC): Ditto.
24035 (GOMP_SIMT_EXIT): Ditto.
24036 * target-insns.def (omp_simt_enter): New insn.
24037 (omp_simt_exit): Ditto.
24038 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
24040 (lower_rec_simd_input_clauses): Implement SIMT privatization.
24041 (lower_rec_input_clauses): Likewise.
24042 (lower_lastprivate_clauses): Handle SIMT privatization.
24044 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
24045 (ompdevlow_adjust_simt_enter): New.
24046 (find_simtpriv_var_op): New.
24047 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
24048 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
24050 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
24051 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
24052 (copy_decl_for_dup_finish): Ditto.
24054 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
24056 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
24059 * config/i386/i386.c (ix86_option_override_internal): Always
24060 allow -mpreferred-stack-boundary=3 for 64-bit targets.
24062 2017-03-28 Bin Cheng <bin.cheng@arm.com>
24064 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
24066 2017-03-28 Bin Cheng <bin.cheng@arm.com>
24068 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
24069 mark new edge's irreducible flag accordign to it.
24070 (vect_do_peeling): Check loop preheader edge's irreducible flag
24071 and pass it to function slpeel_add_loop_guard.
24073 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
24075 PR tree-optimization/80218
24076 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
24077 Update block frequencies and counts.
24079 2017-03-28 Richard Biener <rguenther@suse.de>
24081 PR tree-optimization/78644
24082 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
24083 of a simplification result we may not use it at all.
24085 2017-03-28 Richard Biener <rguenther@suse.de>
24088 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
24089 without arguments, generate default definition of a SSA name.
24091 2017-03-28 Richard Biener <rguenther@suse.de>
24093 PR middle-end/80222
24094 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
24095 TYPE_REF_CAN_ALIAS_ALL references.
24096 * fold-const.c (fold_indirect_ref_1): Likewise.
24098 2017-03-28 Martin Liska <mliska@suse.cz>
24101 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
24102 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
24104 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
24105 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
24107 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
24108 (EXTRA_SPECS): Define.
24109 (SUBTARGET_EXTRA_SPECS): Likewise.
24110 (SUBTARGET_CPP_SPEC): Likewise.
24111 * config/arc/elf.h (EXTRA_SPECS): Renamed to
24112 SUBTARGET_EXTRA_SPECS.
24113 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
24115 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
24117 * config/arc/simdext.md (vst64_insn): Update pattern.
24118 (vld32wh_insn): Likewise.
24119 (vld32wl_insn): Likewise.
24120 (vld64_insn): Likewise.
24121 (vld32_insn): Likewise.
24123 2017-03-28 Marek Polacek <polacek@redhat.com>
24126 * fold-const.c (fold_comparison): Use protected_set_expr_location
24127 instead of SET_EXPR_LOCATION.
24129 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
24131 * tree.c (add_expr): Avoid name lookup warning.
24133 2017-03-27 Jeff Law <law@redhat.com>
24135 PR tree-optimization/80216
24136 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
24137 function name. Limit recursion depth.
24138 (record_temporary_equivalences): Corresponding changes.
24140 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
24142 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
24145 2017-03-27 Jakub Jelinek <jakub@redhat.com>
24148 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
24150 * cfgcleanup.c (reg_note_cfa_p): New array.
24151 (insns_have_identical_cfa_notes): New function.
24152 (old_insns_match_p): Don't cross-jump in between /f
24153 and non-/f instructions. If both i1 and i2 are frame related,
24154 verify all CFA notes, their order and content.
24156 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
24159 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
24160 HImode and SImode with zero extend to DImode to one insn.
24161 (bswap<mode>2_extenddi): Likewise.
24162 (bswapsi2_extenddi): Likewise.
24163 (bswaphi2_extendsi): Likewise.
24164 (bswaphi2): Combine bswap HImode and SImode into one insn.
24165 Separate memory insns from swapping register.
24166 (bswapsi2): Likewise.
24167 (bswap<mode>2): Likewise.
24168 (bswaphi2_internal): Delete, no longer used.
24169 (bswapsi2_internal): Likewise.
24170 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
24171 store, and gpr<-gpr swap insns.
24172 (bswap<mode>2_store): Likewise.
24173 (bswaphi2_reg): Register only splitter, combine with the splitter.
24174 (bswaphi2 splitter): Likewise.
24175 (bswapsi2_reg): Likewise.
24176 (bswapsi2 splitter): Likewise.
24177 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
24178 the insns into load, store, and register/register insns.
24179 (bswapdi2_ldbrx): Likewise.
24180 (bswapdi2_load): Likewise.
24181 (bswapdi2_store): Likewise.
24182 (bswapdi2_reg): Likewise.
24184 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
24186 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
24187 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
24189 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
24192 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
24194 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24195 special handling for target option conflicts between dform
24196 options (-mpower9-dform, -mpower9-dform-vector,
24197 -mpower9-dform-scalar) and -mno-direct-move.
24199 2017-03-27 Richard Biener <rguenther@suse.de>
24201 PR tree-optimization/80181
24202 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
24204 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24206 * config/arc/predicates.md (move_double_src_operand): Replace the
24207 call to move_double_src_operand with a call to address_operand.
24209 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24211 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
24212 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
24213 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
24215 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24217 * config/arc/predicates.md (long_immediate_loadstore_operand):
24218 Consider scaled addresses cases.
24220 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24222 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
24223 restored when in interrupt.
24224 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
24225 doesn't have delay slot.
24227 2017-03-27 Richard Biener <rguenther@suse.de>
24230 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
24231 inlined thunk clones.
24233 2017-03-27 Jakub Jelinek <jakub@redhat.com>
24236 * asan.c (instrument_derefs): Copy over last operand from
24237 original COMPONENT_REF to the new COMPONENT_REF with
24238 DECL_BIT_FIELD_REPRESENTATIVE.
24239 * ubsan.c (instrument_object_size): Likewise.
24241 2017-03-27 Richard Biener <rguenther@suse.de>
24243 PR tree-optimization/80170
24244 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
24245 sure DR/SCEV didnt fold in constants we do not see when looking
24246 at the reference base alignment.
24248 2017-03-27 Richard Biener <rguenther@suse.de>
24250 PR middle-end/80171
24251 * gimple-fold.c (fold_ctor_reference): Properly guard against
24252 NULL return value from canonicalize_constructor_val.
24254 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
24257 * config/i386/i386.c (ix86_expand_builtin)
24258 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
24259 flags reg setting and flags reg using instructions.
24260 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
24261 clobbering instructions to zero extend op2.
24263 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
24265 * doc/install.texi (Configuration) <--with-aix-soname>:
24266 Update link to AIX ld.
24268 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
24270 PR rtl-optimization/80160
24271 PR rtl-optimization/80159
24272 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
24273 reg_alternate_class into account.
24275 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
24278 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
24279 to consider in curr_insn_transform.
24281 2017-03-24 Jakub Jelinek <jakub@redhat.com>
24283 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
24284 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
24285 and emit_mode_inner.
24287 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24289 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
24290 argument to the overloaded builtin variants. Use the new flag to
24291 deprecate certain builtin variants.
24292 * config/s390/s390-builtin-types.def: Add new builtin types.
24293 * config/s390/s390-builtins.h: Support new flags field for
24294 overloaded builtins.
24295 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
24296 (s390_macro_to_expand): Enable vector float data type.
24297 (s390_cpu_cpp_builtins_internal): Indicate support of the new
24298 builtins by incrementing the __VEC__ version number.
24299 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
24301 (s390_resolve_overloaded_builtin): Emit error messages depending
24302 on the builtin flags.
24303 * config/s390/s390.c (s390_expand_builtin): Support additional
24304 flags argument. Change error message to match the messages
24305 emitted in s390-c.c.
24306 * config/s390/s390.md: New UNSPEC_* constants.
24307 (op_type): Add new instruction types.
24308 * config/s390/vecintrin.h: Add new builtins and test data class
24310 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
24311 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
24312 (VEC_INEXACT, VEC_NOINEXACT): New constants.
24313 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
24314 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
24315 ("vec_mergel<mode>"): V_HW -> VEC_HW.
24317 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
24318 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
24319 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
24320 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
24322 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
24323 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
24324 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
24325 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
24327 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
24328 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
24329 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
24330 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
24331 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
24332 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
24333 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
24335 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
24336 ("vec_scatter_element<V_HW_4:mode>_DI")
24337 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
24338 ("vec_fpint<mode>", "vflls")
24339 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
24340 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
24341 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
24342 ("*vec_cmphe<mode>_cc"): ... these.
24344 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
24345 mode constant instead of magic value.
24347 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24349 * config/s390/s390.c (s390_expand_vec_compare): Support other
24350 vector floating point modes than just V2DF.
24351 (s390_expand_vcond): Likewise.
24352 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
24353 (s390_cannot_change_mode_class): Prevent mode changes between TF
24354 and V1TF in vector registers.
24355 * config/s390/s390.md (DF, SF): New mode attributes.
24356 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
24357 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
24358 SFmode support for VRs.
24359 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
24361 (VFT, VF_HW): New mode iterators.
24362 (vw, sdx): New mode attributes.
24363 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
24364 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
24365 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
24366 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
24367 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
24368 also the new vector floating point modes. Renaming to ...
24370 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
24371 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
24372 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
24373 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
24374 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
24375 ("vec_unordered<mode>"): ... these.
24377 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
24378 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
24379 ("*vec_extendv2df"): New insn definitions.
24381 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24383 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
24384 ("mulditi3_2", "*muldi3_sign"): New patterns.
24385 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
24386 rename the pattern definition.
24388 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24390 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
24392 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
24394 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24396 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
24397 instruction if possible.
24398 * config/s390/vector.md (vec_halfnumelts): New mode
24400 ("*vec_vllezlf<mode>"): New pattern.
24402 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24404 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
24405 ("popcountv4si2", "popcountv2di2"): Rename to ...
24406 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
24407 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
24409 ("popcount<mode>2_vxe"): New pattern.
24411 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24413 * common/config/s390/s390-common.c (processor_flags_table): Add
24415 * config.gcc: Add arch12.
24416 * config/s390/driver-native.c (s390_host_detect_local_cpu):
24417 Default to arch12 for unknown CPU model numbers.
24418 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
24419 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
24420 PROCESSOR_max sanity check.
24421 * config/s390/s390-opts.h (enum processor_type): Add
24423 * config/s390/s390.c (processor_table): Add arch12.
24424 (s390_expand_builtin): Add check for B_VXE flag.
24425 (s390_issue_rate): Add PROCESSOR_ARCH12.
24426 (s390_get_sched_attrmask): Likewise.
24427 (s390_get_unit_mask): Likewise.
24428 (s390_sched_score): Enable z13 scheduling for arch12.
24429 (s390_sched_reorder): Likewise.
24430 (s390_sched_variable_issue): Likewise.
24431 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
24433 (s390_tune_attr): Use z13 scheduling also for arch12.
24434 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
24435 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
24436 (TARGET_VXE_P): New macros.
24437 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
24438 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
24439 * config/s390/s390.opt: Add arch12 as processor_type.
24441 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24443 * config/s390/s390.md
24444 ("fixuns_truncdddi2", "fixuns_trunctddi2")
24445 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
24446 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
24448 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
24449 Rename expanders to ...
24451 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
24452 ("fixuns_truncdddi2_emu"): ... these.
24454 ("fixuns_trunc<mode>si2_emu"): New expander.
24456 ("*fixuns_truncdfdi2_z13"): Rename to ...
24457 ("*fixuns_truncdfdi2_vx"): ... this.
24459 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24461 * config/s390/2964.md: Remove the single element vector compare
24462 instructions which are no longer used.
24463 * config/s390/s390.c (s390_select_ccmode): Remove handling of
24465 (s390_canonicalize_comparison): Remove handling of DFmode
24467 (s390_expand_vec_compare_scalar): Remove function.
24468 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
24469 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
24471 ("*cmp<mode>_ccs"): Add wfcdb instruction.
24473 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24475 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
24477 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
24478 will anyway by matched by mov<mode>_64dfp.
24480 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24482 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
24483 vlef/vstef. Add missing operand to vleif.
24485 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24487 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
24488 pair for all vector types with 64 bit elements.
24489 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
24490 * config/s390/vector.md (V_HW_64): ... here.
24491 (V_128_NOSINGLE): New mode iterator.
24492 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
24493 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
24494 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
24495 ("*vec_load_pairv2di"): Change to ...
24496 ("*vec_load_pair<mode>"): ... this one.
24498 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24500 * config/s390/constraints.md: Add comments.
24501 (jKK): Reject element sizes > 8 bytes.
24502 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
24504 * config/s390/s390.md: Add the s_operand checks formerly in
24505 s390_split_ok_p to various splitters where they are still
24507 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
24508 for 128 bit vectors. Plus two splitters.
24510 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24512 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
24515 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24518 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
24519 error if the boundary argument is not constant.
24521 2017-03-24 Jakub Jelinek <jakub@redhat.com>
24523 PR rtl-optimization/80112
24524 * loop-doloop.c (doloop_condition_get): Don't check condition
24525 if cmp isn't SET with IF_THEN_ELSE src.
24527 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24529 PR tree-optimization/80158
24530 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
24531 replacing a candidate statement, also replace it for the
24532 candidate's alternate interpretation.
24533 (replace_rhs_if_not_dup): Likewise.
24534 (replace_one_candidate): Likewise.
24536 2017-03-24 Richard Biener <rguenther@suse.de>
24538 PR tree-optimization/80167
24539 * graphite-isl-ast-to-gimple.c
24540 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
24542 (translate_isl_ast_to_gimple::get_rename): Likewise.
24544 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
24546 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
24547 handling of certain combinations of target options, including the
24548 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
24549 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
24551 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24554 * config/arm/arm.md (*load_multiple): Add reload_completed to
24555 matching condition.
24557 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24558 Richard Biener <rguenth@suse.de>
24560 PR tree-optimization/79908
24561 PR tree-optimization/80136
24562 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
24563 been cast away, gimplify_and_add suffices.
24565 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
24567 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
24569 2017-03-23 Richard Biener <rguenther@suse.de>
24571 PR tree-optimization/80032
24572 * gimplify.c (gimple_push_cleanup): Forced unconditional
24573 cleanups still have to go to the conditional_cleanups
24576 2017-03-22 Jakub Jelinek <jakub@redhat.com>
24578 PR tree-optimization/80072
24579 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
24581 (next_operand_entry_id): Change type to unsigned int.
24582 (sort_by_operand_rank): Make sure to return the right return value
24583 even if unsigned fields are bigger than INT_MAX.
24584 (struct oecount): Change cnt and id type to unsigned int.
24585 (oecount_hasher::equal): Formatting fix.
24586 (oecount_cmp): Make sure to return the right return value
24587 even if unsigned fields are bigger than INT_MAX.
24588 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
24591 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
24592 TREE_READONLY on result if writing it more than once.
24595 * doc/invoke.texi (-fsanitize=thread): Document that with
24596 -fnon-call-exceptions atomics are not able to throw
24600 * tsan.c: Include tree-eh.h.
24601 (instrument_builtin_call): Call maybe_clean_eh_stmt or
24602 maybe_clean_or_replace_eh_stmt where needed.
24603 (instrument_memory_accesses): Add cfg_changed argument.
24604 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
24605 if it returned true.
24606 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
24608 PR rtl-optimization/63191
24609 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
24610 wrapper function, moved the whole old content into ...
24611 (ix86_delegitimize_address_1): ... this. New inline function.
24612 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
24613 true as last argument instead of ix86_delegitimize_address.
24615 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
24617 * config/aarch64/aarch64.c (generic_branch_cost): Copy
24618 cortexa57_branch_cost.
24620 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
24622 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
24624 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24627 * doc/md.texi (Constraints): Document wA constraint.
24628 * config/rs6000/constraints.md (wA): New.
24629 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
24630 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
24631 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
24632 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
24634 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
24637 * gimplify.c (is_oacc_declared): New function.
24638 (oacc_default_clause): Use it to set default flags for acc declared
24639 variables inside parallel regions.
24640 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
24641 declared variables.
24642 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
24643 declare attribute to any decl as necessary.
24645 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
24648 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
24649 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
24650 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
24651 (arm_arch_lpae): This.
24652 * config/arm/arm.c (arm_arch7ve): Rename into ...
24653 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
24654 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
24657 2017-03-22 Martin Liska <mliska@suse.cz>
24660 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
24661 error message instead of an ICE.
24663 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24665 * doc/extend.texi (6.11 Additional Floating Types): Revise.
24667 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
24669 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
24671 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24674 2017-03-21 Martin Sebor <msebor@redhat.com>
24676 * doc/extend.texi: Use "cannot" instead of "can't."
24677 * doc/hostconfig.texi: Same.
24678 * doc/install.texi: Same.
24679 * doc/invoke.texi: Same.
24680 * doc/loop.texi: Same.
24681 * doc/md.texi: Same.
24682 * doc/objc.texi: Same.
24683 * doc/rtl.texi: Same.
24684 * doc/tm.texi: Same.
24685 * doc/tm.texi.in: Same.
24686 * doc/trouble.texi: Same.
24688 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
24691 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
24692 (collect_checksum_attributes): Set it.
24693 (die_checksum_ordered): Use it.
24695 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24697 PR tree-optimization/79908
24698 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
24699 change: For a VA_ARG whose LHS has been cast away, use
24700 force_gimple_operand to construct the side effects.
24702 2017-03-21 David Malcolm <dmalcolm@redhat.com>
24704 PR translation/80001
24705 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
24706 more amenable to translation.
24707 (oacc_loop_auto_partitions): Likewise.
24709 2017-03-21 Marek Polacek <polacek@redhat.com>
24710 Martin Sebor <msebor@redhat.com>
24712 PR tree-optimization/80109
24713 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
24714 on INTEGRAL_TYPE_P.
24716 2017-03-21 Jakub Jelinek <jakub@redhat.com>
24717 Segher Boessenkool <segher@kernel.crashing.org>
24720 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
24721 check reg_used_between_p between insn and one of succ or succ2
24722 depending on if succ is artificial insn not inserted into insn
24725 2017-03-21 Martin Liska <mliska@suse.cz>
24727 PR gcov-profile/80081
24728 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
24729 * doc/gcc.texi: Include gcov-dump stuff.
24730 * doc/gcov-dump.texi: New file.
24732 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
24734 PR rtl-optimization/79150
24735 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
24736 conditional jump, if the jump is the last insn of the loop.
24738 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24739 Richard Biener <rguenth@suse.de>
24741 PR tree-optimization/79908
24742 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
24743 been cast away, use force_gimple_operand to construct the side
24746 2017-03-21 Martin Liska <mliska@suse.cz>
24748 PR libfortran/79956
24749 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
24752 2017-03-21 Brad Spengler <spender@grsecurity.net>
24755 * plugin.c (htab_hash_plugin): New function.
24756 (add_new_plugin): Use it and adjust.
24757 (parse_plugin_arg_opt): Adjust.
24758 (init_one_plugin): Likewise.
24760 2017-03-21 Richard Biener <rguenther@suse.de>
24762 PR tree-optimization/80032
24763 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
24764 if set force the cleanup to happen unconditionally.
24765 (gimplify_target_expr): Push inserted clobbers with force_uncond
24766 to avoid them being removed by control-dependent DCE.
24768 2017-03-21 Richard Biener <rguenther@suse.de>
24770 PR tree-optimization/80122
24771 * tree-inline.c (copy_bb): Do not expans va-arg packs or
24772 va_arg_pack_len when the inlined call stmt requires pack
24774 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
24776 2017-03-21 Jakub Jelinek <jakub@redhat.com>
24779 * tsan.c (instrument_builtin_call): If the memory model argument
24780 is not a constant, assume it is valid.
24783 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
24786 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
24788 PR rtl-optimization/79910
24789 * combine.c (can_combine_p): Do not allow combining an I0 or I1
24790 if its dest is used by an insn before I2 (other than the combined
24791 insns themselves, which are properly handled already).
24793 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
24796 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
24798 * combine.c (record_used_regs): New static function.
24799 (try_combine): Handle situations where there is an additional
24800 instruction between I2 and I3 which needs to have a LOG_LINK
24804 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
24806 * combine.c (try_combine): Delete redundant i1 test. Call
24807 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
24809 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24812 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
24813 alternatives 13/14.
24815 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24817 PR tree-optimization/80054
24818 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
24819 the optimization if a PHI or any of its arguments is not dominated
24820 by the candidate's basis. Use gphi* rather than gimple* as
24822 (replace_profitable_candidates): Clean up a gimple* variable that
24823 should be a gphi* variable.
24825 2017-03-20 Martin Sebor <msebor@redhat.com>
24828 * doc/extend.texi (attribute constructor): Document present limitation.
24830 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
24833 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
24834 __POWER9_VECTOR__ #ifdef control, change template definition to
24835 use Power9-specific built-in function.
24836 (vec_any_eq): Likewise.
24837 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
24838 to control outcomes from this test.
24839 (vector_ae_<mode>p): For VEC_F modes, likewise.
24841 2017-03-20 Ian Lance Taylor <iant@google.com>
24843 * config/i386/i386.c (ix86_function_regparm): Save an extra
24844 register for -fsplit-stack with DECL_STATIC_CHAIN.
24846 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
24849 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
24850 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
24852 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
24854 * config/riscv/riscv.c (riscv_print_operand): Use "fence
24856 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
24859 2017-03-20 Marek Polacek <polacek@redhat.com>
24862 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
24864 2017-03-20 Richard Biener <rguenther@suse.de>
24866 PR tree-optimization/80113
24867 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
24868 allocate extra SSA name for PHI def.
24869 (add_close_phis_to_outer_loops): Likewise.
24870 (add_close_phis_to_merge_points): Likewise.
24871 (copy_loop_close_phi_args): Likewise.
24872 (copy_cond_phi_nodes): Likewise.
24874 2017-03-20 Martin Liska <mliska@suse.cz>
24876 PR middle-end/79753
24877 * tree-chkp.c (chkp_build_returned_bound): Do not build
24878 returned bounds for a LHS that's not a BOUNDED_P type.
24880 2017-03-20 Martin Liska <mliska@suse.cz>
24884 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
24885 COMPLEX_CST and VECTOR_CST.
24887 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24890 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
24891 target operand. A new splitter adds the clobber statement in case
24892 the target operand is dead anyway.
24894 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
24896 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
24897 to age-old versions of binutils and glibc.
24899 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
24901 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
24903 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
24905 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
24907 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
24909 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
24910 requirement for binutils 2.13.
24912 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
24914 * combine.c (try_combine): Delete redundant i1 test. Call
24915 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
24917 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
24919 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
24920 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
24922 <riscv64-*-elf>: Re-arrange section
24923 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
24924 <riscv32-*-linux>: Likewise.
24925 <riscv64-*-elf>: Likewise
24926 <riscv64-*-linux>: Likewise.
24928 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
24931 * aarch64.opt(verbose-cost-dump): Fix typo.
24933 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
24936 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
24937 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
24939 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
24941 * reload.c (find_reloads): When reloading a nonoffsettable address,
24942 use RELOAD_OTHER for it and its address reloads.
24944 PR rtl-optimization/79910
24945 * combine.c (record_used_regs): New static function.
24946 (try_combine): Handle situations where there is an additional
24947 instruction between I2 and I3 which needs to have a LOG_LINK
24950 2017-03-17 Jeff Law <law@redhat.com>
24952 PR tree-optimization/71437
24953 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
24954 conditional in the hash table first.
24955 (vrp_dom_walker::before_dom_children): Extract condition from
24956 ASSERT_EXPR. Record condition, its inverion and any implied
24957 conditions as well.
24959 2017-03-17 Marek Polacek <polacek@redhat.com>
24960 Markus Trippelsdorf <markus@trippelsdorf.de>
24962 PR tree-optimization/80079
24963 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
24966 2017-03-17 Richard Biener <rguenther@suse.de>
24968 PR middle-end/80075
24969 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
24970 Properly verify the LHS before the RHS possibly claims to be
24972 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
24975 2017-03-17 Martin Jambor <mjambor@suse.cz>
24977 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
24978 (List of -O2 options): Likewise.
24979 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
24982 2017-03-17 Tom de Vries <tom@codesourcery.com>
24984 * gcov-dump.c (print_usage): Print bug_report_url.
24986 2017-03-17 Richard Biener <rguenther@suse.de>
24988 PR middle-end/80050
24989 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
24990 (parser::peek): Likewise.
24992 2017-03-17 Richard Biener <rguenther@suse.de>
24994 PR tree-optimization/80048
24995 * sese.c (free_sese_info): Properly release rename_map and
24996 copied_bb_map elements.
24998 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
25000 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
25001 Add linked-list forward and backlinks. Insert on
25002 construction, remove on destruction.
25003 (class pass_store_merging): Add m_stores_head field.
25004 (pass_store_merging::terminate_and_process_all_chains):
25005 Iterate over m_stores_head list.
25006 (pass_store_merging::terminate_all_aliasing_chains):
25008 (pass_store_merging::execute): Check for debug stmts first.
25009 Push new chains onto the m_stores_head stack.
25011 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
25014 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
25015 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
25016 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
25018 2017-03-16 Jeff Law <law@redhat.com>
25020 PR tree-optimization/71437
25021 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
25022 member function. Implementation moved into after_dom_children
25023 member function and into the threader's thread_outgoing_edges
25025 (dom_opt_dom_walker::after_dom_children): Simplify by moving
25026 some code into new thread_outgoing_edges.
25027 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
25028 definition. Simplify marker handling (do it here). Assume we always
25029 have the available expression and the const/copies tables.
25030 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
25032 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
25033 * tree-vrp.c (equiv_stack): No longer file scoped.
25034 (vrp_dom_walker): New class.
25035 (vrp_dom_walker::before_dom_children): New member function.
25036 (vrp_dom_walker::after_dom_children): Likewise.
25037 (identify_jump_threads): Setup domwalker. Use it rather than
25038 walking edges in a random order by hand. Simplify setup/finalization.
25039 (finalize_jump_threads): Remove.
25040 (vrp_finalize): Do not call identify_jump_threads here.
25041 (execute_vrp): Do it here instead and call thread_through_all_blocks
25044 PR tree-optimization/71437
25045 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
25047 (simplify_stmt_for_jump_threading): Add basic_block argument. All
25049 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
25050 (dom_opt_dom_walker::thread_across_edge): Remove
25051 handle_dominating_asserts argument. All callers changed.
25052 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
25053 changes. Remove calls to lhs_of_dominating_assert. Other
25054 uses of handle_dominating_asserts turn into unconditional code
25055 (simplify_control_stmt_condition_1): Likewise.
25056 (simplify_control_stmt_condition): Likewise.
25057 (thread_through_normal_block, thread_across_edge): Likewise.
25058 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
25059 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
25060 object if it is not an SSA_NAME.
25061 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
25062 before calling into the VRP specific simplifiers.
25063 (identify_jump_threads): Remove handle_dominating_asserts
25066 2017-03-16 Jakub Jelinek <jakub@redhat.com>
25069 * tree-diagnostic.c (default_tree_printer): No longer static.
25070 * tree-diagnostic.h (default_tree_printer): New prototype.
25072 2017-03-16 Tamar Christina <tamar.christina@arm.com>
25074 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
25075 Change ins into fmov.
25077 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25079 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
25080 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
25081 Use h_con constraint for operand 1.
25082 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
25083 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
25085 2017-03-15 Jeff Law <law@redhat.com>
25087 PR tree-optimization/71437
25088 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
25089 (record_temporary_equivalences): Use it.
25091 PR tree-optimization/71437
25092 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
25093 tree-ssa-scopedtables.
25094 (lookup_avail_expr, build_and_record_new_cond): Likewise.
25095 (record_conditions, record_cond, vuse_eq): Likewise.
25096 (record_edge_info): Adjust to API tweak of record_conditions.
25097 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
25098 (record_temporary_equivalences, optimize_stmt): Likewise.
25099 (eliminate_redundant_computations): Likewise.
25100 (record_equivalences_from_stmt): Likewise.
25101 * tree-ssa-scopedtables.c: Include options.h and params.h.
25102 (vuse_eq): New function, moved from tree-ssa-dom.c
25103 (build_and_record_new_cond): Likewise.
25104 (record_conditions): Likewise. Accept vector of conditions rather
25105 than edge_equivalence structure for first argument.
25106 for the first argument.
25107 (avail_exprs_stack::lookup_avail_expr): New member function, moved
25108 from tree-ssa-dom.c.
25109 (avail_exprs_stack::record_cond): Likewise.
25110 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
25111 from tree-ssa-dom.c.
25112 (avail_exprs_stack): Add new member functions lookup_avail_expr
25114 (record_conditions): Declare.
25116 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
25119 * lra-constraints.c (process_alt_operands): Increase reject for
25120 reloading an input/output operand.
25122 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
25125 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
25126 insns to convert from signed/unsigned char/short to IEEE 128-bit
25128 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
25130 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
25133 * config/i386/i386.c (ix86_vector_duplicate_value): Create
25134 subreg of inner mode for values already in registers.
25136 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
25138 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
25139 iteration reg is used after the loop.
25141 2017-03-14 Martin Sebor <msebor@redhat.com>
25143 PR tree-optimization/79800
25144 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
25145 precision in negative-positive range.
25146 (format_floating): Call non-const overload with adjusted precision.
25148 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25151 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
25154 2017-03-14 Martin Sebor <msebor@redhat.com>
25156 PR middle-end/80020
25157 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
25158 * builtins.def (aligned_alloc): Use it.
25161 * Makefile.in (GTFILES): Add calls.c.
25162 * calls.c: Include "gt-calls.h".
25164 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
25166 PR rtl-optimization/79728
25167 * regs.h (struct target_regs): New field
25168 x_contains_allocatable_regs_of_mode.
25169 (contains_allocatable_regs_of_mode): New macro.
25170 * reginfo.c (init_reg_sets_1): Initialize it, and change
25171 contains_reg_of_mode so it includes global regs as well.
25172 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
25173 rather than contains_regs_of_mode.
25175 2017-03-14 Martin Liska <mliska@suse.cz>
25177 * doc/invoke.texi: Document options that can't be combined with
25178 -fcheck-pointer-bounds.
25180 2017-03-14 Martin Liska <mliska@suse.cz>
25182 PR middle-end/79831
25183 * doc/invoke.texi (-Wchkp): Document the option.
25185 2017-03-14 Martin Liska <mliska@suse.cz>
25187 * Makefile.in: Install gcov-dump.
25189 2017-03-14 Martin Liska <mliska@suse.cz>
25191 * multiple_target.c (expand_target_clones): Bail out for
25192 an invalid attribute.
25194 2017-03-14 Richard Biener <rguenther@suse.de>
25196 * alias.c (struct alias_set_entry): Pack properly.
25197 * cfgloop.h (struct loop): Likewise.
25198 * cse.c (struct set): Likewise.
25199 * ipa-utils.c (struct searchc_env): Likewise.
25200 * loop-invariant.c (struct invariant): Likewise.
25201 * lra-remat.c (struct cand): Likewise.
25202 * recog.c (struct change_t): Likewise.
25203 * rtl.h (struct address_info): Likewise.
25204 * symbol-summary.h (function_summary): Likewise.
25205 * tree-loop-distribution.c (struct partition): Likewise.
25206 * tree-object-size.c (struct object_size_info): Likewise.
25207 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
25208 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
25209 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
25210 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
25211 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
25212 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
25213 (struct _stmt_vec_info): Likewise.
25215 2017-03-14 Martin Liska <mliska@suse.cz>
25218 * multiple_target.c (create_dispatcher_calls): Check that
25219 a target can create a function dispatcher.
25221 2017-03-14 Martin Liska <mliska@suse.cz>
25224 * multiple_target.c (expand_target_clones): Drop local.local
25225 flag for default implementation.
25227 2017-03-14 Richard Biener <rguenther@suse.de>
25229 PR tree-optimization/80030
25230 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
25232 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
25234 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
25235 gcc_fallthrough() instead of __attribute__((fallthrough));
25237 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
25239 * doc/gcc.texi: Remove "up" link to (DIR).
25240 * doc/gccint.texi: Ditto.
25242 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
25244 * doc/install.texi (Specific) <avr>: Remove reference to
25247 2017-03-13 Jeff Law <law@redhat.com>
25249 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
25250 attribute rather than comments.
25252 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
25253 match_scratch operand is highest.
25255 2017-03-13 Martin Liska <mliska@suse.cz>
25257 PR middle-end/78339
25258 * ipa-pure-const.c (warn_function_noreturn): If the declarations
25259 is a CHKP clone, use original declaration.
25261 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25263 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
25264 (arc_conditional_register_usage): Use a different allocation order
25265 when optimizing for size.
25266 * common/config/arc/arc-common.c (arc_option_optimization_table):
25267 Section anchors default on when optimizing for size.
25269 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25271 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
25273 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25275 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
25276 * config/arc/arc.md (cpu_facility): Add cd variant.
25277 (*movqi_insn): Add code density variant.
25278 (*movhi_insn): Likewise.
25279 (*movqi_insn): Likewise.
25280 (*addsi3_mixed): Likewise.
25281 (subsi3_insn): Likewise.
25283 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25285 * config/arc/arc.md (movsi_cond_exec): Update constraint.
25287 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25289 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
25290 expressions with MINUS and UNARY ops.
25292 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25295 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
25297 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
25298 between vec_select and vector argument.
25299 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
25300 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
25301 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
25302 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
25303 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
25304 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
25306 2017-03-13 Richard Biener <rguenther@suse.de>
25309 * params.def (vect-max-peeling-for-alignment): Fix typo.
25311 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25313 * doc/install.texi (Specific) <mips-*-*>: Remove description of
25314 issue that only occurred with binutils below 2.18.
25316 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25318 * doc/install.texi (Specific) <cris-axis-elf>: No longer
25319 refer to binutils 2.11/2.12 minimum.
25321 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25323 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
25324 ftp.kernel.org and simplify binutils requirement.
25326 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
25328 * doc/invoke.texi (Warning Options): Fix spelling of link-time
25330 (Optimize Options): Ditto. Also remove redundancy.
25332 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25334 PR translation/79848
25335 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
25337 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
25338 to G_ to avoid double translation.
25340 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25342 PR translation/79923
25343 * auto-profile.c (get_combined_location): Convert leading
25344 character of diagnostics to lower case and remove trailing period.
25345 (read_profile): Likewise for various diagnostics.
25346 * config/arm/arm.c (arm_option_override): Remove trailing period
25347 from various diagnostics.
25348 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
25349 (msp430_expand_delay_cycles): Likewise.
25351 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25354 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
25355 full command-line argument, rather than just "str".
25356 (aarch64_validate_march): Likewise.
25357 (aarch64_validate_mtune): Likewise.
25359 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
25361 PR rtl-optimization/78911
25362 * lra-assigns.c (must_not_spill_p): New function.
25363 (spill_for): Use it.
25365 2017-03-10 Jakub Jelinek <jakub@redhat.com>
25367 PR tree-optimization/79981
25368 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
25369 ATOMIC_COMPARE_EXCHANGE ifn result.
25370 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
25371 IFN_ATOMIC_COMPARE_EXCHANGE.
25373 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25376 * opts.c (parse_sanitizer_options): Add missing question mark to
25377 "did you mean" message.
25379 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25381 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
25383 (VMULEUH_UNS): Likewise.
25384 (VMULOUB_UNS): Likewise.
25385 (VMULOUH_UNS): Likewise.
25386 * config/rs6000/rs6000.c (builtin_function_type): Remove
25387 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
25389 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25392 * read-rtl-function.c (function_reader::read_rtx_operand): Update
25393 x with result of extra_parsing_for_operand_code_0.
25394 (function_reader::extra_parsing_for_operand_code_0): Convert
25395 return type from void to rtx, returning x. When reading
25396 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
25397 larger size containing struct block_symbol.
25399 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
25401 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
25402 -mfloat128-hardware without -m64.
25404 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
25407 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
25408 entries to the case statement that marks unsigned arguments to
25409 overloaded functions.
25411 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25413 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
25414 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
25416 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
25419 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
25420 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
25422 2017-03-10 Martin Liska <mliska@suse.cz>
25426 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
25427 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
25430 2017-03-10 Olivier Hainque <hainque@adacore.com>
25432 * tree-switch-conversion (array_value_type): Start by resetting
25433 candidate type to it's main variant.
25435 2017-03-10 Jakub Jelinek <jakub@redhat.com>
25437 PR rtl-optimization/79909
25438 * combine.c (try_combine): Use simplify_replace_rtx on individual
25439 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
25440 of the whole CALL_INSN_FUNCTION_USAGE.
25442 PR tree-optimization/79972
25443 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
25444 get_range_info on SSA_NAMEs. Formatting fixes.
25446 2017-03-10 Richard Biener <rguenther@suse.de>
25447 Jakub Jelinek <jakub@redhat.com>
25449 PR tree-optimization/77975
25450 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
25451 edge to be constant.
25452 (get_val_for): For constant x return it. Formatting fix.
25453 (loop_niter_by_eval): Avoid pointless looping if the next iteration
25454 would use the same bases as the current one.
25456 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25458 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
25459 instead of vec_select for V1TImode.
25460 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
25462 (VSX_LE_128): Add V1TI to this mode iterator.
25463 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
25464 (*vsx_le_perm_store_<mode>): Likewise.
25465 (pre-reload splitter for VSX stores): Likewise.
25466 (post-reload splitter for VSX stores): Likewise.
25467 (*vsx_xxpermdi2_le_<mode>): Likewise.
25468 (*vsx_lxvd2x2_le_<mode>): Likewise.
25469 (*vsx_stxvd2x2_le_<mode>): Likewise.
25471 2017-03-09 Michael Eager <eager@eagercon.com>
25473 Correct failures with --enable-checking=yes,rtl.
25475 * config/microblaze/microblaze.c (microblaze_expand_shift):
25476 Replace GET_CODE test with CONST_INT_P and INTVAL test with
25477 test for const0_rtx.
25478 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
25479 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
25481 2017-03-09 Richard Biener <rguenther@suse.de>
25483 PR tree-optimization/79977
25484 * graphite-scop-detection.c (scop_detection::merge_sese):
25485 Handle the case of extra exits to blocks dominating the entry.
25487 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
25489 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25492 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
25494 PR rtl-optimization/79949
25495 * lra-constraints.c (process_alt_operands): Check memory when
25496 trying to predict a cycle. Print about the overall increase.
25498 2017-03-09 Richard Biener <rguenther@suse.de>
25500 PR middle-end/79971
25501 * gimple-expr.c (useless_type_conversion_p): Preserve
25502 TYPE_SATURATING for fixed-point types.
25504 2017-03-09 Richard Biener <rguenther@suse.de>
25507 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
25508 alignment of BLKmode params.
25510 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25513 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
25514 (VALL_NO_V2Q): Likewise.
25516 * config/aarch64/aarch64-simd.md
25517 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
25519 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
25520 VALL_NO_V2Q mode iterator.
25521 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
25523 2017-03-09 Martin Liska <mliska@suse.cz>
25525 PR tree-optimization/79631
25526 * tree-chkp-opt.c (chkp_is_constant_addr): Call
25527 tree_int_cst_sign_bit just for INTEGER constants.
25529 2017-03-09 Martin Liska <mliska@suse.cz>
25533 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
25536 2017-03-09 Marek Polacek <polacek@redhat.com>
25539 * tree.c (inchash::add_expr): Handle TREE_VEC.
25541 2017-03-09 Martin Liska <mliska@suse.cz>
25544 (chkp_narrow_size_and_offset): New function.
25545 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
25546 (void chkp_parse_bit_field_ref): New function.
25547 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
25548 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
25550 2017-03-09 Martin Liska <mliska@suse.cz>
25553 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
25554 (chkp_find_bounds_1): Remove gcc_unreachable.
25556 2017-03-09 Jakub Jelinek <jakub@redhat.com>
25559 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
25560 BUILT_IN_SYNC*, determine the access type from the size suffix and
25561 always build a MEM_REF with that type. Handle forgotten
25562 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
25565 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
25566 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
25567 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
25568 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
25569 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
25570 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
25571 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
25572 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
25573 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
25574 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
25575 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
25576 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
25577 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
25578 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
25579 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
25580 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
25581 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
25582 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
25583 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
25584 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
25585 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
25586 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
25587 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
25588 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
25589 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
25590 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
25591 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
25592 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
25593 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
25594 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
25595 definitions outside of __OPTIMIZE__ guarded section.
25598 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
25599 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
25600 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
25601 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
25604 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25606 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
25607 ("vfenez<mode>"): Add missing constraints.
25609 2017-03-08 Martin Sebor <msebor@redhat.com>
25612 * config/nds32/nds32.c (nds32_option_override):
25613 Fix misspelled diagnostic.
25615 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25618 * gimplify.c (gimplify_omp_for): Replace index var in outer
25619 taskloop statement with an artificial variable and add
25620 OMP_CLAUSE_PRIVATE clause for it.
25622 2017-03-08 Richard Biener <rguenther@suse.de>
25624 PR tree-optimization/79955
25625 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
25626 for accesses that are completely outside of the variable.
25628 2017-03-08 Andrew Haley <aph@redhat.com>
25630 PR tree-optimization/79943
25631 * tree-ssa-loop-split.c (compute_new_first_bound): When
25632 calculating the new upper bound, (END-BEG) should be added, not
25635 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25637 * config/avr/avr.md (setmemhi): Make sure match_dup
25638 operand number comes before match_scratch.
25640 2017-03-08 Richard Biener <rguenther@suse.de>
25642 PR tree-optimization/79920
25643 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
25644 with ncopies == 1 to ...
25645 (vect_transform_slp_perm_load): ... here. Properly compute
25646 all element loads by iterating VF times over the group. Do
25647 not handle ncopies (computed in a broken way) in
25648 vect_create_mask_and_perm.
25650 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25653 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
25654 is a uniform vector, use uniform_vector_p return value instead of
25655 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
25657 2017-03-07 Marek Polacek <polacek@redhat.com>
25659 PR middle-end/79809
25660 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
25661 (alloca_call_type): Likewise.
25663 2017-03-07 Martin Liska <mliska@suse.cz>
25665 * gcov.c (process_args): Put comment to correct location.
25667 2017-03-07 Martin Liska <mliska@suse.cz>
25669 PR middle-end/68270
25670 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
25671 Use array_at_struct_end_p instead of DECL_CHAIN (field).
25672 (chkp_narrow_bounds_for_field): Likewise.
25673 (chkp_parse_array_and_component_ref): Pass one more argument to
25676 2017-03-07 Richard Biener <rguenther@suse.de>
25678 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
25681 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
25683 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
25684 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
25686 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25689 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
25690 to end of description.
25691 (PARAM_MAX_STORES_TO_MERGE): Likewise.
25693 2017-03-07 Jakub Jelinek <jakub@redhat.com>
25695 PR rtl-optimization/79901
25696 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
25698 (*avx512f_<code><mode>3<mask_name>): ... this.
25699 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
25700 iterator instead of VI8_AVX2_AVX512BW.
25702 PR rtl-optimization/79901
25703 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
25704 min/max expander, expand it using expand_vec_cond_expr.
25707 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
25710 2017-03-06 Jakub Jelinek <jakub@redhat.com>
25713 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
25714 to void * for PCH reasons.
25715 * dwarf2out.c (output_loc_operands, output_die): Cast
25716 v.val_vec.array to unsigned char *.
25718 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
25721 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
25724 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
25726 PR rtl-optimization/79571
25727 * lra-constraints.c (process_alt_operands): Calculate static
25728 reject and subtract it from overall when only addresses will be
25731 2017-03-06 Julia Koval <julia.koval@intel.com>
25734 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
25735 incoming stack boundary to 128 for 64-bit targets.
25737 2017-03-06 Richard Biener <rguenther@suse.de>
25739 PR tree-optimization/79894
25740 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
25741 to NULL after folding it.
25743 2017-03-06 Richard Biener <rguenther@suse.de>
25745 PR tree-optimization/79824
25746 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
25747 check disabling peeling for gaps.
25749 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
25751 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
25752 attributes): Document gettimeofday.
25754 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
25756 * config/s390/s390.c (s390_option_override_internal): Set
25757 PARAM_MIN_VECT_LOOP_BOUND
25759 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
25761 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
25762 * config/s390/s390.md: Likewise.
25764 2017-03-06 Jakub Jelinek <jakub@redhat.com>
25767 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
25768 (<avx2_avx512>_perm<mode>): Rename to ...
25769 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
25771 (<avx512>_perm<mode>_mask): Rename to ...
25772 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
25774 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
25775 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
25776 instead of VI8F_256_512.
25777 (avx512f_perm<mode>): New define_expand.
25778 (avx512f_perm<mode>_mask): Likewise.
25779 (avx512f_perm<mode>_1<mask_name>): New define_insn.
25780 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
25782 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25784 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
25785 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
25787 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
25789 2017-03-06 Martin Liska <mliska@suse.cz>
25792 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
25793 when having a SSA NAME w/o VAR_DECL assigned to it.
25795 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25797 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
25798 msa_dpsub_<su>_d): Fix MODE for vec_select.
25800 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
25802 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
25804 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
25806 2017-03-06 Richard Biener <rguenther@suse.de>
25808 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
25809 * plugin.c (register_plugin_info): Likewise.
25810 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
25812 2017-03-05 Jakub Jelinek <jakub@redhat.com>
25814 * config/i386/sse.md (sse_storehps, sse_storelps,
25815 avx_<castmode><avxsizesuffix>_<castmode>,
25816 avx512f_<castmode><avxsizesuffix>_<castmode>,
25817 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
25818 in condition that at least one operand is not a MEM.
25820 2017-03-03 Jakub Jelinek <jakub@redhat.com>
25822 PR middle-end/79805
25823 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
25824 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
25826 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
25827 gimple_call_nothrow_p flag based on whether original builtin can throw.
25828 If it can, emit following stmts on the fallthrough edge.
25829 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
25830 don't create new bb if inserting just debug stmts on the edge, try to
25831 insert them on the fallthru bb or just reset debug stmts.
25833 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
25836 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
25837 restore recog_data (including the operand rtxes inside it) around
25838 the call to get_insn_template.
25840 2017-03-03 Martin Sebor <msebor@redhat.com>
25842 PR tree-optimization/79699
25843 * context.c (context::~context): Free MPFR caches to avoid
25844 a memory leak on program exit.
25846 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25848 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
25849 Use wide_int::ulow () instead of .elt (0).
25851 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
25853 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
25854 (*pushxf): Limit oF constraint to 32bit targets and add oC
25855 constraint for 64bit targets.
25856 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
25857 (*pushdf): Change rmF constraint to rmC.
25859 2017-03-03 Martin Liska <mliska@suse.cz>
25861 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
25862 Remove unused variable.
25864 2017-03-03 Jakub Jelinek <jakub@redhat.com>
25867 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
25868 is a memory operand, increase num_memory.
25869 (ix86_expand_args_builtin): Likewise.
25871 2017-03-03 Jan Hubicka <jh@suse.cz>
25874 * ipa-devirt.c (maybe_record_node): Properly handle
25875 __cxa_pure_virtual visibility.
25877 2017-03-03 Martin Liska <mliska@suse.cz>
25879 PR tree-optimization/79803
25880 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
25882 (pass_loop_prefetch::execute): Disabled optimization if an
25883 assumption about L1 cache size is not met.
25885 2017-03-03 Martin Liska <mliska@suse.cz>
25887 PR rtl-optimization/79574
25888 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
25889 (hash_scan_set): Likewise.
25890 (dump_hash_table): Likewise.
25891 (hoist_code): Likewise.
25893 2017-03-03 Richard Biener <rguenther@suse.de>
25895 * fixed-value.c (fixed_from_string): Restore use of elt (1)
25896 in place of uhigh ().
25897 (fixed_convert_from_real): Likewise.
25899 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
25902 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
25904 2017-03-03 Richard Biener <rguenther@suse.de>
25906 PR middle-end/79818
25907 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
25908 TYPE_OVERFLOW_UNDEFINED check.
25910 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25912 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
25914 (vector_ae_<mode>_p): Likewise.
25915 (vector_nez_<mode>_p): Likewise.
25916 (vector_ne_v2di_p): Likewise.
25917 (vector_ae_v2di_p): Likewise.
25918 (vector_ne_<mode>_p): Likewise.
25919 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
25921 (vsx_tsqrt<mode>2_fe): Likewise.
25923 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
25926 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
25928 2017-03-02 Jakub Jelinek <jakub@redhat.com>
25930 PR rtl-optimization/79780
25931 * cprop.c (one_cprop_pass): When second and further conditional trap
25932 in a single basic block is turned into an unconditional trap, turn it
25933 into a deleted note to avoid RTL verification failures.
25935 2017-03-02 Richard Biener <rguenther@suse.de>
25937 * fold-const.c (const_binop): Use ulow () instead of elt (0).
25939 2017-03-02 Richard Biener <rguenther@suse.de>
25941 PR tree-optimization/79345
25943 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
25944 param and abort the walk, returning -1 if it is hit.
25945 (walk_aliased_vdefs): Take a limit param and pass it on.
25946 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
25947 defaulting to 0 and return a signed int.
25948 * tree-ssa-uninit.c (struct check_defs_data): New struct.
25949 (check_defs): New helper.
25950 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
25951 about uninitialized memory.
25952 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
25953 bogus uninitialized warning.
25954 (fixed_convert_from_real): Likewise.
25956 2017-03-02 Bin Cheng <bin.cheng@arm.com>
25958 PR tree-optimization/66768
25959 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
25960 iv_use if base object can't be determined.
25962 2017-03-02 Jakub Jelinek <jakub@redhat.com>
25964 PR tree-optimization/79345
25965 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
25966 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
25967 (get_pattern_stats): Initialize it.
25968 * genemit.c (gen_expand): Verify match_scratch numbers come after
25969 match_operand/match_dup numbers.
25970 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
25971 match_scratch numbers.
25972 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
25974 * config/s390/s390.md (trunctdsd2): Likewise.
25976 2017-03-02 Richard Biener <rguenther@suse.de>
25978 * wide-int.h (wide_int_storage::operator=): Implement in terms
25981 2017-03-02 Richard Biener <rguenther@suse.de>
25983 PR tree-optimization/79777
25984 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
25985 the to insert expression to sth existing.
25987 2017-03-01 Martin Sebor <msebor@redhat.com>
25989 PR middle-end/79692
25990 * gimple-ssa-sprintf.c
25991 (directive::known_width_and_precision): New function.
25992 (format_integer): Use it.
25993 (get_mpfr_format_length): Consider the full range of precision
25994 when computing %g output with the # flag. Set the likely byte
25995 count to 3 rather than 1 when precision is indeterminate.
25996 (format_floating): Correct the lower bound of precision.
25998 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26000 * doc/invoke.texi: Document default code model for 64-bit Linux.
26002 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26005 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
26006 udiv rather than div since input pattern is unsigned.
26008 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
26010 * config/i386/i386.c (print_reg): Warn for values of
26011 unsupported size in integer register.
26013 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
26016 * config/rs6000/predicates.md (current_file_function_operand): Do
26017 not allow self calls to be local if the function is replaceable.
26019 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
26022 * config/rs6000/altivec.h (vec_ctz and others): Change the
26023 preprocessor macro that controls conditional compilation from
26024 _ARCH_PWR9 to __POWER9_VECTOR__.
26025 (vec_all_ne): Change parameterization of __altivec_scalar_pred
26026 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
26027 control (instead of _ARCH_PWR9 control) so that template
26028 definition uses power9-specific function.
26029 (vec_any_eq): Likewise.
26030 (vec_all_ne): Change macro definition to use a power9-specific
26031 expansion under #ifdef __POWER9_VECTOR__ control (instead of
26032 _ARCH_PWR9 control).
26033 (vec_any_eq) Likewise.
26034 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
26035 expansion for CMPNEF to remove support for xvcmpnesp instruction.
26036 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
26037 support for xvcmpnedp instruction.
26038 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
26039 macro expansion so that Power9 implementation of vec_all_ne does
26040 not use the AltiVec predicate framework.
26041 (VCMPNEH_P): Likewise.
26042 (VCMPNEW_P): Likewise.
26043 (VCMPNED_P): Likewise.
26044 (VCMPNEFP_P): Likewise.
26045 (VCMPNEDP_P): Likewise.
26046 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
26047 implementation of vec_any_eq to not use AltiVec predicate
26049 (VCMPAEH_P): Likewise.
26050 (VCMPAEW_P): Likewise.
26051 (VCMPAED_P): Likewise.
26052 (VCMPAEFP_P): Likewise.
26053 (VCMPAEDP_P): Likewise.
26054 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
26055 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
26056 not use the AltiVec predicate framework.
26057 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
26058 of vec_any_eq to not use AltiVec predicate framework.
26059 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
26060 support for predefined __POWER9_VECTOR__ macro to indicate that
26061 Power9 instruction selection is enabled.
26062 (altivec_overloaded_builtins): Remove extraneous
26063 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
26064 function argument types RS6000_BTI_bool_V16QI and
26065 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
26066 entry for overloaded function argument types RS6000_BTI_bool_V4SI
26067 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
26068 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
26069 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
26070 Power9 for implementations of vec_cmpne. Change the signature for
26071 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
26072 (representing vec_all_ne) to remove the previously described first
26073 argument of type RS6000_BTI_INTSI, as this was an artifact of
26074 reliance on the AltiVec predicate framework, which is no longer
26075 used in the implementation of these functions. Add
26076 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
26077 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
26078 since, unlike the AltiVec predicate framework implementation, we
26079 do not share function descriptors between vec_alle and vec_anyeq.
26080 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
26081 set of modes that receive special treatment even when
26082 TARGET_P9_VECTOR is true. The special treatment emits code that
26083 does not depend on Power9 instructions.
26084 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
26085 define_expand to not rely on AltiVec predicate framework.
26086 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
26088 (vector_ne_v2di_p): Change this define_expand to not rely on
26089 AltiVec predicate framework.
26090 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
26092 (vector_ne_<mode>_p): Change this define_expand to not rely on
26093 AltiVec predicate framework.
26094 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
26096 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
26097 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
26098 define_insn pattern.
26099 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
26100 define_insn pattern because the xvcmpne<VSs>. instruction is not
26102 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
26103 instruction is not supported.
26105 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26107 * config/nvptx/nvptx.c: Include intl.h.
26109 2017-03-01 Martin Jambor <mjambor@suse.cz>
26112 * ipa-prop.h (ipa_bits): Removed field known.
26113 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
26114 to pointers. Adjusted their comments to warn about their sharing.
26115 (ipcp_transformation_summary): Change bits to a vector of pointers.
26116 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
26117 (ipa_get_ipa_bits_for_value): Declare.
26118 * tree-vrp.h (value_range): Mark as GTY((for_user)).
26119 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
26120 (ipa_bits_hash_table): Likewise.
26121 (ipa_vr_ggc_hash_traits): Likewise.
26122 (ipa_vr_hash_table): Likewise.
26123 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
26124 being pointers and vr_known being removed.
26125 (ipa_set_jf_unknown): Likewise.
26126 (ipa_get_ipa_bits_for_value): New function.
26127 (ipa_set_jfunc_bits): Likewise.
26128 (ipa_get_value_range): New overloaded functions.
26129 (ipa_set_jfunc_vr): Likewise.
26130 (ipa_compute_jump_functions_for_edge): Use the above functions to
26131 construct bits and vr parts of jump functions.
26132 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
26133 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
26135 (ipcp_grow_transformations_if_necessary): Also allocate
26136 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
26138 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
26139 them. Fix too long lines.
26140 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
26141 vr_known being removed.
26142 (ipa_read_jump_function): Use new setter functions to construct bits
26143 and vr parts of jump functions or set them to NULL.
26144 (write_ipcp_transformation_info): Adjust for bits being pointers.
26145 (read_ipcp_transformation_info): Likewise.
26146 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
26148 Include gt-ipa-prop.h.
26149 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
26151 (ipcp_store_bits_results): Likewise.
26152 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
26153 Do not write to existing jump functions but use a temporary instead.
26155 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26158 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
26159 attempt to use its first operand as BIT_FIELD_REF base.
26161 2017-03-01 Richard Biener <rguenther@suse.de>
26163 PR middle-end/79721
26164 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
26165 interpolating formula in wrapping arithmetic.
26166 (chrec_apply): Convert chrec_evaluate return value to wanted type.
26168 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26170 PR tree-optimization/79734
26171 * tree-vect-generic.c (expand_vector_condition): Optimize
26172 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
26173 Handle VEC_COND_EXPR where comparison has different inner width from
26174 type's inner width.
26176 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
26178 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
26179 markup, and similar issues. Remove @opindex entries for things
26180 that aren't options. Add missing -mmpy-option entries.
26182 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26184 PR tree-optimization/79737
26185 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
26186 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
26187 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
26188 instead of byte_size. Formatting fix.
26189 (shift_bytes_in_array_right): Formatting fix.
26191 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
26194 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
26195 condition on optimize for the leaf function test.
26197 2017-02-28 Martin Liska <mliska@suse.cz>
26200 * read-rtl-function.c (function_reader::handle_unknown_directive):
26201 Bail out when one uses -flto.
26203 2017-02-28 Martin Liska <mliska@suse.cz>
26205 * common.opt: Replace space with tabular for options of <number>
26207 * config/i386/i386.opt: Show <number> value for
26208 -mlarge-data-threshold.
26209 * opts.c (print_filtered_help): Do not display number in hexadecimal
26212 2017-02-28 Martin Liska <mliska@suse.cz>
26214 * common.opt: Fix --help=option -Q for options which are of
26217 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
26219 * config/i386/i386.c (print_reg): Error out for values
26220 of 8-bit size in invalid integer register.
26222 2017-02-28 Martin Sebor <msebor@redhat.com>
26224 PR tree-optimization/79691
26225 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
26227 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26230 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
26231 gcc_unreachable with output_operand_lossage.
26233 2017-02-28 Richard Biener <rguenther@suse.de>
26235 PR tree-optimization/79740
26236 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
26238 (visit_nary_op): Insert the nary into the hashtable if we
26239 pattern-matched sth.
26240 * tree-ssa-pre.c (eliminate_insert): Robustify.
26242 2017-02-28 Richard Biener <rguenther@suse.de>
26244 PR middle-end/79731
26245 * fold-const.c (decode_field_reference): Reject out-of-bound
26248 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26250 * config/i386/i386.c: Include intl.h.
26251 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
26252 instead of just cond ? "..." : "...".
26253 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
26254 * coverage.c (read_counts_file): Likewise.
26255 * omp-offload.c: Include intl.h.
26256 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
26257 of just cond ? "..." : "...".
26258 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
26259 of just cond ? "..." : "...".
26261 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
26264 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
26266 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
26267 'tune for' CPU name.
26268 * config/arm/arm-cpu-data.h: Regenerated.
26270 2017-02-28 Richard Biener <rguenther@suse.de>
26272 PR tree-optimization/79732
26273 * tree-inline.c (expand_call_inline): Do not shadow var.
26275 2017-02-28 Richard Biener <rguenther@suse.de>
26277 PR tree-optimization/79723
26278 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
26279 address-space properly.
26281 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
26283 * doc/optinfo.texi (Optimization groups): Fix option used for
26285 * doc/invoke.texi (-fopt-info): Document "omp".
26286 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
26287 (OPTGROUP_ALL): Add OPTGROUP_OMP.
26288 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
26289 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
26290 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
26292 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
26294 * dumpfile.c (optgroup_options): Instead of "openmp", associate
26295 OPTGROUP_OMP with "omp".
26297 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
26300 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
26301 for arithmetic shift of unsigned V2DI.
26303 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
26305 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
26306 arc/linux.h headers.
26307 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
26308 (LINK_SPEC): Likewise.
26309 (ARC_TLS_EXTRA_START_SPEC): Likewise.
26310 (EXTRA_SPECS): Likewise.
26311 (STARTFILE_SPEC): Likewise.
26312 (ENDFILE_SPEC): Likewise.
26313 (LIB_SPEC): Likewise.
26314 (TARGET_SDATA_DEFAULT): Likewise.
26315 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
26316 (MULTILIB_DEFAULTS): Likewise.
26317 (DWARF2_UNWIND_INFO): Likewise.
26318 * config/arc/big.h: New file.
26319 * config/arc/elf.h: Likewise.
26320 * config/arc/linux.h: Likewise.
26321 * config/arc/t-uClibc: Remove.
26323 2017-02-27 Bin Cheng <bin.cheng@arm.com>
26325 PR tree-optimization/77536
26326 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
26327 (tree_transform_and_unroll_loop): Use above function to compute the
26328 estimated niter of unrolled loop and use it when scaling profile.
26329 Also use count info rather than frequency if it's non-zero.
26330 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
26331 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
26332 (vect_transform_loop): Call above function.
26334 2017-02-27 Richard Biener <rguenther@suse.de>
26336 PR tree-optimization/45397
26337 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
26338 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
26339 (visit_nary_op): Add pattern matching for CSEing sign-changed
26340 or truncated operations with wider ones.
26342 2017-02-27 Richard Biener <rguenther@suse.de>
26344 PR tree-optimization/79690
26345 * tree-vect-stmts.c (vectorizable_store): Use vector type
26346 built from the DR with address-space.
26348 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
26350 * doc/invoke.texi (Optimize Options): Refine the description
26351 of asan-use-after-return.
26353 2017-02-25 Alan Modra <amodra@gmail.com>
26355 PR rtl-optimization/79584
26356 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
26357 base, not ad->base_term, the reg within base. Remove assertion
26358 that ad->base == ad->base_term. Replace gen_int_mode using
26359 bogus mode with const0_rtx.
26361 2017-02-25 Jakub Jelinek <jakub@redhat.com>
26363 PR middle-end/79396
26364 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
26365 FMA_EXPR like tcc_binary or tcc_unary.
26367 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
26370 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
26372 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
26373 (output_loc_operands): Handle DW_OP_call_ref and
26374 DW_OP_GNU_variable_value.
26375 (struct variable_value_struct): New type.
26376 (struct variable_value_hasher): Likewise.
26377 (variable_value_hash): New variable.
26378 (string_types): Remove.
26379 (copy_loc_descr): New function.
26380 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
26381 (prepend_loc_descr_to_each): New function.
26382 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
26383 instead of add_loc_descr_to_each if the first argument is single
26384 location list and the second has multiple.
26385 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
26386 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
26387 when looking for variable value which doesn't have other location info.
26388 (loc_list_from_tree): Formatting fix.
26389 (gen_array_type_die): Simplify DW_AT_string_length handling.
26390 (adjust_string_types): Remove.
26391 (gen_subprogram_die): Don't call adjust_string_types nor test/set
26392 string_types. Call resolve_variable_values.
26393 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
26394 (resolve_addr_in_expr): Likewise. Add A argument.
26395 (copy_deref_exprloc): Remove deref argument. Adjust for the
26396 original expression being DW_OP_GNU_variable_value with optionally
26397 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
26398 optionally after it.
26399 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
26400 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
26401 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
26402 (variable_value_hasher::hash, variable_value_hasher::equal): New
26404 (resolve_variable_value_in_expr, resolve_variable_value,
26405 resolve_variable_values, note_variable_value_in_expr,
26406 note_variable_value): New functions.
26407 (dwarf2out_early_finish): Call note_variable_value on all toplevel
26410 2017-02-24 Jakub Jelinek <jakub@redhat.com>
26413 * opts.h (handle_generated_option): Add GENERATED_P argument.
26414 * opts-common.c (handle_option): Adjust function comment.
26415 (handle_generated_option): Add GENERATED_P argument, pass it to
26417 (control_warning_option): Pass false to handle_generated_option
26419 * opts.c (maybe_default_option): Pass true to handle_generated_option
26421 * optc-gen.awk: Likewise.
26423 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26425 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
26426 a REG, look at the REG it is a SUBREG of.
26427 (splitter for cmpeqsi_t): Ditto.
26429 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26431 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
26432 the special USEs with the pattern of the insn, not the insn itself.
26434 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
26437 * doc/invoke.texi: Document -mload-store-pairs.
26439 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26440 Sandra Loosemore <sandra@codesourcery.com>
26442 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
26443 argument isn't a CONST_INT.
26444 (nios2_alternate_compare_const): Assert op is a CONST_INT.
26445 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
26446 (nios2_validate_compare): Bypass alternate compare logic if *op2
26447 is not a CONST_INT.
26448 (ldstwm_operation_p): Return false if first_base is not a REG or
26449 if first_offset is not a CONST_INT.
26451 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26453 * config/cris/cris.md: Use correct operand in a define_peephole2.
26455 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26457 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
26459 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26461 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
26462 this_insn if it is an INSN or JUMP_INSN.
26463 (force_offsettable): Look at base, not at addr.
26464 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
26465 on things that aren't necessarily CONST_INTs.
26467 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
26469 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
26470 -mfpmath=sse is the default also for x86-32 targets with SSE2
26471 instruction set when @option{-ffast-math} is enabled
26473 2017-02-24 Jeff Law <law@redhat.com>
26475 PR rtl-optimizatoin/79286
26476 * ira.c (update_equiv_regs): Drop may_trap_p exception to
26479 2017-02-24 Richard Biener <rguenther@suse.de>
26481 PR tree-optimization/79389
26482 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
26485 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
26487 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
26488 function comment to reflect reality.
26489 (loop_exits_before_overflow): Fix typo in function description.
26491 2017-02-24 Richard Biener <rguenther@suse.de>
26493 PR tree-optimization/79389
26494 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
26495 properly that a threading opportunity exists. Detect conditional
26496 copy/constant propagation opportunities.
26498 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
26500 * config/visium/visium.md (type): Add trap.
26501 (b): New mode attribute.
26502 (*btst): Rename into...
26503 (*btst<mode>): ...this and adjust.
26504 (*cbranchsi4_btst_insn): Rename into...
26505 (*cbranch<mode>4_btst_insn): ...this and adjust.
26506 (trap): New define_insn.
26508 2017-02-23 Jakub Jelinek <jakub@redhat.com>
26510 PR tree-optimization/79389
26511 * ifcvt.c (struct noce_if_info): Add rev_cond field.
26512 (noce_reversed_cond_code): New function.
26513 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
26514 reversed_comparison_code. Formatting fix.
26515 (noce_try_store_flag): Test rev_cond != NULL in addition to
26516 reversed_comparison_code.
26517 (noce_try_store_flag_constants): Likewise.
26518 (noce_try_store_flag_mask): Likewise.
26519 (noce_try_addcc): Use rev_cond if non-NULL instead of
26520 reversed_comparison_code.
26521 (noce_try_cmove_arith): Likewise. Formatting fixes.
26522 (noce_try_minmax, noce_try_abs): Clear rev_cond.
26523 (noce_find_if_block): Initialize rev_cond.
26524 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
26525 instead of false as last argument never attempt to reverse it
26528 2017-02-23 Bin Cheng <bin.cheng@arm.com>
26530 PR tree-optimization/79663
26531 * tree-predcom.c (combine_chains): Process refs in reverse order
26532 only for ZERO length chains, and add explaining comment.
26534 2017-02-23 Jeff Law <law@redhat.com>
26536 PR tree-optimization/79578
26537 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
26538 in call to operand_equal_p.
26540 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
26543 * config/i386/cpuid.h: Fix another undefined behavior.
26545 2017-02-23 Richard Biener <rguenther@suse.de>
26547 PR tree-optimization/79683
26548 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
26549 vector types for data-refs.
26551 2017-02-23 Martin Liska <mliska@suse.cz>
26553 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
26555 2017-02-23 Jakub Jelinek <jakub@redhat.com>
26557 PR middle-end/79665
26558 * internal-fn.c (get_range_pos_neg): Moved to ...
26559 * tree.c (get_range_pos_neg): ... here. No longer static.
26560 * tree.h (get_range_pos_neg): New prototype.
26561 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
26562 are known to be in between 0 and signed maximum inclusive, try to
26563 expand both unsigned and signed divmod and use the cheaper one from
26566 2017-02-22 Jeff Law <law@redhat.com>
26568 PR tree-optimization/79578
26569 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
26570 to compare base operands.
26572 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
26575 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
26576 gpc_reg_operand instead of fpr_reg_operand.
26578 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
26580 * config/mips/mips.c (mips_return_in_memory): Force FP
26581 vector types to be returned in memory for o32 ABI.
26583 2017-02-22 Jakub Jelinek <jakub@redhat.com>
26585 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
26586 instead of DW_TAG_member for static data member declarations and don't
26587 set no_linkage_name for static inline data members.
26588 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
26591 2017-02-22 Martin Liska <mliska@suse.cz>
26593 * doc/invoke.texi: Replace inequality signs with square brackets
26596 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26598 PR tree-optimization/68644
26599 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
26601 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
26604 * lra-constraints.c (simplify_operand_subreg): Handle
26605 WORD_REGISTER_OPERATIONS targets.
26607 2017-02-22 Jakub Jelinek <jakub@redhat.com>
26610 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
26611 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
26612 elimination by swapping fld*.
26614 2017-02-22 Richard Biener <rguenther@suse.de>
26616 PR tree-optimization/79673
26617 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
26618 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
26619 irrelevant address-space qualifiers and avoiding a
26620 ADDR_SPACE_CONVERT_EXPR from fold_convert.
26622 2017-02-22 Richard Biener <rguenther@suse.de>
26624 PR tree-optimization/79666
26625 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
26626 to not symbolically negate if that may introduce undefined
26629 2017-02-22 Martin Liska <mliska@suse.cz>
26632 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
26633 * data-streamer-out.c (streamer_write_gcov_count_stream):
26635 * value-prof.c (stream_out_histogram_value): Make assert more
26636 precise based on type of counter.
26638 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
26641 * config/i386/i386.md (standard_x87sse_constant_load splitter):
26642 Use nonimmediate_operand instead of memory_operand for operand 1.
26643 (float-extend standard_x87sse_constant_load splitter): Ditto.
26645 2017-02-21 Jeff Law <law@redhat.com>
26647 PR tree-optimization/79621
26648 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
26649 blocks with edges to themselves.
26651 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26654 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
26655 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
26656 Use gimple_call_builtin_p.
26659 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
26660 on temporarily removed DEBUG_INSNs.
26662 PR tree-optimization/79649
26663 * tree-loop-distribution.c (classify_partition): Give up on
26664 non-generic address space loads/stores.
26666 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
26668 * doc/loop.texi (Loop manipulation): Remove nonexistent
26669 tree_ssa_loop_version from the documentation.
26670 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
26672 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26675 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
26676 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
26677 * config/rs6000/rs6000.c: Include except.h.
26678 (rs6000_expand_split_stack_prologue): Call
26679 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
26681 2017-02-21 Martin Jambor <mjambor@suse.cz>
26684 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
26685 have been analyzed.
26687 2017-02-21 Martin Jambor <mjambor@suse.cz>
26689 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
26690 for backward compatibility only.
26691 * doc/invoke.texi (Option Summary): Remove all references to
26692 -fipa-cp-alignment.
26694 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
26698 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26700 * lra-constraints.c (curr_insn_transform): Handle
26701 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
26703 2017-02-21 Martin Liska <mliska@suse.cz>
26705 * config/i386/i386.opt: Replace -masm-dialect with -masm.
26707 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
26709 PR translation/79638
26710 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
26712 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
26715 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
26716 (arm_function_ok_for_sibcall): Return false for an indirect call by
26717 descriptor if all the argument registers are used.
26718 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
26719 alignment of the function.
26721 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26723 PR tree-optimization/61441
26724 * simplify-rtx.c (simplify_const_unary_operation): For
26725 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
26726 the sNaN unmodified.
26728 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26730 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26731 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26732 instead of SYSTEM_HEADER_DIR.
26734 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
26735 Martin Liška <mliska@suse.cz>
26737 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
26738 Fix typos and grammar, use active voice, and clarify.
26740 2017-02-20 Marek Polacek <polacek@redhat.com>
26742 PR middle-end/79537
26743 * gimplify.c (gimplify_expr): Handle unused *&&L;.
26746 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
26748 2017-02-20 Jakub Jelinek <jakub@redhat.com>
26751 * config/i386/i386.c (ix86_expand_builtin): Handle
26752 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
26753 ix86_builtins_isa[fcode].isa as a requirement of those
26754 flags and any other flag in the bitmask.
26755 (ix86_init_mmx_sse_builtins): Use 0 instead of
26756 ~OPTION_MASK_ISA_64BIT as mask.
26757 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
26758 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
26759 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
26760 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
26762 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26765 * lra-constraints.c (split_reg): Check requested split mode
26766 is supported by the register.
26768 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26770 * lra-constraints.c (simplify_operand_subreg): Remove early
26773 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26776 * lra-constraints.c (curr_insn_transform): Tighten condition
26777 for converting SUBREG reloads from OP_OUT to OP_INOUT.
26779 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26782 * lra-constraints.c (curr_insn_transform): Handle
26783 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
26785 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
26788 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
26790 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
26792 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
26795 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
26798 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26800 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
26803 2017-02-18 Jakub Jelinek <jakub@redhat.com>
26805 * final.c (last_columnnum, override_columnnum): New variables.
26806 (final_start_function): Set last_columnnum, pass it to begin_prologue
26807 hook and pass 0 to dwarf2out_begin_prologue.
26808 (final_scan_insn): Update override_columnnum. Pass last_columnnum
26809 to source_line debug hook.
26810 (notice_source_line): Compute last_columnnum and for debug_column_info
26811 return true on column changes.
26812 * debug.h (struct gcc_debug_hooks): Add column argument to
26813 source_line and begin_prologue hooks.
26814 (debug_nothing_int_charstar_int_bool): Remove prototype.
26815 (debug_nothing_int_int_charstar,
26816 debug_nothing_int_int_charstar_int_bool): New prototypes.
26817 (dwarf2out_begin_prologue): Add column argument.
26818 * debug.c (do_nothing_debug_hooks): Adjust source_line and
26819 begin_prologue hooks.
26820 (debug_nothing_int_charstar_int_bool): Remove.
26821 (debug_nothing_int_int_charstar,
26822 debug_nothing_int_int_charstar_int_bool): New functions.
26823 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
26824 through to dwarf2out_source_line.
26825 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
26826 (dwarf2out_source_line): Add column argument, emit it if requested.
26827 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
26829 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
26830 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
26831 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
26832 through to dwarf2out_begin_prologue.
26833 (vmsdbgout_source_line): Add column argument, pass it through to
26834 dwarf2out_source_line.
26835 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
26836 dbxout_source_line caller.
26837 (dbxout_source_line): Add column argument.
26839 * common.opt (gno-column-info, gcolumn-info): New options.
26840 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
26841 (check_die): Also test for multiple DW_AT_decl_column attributes.
26842 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
26843 DW_AT_decl_column if requested.
26844 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
26846 (gen_variable_die): Likewise.
26847 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
26848 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
26851 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
26852 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
26853 (ix86_handle_option): Handle OPT_m3dnowa.
26854 * doc/invoke.texi (-m3dnowa): Document.
26855 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
26856 -m3dnowa instead of -m3dnow -march=athlon.
26859 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
26860 instead of gcc_assert for K, r and R code checks. Formatting fixes.
26862 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26865 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
26866 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
26867 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
26868 generator for vsx_xxpermdi_<mode>_be.
26869 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
26870 force big-endian semantics.
26871 (vsx_xxpermdi_<mode>_be): New define_expand with same
26872 implementation as previous version of vsx_xxpermdi_<mode>.
26874 2017-02-17 Jakub Jelinek <jakub@redhat.com>
26876 PR tree-optimization/79327
26877 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
26878 variable, its initialization and use.
26880 2017-02-17 Julia Koval <julia.koval@intel.com>
26882 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
26883 (OPTION_MASK_ISA_PKU_UNSET): New.
26884 (ix86_handle_option): Handle -mrdpid.
26885 * config/i386/cpuid.h (bit_RDPID): New.
26886 * config/i386/driver-i386.c (host_detect_local_cpu):
26887 Detect RDPID feature.
26888 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
26889 * config/i386/i386-c.c (ix86_target_macros_internal):
26891 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
26892 (ix86_valid_target_attribute_inner_p): Add "rdpid".
26893 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
26894 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
26895 * config/i386/i386.md (define_insn "rdpid"): New.
26896 * config/i386/i386.opt Add -mrdpid.
26897 * config/i386/immintrin.h (_rdpid_u32): New.
26899 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
26901 PR rtl-optimization/79541
26902 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
26903 instead of transforming it into USE.
26905 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
26907 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
26908 If HONOR_SNANS (SFmode) force the input to a register.
26909 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
26910 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
26911 an frsp or similar insn.
26913 2017-02-17 Martin Liska <mliska@suse.cz>
26915 PR rtl-optimization/79577
26916 * params.def (selsched-max-sched-times): Increase minimum to 1.
26918 2017-02-17 Martin Liska <mliska@suse.cz>
26920 PR rtl-optimization/79574
26921 * gcse.c (want_to_gcse_p): Prevent integer overflow.
26923 2017-02-17 Martin Liska <mliska@suse.cz>
26925 PR tree-optimization/79529
26926 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
26927 ssa_defined_default_def_p to handle cases which are implicitly
26929 * tree-ssa.c (ssa_defined_default_def_p): New function.
26930 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
26931 which are implicitly defined.
26932 * tree-ssa.h (ssa_defined_default_def_p): Declare.
26934 2017-02-17 Richard Biener <rguenther@suse.de>
26936 PR middle-end/79576
26937 * params.def (max-ssa-name-query-depth): Limit to 10.
26939 2017-02-17 Richard Biener <rguenther@suse.de>
26941 PR tree-optimization/79552
26942 * tree-ssa-structalias.c (visit_loadstore): Properly verify
26945 2017-02-17 Richard Biener <rguenther@suse.de>
26948 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
26950 2017-02-17 Marek Polacek <polacek@redhat.com>
26952 PR middle-end/79536
26953 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
26954 (fold_negate_expr): New wrapper.
26956 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
26958 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
26959 Correct terminology and de-emphasize pre-standard behavior.
26961 2017-02-16 Alan Modra <amodra@gmail.com>
26963 PR rtl-optimization/79286
26964 * ira.c (def_dominates_uses): New function.
26965 (update_equiv_regs): Don't create an equivalence for insns that
26966 may trap where the register def does not dominate the use.
26968 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
26970 PR rtl-optimization/78127
26971 * lra.c (lra): Call lra_eliminate before finish the loop after
26974 2017-02-16 Richard Biener <rguenther@suse.de>
26976 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
26978 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
26979 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
26980 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
26981 (isl_val_int_from_wi): New function.
26982 (extract_affine_gmp): Rename to ...
26983 (extract_affine_wi): ... this, take a widest_int.
26984 (extract_affine_int): Just wrap extract_affine_wi.
26985 (add_param_constraints): Use isl_val_int_from_wi.
26986 (add_loop_constraints): Likewise, and extract_affine_wi.
26988 2017-02-15 Jeff Law <law@redhat.com>
26990 PR middle-end/79521
26991 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
26992 ira_init_register_move_cost_if_necessary.
26994 2017-02-15 Martin Sebor <msebor@redhat.com>
26996 PR middle-end/32003
26997 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
26998 removed in a prior commit.
27000 2017-02-15 Bin Cheng <bin.cheng@arm.com>
27002 PR tree-optimization/79347
27003 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
27004 counters during peeling.
27006 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
27008 * Makefile.in (site.exp): Remove "set ISLVER".
27010 2017-02-15 Jakub Jelinek <jakub@redhat.com>
27013 * real.c (real_from_integer): Call real_convert even for decimal.
27015 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
27018 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
27020 2017-02-14 Andrew Pinski <apinski@cavium.com>
27022 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
27023 cores and change the partno/implementer to be correct.
27024 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
27025 the 'B" as the implementer.
27026 * config/aarch64/aarch64-tune.md: Regenerate.
27028 2017-02-14 Carl Love <cel@us.ibm.com>
27030 * config/rs6000/rs6000.c: Add case statement entry to make the
27031 xvcvuxdsp built-in argument unsigned.
27032 * config/rs6000/vsx.md: Fix the source and return operand types so they
27033 match the instruction definitions from the ISA document. Fix typo
27034 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
27037 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
27040 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
27041 member early_clobber_alts.
27042 * lra-lives.c (reg_early_clobber_p): New.
27043 (process_bb_lives): Use it.
27044 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
27045 (debug_operand_data): Initialize early_clobber_alts.
27046 (setup_operand_alternative): Set up early_clobber_alts.
27047 (collect_non_operand_hard_regs): Ditto. Pass early clobber
27048 alternatives to new_insn_reg.
27049 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
27051 (lra_update_insn_regno_info): Pass the new arg.
27053 2017-02-14 Jakub Jelinek <jakub@redhat.com>
27055 PR middle-end/79505
27056 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
27057 (new_oacc_loop_raw): Don't clear already cleared fields.
27060 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
27061 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
27062 _mm512_prefetch_i64gather_ps): New inline functions and macros.
27064 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
27067 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
27069 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
27072 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
27073 the extra instruction to the right place to store 128-bit constant
27076 2017-02-14 Martin Sebor <msebor@redhat.com>
27078 PR middle-end/79448
27079 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
27080 warning for strings of unknown length.
27082 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
27084 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
27086 2017-02-14 Jeff Law <law@redhat.com>
27089 * ira-costs.c (scan_one_insn): Initialize register move costs
27090 for pseudos seen in USE/CLOBBER insns.
27092 PR tree-optimization/79095
27093 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
27094 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
27095 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
27096 if the operands are known to be not equal, then the resulting range
27098 (intersect_ranges): If the new range is ~[0,0] and the old range is
27099 wide, then prefer ~[0,0].
27100 * tree-vrp.c (overflow_comparison_p_1): New function.
27101 (overflow_comparison_p): New function.
27102 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
27103 if NAME is used in an overflow test.
27104 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
27105 overflow check that can be expressed as an equality test, then adjust
27106 ops to be that equality test.
27108 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27110 * config/s390/s390-builtin-types.def: Remove flags argument.
27111 * config/s390/s390.c (s390_init_builtins): Likewise.
27113 2017-02-14 Martin Liska <mliska@suse.cz>
27115 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
27116 vector. Fix trailing white spaces.
27118 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
27120 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
27123 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27125 PR rtl-optimization/68664
27126 * config/arm/arm.c (arm_sched_can_speculate_insn):
27127 New function. Declare prototype.
27128 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
27130 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27132 PR rtl-optimization/68664
27133 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
27135 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
27137 2017-02-14 Amit Pawar <amit.pawar@amd.com>
27139 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
27140 max skip bytes for function, loop and jump.
27142 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27144 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
27145 ABS_EXPR for gimple dump.
27147 2017-02-14 Jakub Jelinek <jakub@redhat.com>
27150 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
27152 PR tree-optimization/79408
27153 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
27154 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
27155 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
27156 also if rhs1 is INTEGER_CST.
27158 2017-02-14 Richard Biener <rguenther@suse.de>
27160 PR middle-end/79432
27161 * tree-into-ssa.c (insert_phi_nodes): When the function can
27162 have abnormal edges rewrite SSA names with broken use-def
27163 dominance out of SSA and register them for PHI insertion.
27165 2017-02-13 Martin Sebor <msebor@redhat.com>
27167 PR middle-end/79496
27168 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
27169 clearing info.nowrite flag when snprintf size argument is a range.
27171 2017-02-13 Jakub Jelinek <jakub@redhat.com>
27173 * cprop.c (cprop_jump): Add missing space in string literal.
27174 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
27175 (get_constraint_for_component_ref): Likewise.
27176 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
27177 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
27178 * lra-constraints.c (process_alt_operands): Likewise.
27179 * ipa-inline.c (inline_small_functions): Likewise.
27180 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
27181 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
27182 * trans-mem.c (diagnose_tm_1_op): Likewise.
27183 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
27184 (grid_parallel_clauses_gridifiable): Likewise.
27186 * config/nvptx/mkoffload.c (process): Add space in between
27189 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
27190 "MOD4_SSE_REGS" and "ALL_REGS".
27192 * spellcheck.c (test_data): Add , in between "foo" and "food".
27194 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27197 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
27198 boundary crossing check and subsequent code generation agree.
27200 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27202 * config/aarch64/aarch64.c (has_memory_op): Delete.
27203 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
27206 2017-02-13 Jakub Jelinek <jakub@redhat.com>
27208 PR rtl-optimization/79388
27209 PR rtl-optimization/79450
27210 * combine.c (distribute_notes): When removing TEM_INSN for which
27211 corresponding dest has last value recorded, invalidate that last
27214 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27216 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
27217 of explicit '@'. Add missing assembly comment marker on branch costs
27220 2017-02-13 Nathan Sidwell <nathan@acm.org>
27222 * gengtype-lex.l (<in_struct>): Add '/'.
27224 2017-02-13 Martin Liska <mliska@suse.cz>
27227 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
27229 2017-02-13 Richard Biener <rguenther@suse.de>
27231 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
27233 * configure: Re-generate.
27234 * config.in: Likewise.
27235 * graphite-dependences.c: Simplify as if
27236 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
27237 * graphite-isl-ast-to-gimple.c: Likewise.
27238 * graphite-optimize-isl.c: Likewise.
27239 * graphite-poly.c: Likewise.
27240 * graphite-sese-to-poly.c: Likewise.
27241 * graphite.h: Likewise.
27242 * toplev.c: Include isl/version.h and use isl_version () for
27243 printing the ISL version.
27244 * doc/install.texi: Update ISL requirement.
27246 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
27248 * doc/standards.texi (Standards): Update reference to
27251 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
27253 * doc/extend.texi (Named Address Spaces): sourceware.org now
27255 * doc/install.texi (Binaries): Ditto.
27258 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
27260 * doc/cpp.texi: Replace "stringify"/"stringification" with C
27261 standard terminology "stringize"/"stringizing" throughout.
27262 * doc/cppinternals.texi: Likewise.
27264 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
27266 * doc/extend.texi: Fix some spelling mistakes and typos.
27267 * doc/invoke.texi: Likewise.
27269 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
27272 * params.def (inline-min-speedup) Change from 10 to 8.
27274 2017-02-11 Jakub Jelinek <jakub@redhat.com>
27276 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
27279 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
27282 * ipa-inline-analysis.c (get_minimal_bb): New function.
27283 (record_modified): Use it.
27284 (remap_edge_change_prob): Handle also ancestor functions.
27286 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
27288 * doc/contrib.texi (Contributors): Remove broken link into
27289 the Mauve CVS repository.
27291 2017-02-11 Jakub Jelinek <jakub@redhat.com>
27293 PR middle-end/79454
27294 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
27295 result computation whenever lhs doesn't have vector mode, not
27296 just when it has BLKmode.
27298 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
27300 * doc/makefile.texi (profiledbootstrap): Refer to the
27301 installation instructions only in textual form.
27303 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27306 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
27308 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
27310 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
27311 (Specific): Update mingw-w64 reference.
27313 (Specific): Remove broken link to Renesas RX processor.
27315 2017-02-10 Richard Biener <rguenther@suse.de>
27317 * toplev.c (process_options): Do not mention obsolete graphite
27318 options when printing sorry message about missing graphite support.
27319 Mention -floop-nest-optimize.
27321 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
27323 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
27324 (vtst_p16): Likewise.
27325 (vtstq_p8): Likewise.
27326 (vtstq_p16): Likewise.
27328 (vtstq_p64): Likewise.
27329 * config/arm/arm_neon.h (vgetq_lane_p64): New.
27330 (vset_lane_p64): New.
27331 (vsetq_lane_p64): New.
27333 2017-02-10 Jakub Jelinek <jakub@redhat.com>
27335 PR tree-optimization/79411
27336 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
27337 stmt operands are SSA_NAMEs used in abnormal phis.
27338 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
27341 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
27344 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
27347 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
27349 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
27351 2017-02-09 Jakub Jelinek <jakub@redhat.com>
27353 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
27357 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
27358 not arbitrary TREE_CONSTANT.
27361 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
27362 "omp declare target link" attribute unless is_global_var.
27363 * omp-offload.c (find_link_var_op): Likewise.
27365 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
27366 Chung-Lin Tang <cltang@codesourcery.com>
27368 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
27370 (gimplify_adjust_omp_clauses): Don't delete TILE.
27371 (gimplify_omp_for): Deal with TILE.
27372 * internal-fn.c (expand_GOACC_TILE): New function.
27373 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
27375 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
27376 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
27378 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
27379 avoid DIV for outermost collapse var.
27380 (expand_oacc_for): Insert tile element loop as needed. Adjust.
27381 Remove out of date comments, fix whitespace.
27382 * omp-general.c (omp_extract_for_data): Deal with tiling.
27383 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
27384 adjust OLF_DIM_BASE value.
27385 (struct omp_for_data): Add tiling field.
27386 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
27387 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
27388 for auto loops. Remove default auto determining, moved to
27389 oacc_loop_fixed_partitions.
27390 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
27391 stmts, add e_mask field.
27392 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
27393 (oacc_thread_numbers): Use oacc_dim_call.
27394 (oacc_xform_tile): New.
27395 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
27396 (finish_oacc_loop): Adjust for ifns vector.
27397 (oacc_loop_discover_walk): Append loop abstraction sites to list,
27398 add case for GOACC_TILE fns.
27399 (oacc_loop_xform_loop): Delete.
27400 (oacc_loop_process): Iterate over call list directly, and add
27401 handling for GOACC_TILE fns.
27402 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
27404 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
27405 vector partitioning to outer loops. Assign 2 partitions to loops
27406 when available. Add TILE handling.
27407 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
27408 (execite_oacc_device_lower): Process GOACC_TILE fns,
27409 ignore unknown specs.
27410 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
27411 * tree.c (omp_clause_num_ops): Adjust TILE ops.
27412 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
27414 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
27416 * configure.ac (ACX_BUGURL): Update.
27417 * configure: Regenerate.
27419 2017-02-09 Richard Biener <rguenther@suse.de>
27421 PR tree-optimization/69823
27422 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
27423 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
27425 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
27427 * config/arc/arc-c.def: Add __NPS400__ definition.
27428 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
27429 (TARGET_NPS400): Define.
27431 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
27433 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
27435 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
27436 pointer, arch_info.
27437 (arc_cpu_types): Fill the arch_info field with a pointer into the
27438 arc_arch_types table.
27439 (arc_selected_cpu): Declare.
27440 * config/arc/arc.c (arc_selected_cpu): Make global.
27441 (arc_selected_arch): Delete.
27442 (arc_base_cpu): Delete.
27443 (arc_override_options): Remove references to deleted variables,
27444 update access to arch information.
27445 (ARC_OPT): Update access to arch information.
27446 (ARC_OPTX): Likewise.
27447 * config/arc/arc.h (arc_base_cpu): Remove declaration.
27448 (TARGET_ARC600): Update access to arch information.
27449 (TARGET_ARC601): Likewise.
27450 (TARGET_ARC700): Likewise.
27451 (TARGET_EM): Likewise.
27452 (TARGET_HS): Likewise.
27453 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
27456 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
27459 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
27460 condition/operands for integer GE/LE/GEU/LEU operations.
27462 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
27464 PR translation/79397
27465 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
27468 2017-02-08 Martin Jambor <mjambor@suse.cz>
27471 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
27472 whether allocation happened.
27473 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
27474 nothing was allocated.
27476 2017-02-08 Jakub Jelinek <jakub@redhat.com>
27478 PR tree-optimization/79408
27479 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
27480 constant, but SSA_NAME with a known integer range, use the minimum
27481 of that range instead of op1 to determine if modulo can be replaced
27482 with its first operand.
27484 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27486 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
27488 2017-02-08 Richard Biener <rguenther@suse.de>
27490 PR tree-optimization/71824
27491 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
27492 Check all loops contained in the merged region.
27494 2017-02-07 Andrew Pinski <apinski@cavium.com>
27496 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
27498 2017-02-07 Andrew Pinski <apinski@cavium.com>
27500 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
27501 (thunderxt88): Likewise.
27502 (thunderxt81): Disable LSE and change v8.1 to v8.
27503 (thunderxt83): Likewise.
27505 2017-02-07 Jakub Jelinek <jakub@redhat.com>
27506 Richard Biener <rguenther@suse.de>
27508 PR middle-end/79399
27509 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
27510 type from int to size_t.
27511 * ira-costs.c (struct_costs_size): Change type from int to size_t.
27513 2017-02-07 Jakub Jelinek <jakub@redhat.com>
27515 PR rtl-optimization/79386
27516 * cprop.c (bypass_conditional_jumps): Initialize
27517 bypass_last_basic_block already before splitting bbs after
27518 unconditional traps...
27519 (bypass_conditional_jumps): ... rather than here.
27522 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
27523 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
27524 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
27525 fix -masm=intel patterns.
27527 2017-02-07 Richard Biener <rguenther@suse.de>
27529 PR tree-optimization/79256
27530 PR middle-end/79278
27531 * builtins.c (get_object_alignment_2): Use min_align_of_type
27532 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
27533 and ADJUST_FIELD_ALIGN.
27535 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
27537 * doc/tm.texi: Regenerate.
27538 * stor-layout.c (layout_decl): Adjust.
27539 (update_alignment_for_field): Likewise.
27540 (place_field): Likewise.
27541 (min_align_of_type): Likewise.
27542 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
27543 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
27544 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
27545 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
27546 * config/frv/frv.c (frv_adjust_field_align): Likewise.
27547 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
27548 * config/i386/i386.c (x86_field_alignment): Likewise.
27549 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
27550 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
27551 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
27552 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
27553 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
27554 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
27558 2017-01-30 Richard Biener <rguenther@suse.de>
27560 PR tree-optimization/79256
27561 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
27562 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
27565 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
27567 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
27568 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
27569 builtins to SImode and emit a zero-extend, if necessary.
27571 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27573 * docs/invoke.texi (RISC-V Options): Alphabetize.
27575 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27577 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
27580 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27582 * config/riscv/riscv.c: New file.
27583 * common/config/riscv/riscv-common.c: Likewise.
27584 * config.gcc: Likewise.
27585 * config/riscv/constraints.md: Likewise.
27586 * config/riscv/elf.h: Likewise.
27587 * config/riscv/generic.md: Likewise.
27588 * config/riscv/linux.h: Likewise.
27589 * config/riscv/multilib-generator: Likewise.
27590 * config/riscv/peephole.md: Likewise.
27591 * config/riscv/pic.md: Likewise.
27592 * config/riscv/predicates.md: Likewise.
27593 * config/riscv/riscv-builtins.c: Likewise.
27594 * config/riscv/riscv-c.c: Likewise.
27595 * config/riscv/riscv-ftypes.def: Likewise.
27596 * config/riscv/riscv-modes.def: Likewise.
27597 * config/riscv/riscv-opts.h: Likewise.
27598 * config/riscv/riscv-protos.h: Likewise.
27599 * config/riscv/riscv.h: Likewise.
27600 * config/riscv/riscv.md: Likewise.
27601 * config/riscv/riscv.opt: Likewise.
27602 * config/riscv/sync.md: Likewise.
27603 * config/riscv/t-elf-multilib: Likewise.
27604 * config/riscv/t-linux: Likewise.
27605 * config/riscv/t-linux-multilib: Likewise.
27606 * config/riscv/t-riscv: Likewise.
27607 * configure.ac: Likewise.
27608 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
27609 Waterman as RISC-V maintainers.
27610 * doc/install.texi: Add RISC-V entries.
27611 * doc/invoke.texi: Add RISC-V options section.
27612 * doc/md.texi: Add RISC-V constraints section.
27613 * configure: Regenerated.
27615 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
27618 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
27619 false values to be constant vectors with all 0 or all 1 bits set.
27620 (vcondu<mode><mode>): Likewise.
27621 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
27623 (fpmask_comparison_operator): Update comment.
27624 (vecint_comparison_operator): New predicate.
27625 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
27626 vector conditionals when the true and false values are constant
27627 vectors with all 0 bits or all 1 bits set.
27629 2017-02-06 Martin Sebor <msebor@redhat.com>
27631 PR tree-optimization/79376
27632 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
27634 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
27636 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
27637 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
27638 to simplify split condition.
27640 2017-02-06 Jakub Jelinek <jakub@redhat.com>
27642 * omp-expand.c (oxpand_omp_atomic_fetch_op,
27643 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
27646 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
27648 PR rtl-optimization/68664
27649 * target.def (can_speculate_insn): New hook.
27650 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
27651 * doc/tm.texi: Regenerate.
27652 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
27653 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
27654 (rs6000_sched_can_speculate_insn): New function.
27656 2017-02-06 Jakub Jelinek <jakub@redhat.com>
27658 PR tree-optimization/79284
27659 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
27660 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
27661 vectorizable_mask_load_store, vectorizable_operation,
27662 vect_is_simple_cond, get_same_sized_vectype): Use it instead
27663 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
27664 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
27665 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
27666 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
27667 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
27668 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
27669 is_gimple_assign (stmt). Replace another such test with
27670 is_gimple_assign (stmt).
27672 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
27675 * config/avr/avr.c (rtl-iter.h): Include it.
27676 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
27677 (avr_legitimate_combined_insn): ...and implementation.
27679 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
27681 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
27682 * config/s390/s390.c (s390_const_operand_ok)
27683 (s390_canonicalize_comparison, s390_extract_part)
27684 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
27685 (s390_contiguous_bitmask_p, s390_rtx_costs)
27686 (legitimize_pic_address): Likewise.
27687 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
27688 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
27689 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
27690 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
27691 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
27693 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
27695 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
27696 REGNO($0) == REGNO($1).
27698 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27700 * config/s390/linux.h(SIZE_TYPE): Add comment.
27702 2017-02-06 Julian Brown <julian@codesourcery.com>
27703 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27704 Virendra Pathak <virendra.pathak@broadcom.com>
27706 * config/aarch64/aarch64-cores.def: Change the scheduler
27708 * config/aarch64/aarch64.md: Include thunderx2t99.md.
27709 * config/aarch64/thunderx2t99.md: New file.
27711 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27713 * doc/standards.texi (Go Language): Update link to language
27716 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
27718 * tree-eh.c (lower_resx): Sanitize profile.
27719 (cleanup_empty_eh_move_lp): Likewise.
27721 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
27724 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
27726 * cfgloopmanip.h (loop_version): Update prototype.
27727 * modulo-sched.c (sms_schedule): Update call of loop_version.
27728 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
27729 * tree-parloops.c (gen_parallel_loop): Likewise.
27730 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
27731 * tree-ssa-loop-split.c (split_loop): Likewise.
27732 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
27733 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
27735 2017-02-05 Martin Liska <mliska@suse.cz>
27738 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
27740 (print_operand_address): Initialize a struct to zero.
27742 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27744 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
27745 garbage collector only in textual form.
27747 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27749 * doc/extend.texi (x86 specific memory model extensions for
27750 transactional memory): Simplify a phrase.
27752 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
27755 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
27756 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
27757 (atomic_storedi_1): Likewise.
27759 2017-02-04 Jakub Jelinek <jakub@redhat.com>
27761 PR tree-optimization/79338
27762 * tree-parloops.c (gather_scalar_reductions): Don't call
27763 vect_analyze_loop_form for loop->inner before destroying loop's
27766 2017-02-03 Martin Sebor <msebor@redhat.com>
27768 PR tree-optimization/79327
27769 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
27770 when precision has resulted in leading zeros.
27771 (format_integer): Adjust the likely counter to assume an unknown
27772 argument that may be zero is non-zero.
27774 2017-02-03 Jason Merrill <jason@redhat.com>
27777 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
27778 avoid copying non-taken branch.
27780 2017-02-03 Jakub Jelinek <jakub@redhat.com>
27782 PR tree-optimization/79340
27783 * tree-vect-loop.c (vectorizable_reduction): Release
27784 vec_defs elements after safe_splicing them into other vectors.
27787 PR tree-optimization/79327
27788 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
27789 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
27791 (format_integer): Use wide_int_to_tree instead of build_int_cst
27792 + to_?hwi. If argmin is NULL, just set argmin and argmax to
27793 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
27794 of shortest and longest sequence.
27796 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
27798 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
27799 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
27801 2017-02-03 Walter Lee <walt@tilera.com>
27804 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
27805 after initial stackframe link reg save.
27806 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
27808 2017-02-03 Jakub Jelinek <jakub@redhat.com>
27811 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
27812 wu for stxssp alternative.
27814 2017-02-03 Martin Sebor <msebor@redhat.com>
27816 PR tree-optimization/79352
27817 * gimple-fold.c (get_range_strlen): Add argument.
27818 (get_range_strlen): Change return type to bool.
27819 (get_maxval_strlen): Pass in a dummy argument.
27820 * gimple-fold.h (get_range_strlen): Change return type to bool.
27821 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
27822 * tree.h (array_at_struct_end_p): Add argument.
27823 * tree.c (array_at_struct_end_p): Handle it.
27825 2017-02-03 Martin Liska <mliska@suse.cz>
27828 * multiple_target.c (create_dispatcher_calls): Redirect edge
27829 from a caller of a dispatcher.
27830 (expand_target_clones): Make the clones local.
27831 (ipa_target_clone): Do both target clones and resolvers.
27832 (ipa_dispatcher_calls): Remove the pass.
27833 (pass_dispatcher_calls::gate): Likewise.
27834 (make_pass_dispatcher_calls): Likewise.
27835 * passes.def (pass_target_clone): Put as very first IPA early
27838 2017-02-03 Martin Liska <mliska@suse.cz>
27840 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
27841 in case of a function with ifunc attribute.
27843 2017-02-03 Martin Liska <mliska@suse.cz>
27845 * cgraph.c (cgraph_node::dump): Dump function version info.
27846 * symtab.c (symtab_node::dump_base): Add missing new line.
27848 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
27850 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
27851 (ifcombine_ifandif): Use it.
27853 2017-02-03 Martin Liska <mliska@suse.cz>
27855 * doc/invoke.texi: Document default value for
27856 use-after-scope-direct-emission-threshold.
27858 2017-02-03 Martin Liska <mliska@suse.cz>
27860 PR tree-optimization/79339
27861 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
27862 (format_floating): Likewise.
27864 2017-02-03 Martin Liska <mliska@suse.cz>
27867 * ipa-prop.c (ipa_node_params_t::insert): Remove current
27869 (ipa_node_params_t::remove): Likewise.
27870 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
27871 initialization from removed ipa_node_params_t::insert.
27872 (ipa_node_params::~ipa_node_params): Move from removed
27873 ipa_node_params_t::release.
27874 * symbol-summary.h (symbol_summary::m_released): New member.
27875 Do not release a summary twice. Do not allow to call finalizer
27876 for types of a summary that live in GGC memory.
27878 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27880 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
27883 2017-02-02 Martin Sebor <msebor@redhat.com>
27885 PR middle-end/79275
27886 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
27887 (format_string): Tighten up the range of output for non-constant
27888 strings and correct the expected range for wide non-constant strings.
27890 2017-02-02 Martin Sebor <msebor@redhat.com>
27892 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
27894 PR middle-end/32003
27895 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
27897 (-fdump-tree-@var): Add to index and document how to come up
27898 with pass-specific option and dump file names.
27899 (-fdump-passes): Clarify where to look for output.
27901 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
27903 PR middle-end/77445
27904 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
27905 statistics of the analyzed path; allow threading for speed when
27906 any of BBs along the path are optimized for speed.
27908 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
27910 PR middle-end/78468
27911 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
27912 settings of the virtual registers.
27915 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
27917 * explow.c (get_dynamic_stack_size): Take known alignment of stack
27918 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
27921 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27923 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
27924 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
27926 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27928 * config/s390/s390.md: Add missing comments with the expanded
27930 * config/s390/vector.md: Likewise.
27931 * config/s390/vx-builtins.md: Likewise.
27933 2017-02-02 Jakub Jelinek <jakub@redhat.com>
27936 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
27937 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
27938 conditions on a single line.
27940 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27942 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
27943 __S390_VX__ to __VX__.
27945 2017-02-01 Andrew Pinski <apinski@cavium.com>
27947 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
27948 stmt_info to record_stmt_cost.
27949 (vect_get_known_peeling_cost): Pass stmt_info if known to
27951 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
27952 cpu_vector_cost field into
27953 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
27954 field into vec_int_stmt_cost and vec_fp_stmt_cost.
27955 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
27956 splitting of scalar_stmt_cost and vec_stmt_cost.
27957 (thunderx_vector_cost): Likewise.
27958 (cortexa57_vector_cost): LIkewise.
27959 (exynosm1_vector_cost): Likewise.
27960 (xgene1_vector_cost): Likewise.
27961 (thunderx2t99_vector_cost): Improve after the splitting of the two
27963 (aarch64_builtin_vectorization_cost): Update for the splitting of
27964 scalar_stmt_cost and vec_stmt_cost.
27966 2017-02-01 Torvald Riegel <triegel@redhat.com>
27967 Richard Henderson <rth@redhat.com>
27969 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
27970 conditional on existance of a fast atomic load.
27971 * optabs-query.c (can_atomic_load_p): New function.
27972 * optabs-query.h (can_atomic_load_p): Declare it.
27973 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
27974 no fast atomic load is available for the particular size of access.
27975 (expand_atomic_compare_and_swap): Likewise.
27976 (expand_atomic_load): Likewise.
27977 (expand_atomic_store): Likewise.
27978 (expand_atomic_fetch_op): Likewise.
27979 * testsuite/lib/target-supports.exp
27980 (check_effective_target_sync_int_128): Remove x86 because it provides
27981 no fast atomic load.
27982 (check_effective_target_sync_int_128_runtime): Likewise.
27984 2017-02-01 Richard Biener <rguenther@suse.de>
27986 * graphite.c: Include tree-vectorizer.h for find_loop_location.
27987 (graphite_transform_loops): Provide opt-info for optimized nests.
27988 * tree-parloop.c (parallelize_loops): Provide opt-info for
27989 parallelized loops.
27991 2017-02-01 Richard Biener <rguenther@suse.de>
27993 PR middle-end/79315
27994 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
27995 was not set before.
27997 2017-02-01 Richard Biener <rguenther@suse.de>
27999 PR tree-optimization/71824
28000 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
28001 Verify the loops are valid in the merged SESE region.
28002 (scop_detection::can_represent_loop_1): Check analyzing the
28003 evolution of the number of iterations in the region succeeds.
28005 2017-01-31 Ian Lance Taylor <iant@golang.org>
28007 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
28008 REG_ARGS_SIZE note to 32-bit push insns and call insn.
28010 2017-01-31 David Malcolm <dmalcolm@redhat.com>
28012 PR preprocessor/79210
28013 * input.c (get_substring_ranges_for_loc): Replace line_width
28014 assertion with error-handling.
28016 2017-01-31 Richard Biener <rguenther@suse.de>
28018 PR tree-optimization/77318
28019 * graphite-sese-to-poly.c (extract_affine): Fix assert.
28020 (create_pw_aff_from_tree): Take loop parameter.
28021 (add_condition_to_pbb): Pass loop of the condition to
28022 create_pw_aff_from_tree.
28024 2017-01-31 Jakub Jelinek <jakub@redhat.com>
28026 * config/s390/s390.c (s390_asan_shadow_offset): New function.
28027 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
28029 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
28033 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
28035 (convert_int_to_float128): Likewise.
28036 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
28037 (convert_int_to_float128): Likewise.
28038 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
28039 (UNSPEC_IEEE128_CONVERT): Likewise.
28040 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
28041 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
28042 Use local variables for IBM extended format.
28043 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
28044 (fix_trunc<mode>si2_fprs): Likewise.
28045 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
28046 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
28047 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
28048 to know that we can now have integers of all sizes in vector
28050 (fix<uns>_<mode>di2_hw): Likewise.
28051 (float<uns>_<mode>si2_hw): Likewise.
28052 (fix_<mode>si2_hw): Likewise.
28053 (fixuns_<mode>si2_hw): Likewise.
28054 (float<uns>_<mode>di2_hw): Likewise.
28055 (float_<mode>di2_hw): Likewise.
28056 (float_<mode>si2_hw): Likewise.
28057 (floatuns_<mode>di2_hw): Likewise.
28058 (floatuns_<mode>si2_hw): Likewise.
28059 (xscvqp<su>wz_<mode>): Delete, no longer used.
28060 (xscvqp<su>dz_<mode>): Likewise.
28061 (xscv<su>dqp_<mode>): Likewise.
28062 (ieee128_mfvsrd_64bit): Likewise.
28063 (ieee128_mfvsrd_32bit): Likewise.
28064 (ieee128_mfvsrwz): Likewise.
28065 (ieee128_mtvsrw): Likewise.
28066 (ieee128_mtvsrd_64bit): Likewise.
28067 (ieee128_mtvsrd_32bit): Likewise.
28069 2017-01-31 Martin Liska <mliska@suse.cz>
28072 * ipa-prop.c (ipa_free_all_node_params): Call release method
28073 instead of ~sumbol_summary to not to trigger double times
28076 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
28078 PR tree-optimization/71691
28079 * bitmap.h (class auto_bitmap): New.
28080 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
28081 is_maybe_undefined instead of ssa_undefined_value_p.
28083 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28085 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
28086 __S390_ARCH_LEVEL__ to __ARCH__.
28088 2017-01-31 Jakub Jelinek <jakub@redhat.com>
28090 PR tree-optimization/79267
28091 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
28092 if should_remove_lhs_p is true.
28094 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
28097 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
28098 (add_alignment_attribute): New.
28099 (base_type_die): Add alignment attribute.
28100 (subrange_type_die): Likewise.
28101 (modified_type_die): Likewise.
28102 (gen_array_type_die): Likewise.
28103 (gen_descr_array_type_die: Likewise.
28104 (gen_enumeration_type_die): Likewise.
28105 (gen_subprogram_die): Likewise.
28106 (gen_variable_die): Likewise.
28107 (gen_field_die): Likewise.
28108 (gen_ptr_to_mbr_type_die): Likewise.
28109 (gen_struct_or_union_type_die): Likewise.
28110 (gen_subroutine_type_die): Likewise.
28111 (gen_typedef_die): Likewise.
28112 (base_type_cmp): Compare alignment attribute.
28114 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28117 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
28118 (setb_unsigned) New pattern for setb with CCUNS.
28119 * config/rs6000/rs6000.c (expand_block_compare): Use a different
28120 subfc./subfe sequence to avoid overflow problems. Generate a
28121 shorter sequence with cmpld/setb for power9.
28122 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
28123 for generating subfc. instruction.
28124 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
28125 now uses this instruction.
28127 2017-01-30 Ian Lance Taylor <iant@google.com>
28130 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
28131 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
28133 2017-01-30 Martin Sebor <msebor@redhat.com>
28135 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
28136 Move constant to the right of a relational operator.
28137 (get_mpfr_format_length, format_character, format_string): Ditto.
28138 (should_warn_p, maybe_warn): Same.
28140 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
28142 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
28145 * asan.c (get_translation_unit_decl): Remove function.
28146 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
28148 2017-01-30 Martin Liska <mliska@suse.cz>
28150 PR gcov-profile/79259
28151 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
28152 -fprofile-generate.
28154 2017-01-30 Martin Liska <mliska@suse.cz>
28157 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
28158 Initialize variables with NULL value.
28160 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
28163 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
28165 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
28167 2017-01-30 Richard Biener <rguenther@suse.de>
28169 PR tree-optimization/79276
28170 * tree-vrp.c (process_assert_insertions): Properly adjust common
28171 when removing a duplicate.
28173 * gcc.dg/torture/pr79276.c: New testcase.
28175 2017-01-30 Richard Biener <rguenther@suse.de>
28177 PR tree-optimization/79256
28178 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
28179 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
28181 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
28183 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
28186 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
28187 ("*r<noxa>sbg_<mode>_sll_bitmask")
28188 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
28189 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
28190 Use contiguous_bitmask_nowrap_operand.
28192 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28195 * config/rs6000/altivec.h (vec_xl): Revise #define.
28196 (vec_xst): Likewise.
28198 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
28200 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
28202 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
28204 PR rtl-optimization/79194
28205 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
28206 traps before call to bypass_conditional_jumps.
28208 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
28210 PR tree-optimization/71374
28211 * lra-constraints.c (check_conflict_input_operands): New.
28212 (match_reload): Use it.
28214 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
28217 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
28218 account to calculate conflict_set.
28220 2017-01-27 Bin Cheng <bin.cheng@arm.com>
28222 PR rtl-optimization/78559
28223 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
28224 other_insn in combine.
28226 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
28228 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
28229 uint16_type_node for BT_UINT16.
28231 2017-01-27 David Malcolm <dmalcolm@redhat.com>
28233 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
28234 "RTL Tests" to menu.
28235 (GIMPLE Tests): New node.
28236 (RTL Tests): New node.
28238 2017-01-27 Richard Biener <rguenther@suse.de>
28240 PR tree-optimization/79245
28241 * tree-loop-distribution.c (distribute_loop): Apply cost
28242 modeling also to detected patterns.
28244 2017-01-27 Richard Biener <rguenther@suse.de>
28246 PR tree-optimization/71433
28247 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
28248 (compare_assert_loc): New function.
28249 (process_assert_insertions): Sort and optimize assert locations
28250 to remove duplicates and push down identical assertions on
28251 edges to their destination block.
28253 2017-01-27 Richard Biener <rguenther@suse.de>
28255 PR tree-optimization/79244
28256 * tree-vrp.c (remove_range_assertions): Forcefully propagate
28257 out SSA names even if abnormal.
28259 2017-01-27 Jakub Jelinek <jakub@redhat.com>
28261 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
28262 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
28263 instead of MPFR_RNDN.
28265 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
28268 * arm.c (arm_option_override): Don't call build_target_option_node
28269 until after doing all option overrides.
28270 (arm_valid_target_attribute_tree): Likewise.
28272 2017-01-27 Martin Liska <mliska@suse.cz>
28274 * doc/invoke.texi (-fprofile-arcs): Document profiling support
28275 for {cd}tors and C++ {cd}tors.
28277 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
28279 * config/s390/s390.md ("*setmem_long_and")
28280 ("*setmem_long_and_31z"): Use zero_extend instead of and.
28282 2017-01-26 Martin Sebor <msebor@redhat.com>
28284 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
28287 2017-01-26 Martin Sebor <msebor@redhat.com>
28289 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
28290 HAVE_DFmode before using XFmode or DFmode.
28291 (parse_directive): Avoid using the z length modifier to avoid
28292 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
28294 PR middle-end/78703
28295 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
28296 to accept adjustment as an array.
28297 (get_int_range): New function.
28298 (struct directive): Make width and prec arrays.
28299 (directive::set_width, directive::set_precision): Call get_int_range.
28300 (format_integer, format_floating): Handle width and precision ranges.
28301 (format_string, parse_directive): Same.
28303 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28306 * dwarf2out.c (generate_skeleton_bottom_up): For children with
28307 comdat_type_p set, just clone them, but keep the children in the
28311 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
28312 which have direct callers with -fvar-tracking-assignments enabled
28314 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
28315 inside of type units.
28317 2017-01-26 Martin Sebor <msebor@redhat.com>
28319 PR middle-end/78703
28320 * gimple-ssa-sprintf.c (struct result_range): Add likely and
28322 (struct format_result): Replace number_chars, number_chars_min,
28323 and number_chars_max with a single member of struct result_range.
28325 (format_result::operator+=): Adjust.
28326 (struct fmtresult): Remove bounded. Handle likely and unlikely
28328 (fmtresult::adjust_for_width_or_precision): New function.
28329 (fmtresult:type_max_digits): New function.
28330 (bytes_remaining): Handle likely and unlikely counters.
28331 (min_bytes_remaining): Remove.
28332 (format_percent): Simplify.
28333 (format_integer, format_floating): Set likely and unlikely counters.
28334 (get_string_length, format_character, format_string): Same.
28335 (format_plain, should_warn_p): New function.
28336 (maybe_warn): Call should_warn_p. Update diagnostic messages
28337 and handle those for all directives, including plain strings.
28338 (format_directive): Handle likely and unlikely counters.
28339 Remove unnecessary quoting from diagnostics. Add an informational
28341 (add_bytes): Remove.
28342 (pass_sprintf_length::compute_format_length): Simplify.
28343 (try_substitute_return_value): Handle likely and unlikely counters.
28345 2017-01-26 Carl Love <cel@us.ibm.com>
28347 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
28348 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
28350 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
28353 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
28354 endianess for subregs into account.
28355 * lra-constraints.c (lra_constraints): Do risky transformations
28356 always on the first iteration.
28357 * lra-lives.c (check_pseudos_live_through_calls): Add arg
28358 last_call_used_reg_set.
28359 (process_bb_lives): Define and use last_call_used_reg_set.
28360 * lra.c (lra): Always continue after lra_constraints on the first
28363 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
28365 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
28367 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
28369 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28371 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
28372 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
28373 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
28374 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
28375 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
28376 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
28377 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
28378 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
28379 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
28381 2017-01-26 Marek Polacek <polacek@redhat.com>
28384 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
28385 for the third operand.
28387 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28389 PR middle-end/79236
28390 * omp-low.c (struct omp_context): Add simt_stmt field.
28391 (scan_omp_for): Return omp_context *.
28392 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
28393 context to the _simt_ SIMD stmt.
28394 (lower_omp_for): For combined SIMD with sibling _simt_
28395 SIMD, make sure to use the same decls in _looptemp_
28396 clauses as in the sibling.
28398 2017-01-26 David Sherwood <david.sherwood@arm.com>
28400 PR middle-end/79212
28401 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
28404 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28407 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
28408 emit fld b; fld a; if possible.
28410 * brig-builtins.def: Update copyright years.
28411 * config/arm/arm_acle_builtins.def: Update copyright years.
28413 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
28416 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
28417 constraint instead of o for the stxsd instruction.
28419 2017-01-25 Carl Love <cel@us.ibm.com>
28421 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
28422 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
28424 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
28426 * doc/invoke.texi (C++ Dialect Options): Fix typo.
28428 2017-01-25 Richard Biener <rguenther@suse.de>
28430 PR tree-optimization/69264
28431 * target.def (vector_alignment_reachable): Improve documentation.
28432 * doc/tm.texi: Regenerate.
28433 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
28435 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
28436 earlier changes with respect to TYPE_USER_ALIGN.
28437 (vector_alignment_reachable_p): Likewise. Improve dumping.
28439 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28442 * config/arm/arm.md (xordi3): Force constant operand into a register
28445 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28447 * doc/invoke.texi (-fstore-merging): Correct default optimization
28448 levels at which it is enabled.
28449 (-O): Move -fstore-merging from list to...
28452 2017-01-25 Richard Biener <rguenther@suse.de>
28455 * omp-expand.c: Include debug.h.
28456 (expand_omp_taskreg): Make sure to generate early debug before
28457 outlining anything from a function.
28458 (expand_omp_target): Likewise.
28459 (grid_expand_target_grid_body): Likewise.
28461 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
28464 * asan.c (get_translation_unit_decl): New function.
28465 (asan_add_global): Extract modules file name from globals
28466 TRANSLATION_UNIT_DECL name.
28468 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
28471 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
28472 for long calls with APCS frame and VFP.
28474 2017-01-24 David Malcolm <dmalcolm@redhat.com>
28476 * cfg.c (original_copy_tables_initialized_p): New function.
28477 * cfg.h (original_copy_tables_initialized_p): New decl.
28478 * cfgrtl.c (relink_block_chain): Guard the call to
28479 free_original_copy_tables with a call to
28480 original_copy_tables_initialized_p.
28481 * cgraph.h (symtab_node::native_rtl_p): New decl.
28482 * cgraphunit.c (symtab_node::native_rtl_p): New function.
28483 (symtab_node::needed_p): Don't assert for early assembly output
28484 for __RTL functions.
28485 (cgraph_node::finalize_function): Set "force_output" for __RTL
28487 (cgraph_node::analyze): Bail out early for __RTL functions.
28488 (analyze_functions): Update assertion to support __RTL functions.
28489 (cgraph_node::expand): Bail out early for __RTL functions.
28490 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
28492 * function.h (struct function): Update comment for field
28494 * gimple-expr.c: Include "tree-pass.h".
28495 (gimple_has_body_p): Return false for __RTL functions.
28496 * Makefile.in (OBJS): Add run-rtl-passes.o.
28497 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
28499 (gcc::pass_manager::get_clean_slate): New accessor.
28500 * passes.c: Include "insn-addr.h".
28501 (should_skip_pass_p): Add logging. Update logic for running
28502 "expand" to be compatible with both __GIMPLE and __RTL. Guard
28503 property-provider override so it is only done for gimple passes.
28505 (skip_pass): New function.
28506 (execute_one_pass): Call skip_pass when skipping passes.
28507 * read-md.c (md_reader::read_char): Support filtering
28508 the input to a subset of line numbers.
28509 (md_reader::md_reader): Initialize fields
28510 m_first_line and m_last_line.
28511 (md_reader::read_file_fragment): New function.
28512 * read-md.h (md_reader::read_file_fragment): New decl.
28513 (md_reader::m_first_line): New field.
28514 (md_reader::m_last_line): New field.
28515 * read-rtl-function.c (function_reader::create_function): Only
28516 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
28517 curr_properties. Set DECL_INITIAL to a dummy block.
28518 (read_rtl_function_body_from_file_range): New function.
28519 * read-rtl-function.h (read_rtl_function_body_from_file_range):
28521 * run-rtl-passes.c: New file.
28522 * run-rtl-passes.h: New file.
28524 2017-01-24 Jeff Law <law@redhat.com>
28526 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
28529 2017-01-24 Bin Cheng <bin.cheng@arm.com>
28531 PR tree-optimization/79159
28532 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
28533 (record_nonwrapping_iv): Improve boundary using above function if no
28534 value range information.
28536 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
28537 Martin Jambor <mjambor@suse.cz>
28539 * brig-builtins.def: New file.
28540 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
28541 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
28542 (DEF_HSAIL_SAT_BUILTIN): Likewise.
28543 (DEF_HSAIL_INTR_BUILTIN): Likewise.
28544 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
28545 * builtin-types.def (BT_INT8): New.
28546 (BT_INT16): Likewise.
28547 (BT_UINT8): Likewise.
28548 (BT_UINT16): Likewise.
28549 (BT_FN_ULONG): Likewise.
28550 (BT_FN_UINT_INT): Likewise.
28551 (BT_FN_UINT_ULONG): Likewise.
28552 (BT_FN_UINT_LONG): Likewise.
28553 (BT_FN_UINT_PTR): Likewise.
28554 (BT_FN_ULONG_PTR): Likewise.
28555 (BT_FN_INT8_FLOAT): Likewise.
28556 (BT_FN_INT16_FLOAT): Likewise.
28557 (BT_FN_UINT32_FLOAT): Likewise.
28558 (BT_FN_UINT16_FLOAT): Likewise.
28559 (BT_FN_UINT8_FLOAT): Likewise.
28560 (BT_FN_UINT64_FLOAT): Likewise.
28561 (BT_FN_UINT16_UINT32): Likewise.
28562 (BT_FN_UINT32_UINT16): Likewise.
28563 (BT_FN_UINT16_UINT16_UINT16): Likewise.
28564 (BT_FN_INT_PTR_INT): Likewise.
28565 (BT_FN_UINT_PTR_UINT): Likewise.
28566 (BT_FN_LONG_PTR_LONG): Likewise.
28567 (BT_FN_ULONG_PTR_ULONG): Likewise.
28568 (BT_FN_VOID_UINT64_UINT64): Likewise.
28569 (BT_FN_UINT8_UINT8_UINT8): Likewise.
28570 (BT_FN_INT8_INT8_INT8): Likewise.
28571 (BT_FN_INT16_INT16_INT16): Likewise.
28572 (BT_FN_INT_INT_INT): Likewise.
28573 (BT_FN_UINT_FLOAT_UINT): Likewise.
28574 (BT_FN_FLOAT_UINT_UINT): Likewise.
28575 (BT_FN_ULONG_UINT_UINT): Likewise.
28576 (BT_FN_ULONG_UINT_PTR): Likewise.
28577 (BT_FN_ULONG_ULONG_ULONG): Likewise.
28578 (BT_FN_UINT_UINT_UINT): Likewise.
28579 (BT_FN_VOID_UINT_PTR): Likewise.
28580 (BT_FN_UINT_UINT_PTR: Likewise.
28581 (BT_FN_UINT32_UINT64_PTR): Likewise.
28582 (BT_FN_INT_INT_UINT_UINT): Likewise.
28583 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
28584 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
28585 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
28586 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
28587 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
28588 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
28589 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
28590 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
28591 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
28592 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
28593 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
28594 * doc/frontends.texi: List BRIG FE.
28595 * doc/install.texi (Testing): Add BRIG tesring requirements.
28596 * doc/invoke.texi (Overall Options): Mention BRIG.
28597 * doc/standards.texi (Standards): Doucment BRIG HSA version.
28599 2017-01-24 Richard Biener <rguenther@suse.de>
28601 PR translation/79208
28602 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
28604 2017-01-24 Martin Jambor <mjambor@suse.cz>
28607 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
28608 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
28609 and known_contexts.
28611 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
28613 PR middle-end/79123
28614 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
28615 casts from signed to unsigned really don't have a range.
28617 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
28619 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
28620 GMP_RNDx for compatiblity.
28622 2017-01-24 Martin Liska <mliska@suse.cz>
28625 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
28626 that would prevent us to call alloca with -1 as argument.
28628 2017-01-24 Jakub Jelinek <jakub@redhat.com>
28630 * dwarf2out.c (output_compilation_unit_header, output_file_names):
28631 Avoid -Wformat-security warning.
28633 2017-01-23 Andrew Pinski <apinski@cavium.com>
28635 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
28638 2017-01-23 Martin Sebor <msebor@redhat.com>
28640 PR middle-end/78703
28641 * gimple-ssa-sprintf.c (warn_level): New global.
28642 (format_integer): Use it here and throughout the rest of the file.
28643 Use the same switch to compute sign as base.
28644 (maybe_warn): New function.
28645 (format_directive): Factor out warnings into maybe_warn.
28646 Add debugging output. Use warn_level.
28647 (add_bytes): Use warn_level.
28648 (pass_sprintf_length::compute_format_length): Add debugging output.
28649 (try_substitute_return_value): Same.
28650 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
28652 PR middle-end/78703
28653 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
28654 (struct fmtresult, format_integer, format_floating): Adjust.
28655 (fmtresult::fmtresult): Set max correctly in two argument ctor.
28656 (get_string_length, format_string,format_directive): Same.
28657 (pass_sprintf_length::compute_format_length): Same.
28658 (try_substitute_return_value): Simplify slightly.
28660 PR middle-end/78703
28661 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
28662 (fmtresult::operator+=): Outlined.
28663 (struct fmtresult): Add ctors.
28664 (struct conversion_spec): Rename...
28665 (struct directive): ...to this. Add and remove data members.
28666 (directive::set_width, directive::set_precision): New functions.
28667 (format_percent): Use fmtresult ctor.
28668 (get_width_and_precision): Remove.
28669 (format_integer): Make naming changes. Avoid computing width and
28671 (format_floating): Same. Adjust indentation.
28672 (format_character, format_none): New functions.
28673 (format_string): Moved character handling to format_character.
28674 (format_directive): Remove arguments, change return type.
28675 (parse_directive): New function.
28676 (pass_sprintf_length::compute_format_length): Move directive
28677 parsing to parse_directive.
28679 2017-01-23 Jakub Jelinek <jakub@redhat.com>
28681 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
28682 (assign_assembler_name_if_needed): ... this.
28683 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
28684 (assign_assembler_name_if_needed): ... this.
28685 (free_lang_data_in_cgraph): Adjust callers.
28686 * cgraphunit.c (cgraph_node::analyze): Likewise.
28687 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
28690 2017-01-23 Richard Biener <rguenther@suse.de>
28692 PR tree-optimization/79088
28693 PR tree-optimization/79188
28694 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
28695 resetting loop bounds after last path deletion. Reset loop
28696 bounds of the target loop, make code match the comments.
28697 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
28698 Make sure loops need no fixups.
28700 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28702 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
28703 exponent support with double type for first argument.
28704 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
28705 type returned by __builtin_vec_extract_sig,
28706 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
28707 functions from "vector int" to "vector unsigned int" or from
28708 "vector long long int" to "vector unsigned long long int".
28709 Changed type returned by __builtin_vec_extract_exp,
28710 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
28711 functions from "vector int" to "vector unsigned int" or from
28712 "vector long long int" to "vector unsigned long long int".
28713 Changed return type of __builtin_vec_test_data_class,
28714 __builtin_vec_test_data_class_sp, and
28715 __builtin_vec_test_data_class_dp from "vector int" to
28716 "vector bool int" or from "vector long long int" to "vector bool
28717 long long int" and changed second argument type from "unsigned
28718 int" to "int". Added new overloaded function forms "vector float
28719 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
28720 "vector float __builtin_vec_insert_exp_sp (vector float, vector
28721 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
28722 double, vector unsigned long long int)" and "vector double
28723 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
28724 long int)". Changed return type of
28725 __builtin_scalar_test_data_class and
28726 __builtin_scalar_test_data_class_sp and
28727 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
28728 int" and changed second argument from "unsigned int" to "int".
28729 Changed type returned by __builtin_scalar_test_neg,
28730 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
28731 from "int" to "bool int". Added new overloaded function form
28732 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
28733 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
28734 exponent double-precision with floating point first argument.
28735 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
28736 documentation of scalar_test_data_class, scalar_test_neg,
28737 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
28738 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
28739 vec_test_data_class built-in functions to reflect refinements in
28740 their type signatures.
28742 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
28744 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
28746 (aarch64_elf_asm_destructor): Likewise.
28748 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
28750 PR rtl-optimization/78634
28751 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
28752 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
28753 * ifcvt.c (noce_try_cmove): Add missing cost check.
28755 PR rtl-optimization/71724
28756 * combine.c (if_then_else_cond): Look for situations where it is
28757 beneficial to undo the work of one of the recursive calls.
28759 2017-01-23 Bin Cheng <bin.cheng@arm.com>
28761 PR tree-optimization/70754
28762 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
28763 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
28764 combined stmt before it if not NULL.
28765 (combine_chains): Process refs reversely and compute dominance point
28768 2017-01-23 Martin Liska <mliska@suse.cz>
28770 PR tree-optimization/79196
28771 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
28772 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
28774 (strlen_optimize_stmt): Call the renamed function.
28776 2017-01-23 Michael Matz <matz@suse.de>
28778 PR tree-optimization/78384
28779 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
28781 2017-01-23 Richard Biener <rguenther@suse.de>
28783 PR tree-optimization/79186
28784 * tree-vrp.c (register_new_assert_for): Make sure we've seen
28785 both incoming edges before moving an assert.
28787 2017-01-23 Martin Jambor <mjambor@suse.cz>
28789 * ipa-prop.c (load_from_param_1): Removed.
28790 (load_from_unmodified_param): Bits from load_from_param_1 put back
28792 (load_from_param): Removed.
28793 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
28794 with stmt. Reverted back to use of load_from_unmodified_param.
28796 2017-01-23 Martin Jambor <mjambor@suse.cz>
28799 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
28800 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
28801 field a pointer to garbage collected vector, mark lattices and
28802 ipcp_orig_node with GTY((skip)).
28803 (ipa_get_param_count): Adjust to descriptors being a pointer.
28804 (ipa_get_param): Likewise.
28805 (ipa_get_type): Likewise.
28806 (ipa_get_param_move_cost): Likewise.
28807 (ipa_set_param_used): Likewise.
28808 (ipa_get_controlled_uses): Likewise.
28809 (ipa_set_controlled_uses): Likewise.
28810 (ipa_is_param_used): Likewise.
28811 (ipa_node_params_t): Move into garbage collector. New methods insert
28813 (ipa_node_params_sum): Annotate wth GTY(()).
28814 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
28816 (ipa_load_from_parm_agg): Adjust declaration.
28817 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
28818 * ipa-profile.c (ipa_profile): Likewise.
28819 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
28820 (ipa_populate_param_decls): Make descriptors parameter garbage
28822 (ipa_dump_param): Adjust to descriptors being a pointer.
28823 (ipa_alloc_node_params): Likewise.
28824 (ipa_initialize_node_params): Likewise.
28825 (load_from_param_1): Make descriptors parameter garbage collected.
28826 (load_from_unmodified_param): Likewise.
28827 (load_from_param): Likewise.
28828 (ipa_load_from_parm_agg): Likewise.
28829 (ipa_node_params::~ipa_node_params): Removed.
28830 (ipa_free_all_node_params): Remove call to delete operator.
28831 (ipa_node_params_t::insert): New.
28832 (ipa_node_params_t::remove): Likewise.
28833 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
28834 copy known_csts and known_contexts vectors.
28835 (ipa_read_node_info): Adjust to descriptors being a pointer.
28836 (ipcp_modif_dom_walker): Make m_descriptors field garbage
28838 (ipcp_transform_function): Make descriptors variable garbage
28841 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
28843 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
28844 * config/i386/avx512dqintrin.h: Ditto.
28845 * config/i386/avx512fintrin.h: Ditto.
28846 * config/i386/i386.c: Handle new builtins.
28847 * config/i386/i386-builtin.def: Add new builtins.
28848 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
28849 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
28851 2017-01-23 Jakub Jelinek <jakub@redhat.com>
28852 Martin Liska <mliska@suse.cz>
28854 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
28855 * asan.c (asan_expand_poison_ifn): Support stores and use
28856 appropriate ASAN report function.
28857 * internal-fn.c (expand_ASAN_POISON_USE): New function.
28858 * internal-fn.def (ASAN_POISON_USE): Declare.
28859 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
28860 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
28861 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
28862 ASAN_POISON calls w/o LHS.
28863 * tree-ssa.c (execute_update_addresses_taken): Create clobber
28864 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
28865 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
28866 * gimplify.c (asan_poison_variables): Add attribute
28867 use_after_scope_memory to variables that really needs to live
28869 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
28870 having the attribute.
28872 2017-01-23 Martin Liska <mliska@suse.cz>
28874 * asan.c (create_asan_shadow_var): New function.
28875 (asan_expand_poison_ifn): Likewise.
28876 * asan.h (asan_expand_poison_ifn): New declaration.
28877 * internal-fn.c (expand_ASAN_POISON): Likewise.
28878 * internal-fn.def (ASAN_POISON): New builtin.
28879 * sanopt.c (pass_sanopt::execute): Expand
28880 asan_expand_poison_ifn.
28881 * tree-inline.c (copy_decl_for_dup_finish): Make function
28883 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
28884 * tree-ssa.c (is_asan_mark_p): New function.
28885 (execute_update_addresses_taken): Rewrite local variables
28886 (identified just by use-after-scope as addressable) into SSA.
28888 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
28890 * doc/install.texi (Specific): opensource.apple.com uses https
28891 now. Remove trailing slash.
28893 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
28895 * README.Portability: Remove note on an Irix compatibility issue.
28897 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
28899 * gcov.c (INCLUDE_ALGORITHM): Define.
28900 (INCLUDE_VECTOR): Define.
28901 No longer include <vector> and <algorithm> directly.
28903 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
28905 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
28907 * doc/invoke.texi (Code Gen Options): Ditto.
28909 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
28912 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
28914 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
28916 rtl-optimization/79125
28917 * cprop.c (local_cprop_pass): Handle cases where we make an
28918 unconditional trap.
28920 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
28924 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
28925 read from, for big endian.
28927 2017-01-20 Jiong Wang <jiong.wang@arm.com>
28929 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
28930 register pauth builtins for LP64 only.
28932 2017-01-20 Marek Polacek <polacek@redhat.com>
28935 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
28938 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
28940 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
28942 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
28943 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
28944 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
28946 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28949 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
28950 in big-endian mode when they are not a single duplicated value.
28952 2017-01-20 Richard Biener <rguenther@suse.de>
28954 * BASE-VER: Bump to 7.0.1.
28956 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
28958 * omp-low.c (omplow_simd_context): New struct. Use it...
28959 (lower_rec_simd_input_clauses): ...here and...
28960 (lower_rec_input_clauses): ...here to hold common data. Adjust all
28961 references to idx, lane, max_vf, is_simt.
28963 2017-01-20 Graham Markall <graham.markall@embecosm.com>
28965 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
28968 2017-01-20 Martin Jambor <mjambor@suse.cz>
28970 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
28971 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
28973 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
28974 (GTFILES): Rename hsa.c to hsa-common.c.
28975 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
28976 * hsa-dump.c: Likewise.
28977 * hsa-gen.c: Likewise.
28978 * hsa-regalloc.c: Likewise.
28979 * ipa-hsa.c: Likewise.
28980 * omp-expand.c: Likewise.
28981 * omp-low.c: Likewise.
28982 * toplev.c: Likewise.
28984 2017-01-20 Marek Polacek <polacek@redhat.com>
28987 * doc/invoke.texi: Document -Wduplicated-branches.
28988 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
28989 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
28990 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
28991 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
28992 return 0 only when not OEP_LEXICOGRAPHIC.
28993 (fold_build_cleanup_point_expr): Use the expression
28994 location when building CLEANUP_POINT_EXPR.
28995 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
28996 * tree.c (add_expr): Handle error_mark_node.
28998 2017-01-20 Martin Liska <mliska@suse.cz>
29001 * tree-profile.c (init_ic_make_global_vars): Do not call
29003 (gimple_init_gcov_profiler): Likewise.
29005 2017-01-20 Martin Liska <mliska@suse.cz>
29008 * cgraph.h (maybe_create_reference): Remove argument and
29010 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
29012 * ipa-cp.c (create_specialized_node): Likewise.
29013 * symtab.c (symtab_node::maybe_create_reference): Handle
29014 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
29016 2017-01-20 Martin Liska <mliska@suse.cz>
29018 * read-rtl-function.c (function_reader::create_function): Use
29019 build_decl instread of build_decl_stat.
29021 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
29023 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
29024 * config/i386/avx512dqintrin.h: Ditto.
29025 * config/i386/avx512fintrin.h: Ditto.
29026 * config/i386/i386-builtin-types.def: Add new types.
29027 * config/i386/i386.c: Handle new types.
29028 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
29029 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
29030 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
29031 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
29032 (__builtin_ia32_kshiftridi): New.
29033 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
29035 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
29039 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
29040 define to rs6000_init_stack_protect_guard.
29041 (rs6000_init_stack_protect_guard): New function.
29043 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
29044 Yunqiang Su <yunqiang.su@imgtec.com>
29046 * config.gcc (supported_defaults): Add madd4.
29047 (with_madd4): Add validation.
29048 (all_defaults): Add madd4.
29049 * config/mips/mips.opt (mmadd4): New option.
29050 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
29052 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
29054 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
29055 (ISA_HAS_FUSED_MADD4): Likewise.
29056 * doc/invoke.texi (-mmadd4): Document the new option.
29057 * doc/install.texi (--with-madd4): Document the new option.
29059 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29061 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
29062 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
29063 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
29064 (aarch64_init_pauth_hint_builtins): New.
29065 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
29066 (aarch64_expand_builtin): Expand new builtins.
29068 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29070 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
29071 * combine-stack-adj.c (no_unhandled_cfa): Handle
29072 REG_CFA_TOGGLE_RA_MANGLE.
29073 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
29074 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
29075 info for return address signing.
29076 (aarch64_expand_epilogue): Likewise.
29078 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29080 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
29081 * config/aarch64/aarch64-protos.h
29082 (aarch64_return_address_signing_enabled): New declaration.
29083 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
29085 (aarch64_expand_prologue): Sign return address before it's pushed onto
29087 (aarch64_expand_epilogue): Authenticate return address fetched from
29089 (aarch64_override_options): Sanity check for ILP32 and ISA level.
29090 (aarch64_attributes): New function attributes for "sign-return-address".
29091 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
29092 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
29093 ("*do_return"): Generate combined instructions according to key index.
29094 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
29095 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
29097 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
29098 * config/aarch64/aarch64.opt (msign-return-address=): New.
29099 * doc/extend.texi (AArch64 Function Attributes): Documents
29100 "sign-return-address=".
29101 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
29103 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
29105 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
29106 overall option summary.
29108 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29110 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
29111 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
29112 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
29113 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
29115 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
29117 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
29118 -mpower9-minmax by default for -mcpu=power9.
29119 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
29120 128-bit floating point.
29122 2017-01-20 Alan Modra <amodra@gmail.com>
29124 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
29125 optimizing for size.
29127 2017-01-20 Alan Modra <amodra@gmail.com>
29130 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
29131 for strcmp and strncmp from corresponding builtin decl.
29133 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
29135 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
29136 instead of i386/rtems-64.h.
29137 * config/i386/rtems-64.h: Remove.
29139 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
29143 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
29145 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
29147 2017-01-19 Tamar Christina <tamar.christina@arm.com>
29149 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
29150 Change int to HOST_WIDE_INT.
29151 * config/aarch64/aarch64-protos.h
29152 (aarch64_simd_gen_const_vector_dup): Likewise.
29153 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
29155 2017-01-19 David Malcolm <dmalcolm@redhat.com>
29157 * langhooks-def.h (lhd_type_for_size): New decl.
29158 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
29159 * langhooks.c (lhd_type_for_size): New function, taken from
29162 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
29164 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
29165 define_bypass for CR latency.
29166 (power9-cracked-alu): Update bypass latency and remove power9-branch.
29167 (power9-alu2): Add define_bypass for CR latency.
29169 (power9-mul): Update insn latency.
29170 (power9-mul-compare): Update insn latency, bypass latency and remove
29173 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29175 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
29177 * config/aarch64/aarch64.md
29178 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
29179 aarch64_nopcrelative_literal_loads.
29180 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
29182 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
29184 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
29185 TARGET_LOONGSON_3A.
29186 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
29188 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
29191 * config.gcc (supported_defaults): Add lxc1-sxc1.
29192 (with_lxc1_sxc1): Add validation.
29193 (all_defaults): Add lxc1-sxc1.
29194 * config/mips/mips.opt (mlxc1-sxc1): New option.
29195 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
29197 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
29198 __mips_no_lxc1_sxc1.
29199 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
29200 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
29201 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
29203 2017-01-19 Richard Biener <rguenther@suse.de>
29205 PR tree-optimization/72488
29206 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
29207 sure to restore SSA info.
29208 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
29210 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
29212 PR rtl-optimization/79121
29213 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
29214 of the inner type when shifting an extended value.
29216 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
29219 * symtab.c (symtab_node::equal_address_to): Fix comparing of
29220 interposable aliases.
29222 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
29225 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
29226 Use the evmergelohi instruction.
29227 (mov_si<mode>_e500_subreg4_2_le): Likewise.
29228 (mov_sitf_e500_subreg8_2_be): Likewise.
29229 (mov_sitf_e500_subreg12_2_le): Likewise.
29230 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
29231 (mov_si<mode>_e500_subreg4_2_be): Likewise.
29232 (mov_sitf_e500_subreg8_2_le): Likewise.
29233 (mov_sitf_e500_subreg12_2_be): Likewise.
29235 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29237 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
29238 attribute from vecsimple to vecperm.
29239 (altivec_vbpermq2): Likewise.
29241 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29244 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
29246 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29247 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
29248 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
29249 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
29250 case where N arg is SIZE_MAX.
29251 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
29252 (cmpstrsi): Add pattern.
29254 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29256 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
29257 __builtin_vec_revb builtins.
29258 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
29259 built-in functions to support generation of the ISA 3.0 XXBR<x>
29260 vector byte reverse instructions.
29261 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
29262 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
29263 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
29264 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
29265 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
29266 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
29267 (P9V_BUILTIN_VEC_REVB): Likewise.
29268 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
29269 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
29270 (p9_xxbrq_v16qi): Likewise.
29271 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
29272 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
29273 (p9_xxbrh_v8hi): Likewise.
29274 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
29275 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
29276 vec_revb built-in functions.
29278 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
29280 PR rtl-optimization/78952
29281 * config/i386/i386.md (any_extract): New code iterator.
29282 (*insvqi_2): Use any_extract for source operand.
29283 (*insvqi_3): Use any_shiftrt for source operand.
29285 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
29287 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
29289 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
29291 2017-01-18 Matthias Klose <doko@ubuntu.com>
29293 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
29295 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29297 * config/rs6000/altivec.h (vec_bperm): Change #define.
29298 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
29299 (altivec_vbpermq2): New define_insn.
29300 (altivec_vbpermd): Likewise.
29301 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
29302 function interface.
29303 (VBPERMD): Likewise.
29304 (VBPERM): New polymorphic function interface.
29305 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
29306 Add entries for P9V_BUILTIN_VEC_VBPERM.
29307 * doc/extend.texi: Add interfaces for vec_bperm.
29309 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29311 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
29312 first letter of error messages.
29313 (s390_resolve_overloaded_builtin): Likewise.
29314 * config/s390/s390.c (s390_expand_builtin): Likewise.
29315 (s390_invalid_arg_for_unprototyped_fn): Likewise.
29316 (s390_valid_target_attribute_inner_p): Likewise.
29317 * config/s390/s390.md ("tabort"): Likewise.
29319 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
29321 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
29322 (ISA_AVOID_DIV_HILO): New macro.
29323 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
29324 (ISA_HAS_DDIV): Likewise.
29326 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29328 * doc/invoke.texi (fabi-version): Correct number of occurrences.
29330 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29332 * doc/invoke.texi (fabi-version): Spelling fix.
29334 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29337 * doc/invoke.texi (fabi-version): Mention mangling fix for
29340 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29343 * doc/invoke.texi (fabi-version): Document discriminator mangling.
29345 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
29348 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
29349 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
29351 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
29353 (stack_protect_test): Ditto.
29354 * config/rs6000/rs6000.opt (mstack-protector-guard=,
29355 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
29357 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
29358 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
29359 -mstack-protector-guard-offset=.
29360 (RS/6000 and PowerPC Options): Ditto.
29362 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
29364 * config/i386/i386.h (MASK_CLASS_P): New define.
29365 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
29366 there are no registers from different register sets also when
29367 mask registers are used. Update function comment.
29368 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
29369 to (*k/*r) and (*k/*km) alternatives.
29371 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
29373 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
29374 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
29375 (EH_RETURN_HANDLER_RTX): New define.
29376 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
29377 Force frame pointer in EH return functions.
29378 (aarch64_expand_epilogue): Add barrier for eh_return.
29379 (aarch64_final_eh_return_addr): Remove.
29380 (aarch64_eh_return_handler_rtx): New function.
29381 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
29383 (aarch64_eh_return_handler_rtx): New prototype.
29385 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29387 * config/rs6000/altivec.h (vec_rlmi): New #define.
29388 (vec_vrlnm): Likewise.
29389 (vec_rlnm): Likewise.
29390 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
29391 (UNSPEC_VRLNM): Likewise.
29392 (VIlong): New mode iterator.
29393 (altivec_vrl<VI_char>mi): New define_insn.
29394 (altivec_vrl<VI_char>nm): Likewise.
29395 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
29397 (VRLDNM): Likewise.
29398 (RLNM): New polymorphic function entry.
29399 (VRLWMI): New monomorphic function entry.
29400 (VRLDMI): Likewise.
29401 (RLMI): New polymorphic function entry.
29402 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
29403 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
29404 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
29407 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29410 * dwarf2out.c (field_byte_offset): Restore the
29411 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
29412 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
29413 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
29416 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
29419 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
29421 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29424 * dwarf2out.c (add_data_member_location_attribute): For constant
29425 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
29426 instead of DW_AT_data_member_location, DW_AT_bit_offset and
29427 DW_AT_byte_size attributes.
29429 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
29431 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
29432 after forcing to constant memory when the code model is medium.
29434 2017-01-17 Julia Koval <julia.koval@intel.com>
29437 * config/i386/avx512fintrin.h
29438 (_mm512_i32gather_ps): Change __addr type to void const*.
29439 (_mm512_mask_i32gather_ps): Ditto.
29440 (_mm512_i32gather_pd): Ditto.
29441 (_mm512_mask_i32gather_pd): Ditto.
29442 (_mm512_i64gather_ps): Ditto.
29443 (_mm512_mask_i64gather_ps): Ditto.
29444 (_mm512_i64gather_pd): Ditto.
29445 (_mm512_mask_i64gather_pd): Ditto.
29446 (_mm512_i32gather_epi32): Ditto.
29447 (_mm512_mask_i32gather_epi32): Ditto.
29448 (_mm512_i32gather_epi64): Ditto.
29449 (_mm512_mask_i32gather_epi64): Ditto.
29450 (_mm512_i64gather_epi32): Ditto.
29451 (_mm512_mask_i64gather_epi32): Ditto.
29452 (_mm512_i64gather_epi64): Ditto.
29453 (_mm512_mask_i64gather_epi64): Ditto.
29454 (_mm512_i32scatter_ps): Change __addr type to void*.
29455 (_mm512_mask_i32scatter_ps): Ditto.
29456 (_mm512_i32scatter_pd): Ditto.
29457 (_mm512_mask_i32scatter_pd): Ditto.
29458 (_mm512_i64scatter_ps): Ditto.
29459 (_mm512_mask_i64scatter_ps): Ditto.
29460 (_mm512_i64scatter_pd): Ditto.
29461 (_mm512_mask_i64scatter_pd): Ditto.
29462 (_mm512_i32scatter_epi32): Ditto.
29463 (_mm512_mask_i32scatter_epi32): Ditto.
29464 (_mm512_i32scatter_epi64): Ditto.
29465 (_mm512_mask_i32scatter_epi64): Ditto.
29466 (_mm512_i64scatter_epi32): Ditto.
29467 (_mm512_mask_i64scatter_epi32): Ditto.
29468 (_mm512_i64scatter_epi64): Ditto.
29469 (_mm512_mask_i64scatter_epi64): Ditto.
29470 * config/i386/avx512pfintrin.h
29471 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
29472 (_mm512_mask_prefetch_i32gather_ps): Ditto.
29473 (_mm512_mask_prefetch_i64gather_pd): Ditto.
29474 (_mm512_mask_prefetch_i64gather_ps): Ditto.
29475 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
29476 (_mm512_prefetch_i32scatter_ps): Ditto.
29477 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
29478 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
29479 (_mm512_prefetch_i64scatter_pd): Ditto.
29480 (_mm512_prefetch_i64scatter_ps): Ditto.
29481 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
29482 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
29483 * config/i386/avx512vlintrin.h
29484 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
29485 (_mm_mmask_i32gather_ps): Ditto.
29486 (_mm256_mmask_i32gather_pd): Ditto.
29487 (_mm_mmask_i32gather_pd): Ditto.
29488 (_mm256_mmask_i64gather_ps): Ditto.
29489 (_mm_mmask_i64gather_ps): Ditto.
29490 (_mm256_mmask_i64gather_pd): Ditto.
29491 (_mm_mmask_i64gather_pd): Ditto.
29492 (_mm256_mmask_i32gather_epi32): Ditto.
29493 (_mm_mmask_i32gather_epi32): Ditto.
29494 (_mm256_mmask_i32gather_epi64): Ditto.
29495 (_mm_mmask_i32gather_epi64): Ditto.
29496 (_mm256_mmask_i64gather_epi32): Ditto.
29497 (_mm_mmask_i64gather_epi32): Ditto.
29498 (_mm256_mmask_i64gather_epi64): Ditto.
29499 (_mm_mmask_i64gather_epi64): Ditto.
29500 (_mm256_i32scatter_ps): Change __addr type to void*.
29501 (_mm256_mask_i32scatter_ps): Ditto.
29502 (_mm_i32scatter_ps): Ditto.
29503 (_mm_mask_i32scatter_ps): Ditto.
29504 (_mm256_i32scatter_pd): Ditto.
29505 (_mm256_mask_i32scatter_pd): Ditto.
29506 (_mm_i32scatter_pd): Ditto.
29507 (_mm_mask_i32scatter_pd): Ditto.
29508 (_mm256_i64scatter_ps): Ditto.
29509 (_mm256_mask_i64scatter_ps): Ditto.
29510 (_mm_i64scatter_ps): Ditto.
29511 (_mm_mask_i64scatter_ps): Ditto.
29512 (_mm256_i64scatter_pd): Ditto.
29513 (_mm256_mask_i64scatter_pd): Ditto.
29514 (_mm_i64scatter_pd): Ditto.
29515 (_mm_mask_i64scatter_pd): Ditto.
29516 (_mm256_i32scatter_epi32): Ditto.
29517 (_mm256_mask_i32scatter_epi32): Ditto.
29518 (_mm_i32scatter_epi32): Ditto.
29519 (_mm_mask_i32scatter_epi32): Ditto.
29520 (_mm256_i32scatter_epi64): Ditto.
29521 (_mm256_mask_i32scatter_epi64): Ditto.
29522 (_mm_i32scatter_epi64): Ditto.
29523 (_mm_mask_i32scatter_epi64): Ditto.
29524 (_mm256_i64scatter_epi32): Ditto.
29525 (_mm256_mask_i64scatter_epi32): Ditto.
29526 (_mm_i64scatter_epi32): Ditto.
29527 (_mm_mask_i64scatter_epi32): Ditto.
29528 (_mm256_i64scatter_epi64): Ditto.
29529 (_mm256_mask_i64scatter_epi64): Ditto.
29530 (_mm_i64scatter_epi64): Ditto.
29531 (_mm_mask_i64scatter_epi64): Ditto.
29532 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
29533 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
29534 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
29535 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
29536 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
29537 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
29538 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
29539 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
29540 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
29541 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
29542 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
29543 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
29544 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
29545 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
29546 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
29547 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
29548 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
29549 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
29550 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
29551 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
29552 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
29553 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
29554 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
29555 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
29556 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
29557 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
29558 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
29559 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
29560 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
29561 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
29562 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
29563 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
29564 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
29565 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
29566 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
29567 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
29568 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
29569 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
29570 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
29571 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
29572 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
29573 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
29574 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
29575 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
29576 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
29577 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
29578 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
29579 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
29580 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
29581 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
29582 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
29583 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
29584 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
29585 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
29586 definitions accordingly.
29588 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
29589 Kuan-Lin Chen <kuanlinchentw@gmail.com>
29592 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
29595 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
29598 * ira-conflicts.c (ira_build_conflicts): Update total conflict
29599 hard regs for inner regno.
29601 2017-01-17 Martin Liska <mliska@suse.cz>
29604 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
29605 assumption and add comment.
29607 2017-01-17 Nathan Sidwell <nathan@acm.org>
29609 * ipa-visibility.c (localize_node): New function, broken out of ...
29610 (function_and_variable_visibility): ... here. Call it.
29612 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
29614 PR middle-end/77445
29615 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
29616 correctly set frequency of oudgoing edge.
29617 (duplicate_thread_path): Fix profile updating.
29619 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29622 * configure.ac: Add GCC_BASE_VER.
29623 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
29624 version from BASE-VER file.
29625 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
29626 (gcc.o): Depend on $(BASEVER).
29627 * common.opt (dumpfullversion): New option.
29628 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
29629 * doc/invoke.texi: Document -dumpfullversion.
29630 * doc/install.texi: Document --with-gcc-major-version-only.
29631 * configure: Regenerated.
29633 2017-01-17 Richard Biener <rguenther@suse.de>
29635 PR tree-optimization/71433
29636 * tree-vrp.c (register_new_assert_for): Merge same asserts
29637 on all incoming edges.
29638 (process_assert_insertions_for): Handle insertions at the
29641 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
29643 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
29644 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
29646 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
29649 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
29652 2017-01-17 Alan Modra <amodra@gmail.com>
29655 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
29656 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
29657 symbolic stack limit when pic.
29659 2017-01-16 Martin Sebor <msebor@redhat.com>
29661 PR tree-optimization/78608
29662 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
29664 2017-01-16 Jeff Law <law@redhat.com>
29667 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
29668 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
29669 for several include directories that may be relative to sysroot.
29670 * config/i386/x-mingw32 (gplus_includedir): Define.
29671 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
29672 (native_system_includedir): Likewise.
29673 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
29674 override if TARGET_SYSTEM_ROOT is defined.
29675 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
29677 PR tree-optimization/79090
29678 PR tree-optimization/33562
29679 PR tree-optimization/61912
29680 PR tree-optimization/77485
29681 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
29682 and computed trims into the dump file.
29684 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
29686 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
29688 2017-01-16 Jakub Jelinek <jakub@redhat.com>
29691 * gimplify.c (gimplify_init_constructor): If want_value and
29692 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
29696 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
29697 sequence. Formatting fixes.
29698 (doloop_optimize): Formatting fixes.
29701 * gcc.c (debug_level_greater_than_spec_func): New function.
29702 (static_spec_functions): Add debug-level-gt spec function.
29703 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
29705 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
29706 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
29707 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
29708 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
29709 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
29710 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
29712 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
29714 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
29715 QImode fixups to general and mask registers only.
29717 2017-01-16 Carl Love <cel@us.ibm.com>
29719 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
29720 for built-in functions
29721 vector signed char vec_nabs (vector signed char)
29722 vector signed short vec_nabs (vector signed short)
29723 vector signed int vec_nabs (vector signed int)
29724 vector signed long long vec_nabs (vector signed long long)
29725 vector float vec_nabs (vector float)
29726 vector double vec_nabs (vector double)
29727 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
29729 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
29730 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
29731 * doc/extend.texi: Update the documentation file for the new built-in
29734 2017-01-16 Martin Sebor <msebor@redhat.com>
29736 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
29739 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29741 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
29742 UNSPEC_VSX__XXSPLTD to require special splat handling.
29744 2017-01-16 David Malcolm <dmalcolm@redhat.com>
29747 * system.h: Poison strndup.
29749 2017-01-16 Alan Modra <amodra@gmail.com>
29752 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
29755 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
29757 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
29759 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
29761 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
29762 call recog here. Assert that INSN_CODE (insn) is non-negative.
29764 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
29767 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
29769 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
29770 in the currently scheduled RTL fragment.
29772 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
29774 PR rtl-optimization/78751
29775 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
29778 2017-01-14 Jeff Law <law@redhat.com>
29780 PR tree-optimization/79090
29781 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
29782 variable length stores.
29783 (compute_trims): Delete dead assignment to *trim_tail.
29784 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
29787 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
29789 PR rtl-optimization/78626
29790 PR rtl-optimization/78727
29791 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
29792 of a block, and split such blocks after everything else is finished.
29794 2017-01-14 Alan Modra <amodra@gmail.com>
29797 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
29798 target legitimate_combined_insn.
29799 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
29800 (rs6000_legitimate_combined_insn): New function.
29801 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
29803 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
29804 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
29805 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
29807 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
29809 * doc/frontends.texi (G++ and GCC): Remove references to Java.
29811 2017-01-13 Jeff Law <law@redhat.com>
29813 PR tree-optimization/33562
29814 PR tree-optimization/61912
29815 PR tree-optimization/77485
29816 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
29818 (delete_dead_assignment): Likewise.
29819 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
29820 statement to delete_dead_call and delete_dead_assignment.
29822 2017-01-13 David Malcolm <dmalcolm@redhat.com>
29825 * substring-locations.c (format_warning_va): Strengthen case 1 so
29826 that both endpoints of the substring must be within the format
29827 range for just the substring to be printed.
29829 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
29831 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
29832 * config/i386/i386.c (ix86_target_string): Add missing options
29833 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
29834 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
29835 flags_other and ix86_target_other to flags2_other. Display unknown
29837 (ix86_valid_target_attribute_inner_p): Add missing options and
29838 reorder options by implied ISAs, as in ix86_target_string.
29840 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
29842 * hash-table.h (hash_table::too_empty_p): New function.
29843 (hash_table::expand): Use it.
29844 (hash_table::traverse): Likewise.
29845 (hash_table::empty_slot): Use sizeof (value_type) instead of
29846 sizeof (PTR) to convert bytes to elements. Shrink the table
29847 if the current size is excessive for the current number of
29850 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
29852 * ira-costs.c (record_reg_classes): Break from the inner loop
29853 early once alt_fail is known to be true. Update outer loop
29854 handling accordingly.
29856 2017-01-13 Jeff Law <law@redhat.com>
29858 * tree-ssa-dse.c (decrement_count): New function.
29859 (increment_start_addr, maybe_trim_memstar_call): Likewise.
29860 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
29861 when we know the partially dead statement is a mem* function.
29863 PR tree-optimization/61912
29864 PR tree-optimization/77485
29865 * tree-ssa-dse.c: Include expr.h.
29866 (maybe_trim_constructor_store): New function.
29867 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
29869 PR tree-optimization/33562
29870 PR tree-optimization/61912
29871 PR tree-optimization/77485
29872 * doc/invoke.texi: Document new dse-max-object-size param.
29873 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
29874 * tree-ssa-dse.c: Include params.h.
29875 (dse_store_status): New enum.
29876 (initialize_ao_ref_for_dse): New, partially extracted from
29878 (valid_ao_ref_for_dse, normalize_ref): New.
29879 (setup_live_bytes_from_ref, compute_trims): Likewise.
29880 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
29881 (maybe_trim_partially_dead_store): Likewise.
29882 (maybe_trim_complex_store): Likewise.
29883 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
29884 Track what bytes live from the original store. Return tri-state
29885 for dead, partially dead or live.
29886 (dse_dom_walker): Add constructor, destructor and new private members.
29887 (delete_dead_call, delete_dead_assignment): New extracted from
29889 (dse_optimize_stmt): Make a member of dse_dom_walker.
29890 Use initialize_ao_ref_for_dse.
29892 PR tree-optimization/33562
29893 PR tree-optimization/61912
29894 PR tree-optimization/77485
29895 * sbitmap.h (bitmap_count_bits): Prototype.
29896 (bitmap_clear_range, bitmap_set_range): Likewise.
29897 * sbitmap.c (bitmap_clear_range): New function.
29898 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
29900 2017-01-13 Martin Liska <mliska@suse.cz>
29903 * function.c (set_cfun): Add new argument force.
29904 * function.h (set_cfun): Likewise.
29905 * ipa-inline-transform.c (inline_call): Use the function when
29906 strict alising from is dropped for function we inline to.
29908 2017-01-13 Richard Biener <rguenther@suse.de>
29910 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
29911 for dumping GIMPLE INTEGER_CSTs.
29913 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29915 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
29916 to 201112L since C++17.
29918 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
29921 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
29922 if -fsanitize=kernel-address is present.
29924 2017-01-13 Richard Biener <rguenther@suse.de>
29926 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
29927 as _Literal ( type ) number in case usual suffixes do not
29928 preserve all information.
29930 2017-01-13 Richard Biener <rguenther@suse.de>
29932 PR tree-optimization/77283
29933 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
29934 and ssa-iterators.h.
29935 (is_feasible_trace): Implement a cost model based on joiner
29938 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
29941 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
29942 char or short to __float128/_Float128 directly.
29944 2017-01-12 Martin Sebor <msebor@redhat.com>
29946 to -Wformat-overflow.
29947 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
29948 (min_bytes_remaining): Same.
29949 (get_string_length): Same.
29950 (format_string): Same.
29951 (format_directive): Same.
29953 (pass_sprintf_length::handle_gimple_call): Same.
29955 2017-01-12 Jakub Jelinek <jakub@redhat.com>
29957 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
29958 info.nowrite calls with no lhs that can't throw. Return bool
29959 whether gsi_remove has been called or not.
29960 (pass_sprintf_length::handle_gimple_call): Return bool whether
29961 try_substitute_return_value called gsi_remove. Formatting fix.
29962 (pass_sprintf_length::execute): Don't use gsi_remove if
29963 handle_gimple_call returned true.
29966 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
29967 be removed due to side-effects, don't remove following barrier nor
29968 turn the successor edge into fallthru edge.
29970 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29973 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
29974 element-reversing loads and stores as not swappable.
29976 2017-01-12 Nathan Sidwell <nathan@acm.org>
29977 Nicolai Stange <nicstange@gmail.com>
29979 * combine.c (try_combine): Don't ignore result of overlap checking
29980 loop. Combine overlap & asm check into single loop.
29982 2017-01-12 Richard Biener <rguenther@suse.de>
29984 * tree-pretty-print.c (dump_generic_node): Provide -gimple
29985 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
29987 2017-01-12 Richard Biener <rguenther@suse.de>
29989 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
29990 and TS_TARGET_OPTION directly derive from TS_BASE.
29991 * tree-core.h (tree_optimization_option): Derive from tree_base.
29992 (tree_target_option): Likewise.
29994 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
29996 * config/i386/i386.c (memory_address_length): Increase len
29997 only when rip_relative_addr_p returns false.
29999 2017-01-11 Julia Koval <julia.koval@intel.com>
30001 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
30002 (OPTION_MASK_ISA_SGX_SET): New.
30003 (ix86_handle_option): Handle OPT_msgx.
30004 * config.gcc: Added sgxintrin.h.
30005 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
30006 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
30007 * config/i386/i386.c (ix86_target_string): Add -msgx.
30009 (ix86_option_override_internal): Handle new options.
30010 (ix86_valid_target_attribute_inner_p): Add sgx.
30011 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
30012 * config/i386/i386.opt: Add msgx.
30013 * config/i386/sgxintrin.h: New file.
30014 * config/i386/x86intrin.h: Add sgxintrin.h.
30016 2017-01-11 Jakub Jelinek <jakub@redhat.com>
30019 * fold-const.c (maybe_nonzero_address): Return 1 for function
30021 (tree_single_nonzero_warnv_p): Don't handle function local objects
30025 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
30028 2017-01-11 David Malcolm <dmalcolm@redhat.com>
30031 * opts.c: Include "spellcheck.h"
30032 (struct string_fragment): New struct.
30033 (struct edit_distance_traits<const string_fragment &>): New
30035 (get_closest_sanitizer_option): New function.
30036 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
30038 2017-01-11 Jakub Jelinek <jakub@redhat.com>
30040 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
30042 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
30043 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
30044 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
30045 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
30046 for initial die_offset if dwarf_split_debug_info.
30047 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
30048 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
30050 (output_skeleton_debug_sections): Formatting fix. Use
30051 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
30052 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
30054 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
30056 * config/arm/cortex-a53.md: Add bypasses for
30057 cortex_a53_r2f_cvt.
30058 (cortex_a53_r2f): Only use for transfers.
30059 (cortex_a53_f2r): Likewise.
30060 (cortex_a53_r2f_cvt): Add reservation for conversions.
30061 (cortex_a53_f2r_cvt): Likewise.
30063 2017-01-11 Tamar Christina <tamar.christina@arm.com>
30065 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
30066 to all inlined functions, change static to extern.
30068 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
30071 * config/arm/arm.c (legitimize_pic_address): Handle reference to
30073 (arm_assemble_integer): Likewise.
30075 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
30077 * config.gcc: Use new awk script to check CPU, FPU and architecture
30078 parameters for --with-... options.
30079 * config/arm/parsecpu.awk: New file
30080 * config/arm/arm-cpus.in: New file.
30081 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
30083 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
30085 * config/arm/t-arm: Update dependency rules.
30086 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
30087 of processing .def files.
30088 * config/arm/genopt.sh: Deleted.
30089 * config/arm/gentune.sh: Deleted.
30090 * config/arm/arm-cores.def: Deleted.
30091 * config/arm/arm-arches.def: Deleted.
30092 * config/arm/arm-fpus.def: Deleted.
30093 * config/arm/arm-tune.md: Regenerated.
30094 * config/arm/arm-tables.opt: Regenerated.
30095 * config/arm/arm-cpu.h: New generated file.
30096 * config/arm/arm-cpu-data.h: New generated file.
30097 * config/arm/arm-cpu-cdata.h: New generated file.
30099 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
30102 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
30104 (input_varpool_node): Unpack dynamically_initialized bit.
30106 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
30108 PR rtl-optimization/79032
30109 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
30110 the alignment of the adjusted memory reference against that of MODE,
30111 instead of the alignment of the original memory reference.
30113 2017-01-11 Martin Jambor <mjambor@suse.cz>
30115 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
30117 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
30118 decorated functions.
30120 2017-01-11 Richard Biener <rguenther@suse.de>
30122 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
30123 set range/nonnull info for PHI results. Do not set it on
30124 stmts marked for removal.
30126 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
30128 * expr.c (store_field): In the bitfield case, fetch the return value
30129 from the registers before applying a single big-endian adjustment.
30130 Always do a final load for a BLKmode value not larger than a word.
30132 2017-01-10 David Malcolm <dmalcolm@redhat.com>
30135 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
30136 that we correctly handle column numbers greater than
30137 LINE_MAP_MAX_COLUMN_NUMBER.
30139 2017-01-10 Martin Sebor <msebor@redhat.com>
30141 PR middle-end/78245
30142 * gimple-ssa-sprintf.c (get_destination_size): Call
30143 {init,fini}object_sizes.
30144 * tree-object-size.c (addr_object_size): Adjust.
30145 (pass_through_call): Adjust.
30146 (pass_object_sizes::execute): Adjust.
30147 * tree-object-size.h (fini_object_sizes): Declare.
30149 2017-01-10 Martin Sebor <msebor@redhat.com>
30151 PR tree-optimization/78775
30152 * builtins.c (get_size_range): Move...
30153 * calls.c: ...to here.
30154 (alloc_max_size): Accept zero argument.
30155 (operand_signed_p): Remove.
30156 (maybe_warn_alloc_args_overflow): Call get_size_range.
30157 * calls.h (get_size_range): Declare.
30159 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
30161 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
30162 from TI's devices.csv file as of September 2016.
30163 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
30165 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
30167 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
30168 * doc/invoke.texi: Likewise.
30169 * doc/md.texi: Likewise.
30170 * doc/objc.texi: Likewise.
30172 2017-01-10 Joshua Conner <joshconner@google.com>
30174 * config/arm/fuchsia-elf.h: New file.
30175 * config/fuchsia.h: New file.
30176 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
30177 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
30179 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
30181 2016-01-10 Richard Biener <rguenther@suse.de>
30183 PR tree-optimization/79034
30184 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
30185 Propagate out degenerate PHIs in the joiner.
30187 2017-01-10 Martin Liska <mliska@suse.cz>
30189 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
30190 (sort_congruence_classes_by_decl_uid): Likewise.
30191 (sort_congruence_class_groups_by_decl_uid): Likewise.
30192 (sem_item_optimizer::merge_classes): Sort class, groups in these
30193 classes and members in the groups by DECL_UID of declarations.
30194 This would make merge operations stable.
30196 2017-01-10 Martin Liska <mliska@suse.cz>
30198 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
30199 usage of m_classes_vec.
30200 (sem_item_optimizer::~sem_item_optimizer): Likewise.
30201 (sem_item_optimizer::get_group_by_hash): Likewise.
30202 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
30203 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
30204 (sem_item_optimizer::verify_classes): Likewise.
30205 (sem_item_optimizer::process_cong_reduction): Likewise.
30206 (sem_item_optimizer::dump_cong_classes): Likewise.
30207 (sem_item_optimizer::merge_classes): Likewise.
30208 * ipa-icf.h (congruence_class_hash): Rename from
30209 congruence_class_group_hash. Remove declaration of m_classes_vec.
30211 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
30213 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
30214 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
30215 * config.gcc: Add avx512vpopcntdqintrin.h.
30216 * config/i386/avx512vpopcntdqintrin.h: New.
30217 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
30218 * config/i386/i386-builtin-types.def: Add new types.
30219 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
30220 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
30221 __builtin_ia32_vpopcountq_v8di_mask): New.
30222 * config/i386/i386-c.c (ix86_target_macros_internal): Define
30223 __AVX512VPOPCNTDQ__.
30224 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
30225 (PTA_AVX512VPOPCNTDQ): Define.
30226 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
30227 TARGET_AVX512VPOPCNTDQ_P): Define.
30228 * config/i386/i386.opt: Add mavx512vpopcntdq.
30229 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
30230 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
30232 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30234 PR middle-end/77484
30235 * predict.def (PRED_CALL): Set to 67.
30237 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
30239 * expr.c (store_field): In the bitfield case, if the value comes from
30240 a function call and is of an aggregate type returned in registers, do
30241 not modify the field mode; extract the value in all cases if the mode
30242 is BLKmode and the size is not larger than a word.
30244 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
30247 * config/i386/cpuid.h: Fix undefined behavior.
30249 2017-01-04 Jeff Law <law@redhat.com>
30251 PR tree-optimization/79007
30252 PR tree-optimization/67955
30253 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
30254 conservative for pt.null when flag_non_call_exceptions is on.
30256 2017-01-09 Jakub Jelinek <jakub@redhat.com>
30258 PR translation/79019
30259 PR translation/79020
30260 * params.def (PARAM_INLINE_MIN_SPEEDUP,
30261 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
30262 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
30264 * config/avr/avr.opt (maccumulate-args): Likewise.
30265 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
30266 * common.opt (freport-bug): Likewise.
30267 * cif-code.def (CIF_FINAL_ERROR): Likewise.
30268 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
30269 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
30270 translatable string.
30271 * config/i386/i386.c (function_value_32): Likewise.
30272 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
30273 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
30275 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
30276 * common/config/msp430/msp430-common.c (msp430_handle_option):
30278 * symtab.c (symtab_node::verify_base): Likewise.
30279 * opts.c (set_debug_level): Likewise.
30280 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
30281 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
30282 missing whitespace to translatable strings.
30283 * config/avr/avr.md (bswapsi2): Fix typo in comment.
30284 * config/sh/superh.h: Likewise.
30285 * config/i386/xopintrin.h: Likewise.
30286 * config/i386/znver1.md: Likewise.
30287 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
30288 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
30289 * double-int.h (struct double_int): Likewise.
30290 * double-int.c (div_and_round_double): Likewise.
30291 * wide-int.cc: Likewise.
30292 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
30293 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
30294 * cfgcleanup.c (crossjumps_occured): Renamed to ...
30295 (crossjumps_occurred): ... this.
30296 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
30299 PR tree-optimization/78899
30300 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
30301 returning bool return struct loop *, NULL for failure and the new
30303 (versionable_outer_loop_p): Don't version outer loop if it has
30304 dont_vectorized bit set.
30305 (tree_if_conversion): When versioning outer loop, ensure
30306 tree_if_conversion is performed also on the inner loop of the
30307 non-vectorizable outer loop copy.
30308 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
30309 LOOP_VECTORIZED in inner loop of the scalar outer loop and
30310 prevent vectorization of it.
30311 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
30312 the outer loop vectorization of the non-scalar version is attempted
30313 before vectorization of the inner loop in scalar version. If
30314 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
30315 vectorization of its inner loop.
30316 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
30317 has 2 inner loops, rename also on edges from bb whose single pred
30318 is outer_loop->header. Fix typo in function comment.
30320 2017-01-09 Martin Sebor <msebor@redhat.com>
30323 * asan.c (asan_emit_stack_protection): Increase local buffer size
30324 to avoid snprintf truncation warning.
30326 2017-01-09 Andrew Pinski <apinski@cavium.com>
30328 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
30329 to reference thunderx2t99 for the tuning structure
30330 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
30332 (thunderx2t99_extra_costs): This.
30333 * config/aarch64/aarch64-tune.md: Regenerate.
30334 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
30335 (vulcan_addrcost_table): This.
30336 (vulcan_regmove_cost): Rename to ...
30337 (thunderx2t99_regmove_cost): This.
30338 (vulcan_vector_cost): Rename to ...
30339 (thunderx2t99_vector_cost): this.
30340 (vulcan_branch_cost): Rename to ...
30341 (thunderx2t99_branch_cost): This.
30342 (vulcan_tunings): Rename to ...
30343 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
30344 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
30346 2017-01-09 Martin Jambor <mjambor@suse.cz>
30350 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
30351 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
30352 (propagate_vr_accross_jump_function): Use the above function for all
30353 value range computations for pass-through jump functions and type
30354 converasion from explicit value range values.
30355 (ipcp_propagate_stage): Do not attempt to deduce types of formal
30356 parameters from TYPE_ARG_TYPES.
30357 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
30358 (ipa_write_node_info): Stream type of the actual argument.
30359 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
30361 2017-01-09 Martin Liska <mliska@suse.cz>
30364 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
30365 (lookup_compiler): Do not show error message with have_E.
30367 2017-01-09 Jakub Jelinek <jakub@redhat.com>
30369 PR tree-optimization/78938
30370 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
30371 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
30372 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
30373 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
30376 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30378 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
30381 2017-01-09 Richard Biener <rguenther@suse.de>
30383 PR tree-optimization/78997
30384 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
30385 name condition properly.
30387 2017-01-09 Richard Biener <rguenther@suse.de>
30390 * dwarf2out.c (is_cxx): New overload with context.
30391 (is_naming_typedef_decl): Use it.
30393 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30395 * invoke.texi (Option Summary): Correct spacing in option lists
30396 and add line breaks to fix over-long lines.
30398 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30400 PR middle-end/17660
30402 * extend.texi (Common Variable Attributes): Add xref to GCC
30403 Internals manual to explain mode attribute keywords.
30405 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30408 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
30409 and Preprocessor Options.
30410 (Options for Linking): Document -pthread here....
30411 (RS/6000 and PowerPC Options): ...not here.
30412 (Solaris 2 Options): ...or here.
30413 * doc/cppopts.texi: Document -pthread.
30415 2017-01-08 Martin Sebor <msebor@redhat.com>
30417 PR middle-end/77708
30418 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
30419 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
30420 New member functions.
30421 (format_directive): Used them.
30423 (pass_sprintf_length::handle_gimple_call): Same.
30424 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
30425 to avoid truncation for any argument.
30426 (extract_affine_mul): Same.
30427 * tree.c (get_file_function_name): Same.
30429 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30431 PR middle-end/77484
30432 * predict.def (PRED_INDIR_CALL): Set to 86.
30434 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30436 PR preprocessor/54124
30437 * doc/cppopts.texi: Reformat -d subtable to list the full name
30438 of the options. Add cross-reference to the docs for the general
30439 compiler -d options.
30440 * doc/invoke.texi (Developer Options): Add cross-reference to the
30441 preprocessor-specific -d option documentation.
30443 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30445 PR preprocessor/13498
30446 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
30447 redudant material, and reflect new command-line options.
30448 (System Headers): Likewise.
30450 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30452 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
30453 -isystem, and -idirafter. Copy-edit.
30454 * doc/cppopts.texi: Copy-edit. Remove contradiction about
30455 default for -ftrack-macro-expansion. Delete obsolete and
30456 badly-formatted implementation details about -fdebug-cpp output.
30457 * doc/cppwarnopts.texi: Copy-edit.
30459 2017-01-07 David Malcolm <dmalcolm@redhat.com>
30462 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
30463 that the transition from a max line width >= 1<<10 to narrower
30464 lines works correctly.
30466 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
30468 * doc/options.texi (PerFunction): New.
30469 * opt-functions.awk (switch_flags): Map both Optimization and
30470 PerFunction to CL_OPTIMIZATION.
30471 * opth-gen.awk: Test for PerFunction flag along with
30473 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
30474 it only when the latter is present. Skip those that don't in
30475 the hash function generator.
30476 * common.opt (fvar-tracking): Mark as PerFunction instead of
30478 (fvar-tracking-assignments): Likewise.
30479 (fvar-tracking-assignments-toggle): Likewise.
30480 (fvar-tracking-uninit): Likewise.
30482 2017-01-07 Jakub Jelinek <jakub@redhat.com>
30484 PR translation/79018
30485 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
30488 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
30491 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
30492 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
30493 TARGET_LONG_JUMP_TABLE_OFFSETS.
30494 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
30495 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
30496 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
30497 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
30498 * config/m68k/m68k.md (tablejump expander): Likewise.
30499 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
30500 TARGET_LONG_JUMP_TABLE_OFFSETS.
30501 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
30502 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
30504 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
30505 David Holsgrove <david.holsgrove@xilinx.com>
30507 * common/config/microblaze/microblaze-common.c
30508 (TARGET_EXCEPT_UNWIND_INFO): Remove.
30509 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
30511 * config/microblaze/microblaze.c (microblaze_must_save_register)
30512 (microblaze_expand_epilogue, microblaze_return_addr): Handle
30514 (microblaze_eh_return): New function.
30515 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
30516 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
30517 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
30518 * config/microblaze/microblaze.md (eh_return): New pattern.
30520 2017-01-06 Jakub Jelinek <jakub@redhat.com>
30522 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
30523 GCC_DIAGNOSTIC_STRINGIFY): Define.
30525 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
30527 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30529 * config/arm/arm.md (<mcrr>): New.
30531 * config/arm/arm.c (arm_arch5te): New.
30532 (arm_option_override): Set arm_arch5te.
30533 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
30535 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
30536 (arm_mcrr_qualifiers): ... this. New.
30537 (MRRC_QUALIFIERS): Define to...
30538 (arm_mrrc_qualifiers): ... this. New.
30539 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
30541 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
30542 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
30543 (MRRCI, mrrc, MRRC): New.
30544 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
30545 VUNSPEC_MRRC2): New.
30547 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30549 * config/arm/arm.md (<mcr>): New.
30551 * config/arm/arm.c (arm_coproc_builtin_available): Add
30552 support for mcr, mrc, mcr2 and mrc2.
30553 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
30554 (arm_mcr_qualifiers): ... this. New.
30555 (MRC_QUALIFIERS): Define to ...
30556 (arm_mrc_qualifiers): ... this. New.
30557 (MCR_QUALIFIERS): Define to ...
30558 (arm_mcr_qualifiers): ... this. New.
30559 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
30561 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
30562 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
30563 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
30564 VUNSPEC_MRC2): New.
30566 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30568 * config/arm/arm.md (*ldc): New.
30572 * config/arm/arm.c (arm_coproc_builtin_available): Add
30573 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
30574 (arm_coproc_ldc_stc_legitimate_address): New.
30575 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
30576 'qualifier_const_pointer'.
30577 (LDC_QUALIFIERS): Define to...
30578 (arm_ldc_qualifiers): ... this. New.
30579 (STC_QUALIFIERS): Define to...
30580 (arm_stc_qualifiers): ... this. New.
30581 * config/arm/arm-protos.h
30582 (arm_coproc_ldc_stc_legitimate_address): New.
30583 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
30584 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
30585 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
30586 stc2, stcl, stc2l): New.
30587 * config/arm/constraints.md (Uz): New.
30588 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
30589 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
30590 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
30591 VUNSPEC_STC2L): New.
30593 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30595 * config/arm/arm.md (<cdp>): New.
30596 * config/arm/arm.c (neon_const_bounds): Rename this ...
30597 (arm_const_bounds): ... this.
30598 (arm_coproc_builtin_available): New.
30599 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
30600 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
30601 (CDP_QUALIFIERS): Define to...
30602 (arm_cdp_qualifiers): ... this. New.
30604 (arm_expand_builtin_args): Add case for 6 arguments.
30605 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
30606 (arm_const_bounds): ... this.
30607 (arm_coproc_builtin_available): New.
30608 * config/arm/arm_acle.h (__arm_cdp): New.
30610 * config/arm/arm_acle_builtins.def (cdp): New.
30612 * config/arm/iterators.md (CDPI,CDP,cdp): New.
30613 * config/arm/neon.md: Rename all 'neon_const_bounds' to
30614 'arm_const_bounds'.
30615 * config/arm/types.md (coproc): New.
30616 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
30617 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
30618 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
30619 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
30621 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30623 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
30624 (UBINOP_QUALIFIERS): New.
30626 (acle_builtin_data): New. Change comment.
30627 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
30628 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
30629 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
30630 arm_acle_builtins.def.
30631 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
30632 (arm_init_acle_builtins): New.
30633 (CRC32_BUILTIN): Remove.
30634 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
30635 crc32cb, crc32ch and crc32cw.
30636 (arm_init_crc32_builtins): Remove.
30637 (arm_init_builtins): Use arm_init_acle_builtins rather
30638 than arm_init_crc32_builtins.
30639 (arm_expand_acle_builtin): New.
30640 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
30641 * config/arm/arm_acle_builtins.def: New.
30643 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30645 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
30646 (arm_builtin_datum): ... this.
30647 (arm_init_neon_builtin): Rename to ...
30648 (arm_init_builtin): ... this. Add a new parameters PREFIX
30649 and USE_SIG_IN_NAME.
30650 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
30651 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
30652 'arm_builtin_datum'.
30653 (arm_init_vfp_builtins): Likewise.
30654 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
30655 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
30656 (arm_expand_neon_args): Rename to ...
30657 (arm_expand_builtin_args): ... this. Rename builtin_arg
30658 enum values and differentiate between ARG_BUILTIN_MEMORY
30659 and ARG_BUILTIN_NEON_MEMORY.
30660 (arm_expand_neon_builtin_1): Rename to ...
30661 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
30662 values, arm_expand_builtin_args and add bool parameter NEON.
30663 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
30664 (arm_expand_vfp_builtin): Likewise.
30665 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
30667 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30669 PR middle-end/77484
30670 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
30671 * predict.c (tree_estimate_probability_bb): Reverse direction of
30672 polymorphic call predictor.
30674 2017-01-06 David Malcolm <dmalcolm@redhat.com>
30676 * passes.c (execute_one_pass): Split out pass-skipping logic into...
30677 (determine_pass_name_match): ...this new function and...
30678 (should_skip_pass_p): ...this new function.
30680 2017-01-06 Nathan Sidwell <nathan@acm.org>
30682 * ipa-visibility.c (function_and_variable_visibility): Reformat
30683 comments and long lines. Remove extrneous if.
30684 * symtab.c (symtab_node::make_decl_local): Fix code format.
30685 (symtab_node::set_section_for_node): Fix comment typo.
30687 2017-01-06 Martin Liska <mliska@suse.cz>
30690 * lra-constraints.c: Rename invariant to lra_invariant.
30691 * predict.c (set_even_probabilities): Initialize e to NULL.
30693 2017-01-05 Martin Sebor <msebor@redhat.com>
30695 PR tree-optimization/78910
30696 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
30697 (format_integer): Correct off-by-one error in the handling
30698 of precision with negative numbers in signed conversions..
30700 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
30702 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
30704 2017-01-05 Jakub Jelinek <jakub@redhat.com>
30706 PR tree-optimization/71016
30707 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
30708 factor_out_conditional_conversion. Formatting fix.
30709 (factor_out_conditional_conversion): Add cond_stmt argument.
30710 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
30711 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
30714 2017-01-05 David Malcolm <dmalcolm@redhat.com>
30716 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
30717 read-rtl-function.o, and selftest-rtl.o.
30718 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
30719 (selftest::aarch64_test_loading_full_dump): New function.
30720 (selftest::aarch64_run_selftests): New function.
30721 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
30722 selftest::aarch64_run_selftests.
30723 * config/i386/i386.c
30724 (selftest::ix86_test_loading_dump_fragment_1): New function.
30725 (selftest::ix86_test_loading_call_insn): New function.
30726 (selftest::ix86_test_loading_full_dump): New function.
30727 (selftest::ix86_test_loading_unspec): New function.
30728 (selftest::ix86_run_selftests): Call the new functions.
30729 * emit-rtl.c (maybe_set_max_label_num): New function.
30730 * emit-rtl.h (maybe_set_max_label_num): New decl.
30731 * function.c (instantiate_decls): Guard call to
30732 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
30733 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
30735 * gensupport.c (gen_reader::gen_reader): Pass "false"
30736 for new "compact" param of rtx_reader.
30737 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
30738 rather than an empty string for NULL strings.
30739 * read-md.c: Potentially include config.h rather than bconfig.h.
30740 Wrap include of errors.h with #ifdef GENERATOR_FILE.
30741 (have_error): New global, copied from errors.c.
30742 (md_reader::read_name): Rename to...
30743 (md_reader::read_name_1): ...this, adding "out_loc" param,
30744 and converting "missing name or number" to returning false, rather
30746 (md_reader::read_name): Reimplement in terms of read_name_1.
30747 (md_reader::read_name_or_nil): New function.
30748 (md_reader::read_string): Handle "(nil)" by returning NULL.
30749 (md_reader::md_reader): Add new param "compact".
30750 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
30751 (md_reader::read_file): New method.
30752 * read-md.h (md_reader::md_reader): Add new param "compact".
30753 (md_reader::read_file): New method.
30754 (md_reader::is_compact): New accessor.
30755 (md_reader::read_name): Convert return type from void to file_location.
30756 (md_reader::read_name_or_nil): New decl.
30757 (md_reader::read_name_1): New decl.
30758 (md_reader::m_compact): New field.
30759 (noop_reader::noop_reader): Pass "false" for new "compact" param
30761 (rtx_reader::rtx_reader): Add new "compact" param.
30762 (rtx_reader::read_rtx_operand): Make virtual and convert return
30763 type from void to rtx.
30764 (rtx_reader::read_until): New decl.
30765 (rtx_reader::handle_any_trailing_information): New virtual function.
30766 (rtx_reader::postprocess): New virtual function.
30767 (rtx_reader::finalize_string): New virtual function.
30768 (rtx_reader::m_in_call_function_usage): New field.
30769 (rtx_reader::m_reuse_rtx_by_id): New field.
30770 * read-rtl-function.c: New file.
30771 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
30772 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
30773 (selftest::verify_three_block_rtl_cfg): New decl.
30774 * read-rtl-function.h: New file.
30775 * read-rtl.c: Potentially include config.h rather than bconfig.h.
30776 For host, include function.h, memmodel.h, and emit-rtl.h.
30777 (one_time_initialization): New function.
30778 (struct compact_insn_name): New struct.
30779 (compact_insn_names): New array.
30780 (find_code): Handle insn codes in compact dumps.
30781 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
30782 (bind_subst_iter_and_attr): Likewise.
30783 (add_condition_to_string): Likewise.
30784 (add_condition_to_rtx): Likewise.
30785 (apply_attribute_uses): Likewise.
30786 (add_current_iterators): Likewise.
30787 (apply_iterators): Likewise.
30788 (initialize_iterators): Guard usage of apply_subst_iterator with
30789 #ifdef GENERATOR_FILE.
30790 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
30791 (md_reader::read_mapping): Likewise.
30792 (add_define_attr_for_define_subst): Likewise.
30793 (add_define_subst_attr): Likewise.
30794 (read_subst_mapping): Likewise.
30795 (check_code_iterator): Likewise.
30796 (rtx_reader::read_rtx): Likewise. Move one-time initialization
30798 (one_time_initialization): New function.
30799 (rtx_reader::read_until): New method.
30800 (read_flags): New function.
30801 (parse_reg_note_name): New function.
30802 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
30803 Handle reuse_rtx ids.
30804 Wrap iterator lookup within #ifdef GENERATOR_FILE.
30805 Add parsing support for RTL dumps, mirroring the special-cases in
30806 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
30807 values, and calling handle_any_trailing_information.
30808 (rtx_reader::read_rtx_operand): Convert return type from void
30809 to rtx, returning return_rtx. Handle case 'e'. Call
30810 finalize_string on XSTR and XTMPL fields.
30811 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
30812 "(nil)" values were omitted. Call the postprocess vfunc on the
30814 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
30815 class ctor. Initialize m_in_call_function_usage. Call
30816 one_time_initialization.
30817 * rtl-tests.c (selftest::test_uncond_jump): Call
30818 set_new_first_and_last_insn.
30819 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
30820 * selftest-rtl.c: New file.
30821 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
30822 (selftest::get_insn_by_uid): New decl.
30823 * selftest-run-tests.c (selftest::run_tests): Call
30824 read_rtl_function_c_tests.
30825 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
30826 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
30829 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
30831 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
30832 operands in a special way. Assert that pos+len <= mode precision.
30834 2017-01-05 Jakub Jelinek <jakub@redhat.com>
30836 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
30837 3 argument Alias with unlimited for the negative form.
30838 (fno-vect-cost-model): Removed.
30840 2017-01-05 Martin Liska <mliska@suse.cz>
30842 * hsa-gen.c (gen_hsa_divmod): New function.
30843 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
30845 2017-01-05 Martin Liska <mliska@suse.cz>
30848 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
30851 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30853 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
30854 small constant length operands.
30856 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30858 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
30859 between loop iterations.
30861 2017-01-05 Martin Liska <mliska@suse.cz>
30864 * gimplify.c (gimplify_decl_expr): Compare to
30865 asan_poisoned_variables instread of checking flags.
30866 (gimplify_target_expr): Likewise.
30867 (gimplify_expr): Likewise.
30868 (gimplify_function_tree): Conditionally initialize
30869 asan_poisoned_variables.
30871 2017-01-04 Jeff Law <law@redhat.com>
30873 PR tree-optimizatin/78812
30874 * rtl.h (contains_mem_rtx_p): Prototype.
30875 * ifcvt.c (containts_mem_rtx_p): Move from here to...
30876 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
30877 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
30878 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
30879 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
30881 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30883 * input.c (assert_char_at_range): Default-initialize actual_range.
30885 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30887 * df-scan.c (df_ref_create_structure): Make regno unsigned,
30888 to match the caller.
30890 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30892 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
30893 insns after final jump in test to emit dummy move.
30895 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30897 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
30898 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
30900 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30902 * multiple_target.c (create_dispatcher_calls): Init e_next.
30903 * tree-ssa-loop-split.c (split_loop): Init border.
30904 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
30907 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
30912 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
30913 (altivec_register_operand): Do not return true if the operand
30914 contains a SUBREG mixing SImode and SFmode.
30915 (vsx_register_operand): Likewise.
30916 (vsx_reg_sfsubreg_ok): New predicate.
30917 (vfloat_operand): Do not return true if the operand contains a
30918 SUBREG mixing SImode and SFmode.
30919 (vint_operand): Likewise.
30920 (vlogical_operand): Likewise.
30921 (gpc_reg_operand): Likewise.
30922 (int_reg_operand): Likewise.
30923 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
30924 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
30925 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
30927 (rs6000_emit_move_si_sf_subreg): New helper function.
30928 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
30929 fixup SUBREGs involving SImode and SFmode.
30930 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
30931 numbers for the new peephole2 optimization.
30932 (peephole2 for SFmode unions): New peephole2 to optimize cases in
30933 the GLIBC math library that do AND/IOR/XOR operations on single
30934 precision floating point.
30935 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
30936 target macros to say whether we need to avoid SUBREGs mixing
30938 (TARGET_ALLOW_SF_SUBREG): Likewise.
30939 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
30940 (UNSPEC_SI_FROM_SF): Likewise.
30941 (iorxor): Change spacing.
30942 (and_ior_xor): New iterator for AND, IOR, and XOR.
30943 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
30944 (movdi_from_sf_zero_ext): Likewise.
30945 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
30946 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
30947 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
30948 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
30949 (fms<mode>4): Likewise.
30950 (fnma<mode>4): Likewise.
30951 (fnms<mode>4): Likewise.
30952 (nfma<mode>4): Likewise.
30953 (nfms<mode>4): Likewise.
30955 2017-01-04 Marek Polacek <polacek@redhat.com>
30958 * doc/invoke.texi: Document -Wpointer-compare.
30960 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30962 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
30965 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
30966 descriptions for -gdwarf-5 and emit them as uleb128 instead of
30969 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
30972 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
30973 documentation of the powerpc_popcntb_ok attribute.
30974 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30975 code to issue warning messages if a requested CPU configuration is
30976 not supported by the binary (assembler and loader) toolchain.
30977 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
30978 made to define a built-in function that has been disabled.
30979 (paired_init_builtins): Add assertion to prevent ICE if attempt is
30980 made to define a built-in function that has been disabled.
30981 (altivec_init_builtins): Add comment explaining why definition
30982 of the DST built-in functions is not preceded by an assertion
30983 check. Add assertions to prevent ICE if attempts are made to
30984 define an altivec predicate or an abs* built-in function that has
30986 (htm_init_builtins): Add comment explaining why definition of the
30987 htm built-in functions is not preceded by an assertion check.
30989 2017-01-04 Jeff Law <law@redhat.com>
30991 PR tree-optimizatin/67955
30992 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
30993 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
30994 the points-to solution does not include pt_null. Use DECL_PT_UID
30997 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
30999 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
31000 Use gen_int_mode instead of gen_lopwart for const_int operands.
31002 2017-01-04 Jakub Jelinek <jakub@redhat.com>
31004 PR tree-optimization/71563
31005 * match.pd: Simplify X << Y into X if Y is known to be 0 or
31006 out of range value - has low bits known to be zero.
31008 2017-01-04 Alan Modra <amodra@gmail.com>
31010 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
31011 * configure: Regenerate.
31012 * config.in: Regenerate.
31014 2017-01-04 Jakub Jelinek <jakub@redhat.com>
31017 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
31018 a substring of the message, but strcmp with the whole message. Ifdef
31019 ENABLE_NLS, translate the message first using dgettext.
31021 2017-01-03 Jeff Law <law@redhat.com>
31023 PR tree-optimizatin/78856
31024 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
31025 (mark_threaded_blocks): Remove code to truncate thread paths that
31026 cross multiple loop headers. Instead invalidate the cached loop
31027 iteration information and handle case of a thread path walking
31028 into an irreducible region.
31030 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
31033 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
31034 assertions. Add support for doing the signbit if the IEEE 128-bit
31035 floating point value is in a GPR.
31036 * config/rs6000/rs6000.md (Fsignbit): Delete.
31037 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
31038 Update the length attribute if the value is in a GPR.
31039 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
31040 the sign or zero extension instruction, since the value is always 0/1.
31041 (signbit<mode>2_dm2): Delete using <Fsignbit>.
31044 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
31045 extracting SImode to a GPR register so that we can generate a
31046 store, limit the vector to be in a traditional Altivec register
31047 for the vextuwrx instruction.
31049 2017-01-03 Ian Lance Taylor <iant@google.com>
31051 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
31053 2017-01-03 Martin Sebor <msebor@redhat.com>
31055 PR tree-optimization/78696
31056 * gimple-ssa-sprintf.c (format_floating): Correct handling of
31057 precision. Use MPFR for %f for greater fidelity. Correct handling
31059 (pass_sprintf_length::compute_format_length): Set width and precision
31060 specified by asrerisk to void_node for vararg functions.
31061 (try_substitute_return_value): Adjust dump output.
31063 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
31065 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
31067 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
31069 * doc/invoke.texi (SPARC options): Document -mlra as the default.
31070 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
31071 -mlra/-mno-lra was passed to the compiler.
31073 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
31075 PR rtl-optimization/65618
31076 * emit-rtl.c (try_split): Move initialization of "before" and
31077 "after" to just before the call to emit_insn_after_setloc.
31079 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
31081 * doc/md.texi (Standard Names): Remove reference to Java frontend.
31083 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
31085 * dwarf2out.c (gen_enumeration_type_die): When
31086 -gno-strict-dwarf, add a DW_AT_encoding attribute.
31088 2017-01-03 Jakub Jelinek <jakub@redhat.com>
31090 PR tree-optimization/78965
31091 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
31092 Change first argument from const call_info & to call_info &. For %n
31093 set info.nowrite to false.
31095 PR middle-end/78901
31096 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
31097 possibly throwing calls.
31099 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
31100 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
31101 and fns handling, rather than in a separate case SSA_NAME.
31103 2017-01-02 Jeff Law <law@redhat.com>
31105 * config/darwin-driver.c (darwin_driver_init): Const-correctness
31106 fixes for first_period and second_period variables.
31108 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
31111 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
31112 (*insvqi_1): New insn pattern.
31113 (*insvqi_1_mem_rex64): Ditto.
31114 (*insvqi_2): Ditto.
31115 (*insvqi_3): Rename from *insvqi.
31117 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
31119 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
31121 * doc/cfg.texi (Edges): Remove reference to Java.
31122 (Maintaining the CFG): Ditto.
31124 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31126 PR middle-end/77674
31127 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
31128 transparent aliases.
31130 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31132 PR middle-end/77484
31133 * predict.def (PRED_CALL): Update hitrate.
31134 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
31135 * predict.c (tree_estimate_probability_bb): Split CALL predictor
31136 into direct/indirect/polymorphic variants.
31138 2017-01-01 Jakub Jelinek <jakub@redhat.com>
31140 Update copyright years.
31142 * gcc.c (process_command): Update copyright notice dates.
31143 * gcov-dump.c (print_version): Ditto.
31144 * gcov.c (print_version): Ditto.
31145 * gcov-tool.c (print_version): Ditto.
31146 * gengtype.c (create_file): Ditto.
31147 * doc/cpp.texi: Bump @copying's copyright year.
31148 * doc/cppinternals.texi: Ditto.
31149 * doc/gcc.texi: Ditto.
31150 * doc/gccint.texi: Ditto.
31151 * doc/gcov.texi: Ditto.
31152 * doc/install.texi: Ditto.
31153 * doc/invoke.texi: Ditto.
31155 Copyright (C) 2017 Free Software Foundation, Inc.
31157 Copying and distribution of this file, with or without modification,
31158 are permitted in any medium without royalty provided the copyright
31159 notice and this notice are preserved.